PDA

View Full Version : "ALMower" [was: Robotic Lawn Mower]



Pages : [1] 2

revoltage
08-04-2008, 12:39 PM
Hi everyone,


First I'm sorry that I posted this request for my first post. http://www.electro-tech-online.com/images/smilies/redface.gif
I'm planning to build an autonomous lawnmower.
Does anyone has idea how Friendly Robotics (Robomow) system works?

I need your precious help
Thanks a lot for your response http://www.electro-tech-online.com/images/smilies/smile.gif

Adrenalynn
08-04-2008, 12:55 PM
Although he's not using a PIC (actually a full-blown PC), Ooops here is working on a full size autonomous lawnmower. I've been offering up some hardware and software help for him.

If you have some specific questions we can probably be of more help.

And, hey, this is a great first post to the community - welcome to the TRC! Don't stress your dissertation - they always have a habit of coming together at the last possible instant. :)

Electricity
08-04-2008, 01:22 PM
Yeah, check out Ooops' S.A.M.i (http://forums.trossenrobotics.com/showthread.php?t=1856) project. I'm sure you guys can swap ideas!

ooops
08-04-2008, 02:01 PM
Welcome revoltage!
I am in the midst of couple of these mower projects. The first one is an all terrain mower, I live in the mountains where flat land is scarce, and hills are a pain to mow! It is still on the "white board" stages. S.A.M.i on the other hand is a bit further along. Link:( http://forums.trossenrobotics.com/showthread.php?t=1856 ) I am concentrating on the mechanics at this point, more so than the programing. I would be happy to assist anyway possible.

Let me say I think you have come to the right spot for help ... everyone here is great!
Ed

metaform3d
08-04-2008, 02:53 PM
Does anyone has idea how Friendly Robotics (Robomow) system works?I have one. You can go to their site for videos (http://www.friendlyrobotics.com/videos/) on setup and operation of the different models.

The robot has two drive motors and a front caster, two sensors that allow it to sense the perimeter wire, a set of front and back bump sensors, a tilt sensor and a compass. There's also a control module that you can pop out to drive it manually, set settings and so forth, which you stow back on the robot to run it autonomously.

Autonomous operation has two modes. In edging mode you position the robot near the perimeter wire and it finds it and then follows it to mow along the perimeter. When the compass turns 360 degrees so the robot is back where it started it goes into interior mode. In this mode it drives forward and reverse in a zig-zag pattern, reversing each time it hits the perimeter wire or an obstacle like a tree. After each pass it turns to a different angle and makes another pass.

Most of the smarts seem to have to do with how to deal with getting stuck. It's pretty amusing to watch when the mower works itself into a corner and has to try several different patterns of backing up and turning before it finds where the lawn is again.

Adrenalynn
08-04-2008, 02:56 PM
That's awesome, Meta. How well does it do with that method of getting unstuck in actual practice? How frequently do you have to "help it out"?

metaform3d
08-04-2008, 03:20 PM
That's awesome, Meta. How well does it do with that method of getting unstuck in actual practice? How frequently do you have to "help it out"?In edging mode it has a lot of problems. The border plants have grown a lot so they hang into the lawn and it hits them or gets snagged by them. If it hits something it makes a goofy "boing" sound, backs up and tries again with a little swerve. Usually it's not enough though and it hits the plant again and gives up on edging. It also tends to lose traction so it will sit there with one wheel spinning. If it detects that it's no moving long enough it will time out and call for help.

I generally walk around with it as it edges and give it friendly nudges and hold plants out of its way.

When it does the lawn interior it's pretty good. It gets stuck rarely, but if it does it stops and waits for help. It's just a matter of driving it out of the corner manually and setting it off again. It also has a very rare software error that for some reason causes it to drive past the perimeter wire, but it detects that and shuts down as well, which is good since it could do a lot of damage otherwise. Most of the time it just works.

Despite the problems I think it's great. Much better than having to get all dirty doing it myself -- a definite asset in this robotic lifestyle of mine!

darkback2
08-04-2008, 04:20 PM
I was actually thinking about this very problem...navigating in a free space, with an outer boarder marked out for you.

I remember reading about a robot that made nodes every so often. Then when it got stuck, it would back up to the last node, and try again. Seams to me like if your using your own custom setup / full PC based lawnmowing rover, you could drive the perimeter with the robot the first time. It would then set up increased probabilities for finding the right solution based on the path that you originally put in...so...

Go forward for about three feet...then expect at some point about there to fine an obstacle, and turn left...and expect to be able to go forward again for...

Of course that wouldn't help with say things being left out in the yard, but it seams like it would help with edging and stuff where things don't change much from mow to mow. It could also give you a basis by which it could store solutions to problems. I got stuck here, and the directions file was updated to include these changes sort of deal...

DB

Adrenalynn
08-04-2008, 04:59 PM
Some great points, Meta (and DB). Your practical experience actually changes some of what I was thinking about with Ooops mower. I'd originally be thinking about marking the perimeter with an IR laser bounced around the perimeter and anything overlapping it. Plants swaying in the wind and breaking the beam could be catastrophic! Thanks for those pointers! The mower losing traction seems like an unconscionable failure of design to me. That's disappointing, especially looking at pictures of its wheels. They really cheaped-out there, imho.

The SLAM problem you describe DB is a really complex problem - the largest problem facing robotics today, and the one I'm most interested in solving. It gets pretty complex the more you dig into it. Add in a few hundred pounds, a few dozen foot pounds of torque, and a few pounds of spinning blades [eek] and it's a problem that is even more "for keeps".

darkback2
08-04-2008, 05:09 PM
The SLAM problem you describe DB is a really complex problem - the largest problem facing robotics today, and the one I'm most interested in solving. It gets pretty complex the more you dig into it. Add in a few hundred pounds, a few dozen foot pounds of torque, and a few pounds of spinning blades [eek] and it's a problem that is even more "for keeps".

I'm not about solving the problem, just more about doing better each time I do the same task. If the lawnmower has the sense to know when its stuck and call for help, then it should have the sense to remember how you helped it. Something that may work would be to have it drop some sort of RFID chip wherever it got stuck. sensing that chip, and pattern of movements could trigger the stored solution file.

DB

metaform3d
08-04-2008, 06:04 PM
I would definitely recommend using a wire to define the perimeter. You install it by staking it down on the top of the grass, and over time the growing grass causes it to migrate to the bottom of the sod naturally. As long as you use the right kind of crimp connectors for any joins (the wrong kind can fail over time, I learned the hard way!) it's pretty indestructible.


The Robomow does no mapping of any kind. It has no idea where it is and only a rough idea of what direction it's headed. That said, the heuristics for getting out of corners could certainly be better. If it knows, for example, that it hits the perimeter going forward and it hits the perimeter again going backward a few seconds, then it could try turning left and if it hits the perimeter going forward there then it knows the lawn is directly behind it. It would probably have the same problem with closely-spaced trees that the Roomba has with chair legs.

I like the idea of mapping a lawn using RFID tags. Would they survive buried in the dirt? And would they have enough range for the robot to be able to triangulate its position?

Adrenalynn
08-04-2008, 07:04 PM
Passive RFID could be buried, yes. I have one implanted. But they don't have near the range to do it.

Active is where you need to go for that. Controlling the range is the hard part and it wouldn't be buried - not going to punch through that dirt. It would have to be in a water tight PVC (or other material transparent to the frequency) box. Again, though, figuring out how far away from it you are is the hard part.

revoltage
08-05-2008, 05:29 AM
Thanks you all for your suggestions guyz. I really appreciated it. :)
Can you please help make a list of components (sensors, motors,motor drivers etc..compatible with PIC 16f877A)so that I can buy them from the trossenrobotics store?I am planning to build the robot like robomow(friendly robotics), that is, with the perimeter wire and simply a "Go" button ( and the rest of the mechanism is like meta described). The weight of the robot should be no more than 10 kg.

Any one got an idea how to build the perimeter wire circuit and make the microcontrolle sense it when the robot approaches the wire?

Please bear with me coz I'm a debutant in practical electronics/robotics :(

Thank you all :)

Electricity
08-05-2008, 08:54 AM
I'm a beginner myself, but maybe I can offer a suggestion.
You may not need to actually hook the bot up to the ground wire, just run a constant current through it. They install a Current Sensor (http://www.trossenrobotics.com/store/c/2673-Voltage-Current-Sensors.aspx) onto your bot. Assuming you have no other lines in your yard, the robot could then detect that it had reached the edge of your yard once it sensed the current from the buried line.

Electricity
08-05-2008, 08:57 AM
Edit: You could probably do the same thing with a Magnetic Sensor (http://www.trossenrobotics.com/store/p/3245-Magnetic-Sensor.aspx)

And actually, now that I think about it, another interesting idea would be use an onboard IR detector and strategically placed IR emitters to box in your yard. It would probably take some trial and error to get it right, but it would use less electricity, and would be pretty cool! Plus, you could take a picture of it working with an ir lense, and it would look cool.. lol

Adrenalynn
08-05-2008, 11:01 AM
That current sensor would require direct contact with the wire.
That Hall Effect sensor might be usable, but you'd be wanting to bury an electromagnet or one very large ring magnet.

As I pointed out, the IR emitters are a bad idea.


What you want is an inductive sensor. http://en.wikipedia.org/wiki/Inductive_sensor

I can poke at the digikey catalog after I have a soda and wake up. :)

ooops
08-05-2008, 11:12 AM
Can you please help make a list of components (sensors, motors,motor drivers etc..compatible with PIC 16f877A)so that I can buy them from the trossenrobotics store?I am planning to build the robot like robomow(friendly robotics), that is, with the perimeter wire and simply a "Go" button (and the rest of the mechanism is like meta described). The weight of the robot should be no more than 10 kg.

There are so many variables that need to be addressed on the subject of motors and motor controllers that IMHO that will be your first design challenge and will dictate the rest of the project. The "average off the shelf" robot propulsion system simply won't do for yard work.
If these were available back when I started, they would have gotten a serious look http://www.trossenrobotics.com/store/p/5761-Motor-Mount-Wheel-Kit-with-Position-Controller.aspx
Now assuming those would work (for discussion), your next step will be to determine your power needs. My guess is you will have some sort of DC motor spinning the cutting device, and you will need to power it as well as any electronics. So, depending on the "cutting motor" you have a minimum power draw of 12 - 23+ amps during the time spent mowing (lets assume 20 amps average). Ideally depending how big or small your yard is you will need a battery capable of running enough hours (or fractions there of) to mow the whole yard in one cutting. To keep things simple lets say it will only take one hour to mow the whole yard. You now need, at the very least, a 20 amp hour battery. A quick google search and it looks like a small 32AH gel cell will weigh 26lbs or 11kg (if my quick in the head math works) and that is just the battery.

Now I am not suggesting for a second that your original plan to keep things under 10kg is impossible. What I am suggesting, in what now has become a very long winded post, is that you will need to sort out the drive system first and the rest will follow.

Electricity
08-05-2008, 11:17 AM
That current sensor would require direct contact with the wire.
That Hall Effect sensor might be usable, but you'd be wanting to bury an electromagnet or one very large ring magnet.

As I pointed out, the IR emitters are a bad idea.


What you want is an inductive sensor. http://en.wikipedia.org/wiki/Inductive_sensor

I can poke at the digikey catalog after I have a soda and wake up. :)
heh I guess you could use a metal detector, and bury enough wire/metal to be detectable..

Adrenalynn
08-05-2008, 11:32 AM
A metal detector is a form of inductive sensor. But that's painting with too wide a brush for me. It's the rare yard that doesn't have construction debris 5" under the soil, very inductive water pipes <18" down, sprinkler-system wires, ...

I'm advocating a tuned inductive probe running on something like an electric fence transformer scaled down.

One could probably use a capacitance plate and look for the 50/60hz hum too, but having built a 3D sensor using capacitance I can attest just how hard it really is to filter environmental noise from that.

The more I think about it the more certain I am that Meta's toy uses a tuned inductive probe.

revoltage
08-05-2008, 11:43 AM
I'm a bit confused now what to do...
The motor I was planning to used to cut the grass is also a dc one. I planned to use 250 square metres of the available flat lawn area to carry out the test.

From what I see from this website I guess the parallax motor mount wheel kit would work. Please give me some clue guyz :happy:

revoltage
08-05-2008, 11:46 AM
One system I heard that could is the electric dog fence. Anyone got a schematic for it?

Adrenalynn
08-05-2008, 12:06 PM
I worked up numbers/graphs on those motors being offered in the catalog. I should go dig that graph up, I posted it in one of the threads here. Maybe get Alex or Andrew to add it to the catalog page...

Those are pretty cool motors, but they'll be a lot less cool carrying around the weight of a 30Ah gel-cell. ;) The total weight on those motors should probably be kept under ~40lbs/18kg. I think you'll have traction issues with those tires on a lawn, although I don't have that specific tire to make the call for sure.

Now that Robocut that you link is doing some pretty cool stuff with its design mechanically. At first blush it looks like it's using drill motors. Great torque good runtime. My Hitachi Lithium will run nearly 3hrs drilling "normal" holes. My Hitachi NiMH actually had even better runtime.

If *I* were building this project, I'd go down to Harbor Freight (might have a challenge not being in the US, but you could contact them on their website, harborfreight.com or find someone to ship to you) and pick up a few of their cheapie nicad cordless drills. Rip the battery packs out and toss 'em in the gar... in the hazardous waste recycling bin, and pick up some NiMH packs. The 10Ah batteries from Tenergy would be perfect. I would yank the motors and gearboxes out of those drills, and put them on the ammeter to make sure they'd run off of a Syren motor controller, which I suspect they will. (The DeWalt drill motors are just over the rating, so those should be under it).

For wheels, I'd use a design similar to theirs, only I'd go on the cheap and use two or three per side old bicycle chainrings. Those would bite nice. I'd build the frame out of lexan or other poly-carb plastic with corner bolsters similar to the ones on the robocut. My cutter blade would be chain driven from one of the drill motors without the gearbox, because adjusting torque and speed would be cheap and easy by just replacing the cogs. 3500 RPM should be a no brainer if you're not trying to cut little trees or something. ;)


That's my whiteboard design. If I had a lawn, I'd probably build it. ;)

Once we all come to some consensus, we can start putting together a shopping list. :)

Adrenalynn
08-05-2008, 12:15 PM
One system I heard that could is the electric dog fence. Anyone got a schematic for it?

When I want to understand already existing technology, I do a patent search. A patent is required to "make plain to someone reasonably skilled in the art".

http://www.patentgenius.com/patent/7081821.html

ooops
08-05-2008, 12:26 PM
Hmmm...:o
I'm a bit confused now what to do...
The motor I was planning to used to cut the grass is also a dc one. I planned to use 250 square metres of the available flat lawn area to carry out the test. In my university so far no one has ever done a robotic lawn mower and it is also not at sales at hardware stores.
I have come across a link to another website but I am a bit confused about it too due to my lack of technical knowledge.:o I guess you all experienced robotics experimenters would find it simple. Could you please check it out and give me some feedback :)

http://www.robotshop.se/micro/wwwrc_us/indext.htm

From what I see from this website I guess the parallax motor mount wheel kit would work. Please give me some clue guyz :happy:
Thanks you all for your generous help :)

Well, I certainly didn't mean to confuse you. The example from my earlier post was just that, an example! I don't want you to be influenced by my perceptions of what a mower should be, rather you need to envision what your mower should be and work out the engineering from the ground up! The example should give you an idea of the issues to work out, but by no means assume that you have to build it that way. I for one would rather see you "think outside the box" and develop something truly unique:)
The link to the robotshop mower you provided shows an elegant solution to the weight to battery issue and the use of very thin "low drag" all terrain wheels. It looks like it would fall into your 11kg goal. Someone was thinking outside the box on that one.
As you work out your design ideas feel free to keep asking questions!!!

Adrenalynn
08-05-2008, 12:59 PM
Drills: http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=93440

Caster: http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=38709

Batteries: http://www.tenergybattery.com/index.php?page=shop.product_details&flypage=shop.flypage&product_id=33&category_id=1&manufacturer_id=0&option=com_virtuemart&Itemid=27

ooops
08-05-2008, 02:53 PM
When I want to understand already existing technology, I do a patent search. A patent is required to "make plain to someone reasonably skilled in the art".

http://www.patentgenius.com

Great Link !!!!!!!!!!!!!!!!!!!

revoltage
08-06-2008, 01:56 AM
Harborfreight does not ship to my country. So do you think that the parallax motor can be used so that I can make a robot like robotcut?

darkback2
08-06-2008, 02:08 AM
If by the parallax motors your talking about the new right and left hand motors? Then...the short answer is yes. I've had students sit on vivian, my robot, and she can move them. She uses 4 of them and direct drive. The motors do suck amps though, especially when you load them with weight. Vivian weighs about 25-30 lbs and can run for about 20 minutes continuously on a five amp hour battery.

You are planning on using a mulching blade right?

Here is the link to vivian...she started out as gepetto II

http://forums.trossenrobotics.com/showthread.php?t=1439&page=2


Hope this helps.

DB

Adrenalynn
08-06-2008, 02:30 AM
It's for absolutely dead certain those wheels will be worthless for that application - I had an opportunity to see 'em today.

revoltage
08-06-2008, 02:36 AM
Yep exactly I'm going to use mulching blades and two rear wheels (thus two motors) and a front castor wheel. For the PIC 16F877A I'm going to use an olimex PIC-P40 proto board

http://olimex.com/dev/index.html

Then one rear and one front bumpsensors. Then a dc motor (I don't know which one yet) for the mulching blades. Also a tilt sensor so that when somebody lift the mower during it's operation the baldes stop rotating.

Then there is this perimeter wire thing which I have not yet manage to create a schematic for it :(

And yes for the battery I'll use Adenalynn's link

http://www.tenergybattery.com/index.php?page=shop.product_details&flypage=shop.flypage&product_id=33&category_id=1&manufacturer_id=0&option=com_virtuemart&Itemid=27&vmcchk=1

All this plus the weight of aluminium casing it should be less than 10 kg I think

So it is final everyone agrees that I can buy the parallax motor mount &wheel kit with position controller so that the robot can reverse,stop and move forward?

revoltage
08-06-2008, 02:37 AM
oops saw your message after my post Adrenalynn

revoltage
08-06-2008, 02:44 AM
Ok guyz I've got another cool website where there's plenty of motors

http://www.robotstorehk.com/motors/motors.html

You help me please choose the right one. There's one complete wheel,motor +driver set MOTOR-SET-RD01 -- Two Gearhead Motors with driver board -- US$229.95

I think there's probably one worthwhile motor there
What do you think? :)

Adrenalynn
08-06-2008, 02:55 AM
I'm not in agreement on the wheels. I think if the grass is even moderately damp you're going to go nowhere really quick. ;)

Adrenalynn
08-06-2008, 03:00 AM
Why not mount chain-rings to the Parallax/Denso motors? http://www.sheldonbrown.com/harris/chainrings.html

Building a bracket system to attach that wouldn't be tough.

revoltage
08-06-2008, 03:02 AM
I'm not in agreement on the wheels. I think if the grass is even moderately damp you're going to go nowhere really quick. ;)


You found some motors + controllers on the website:
http://www.robotstorehk.com/motors/motors.html

Adrenalynn
08-06-2008, 03:07 AM
Don't know if you missed my post - the motors that Trossen sells will work fine (those Parallax rebranded ones) - it's those wheels I'm concerned about. Building a bracket to attach chainrings to the parallax motors (actually made by Denso) should be quite easy.

darkback2
08-06-2008, 03:21 AM
Don't know if you missed my post - the motors that Trossen sells will work fine (those Parallax rebranded ones) - it's those wheels I'm concerned about. Building a bracket to attach chainrings to the parallax motors (actually made by Denso) should be quite easy.

If you do get those things running chain, tell me...That would be freeking cool. I've always just run four of them because that usually does the trick in terms of not getting stuck...two with chains should work just fine.

DB

revoltage
08-06-2008, 03:29 AM
So is it fine if the buy the parallax motor set and then substitute the wheel with the chainrings?

revoltage
08-06-2008, 09:33 AM
And I think I have just burned my development board :(
I plugged in a 9V adaptor at the PWR. When I measured the voltage at input pin for the LM317 regulator it was 9V. However when I measured the center pin it was 5V (it should have been zero because it's ground) and the output pin was giving 3.76 V (which should have been 5V). Also when I tried to download a program nothing happened.. the software just keeps waiting...guess I have to buy another board :( pffff

(I think I know how it happened; when I was measuring the input side of the regulator the probe of the multimeter probe connected by accident to one of the pin of bridge rectifier G1 B-DIL and there was kind of a small spark) pffff :(

Adrenalynn
08-06-2008, 10:01 AM
Ack! Don't replace your dev board, just the regulator and likely a diode and maybe the cap.

Microcontrollers are very tolerant, at worst you probably smoked one or two pins on the PIC, not the chip, but likely not even that.

ScuD
08-06-2008, 10:40 AM
Seeing that schematic, I'd have to say that the center pint of the LM317 shouldn't be at ground.
There's a feedback line from the output, through the 240k resistor, and another voltage divider between that and ground.

I should check the datasheet on the LM317 to see what influence the center pin has on the voltage, but I'm sure you should read a voltage there, not ground.

Be that as it may, the output voltage is still too low, but that could be caused by other issues, such as a 'short' pulling the output voltage low.

Adrenalynn
08-06-2008, 10:50 AM
The center pin is the adjust pin on the '317. Resistors determining the output voltage.

Another thing that could obviously pull the output down is too much load on the regulator. It can handle about 1.5A.

ScuD
08-06-2008, 10:59 AM
Indeed.

The voltage on the center pin is what determines the output voltage.
The feedback resistor is used to make sure the output voltage stays at the same level, independently of the current drawn.

Question: are you using a 9v AC power supply or a DC power supply?

/edit: just checked the datasheet, there's one thing you could check on the LM317, there should be 1,25V between the output pin and the center pin.

revoltage
08-06-2008, 11:49 AM
I am using a 9VDC. I just check between the output pin and the center pin there is 1.24V and between the input pin and the centre pin there is 4.7V. So is the board ok?

But I am being unable to download a program. I'm using screamerv1.7 to flash the program in. Is it possible that the PIC itself is down?

Adrenalynn so what do you think, I buy the parallax motor kit and then use a chainring instead of the supplied wheels?

Electricity
08-06-2008, 12:14 PM
Adrenalynn so what do you think, I buy the parallax motor kit and then use a chainring instead of the supplied wheels?
I think what she's saying is use both. Attach the chain ring to the wheel, to give it extra grip. You might even consider attaching one on the inside, and one on the outside, for added grippage.

Adrenalynn
08-06-2008, 12:26 PM
I'm heading out soon for a hike, so I'm just shotgunning quickly here.

First - you're using a real serial port and not a USB->Serial converter, right? It's pretty uncommon for PIC programmers to work with USB->Serial converters.

Have you checked your serial port settings and made sure they are correct?

What's the voltage on Pin 2 of the Max232 chip?

Do you have any other test gear available? Oscilloscope?

It's sounding like the board could well be good - the regulator doesn't sound blown. Checking voltage on Pin 2 of the Max232 would tell you for sure. It looks like there's a 3.3v test-point or harness too. Can you take a picture of the board?

I see a lot of problems posted on the Sparkfun board related to getting Screamer to work. Seems people are jumping over to Boostbloader which is a spinoff from Screamer.

http://embeddedadventures.blogspot.com/ and look at the PicPack over on the right sidebar.


The Parallax motor kit would work fine with better traction wheels, yes. Whether you use my idea or some other notion for traction, the motors are great, thanks to DB I have a set of them here. Should be just about the right speed for a small mower too.

Electricity's notion of mounting the chainrings to the wheels sounds like a good one. If you need more traction you could mount them to the inside and outside of the wheels and you'd get a good 1.5"+ of spread which should be even better. It would also be trivial to count the teeth with IR, so you wouldn't even need any fancy encoders.

revoltage
08-06-2008, 12:56 PM
Have a great time Adrenalynn :veryhappy:
Yep its a serial rs232 programmer.

I measured the voltage across pin 2 and it is 3.9V.

I have only a multimeter available at home.

Ok then I'm going to buy the parallax motor kit and then find some traction wheels to replaced the ones obtained in the package.

I know I'm emphasizing again and again over this; that's because I have a planned budget and I cannot go beyond that limit. Hope you understand guyz.

Now the motor driver so that the robot reverse,stop, move forward and vice versa; will that be ok?

http://www.parallax.com/Store/Accessories/MotorServoControllers/tabid/160/CategoryID/22/List/0/SortField/0/Level/a/ProductID/64/Default.aspx


Thanks a lot to you all for being very helpful and comprehensive :):):). I can't thank you enough for your help and I'll be having more and more questions for the next few months :veryhappy:

Hats off to you all :happy:

ScuD
08-06-2008, 02:14 PM
Probably a dumb question, but you are using the ICSP header through a seperate programmer right?
If yes, what programmer are you using?

Or do you have a bootloaded PIC which uses the onboard serial connection?

Cause if you just use the board as-is in the schematic, it's not going to do anything.

revoltage
08-07-2008, 12:45 AM
Nope I am downloading the program using screamer through the serial cable from my pc to the PIC. Hey Adrenalynn do you think the board is alright then?

Adrenalynn
08-07-2008, 01:01 AM
I suspect it is, but what ScuD is getting at is that it doesn't appear that board has an onboard programmer. I haven't looked that closely at the schematic yet - just taking his word for it at the moment. I just got back from another brutal hike, so I'm trying to get caught up.

revoltage
08-07-2008, 01:12 AM
Recover well form the hike Adrenalynn :) Where was that hiking?

Nope the board does not have an onboard programmer. I just place the PIC16F877A(which already have bootloader in it) in the IC holder and power the board with a 9VDC adaptor. Then using my PC I have to translate the hex code and after that using the screamer v1.7 I have to download the hex through serial cable to the PIC16F877A.

Adrenalynn
08-07-2008, 01:20 AM
Up in the Sierras here in California. I'll blog it tomorrow - thanks for the well wishes!

So it's entirely possible the bootloader has been blown away, in which case you wouldn't be able to get code to it with what you have.

ScuD
08-07-2008, 02:00 AM
Seems to me like you have to manually connect the MAX232 to the PIC, since it's a proto board.
Have you done this?

revoltage
08-07-2008, 02:05 AM
Ok guess that I'll have to buy a programmer. There's a relatively good cheap one available here in the local electronics shops for $50.

Now that we have all decided for the motor system (that is the parallax Motor Mount & Wheel Kit with Position Controller), will its corresponding HB-25 Motor Controller be a good one so that the PIC send a signal and triggers appropriate motion? And a rather stupid question how many of these controllers do I need (because I'm not sure whether it can control both motors)?

http://www.parallax.com/Store/Accessories/MotorServoControllers/tabid/160/List/1/ProductID/64/Default.aspx?SortField=ProductName%2cProductName (http://www.parallax.com/Store/Accessories/MotorServoControllers/tabid/160/List/1/ProductID/64/Default.aspx?SortField=ProductName&#37;2cProductName)

Can you help me find out which dc motor and its controller would be suitable for the mulching blades?

Thanks :)

revoltage
08-07-2008, 02:11 AM
Yep the MAX232 is already in its socket. Some time in June I lend friend the protoboard to control a stepper motor. It was working fine. After that I have not used it again until last Monday when I discovered that I could no longer download a program. I guess like Adrenalynn said that I have to redownload bootloader into the PIC16F877A and test it again.:)

ScuD
08-07-2008, 04:46 AM
Actually, I meant that it seems there are no direct connections between the output of the MAX232 and the PIC, you need to wire those up yourself.

But since it worked fine for your friend, you can pretty much rule that out

revoltage
08-07-2008, 05:52 AM
Actually these connections have already been made. Thanks Scud

revoltage
08-07-2008, 09:54 AM
For the controllers?

Adrenalynn
08-07-2008, 10:14 AM
For dual channel controllers in this size range, I have to admit that I'm rather fond of Dimension Engineering's controllers. http://www.trossenrobotics.com/store/p/5466-Sabertooth-dual-25A-motor-driver.aspx

revoltage
08-07-2008, 10:28 AM
Hey guyz I have used a drive motor sizing tool from robotshop and here's the result. Do you think the parallax kit will still rock? I'm sorry if I'm a bit adamant about this.

Hmm:o...the sabertooth dual 25A motor driver is a bit too expensive:o

Will these be fine for dual control?
http://www.parallax.com/Store/Accessories/MotorServoControllers/tabid/160/List/1/ProductID/64/Default.aspx?SortField=ProductName%2cProductName (http://www.parallax.com/Store/Accessories/MotorServoControllers/tabid/160/List/1/ProductID/64/Default.aspx?SortField=ProductName&#37;2cProductName)

Adrenalynn
08-07-2008, 11:49 AM
If you buy two of them.

revoltage
08-07-2008, 12:01 PM
Ok. Guess after all I'll have to buy the sabertooth one. I'll have to sell my ass to buy all these stuffs lol; i'm having a headache :) I have calculated that the shipping itself would cost me more than $350!!! pfff
I feel like vomitting at the thought it and a terrible headache :(

Hey guyz here's a website where the components are relatively cheaper; please help me find equivalent components that we have already discussed. Thanks a lot :)

http://www.robotstorehk.com/

Adrenalynn
08-07-2008, 12:06 PM
Have you sent an email to Trossen Customer Service to see if there's anything they can do on the shipping? That's one of the nice things about working with Trossen - they do tend to do whatever they can.

revoltage
08-07-2008, 10:14 PM
I just sent an email to the store. Hey Adrenalynn do you still think that the parallax motor kit will be good after looking at the drive sizing calculation I have posted before(I'm sorry I'm asking that again and again)?

Adrenalynn
08-07-2008, 11:04 PM
I ran some numbers, and I think you'll fall short with those parameters. I think 5hr runtimes are pretty unrealistic without very expensive battery technology. I think you're looking at more like 0.2m/s for speed assuming a full 18kg load and a 20deg incline.

revoltage
08-08-2008, 12:51 AM
So will 1.5 - 2 hrs runtime be ok for the parallax kit and a suitable battery and the above parameters you gave?

Adrenalynn
08-08-2008, 01:21 AM
2hrs of runtime should require a battery weighing about 16kg by itself, assuming lead acid. 9kg or so for NiMH - but at more expense.

We start getting into a vicious cycle. That's why the smaller mowers only run about 20-30mins before they plug themselves in.

revoltage
08-08-2008, 01:45 AM
Yep I guess I have to lower my expectation and reduce the area to be mow to 80 square metres and then explain in my thesis that this robot can be expanded for larger areas and that the appropriate components have to be used at a greater expense. So I guess I have to find out some way that after the battery is down the robot either send out an alarm to be recharged or it finds its way to a charging station


I modified the motor sizing calculation and that's what I got; so ok then for the parallax kit (I'm a bit doubtful because the kit says that the motor runs at 150 rpm and my calculated speed is 26 rpm)?

revoltage
08-08-2008, 05:00 AM
Adrenalynn what do you say? :)

ooops
08-08-2008, 09:10 AM
I like the look that you are going for.
I will defer to Jodi on the math:)
It seems as though your drive train will have ample power to pull the bigger battery around, how flexible are you on the weight limit?

revoltage
08-08-2008, 09:19 AM
I think it should be 18 kg because I'm planning to use the parallax motor kit (and it's weight limit is 18kg) and from the attached motor sizing calculation for a 30 minutes runtime it says a 0.8Ah hour battery would be fine . What do you think?

Adrenalynn
08-08-2008, 10:51 AM
Where is your motor calculator getting the specifics for that motor?

Every specific motor design is different in its current draw.

Why do you have 0.0 in for acceleration? I'm surprised the calculator would accept that. It will take you infinite time to reach your top speed. Accelerating mass is the biggest gotcha.

revoltage
08-08-2008, 11:05 AM
I used the calculator from from the following website:

http://www.robotshop.ca/learning-center/dynamic-tools/drive-motor-selection.html

Adrenalynn
08-08-2008, 11:09 AM
And the acceleration?

revoltage
08-08-2008, 11:14 AM
I just put the zero value like stupid there...lol
For the traction wheel I'll manufacture one like the robotcut one with plastics or I'll scrap some old bike for the chain rings:)

Electricity
08-08-2008, 11:29 AM
Heres a random thought, what if you took a page from the airforce's book, and had a moble recharge station that, when signaled by your bot would come out to it, and either swap batteries, or attach and recharge the one its got. If it swapped batteries, it could then go back to the plug in station, and recharge the old one.
That would be pretty sweet :)

revoltage
08-08-2008, 11:35 AM
I would like the idea that our robot leaves a decking station and return to it for recharging when the battery is getting low but I've got no clue how to do that...and for now you must all have seen that I'm a real rookie in both electronics and robotics. That's precisely why I'm going to buy off the shelf components and try to get them together to produce that lawnbot.

revoltage
08-08-2008, 01:35 PM
ok guyz cya tomorrow, i'm off to sleep. ciao :)

ooops
08-08-2008, 03:48 PM
I'm going to buy off the shelf components and try to get them together to produce that lawnbot.
That seems a very sensable plan.
I for one am excited to see it come together!

revoltage
08-09-2008, 01:41 AM
Ok guyz, with your help and suggestions I have reviewed the objectives of our robot

revoltage
08-09-2008, 01:58 AM
I don't know if I will be allowed do the robotic lawn mower ...I'm extremely disgusted after all the efforts and research we made. :(

Adrenalynn
08-09-2008, 02:44 AM
Using the Parallax motors with imaginary 15.25cm powered wheels
Using a 7Ah Gel Cell (about 2.75kg in weight)


A good 20A/channel motor controller is a minimum requirement. At stall the motors can just exceed it, but a good controller is insulated against that with auto-shutdown.

I achieve a max velocity of 1.25m/s in simulation on the flat, 0.23m/s @ 25% incline
I accelerate as much as 0.3m/s^2
I can traverse a max of 25% incline

Assuming a worst-case of an average 20% incline, I run for 29min with the motor bogging down to 25RPM
Assuming an absolute disaster case of running right at the stall limit, I run for 11min with the motor down at 12RPM
Assuming a best-case of an average 1% incline, I run for 1:15hr with the motor at 105RPM max, tapering off as the battery wears.

I believe you can meet your specifications with this drivetrain configuration. If your lawn is level, I think you can well exceed your specifications.

Not a very useful shot, but here's your bot in simulation... ;)

revoltage
08-09-2008, 02:55 AM
Wow that's a great work you have done for me Adrenalynn. Thanks a lot. Yep the lawn should be level. :)

Adrenalynn
08-09-2008, 03:13 AM
I hope you can do your project - but if not, we'll all just enter the design and specification phase of your next project. :) Don't sweat it, these things often go that way. I don't think I ever had a first-pass thesis topic approved... Don't let it get you down - just prepares you for the "real world". ;)

This was a fun mental exercise anyway, imho.

revoltage
08-09-2008, 02:44 PM
Hi I have uploaded a draft of my project synopsis. Adrenalynn can you please review it?

revoltage
08-11-2008, 10:10 AM
Ok finally it's for sure I'm doing the lawn mower robot. We can all move forward with the list of components. Thanks guyz for your support :)

ooops
08-11-2008, 12:57 PM
Congrats on the approval:)

revoltage
08-11-2008, 10:04 PM
Thanks Ooops :)

Guyz now that Adrenalynn has given her consent about the parallax motor kit (backed up with great simulation results :)) and that the Sabertooth 25A dual motor controller to be used with the PIC16F877, can you suggest a dc motor up to 5200 rpm (for the cutting baldes)with it's appropriate controller (from trossen store only) to be connected to the PIC?

Hey Adrenalynn what software did you used to simulate the mower?

I am planning to buy all components from trossen only at one single purchasing order because the shIpping will cost me my ass...lol ($350 or perhaps more depending on the weight). And I can't afford to pay that shipping twice.

Also can I use the lead acid battery from trossen? It's only 2.5 kg
http://www.trossenrobotics.com/store/p/3448-Gel-Cell-12V-7-2AH-Zeus.aspx

Do anybody know how to make a charging station such that the robot can go there to "fill its reservoir" when the battery is getting low?

Also what do you think will be better, ultrasonic sensors or bump sensors for obstacle detection? And for these sensors do I have to connect them directly to the PIC or are there some interface boards before? And for these sensor do I need two of them ?

I have uploaded a simulation from robomow on youtube . Can you please check it. I want our mower to mow along the perimeter wire at first. After it returns to it's staring point I then want it to mow inside like in the video. Do I need a magnetic compass for this and any other accessories? Can you please suggest one from trossen?

YouTube - simulation of mow

I am planning to buy the components in two week's time. Do you think it's feasible?

Thanks a lot to all of you :)

revoltage
08-12-2008, 08:33 PM
Hi Guyz,

For the cutter blade motor will these be fine:

http://www.trossenrobotics.com/store/p/5135-4-1-36mm-Planetary-Gearmotor-RS-540.aspx

and for the controller to interface with the PIC:

http://www.trossenrobotics.com/store/p/4267-BaneBots-Motor-Controller-45A-peak-.aspx

For the obstacle sensors:

Will this ultrasonic do?

http://www.trossenrobotics.com/devantech-srf10-ultrasonic-range-sensor.aspx

or bump sensors?

http://www.trossenrobotics.com/store/p/3470-SPDT-Snap-Action-W-Lever-Large-.aspx

Thanks to you all. I need your precious advice :)

Adrenalynn
08-12-2008, 08:45 PM
I really don't know that that motor is going to have the torque. I don't know that there's anything in the Trossen catalog that can meet the speed and torque requirements for spinning a cutting blade... It might, but I can't feel good about saying that it will.

Does your PIC board have servo/PWM outputs? You'd need them to use that motor controller.

I don't care as much for that sonar unit you selected. I'm using the Maxbotics instead.

That switch should be rigable to become a bump-sensor. The bumper design will be more important.

metaform3d
08-12-2008, 08:53 PM
[...]can you suggest a dc motor up to 5200 rpm (for the cutting baldes)with it's appropriate controller (from trossen store only) to be connected to the PIC?Let me point out that the Robomow has three high-RPM cutting blades, each about 6-8 inches across. I'm guessing that's much easier on the batteries than a single large blade.

Adrenalynn
08-12-2008, 08:58 PM
Also easier on both the drive train and trying to find motors with enough speed and torque to cut grass at the end of the blade. Rotational mass is a killer.

revoltage
08-12-2008, 09:00 PM
Nope my board is just a simple protoboard the olimex one I posted lately. Are the Maxbotics available on trossen? I guess I'll go for a sonar sensor. Will I need a circuit to connect the sonar to the a pic input pin ?

One more thing I want our mower to mow along the perimeter wire at first. After it returns to it's staring point I then want it to mow inside like in the video (in a previous post). Do I need a magnetic compass for this and any other accessories? Can you please suggest one from trossen?

Thanks guyz.

Adrenalynn
08-12-2008, 09:12 PM
Your motor controller is going to have to get data from your PIC somehow. Be careful to match the motor controller to what your processor can support.

Yup - the Maxbotics are available in several flavors on the Trossen site.


At some point you're going to have to start doing some of this design work yourself. You tell us: What do you need to make the robot mow the pattern that you want? Ye Gods Man, this is a Thesis Project, and a final year project at that! You're going for a degree in mechtronics, something I don't think any of us possess here that I've seen... I'd hope that you'd be telling us how to design this stuff. ;)

revoltage
08-13-2008, 07:29 AM
To mow the pattern shown on the video we need a magnetic compass (metafrom3d posted how robomow works). So when the robot mows along the perimeter wire and return to its starting point(edging mode) the compass reads 360 degrees so that the robot knows it is at it's starting point. I think this will do:

http://www.trossenrobotics.com/devantech-magnetic-compass-sensor.aspx

I think for a robotic rookie like me I could do without the robot mowing along the edge first. I just place the robot inside the perimeter wire and it just mows randomly there...don't know yet. I think I'll buy the compass in the eventuality I see I can implement it.

Now for the sonar sensor I'll buy two of these
http://www.trossenrobotics.com/maxbotix-lv-maxsonar-ez3.aspx

Now I need your knowledge and experience here. To connect these sensors to the PIC do I need some kind of interface board or do I connect it directly to an ADC pin?

For the battery I think I'll go for a lead-acid one.

sam
08-13-2008, 07:49 AM
The sensors looks nice. I think for a project like this one, espacialy if you don't want to go all out on a battery (Li-po 12 volt 8 amsp are expensive), a lead-acid is a good idea.

Adrenalynn
08-13-2008, 11:48 AM
The compass has two outputs: Output 1 Timing Pulse 1mS to 37mS in 0.1mS increments Output 2 I2C Interface, 0-255 and 0-3599

* From the Trossen catalog page.

So you're either going to need to go I2C with a clock (crystal) or you need a PWM input on the PIC.

If your robot is going to be less than 2ft (~0.61m) wide at the widest, that's probably a good sonar selection. Since you want *instant* notification of obstructions, I'd select my sonar to be as wide as my 'bot or a little wider so it doesn't have to scan. The Maxbotics sonars have a BUNCH of output possibilities, one of which is pure analog so it can be connected fairly direct. You can also connect it to the serial port (RX) on the PIC, or to a PWM input.

If there's any hope of getting batteries locally (a garden supply store, farm supply store, etc), I wouldn't ship them. They take forever through customs and are *very* heavy and therefore expensive. But yes, 7.2Ah should do fine for you.

Is that mowing pattern really the most efficient pattern?

revoltage
08-13-2008, 12:22 PM
The mowing pattern is clearly not efficient :(
But being a rookie I can't see another way to simplify things for the project, you got any idea?

Thanks

Adrenalynn
08-13-2008, 01:09 PM
I'm just wanting you to think through what you're trying to accomplish so you have a killer thesis and killer CV.

You simulated the pseudo-random mowing path - did you give thought to any other path possibilities, and simulate them? Did you measure how long it would take to reliably mow an area with random vs let's say a spiral pattern?

Did you look for other work out there on the most efficient patterns, not necessarily for mowing, but perhaps path-planning or mapping? I've seen a fair amount of academic work in this area of exploration.

What do you hope your thesis will show? What are you trying to prove? Surely it's not that you can just build a device that a thousand other people have built, or a device that a million other people have built, and then toss a spinning blade on it... Isn't there some goal or some theory or some area of exploration?

Different schools, different countries. I may be trying to apply my academic career to yours and I may be totally off-base about what's expected, and if I am, feel free to tell me so...

revoltage
08-13-2008, 07:49 PM
I would certainly have like the the robot to do some kind of mapping once its inside the perimeter area and then from this mapping find the most suitable path to start mowing. But I don't know how and is this kind of mapping possible without computer vision? And I have a project budget of only $1200 (and that includes shipping which is $400).

revoltage
08-13-2008, 08:44 PM
I guess we could do with GPS (but it's kind of expensive) and would need a dedicated laptop?

Adrenalynn
08-13-2008, 08:56 PM
I would certainly have like the the robot to do some kind of mapping once its inside the perimeter area

Then go for it!



But I don't know how

Edison didn't know how to make a light bulb, but that didn't stop him! Also: See my signature line.



and is this kind of mapping possible without computer vision?

Is it? How would you go about doing it without vision?



And I have a project budget of only $1200 (and that includes shipping which is $400).

I've found that $800 spent carefully combined with visits to junk suppliers can replace a million dollars in "organized R&D" if one is clever and motivated. Use your resources carefully. Is shipping 15lbs of toxic waste around the world a careful use of resources? I'm not certain - I try to exhaust local resources first and save my shipping money to buy the cool stuff that I can't get locally for less.

revoltage
08-13-2008, 09:15 PM
Without vision I was thinking about GPS but that would be dependent on a PC or laptop, isn't it?

It would have been good to place the robot inside the perimeter and then it goes around the wire near its edge to "recognise the terrain". After a 360 degree rotation, that is when it return to its starting point it knows that it has to mow inside this region and create some kind of map (coordinate systems) to do this.

Or the robot project some kind of patterns visible only to the mower on the lawn and follow these pattern???

revoltage
08-14-2008, 01:28 AM
For this project can we implement SLAM with no computer vision ( I have wikipediaed localization & mapping)? What components do we need for that apart the sonar sensors we will be using for obstacle dtection ?

Adrenalynn
08-14-2008, 02:48 AM
Try looking into Path Planning. Maybe look into the wavefront path planning algorithm. Wavefront at its simplest can be done with any single sensor capable of providing distance, although having wheel encoders (so cheap as to be almost free) are nearly a requirement - another advantage to a high-traction design.

revoltage
08-14-2008, 03:51 AM
I have read the following:

http://cgi.cse.unsw.edu.au/~cs4411/wiki/index.php?title=Path_Planning

I think that global planning would be great but what I do not understand is whether how the robot gets the map? Is the filling of the grid squares with numbers done by a human and thus the navigation system system is dedicated to a particular area? Or does the robot move round the path for recognition first (and subsequently filling in the grids with numbers) and then does its job?

Adrenalynn
08-14-2008, 04:05 AM
What's going to be most efficient?

How about a combination? What if it starts at point A with pointz being the furthest corner. It navigates back and forth until it hits an obstacle, at which time it has to plan around that obstacle?

revoltage
08-14-2008, 04:18 AM
I was thinking that the robot could move around the perimeter of the boundary wire and then using the global planning system it could move inside that defined boundary to locate obstacle. And during this "terrain recognition time" the cutter blades would be off.


What if it starts at point A with pointz being the furthest corner. It navigates back and forth until it hits an obstacle, at which time it has to plan around that obstacle?I'm a bit lost here :(

ScuD
08-14-2008, 04:49 AM
Without vision I was thinking about GPS but that would be dependent on a PC or laptop, isn't it?



Sorry, haven't read the entire thread, I'm at work.
But glancing over this line I wanted to comment:

You don't need a PC or laptop for GPS. Plenty of GPS receivers have a serial output, which sends out NMEA strings at constant intervals, eg. every second or so, which you could perfectly read out using a simple microcontroller.

In these strings you can find the latitute, longitude, and a bunch of other info which you may or may not need.
So basically, you could just read the string into a buffer, discard the info you don't need, and filter for the latitude and longitude, presto.

Keep in mind though, you're not going to get an accuracy of 2cm. Think more along the lines of a few meters, maybe half a meter if you have a really good receiver.

Oh, here's (http://www.geoaps.com/NMEA.htm) a small explanation of the NMEA strings.

revoltage
08-14-2008, 04:50 AM
Here's a plan of the lawn with perimeter wire, obstacle, etc..

Now let's assume the robot has left it's starting point A and after a complete loop returns there again. What I'm not understanding is how with global planning the robot moves inside and locates obstacles, avoids it and map this into it's grid of square. And the robot has to remember where the obstacles are etc.. What components should be used to do that?

I mean the robot just has to know where are the obstacle and then find the quickest way to mow the lawn...not just find a specific point and go there...how to do this?

revoltage
08-14-2008, 10:42 AM
I was thinking that if we use only one sonar sensor in front and when the robot reach the perimeter wire it just uses differential drive ad rotates onto itself and move again forward...no need to reverse. This will do isn't it?

Adrenalynn
08-14-2008, 10:55 AM
I was thinking that if we use only one sonar sensor in front and when the robot reach the perimeter wire it just uses differential drive ad rotates onto itself and move again forward...no need to reverse. This will do isn't it?

Unless it is boxed in, in which case it will need to reverse.

As far as the other questions, I'm going to wait for you to give them some thought yourself.

revoltage
08-14-2008, 11:15 AM
I'm still trying to master that wavefront navigation concept.

I was thinking that a Sharp Ir Rangefinder would do but after reading more I came across an article that they can be affected by sunlight...so I'm not going to bet on them...

Also I just learn from an article that first the robot scans the area (they used Sharp Ir Rangefinder) and then a map matrix is done by Man and then the rest is programming. So I assume that first I'll let the mower scan the area (I think I'll have to reduce my mowing area to some 20 metre square). Then I let it zigzag between the perimeter wires until it come across the obstacles. Only when an obstacle block its path then it uses the navigation principle of finding the shortest way of going behind the obstacle...and this continues again and again. But this would require some tedious programming. What do you think?

Adrenalynn
08-14-2008, 12:00 PM
No single sensor is everything to everyone at all times.

You probably need to read more. A map can be manually generated, or it can be built on-the-fly. You could simultaneously mow and map the entire surface of the earth if you wanted. No reason to drop your mowing area.

What do I think? I think any problem worth solving has exciting pieces and tedious pieces. Wait until you get a job...

revoltage
08-14-2008, 12:27 PM
I read about navigation from this website:

http://www.societyofrobots.com/programming_wavefront.shtml

I've to read it again.It's just it not getting into my head
There's a technique called Adaptive Mapping. I guess this is the right method...I still have to figure out how this works :(

Also is it feasible to implement this in my $1200 budget?

Thanks

Adrenalynn
08-14-2008, 12:48 PM
It's getting harder and harder...

Also is it feasible to implement this in my $1200 budget?


Yes. It's feasible to implement on a $50 robot, as SOR has demonstrated.

revoltage
08-14-2008, 01:12 PM
hmmm...ok I'll go over it all again. Thanks :)

metaform3d
08-14-2008, 01:12 PM
1) Put the wire sensor in the center of the bot, and inset the wire by the radius of the bot. Then you can mow forward and backward and always hit just the grass area.

2) SLAM for mowing should be possible using only wheel encoders, a compass, and wire and bump sensors. Just drive around hitting obstacles and the wire; mark them on your internal map; see where you haven't been and go there. You only have to be as accurate as the diameter of your cutting circle.

revoltage
08-15-2008, 12:17 PM
Thanks meta.

Sorry I did not get that part:


....and inset the wire by the radius of the botCould you please elaborate on it? Thanks :)

Today I have been googling and yahooing on lawn mowers and navigation system the whole day. I hope to grasp the wavefront concept till tomorrow night so that we can continue with the list of components. :)

metaform3d
08-15-2008, 01:40 PM
Could you please elaborate on it? Thanks :)You basically have to detect the perimeter of the lawn area before the cutting circle of the mower crosses the outside edge. If you inset the perimeter wire by a distance equal to one half the diameter of the cutting circle, and place the detector for the wire at the center of that circle, then as the bot approaches the perimeter moving any direction it will detect the edge just before starting to cut the verges.

At least as a simple idea. The details would probably be more complex depending on the configuration and layout of the bot.

revoltage
08-15-2008, 01:45 PM
Ok thanks meta I got it :)

revoltage
08-17-2008, 02:31 AM
Hey I have been doing some research work over the wavefront algorithm. I don't know if I have understood it well but here's how I think it can be applied to the robotic lawn mower:

The robot starts. Using a sensor (usually a sharp IR scanner ) it scans the path and stores obstacle in an already preprogrammed memory map in the microcontroller. This memory map (or map matrix) can be updated as robot encounters new objects, etc... (adaptive mapping). The memory map consist of grids; one grid unit in our case can represent the centre of the cutting blades.

In the mapping process the robot should have a goal and move to that goal. If we make each grid a goal that would do the job I think. And the robot would automatically avoid the block it considers impassable. The robot then can clear its memory when it reaches the last goal and the process restarts.

But I did not understand how to make each grid a goal. Can someone please help me here?

Did I got the theory right?

Adrenalynn
08-17-2008, 02:43 AM
I think I would set waypoints and let it decide the best path to get there, specifying that in order to reach the final goal it must touch all legal squares - in that way it will iterate the near-best-possible pattern.

Also, I don't think I'd pre-map. That's what the sensors are for. Let it figure out where illegal squares are on its own.

But that's just me.

revoltage
08-17-2008, 04:01 AM
Thanks Adrenalynn :)

But if we set waypoints the robot is bound to work in a specific area isn't it? Or will it adapt to any lawn area enclosed inside the perimeter wire.

sam
08-17-2008, 07:34 AM
Looks like a very nice project!

You can do it! :veryhappy:

revoltage
08-17-2008, 08:08 AM
Thanks Sam for your encouragements. :)
You are welcome to check this thread from time to time and give in your ideas. ;)

revoltage
08-19-2008, 01:08 PM
Hi Guyz,

That's the list of components I'm going to order for to be used with PIC 16F8877A or PIC 18:

Devantech Magnetic Compass Sensor (http://www.trossenrobotics.com/store/ProductDetails.aspx?itemID=4855)

LV-MaxSonar-EZ4 (http://www.trossenrobotics.com/store/ProductDetails.aspx?itemID=5546)

Motor Mount & Wheel Kit with Position Controller (http://www.trossenrobotics.com/store/ProductDetails.aspx?itemID=5761)

Phidget IR Distance Sensor Kit (4-30") - 1 sensor (http://www.trossenrobotics.com/store/ProductDetails.aspx?itemID=5430)

Robotics Connection RangeWizard (http://www.trossenrobotics.com/store/ProductDetails.aspx?itemID=5614)

Sabertooth dual 25A motor driver (http://www.trossenrobotics.com/store/ProductDetails.aspx?itemID=5466)

I need your thumbs up to order them. Have I miss some components (I'm building this robot with off the shelf components only)?

About the parallax motor mount kit I read in it's specification sheet that

"The Position Controller is designed to be interfaced with an HB-25 motor controller to control the position and speed of the motor and wheel. This can be accomplished by connecting a standard three-pin cable from the header pins marked “To HB25” to the three-pin header on the HB-25."

Will the sabertooth work?

Thanks

4mem8
08-19-2008, 01:19 PM
Revoltage: All looks pretty cool to me so far, I have the Parallax motor drive unit [ One of the first 5 of the line] and they are a brilliant piece of engineering, it will be one of my next projects. I love that range wizard also, have not seen that one before, But I like it, might have to investigate that one further. Look forward to the project Revoltage.

Adrenalynn
08-19-2008, 02:00 PM
Yup - the sabertooth will work.

darkback2
08-20-2008, 10:49 AM
Hmmmm navigation.

Ok here are a couple of thing I find possible...but difficult. Absolute positioning when a robot is outside? One idea that I'm using in my book is to have two objects which can be recognized by the robot placed at two locations on the parameter of the lawn. Have two cameras mounted on top of the robot tracking the two objects using a pan/tilt function. As the robot is following the wire, it would create a map of the exterior of the lawn by noting its position relative to the two objects. (the position of the pan tilt servos the camera is mounted on, keep a record of this, and you can draw a map of the lawn. ) Then the robot would plot a path that would cover all of the interior or the lawn, and follow it.

Major problems? The objects would have to be mounted in such a way as to not impede the robots motion, but where they would still be visible from all parts of the lawn....you could create an algorithm to deal with loss of sight of the object...

Writing the actual algorithm...I would check into CNC software to be honest. Think of the robot as the CNC milling head, and the lawn as the shape that it is cutting out.

Not really sure how to do any of this, but I do know it would be hard for a robot to create a map if it didn't know where it was. You can't count on wheel encoders because wheel slip is going to be a serious problem.

Oh...if you ge this working...I want a copy of the software.

DB

Adrenalynn
08-20-2008, 12:50 PM
Hi DB, welcome back!

Did you look at my "wheel design" for him? I don't think slip will bite him too hard, and even if it does, there are many algorithms out there to deal with that.

revoltage
08-20-2008, 12:52 PM
Thanks Db but I think for my first robot it would be difficult :( and I'm slow at programming too.

Later when I finish with this one I can still improved it and implement your suggestions :)

Now that wavefront algorithm thing I'm still stuck at it and has not yet figured out how after the robot has created a memory map of the region it has scanned move over all the grids, avoid obstacles and clear its memory and make sure it does not pass over the same path again. I have read the SOR tutorial like a 100 times but still I am lost. Someone please give me a hint?

Thanks :)

darkback2
08-20-2008, 02:28 PM
Quick question...are you making models to test with? Just wondering.

Oh...and slip is the sort of thing that builds up over time...not to say that the wheels slip more and more, but that the errors accumulate to the point where the robot is nowhere near where it thinks it is. I know it doesn't matter too much because the robot should not be able to leave the inside of the ring...but it just seams if we are trying to "make sure" the whole lawn gets mowed...oh and the mower has to find its way home too...so I guess some sort of location is important.

Could the base track the robot and give it a relative position via wifi or something? Camera on the robots base tracks the robot...sends it coordinates...hmmmmm

DB

metaform3d
08-20-2008, 09:34 PM
Oh...and slip is the sort of thing that builds up over time...not to say that the wheels slip more and more, but that the errors accumulate to the point where the robot is nowhere near where it thinks it is. I know it doesn't matter too much because the robot should not be able to leave the inside of the ring...but it just seams if we are trying to "make sure" the whole lawn gets mowed...oh and the mower has to find its way home too...so I guess some sort of location is important.You've correctly assessed that this is a statistical problem. You know that there will be error, so you cannot take the results directly but must compare them to prior results and a model of what you expect to find. What you know is that the perimeter is a piecewise-smooth closed curve. What you get is discrete samples of this curve. From this you have to estimate the most likely perimeter that gives you the dead-reckoning results that you find empirically.

Ideally you'd use Bayesian analysis, although I bet there are a lot of informal algorithms that can solve for the perimeter curve and the robot position closely enough to mow the lawn. You want to visit each spot more than once, after all.

revoltage
08-21-2008, 01:59 AM
Now I'm actually seeing if I can implement textural analysis using vision for our robot..still doing some research.

Adrenalynn now this is your plate...will that cost much and is the programming ok for a beginner?

Thanks :)

Adrenalynn
08-21-2008, 02:42 AM
On a PIC chip? I wouldn't get in line for that project.

revoltage
08-21-2008, 03:14 AM
Yep you are right

Hmmm after a bit of research and some reading I have concluded..no no vision..I have got a masters paper where they used textural analysis for a lawn mower...wow it took some time and involved training of the robot too and obviously the guy had some good background of computer vision and robotics. hehe

For now I'll stuck to wavefront algorithm only...
Hey Adrenalynn I am a bit lost with that algorithm, can you please give me another hint ? :)

Adrenalynn
08-21-2008, 06:16 AM
What kind of code have you implemented now for it? Where specifically are you stuck? There are examples galore.

I don't have time to do your coding for you, I'm afraid. Too many projects, not enough time.

revoltage
08-21-2008, 07:33 AM
I'm still at the design phase for now...I have not understood how to make the robot move in each grid after the first scan and avoid obstacle.

revoltage
08-22-2008, 11:18 AM
Now I think I should change my project to make a robotic judge or a JudgeBot...so that the Olympic games are more fair....;)

revoltage
08-24-2008, 01:33 AM
Now guyz for the perimeter wire I got this from the net:


"Think of it as two separate devices: the pulse generator and the detector. The pulse generator sends the current in the wire. You have to send your pulse at a specific frequency, let's say 50 KHz. You can do it with a small 555 timer or a microcontroller. The detector is a simple coil attached to the robot. It acts as an antenna and will resonate when the robot pass over the wire perimeter. Your embedded intelligence has to deal with this."


Can someone help me build a schematic for this please?

Adrenalynn
08-24-2008, 03:37 AM
What part of the design are you planning on doing again?

revoltage
08-24-2008, 04:05 AM
The perimeter wire... I haven't got a clue how to design this circuit

Adrenalynn
08-24-2008, 04:15 AM
Ok, so your thesis for your undergraduate degree is really "how to spread a wire out on a lawn"?

Have you considered going with an easier project, something you can accomplish?

revoltage
08-24-2008, 04:44 AM
:o Hmmm what can I say ?:o

sam
08-24-2008, 08:30 AM
...basic electronics stuffs also are expensive. There are no robotic kits, microcontroller have just make its apparition.

And the lecturers don't give projects; you have to go with a project proposal to make sure you can complete your degree within four years. Otherwise you have to extend another semester just for project or just keep wandering from door to door for a project. I had searched a lot during my vacation to manage to get a simple project but could not find one...so I had to jump on the lawn mower itself

:o

there are many books that can help you in this mater. there are also a lot of websites that can give you a lot of information.

As for the price, it's pretty cheap if you know where to go. Where I live, the Radio Shack sells electronic pieces, but over 6 times the prices of a normal electronics store. You probably have a store that specialises in this kind of things.

darkback2
08-24-2008, 04:15 PM
www.allelectroniics.com Great for all the electronics stuff you might need..cheap.

Check out their website. Also, Why make a circuit from scratch? Why not hack a perimeter wire detector from a system like invisible fence...

http://www.invisiblefence.com/

http://www.radiofence.com/dog-fences/innotek-basic-dog-fence-sd2000.htm

I got a system for my dogs for under $100...extra collers about $70...you could probably hook one up to a phidget 8/8/8.

DB

revoltage
08-27-2008, 10:37 PM
Thanks DB and Sam :)

I was wondering if it is necessary to buy the rangewizard to as an interface between my ir and sonar sensors to the PIC?

http://www.trossenrobotics.com/robotics-connection-rangewizard.aspx

Adrenalynn
08-28-2008, 12:16 AM
Nope! In fact, it'd probably be tougher.

revoltage
08-28-2008, 12:19 AM
Ok then I'll remove it from the list. Thanks Adrenalynn.:)
In two weeks time I'll be buying the components from Trossen.

jes1510
08-28-2008, 02:54 PM
www.allelectroniics.com (http://www.allelectroniics.com) Great for all the electronics stuff you might need..cheap.

Check out their website. Also, Why make a circuit from scratch? Why not hack a perimeter wire detector from a system like invisible fence...

http://www.invisiblefence.com/

http://www.radiofence.com/dog-fences/innotek-basic-dog-fence-sd2000.htm

I got a system for my dogs for under $100...extra collers about $70...you could probably hook one up to a phidget 8/8/8.

DB

Actually the radio systems fence might be pretty easy to interface. The receiver pulses a beeper and then a transformer depending on how close to the wire you are. You can look for the pulses on the primary of the transformer and use that to detect the wire. The transformer is probably driven by a FET that varies the pulse width to increase the shock level. Replacing the transformer with a fixed resistor will probably work great. Incidentally I used to work in R&D for a company that designed these systems and I'm assuming the design hasn't changed much in the past 10 years.

jes1510
08-28-2008, 03:10 PM
One more thing to note about these systems:
The shock collar typically stops pulsing the transformer after a certain period of time. This prevents the collar from shocking the dog forever in some circumstances. You may need to periodically reset the collar to prevent this from being a problem.

revoltage
08-31-2008, 02:04 AM
Guys, I have attached circuit for the perimeter wire only (i.e the transmitter). Is it ok?
For the sensor can you help me find one to sense the wire?
Also the parallax motor mount kit's motor are they brushed or brushless dc motors?

Thanks

revoltage
09-01-2008, 11:21 PM
I think that I'll drop the idea of including mapping (wavefront algorithm) due to the high possibility of introducing slip errors. I'll use heuristic planning instead as a first implementation.

revoltage
09-03-2008, 05:00 AM
And a rather stupid question. How is the component I indicated in the picture with a white line called?

jes1510
09-03-2008, 08:01 AM
That looks like a piece of 80/20.
http://www.8020.net/

They have an ebay store as well so you may be able to pick up some pieces cheap if you need them.

revoltage
09-03-2008, 12:21 PM
Thanks Jes1510.

Has anyone check my circuit for the perimeter wire transmitter?

Now for the mowing motor, I have attached the data sheet for the single motor that robocut uses for cutting grass (GR42-40). Can someone suggest a motor from Trossen that I can buy?

I have a motor from a hand-held screwdriver which I have torn apart lately but I don't have its specs; don't know if it will do for the cutting of grass.

For the battery I'm going to get a 12V 9AH lead acid locally.
For the casing I'll be using 6mm thick polypropylene sheets.

revoltage
09-05-2008, 11:13 AM
I have attached the list of components (which you all have helped me to select- thanks a lot ;)) I'll order some time next week.

Are all the quantities (for the sensors) I will order enough to get the job done???

Also has someone check my circuit for the perimeter wire?

jes1510
09-05-2008, 12:56 PM
It looks like your transmitter circuit is going to be problematic. The first problem is that since you are transmitting a square wave you will be receiving an integral of that wave that will probably be a triangle wave form.

How long will the wire be?

revoltage
09-05-2008, 01:22 PM
The wire will probably be 200 m

jes1510
09-05-2008, 01:50 PM
My other concern is that the wire length will directly affect the amount of range on the wire. You may be better off using a current driven system instead of a voltage driven system. That will allow you to vary the wire length and still get the same amount of current through the wire. A constant current source also may allow you to simply drive one end of the wire into the ground so that it doesn't have to be brought back to the transmitter.

Have you explored a sine wave oscillator for the transmitter? I think a simple oscillator with a current drive acting as the transmitter and an LM567 tone decoder as the receiver may work. I think that transmitting around 10kHz would be about right. That is where most of the dog fence systems live.

Here is a page on sine wave oscillators:
http://www.piclist.com/images/www/hobby_elec/e_ckt18.htm


Here is the LM567 datasheet:
http://www.national.com/ds/LM/LM567.pdf

Finally here is a page on constant current sources:
http://en.wikipedia.org/wiki/Current_source

EDIT: Bah I misread your wire length and made some miscalculations. I removed the offending calculations.

Quantum
09-05-2008, 02:06 PM
I just read the back end of this thread and am just jumping in not sure if this was mentioned or not but will read the whole thread when I leave work. Will edit if need to.

I would avoid trying to send a single on wire that long. There is a easier way. Most places that use guided robots have a wire in the ground that simply send current thru it. When you send current thru a wire/cable it creates a magnetic field. You simply could have a sensor that detects magnetic fields and when it hits the field your reverse in the opposite direction. Or you lay the wire and you make pathways for it to follow. Most places whith forklifts use this to prevent the lift from straying of the path and in to racking etc...

This is my winter project to make a robotic lawnmower. I am going to lay the wire in early spring like the cable guys does and a month or so and you shouldnt notice the path ways anymore. Then design a sensor to lock on and follow the path like a maze.

The only problem is if you have to dig in your hard later might be tough to avoid the guidance wire.

Hope this helps

Paul

revoltage
09-06-2008, 05:41 AM
Hi,

Thanks guyz

I have get this link.

http://www.philohome.com/sensors/filoguide.htm

At first glance it seems that this technology will do. Do you think I could use these circuits with a 100 m wire loop?

Adrenalynn
09-06-2008, 06:24 AM
There's a difference between centimeters and hundreds of meters.

Also, since this isn't modulated in any kind of way that leads to certainty, things like sprinkler wiring will drive it crazy.

jes1510
09-07-2008, 02:10 PM
This is purely conjecture but you may be able to use a microcontroller in place of the 555 timer and send serial data down the wire. Of course the amplifier section would have to be changed as is pointed out above. If you use a fairly slow baud rate then you would be able to receive the serial data on on the receiver. Sending a single character down the wire should be fine.

Adrenalynn
09-07-2008, 02:28 PM
Yup! Agreed - that'd be the way to go.

revoltage
09-08-2008, 11:09 AM
Have passed my order for the components :). Now will be the time to test, try (before I have to write on the design section of the thesis) and learn electronics and programming.

revoltage
09-10-2008, 10:55 PM
I was thinking of implementing a base station from where the robot could start and come to charge itself when the battery is down. Also what should I use so that the user can programmed the robot to mow after each 'n' days ? I mean complete autonomy that is the user can input after how many days he wants the robot to mow after each successive mowing. Also he can reset that setting later??

Adrenalynn
09-10-2008, 11:03 PM
You need to add a real time clock (RTC)

revoltage
09-10-2008, 11:16 PM
Thanks Adrenalynn. Just found this:

http://www.sparkfun.com/commerce/product_info.php?products_id=99

I have attached a schematic from the robocut website. I think it solves the perimeter wire and charging station problem. It uses both a microcontroller and a 555 timer for this purpose. Is that right?

jes1510
09-11-2008, 06:42 PM
Did it have a receiver design as well? The tone decoder I mentioned earlier will not work with this. Do you have a link?

revoltage
09-11-2008, 09:45 PM
Ok got the receiver section attached. However the values of the inductance is not known. Will these circuits do?

jes1510
09-12-2008, 11:24 AM
After looking over the schematic I have to say that I'm not entirely sure how it works. It looks like it is supposed to oscillate at a frequency determined by the feedback resistor and cap but I'm not entirely sure. It looks like it is supposed to oscillate until brought near the wire where the induced signal drags the oscillator frequency to the center frequency of the transmitter but I'm not really sure. I'm out of my area of expertise here.

revoltage
09-14-2008, 12:39 AM
I have attached the other schematics I have obtained from the website, perhaps this will help. I'll have to order these components even though I'm not understanding the circuit because the dollar values is rising here.

revoltage
09-15-2008, 11:18 AM
Someone can help me understand the perimeter wire and sensor schematics I uploaded please?

Adrenalynn
09-15-2008, 11:23 AM
Did you try contacting the person that did the work?

revoltage
09-15-2008, 11:43 AM
Yep since the past weeks but no response at all. I think he was trying to sell his robocut but did unfortunately not get a patent for it (I even had ill thoughts of buying the robot but I decided it's better learn robotics myself...:)). So he stopped the production and updating the website altogether. What I read from his website is that his master degree helped him build that circuit...

I even contacted parallax (he used parallax components to build robocut) because the employees there were given a prototype of the robot, but no response as well.

It seems easy to implement the circuit but understanding it is more important

revoltage
09-16-2008, 11:13 AM
Anyone?

darkback2
09-16-2008, 01:27 PM
Ok...two questions.

First, could you build the circuit using the drawings? Is there enough info there? and second, why not just build it, and see what happens?

Really I would love to help on that one, but its way over my head...in ways that I haven't chosen to focus.

DB

Quantum
09-16-2008, 04:53 PM
http://forums.trossenrobotics.com/attachment.php?attachmentid=662&d=1221187500 (http://forums.trossenrobotics.com/attachment.php?attachmentid=662&d=1221187500)

The buried cable sensor uses an inductance coil probably 12 to 20 turns. You might need more or less might work as well, probably will have to experiment with this. And probably 3 to 5 in in diameter samething youll have to play around with this.

The 4069 chip is some sort of current sense amplifier. I look at the spec sheet and from the max4069-4072 series the pin out look wrong in the sketch. According to the sketch pin 4 shows an output where the datasheet show this as an n.c. which a no connect which means this pin is not used in the chip at its just there in the platic package.

He might have used a different manufacture but they usually dont change pin outs.

But the way it works sort of like this. You have a refeence voltage lets say 5 volts and as the the coil is put over the wire the current thru the wire is picked up by the coil. The chip senses this since its connected to the chip thru the small resistors circuit. The chip then outputs a differnet voltage lets say it drops it two 2v when the coil is over the buried wire. Now you know your refence is 5v technically the closer you get the voltge is dropped till your over the wire in the ground and it drops to 2v. With this info you have to convert the analog info to digital thru a adc (analog to digital converter) since the basic stamp cant read analog inputs. You would be able to sense how far you are from the buried wire.

The ir sensor is just in series with each other. Basically hes using one pin to read two sensors.
It saves on pin usage but you might want to put them on individual pins so that you know if something is blocking your mower and its in front and to left or to the right instead oh theres somthing in front of me just back away.

I will look into this a little more since I want to do this as well.

Sorry wrote this really quickly spelling mistakes are guarenteed.

Paul

revoltage
09-16-2008, 09:50 PM
Thanks DB and Quantum

I'm going to buy the components to test it.

I'm going to use a 10uH inductor.

For the 4069 I got this:
http://www.futurlec.com/4000Series/CD4069.shtml

It's some kind of inverter. Is it the same?

Adrenalynn
09-16-2008, 10:21 PM
It looks wrong 'cause it's the wrong chip. :)

The 4069U is a hex inverter, pretty common fair. It's a multi inverter which is why you have matched pins for input and output. http://www.alldatasheet.com/view.jsp?Searchword=4069U&q=4069U

Revoltage is using a PIC chip, I believe, that has a few ADC inputs...

Want to make sure you get a chip with throughhole package, Revoltage - I doubt you're ready for building SMT boards. ;)

revoltage
09-16-2008, 10:32 PM
Want to make sure you get a chip with throughhole package, Revoltage - I doubt you're ready for building SMT boards.

Lol definitely beyond my capability for now :)

Quantum
09-17-2008, 12:13 AM
Yea that makes more sense.

Last post all wrong all of it.

When the inductor passes over cable it goes high invertor makes it low then the second inverter takes the low and brings it high again, Resistors and caps are ther to clean and protect the signal. and to make your processor to read it better.

The ir sensors are in series but they give individual signals back thru p13 and p14.

Hey if your gonna get it wrong might as well mess the whole thing up.

But that other chip that I stumbled on could sense how far you are from the cable. Which you could slow down or increase speeds of the mower. Invertor lests you follow exactly.

The best thting is to buy a chip or two and play around with it. You could also ask for a sample of the chip. I do it when prototyping if there available. Check out the manufactors website.

Paul
I really got to stop jumping in and read this whole post but its like 100 pages by now that would take for ever.

revoltage
09-17-2008, 01:40 AM
Thanks Quantum :),

You can read the whole thread but I guess you can skip some of my posts because sometimes I repeat the same questions ;).

Yep I'm going to buy an excess of these components to play around with first.

Adrenalynn
09-17-2008, 03:23 AM
Quantum - looks about right now - good call! Thanks for helping - +rep.

revoltage
09-18-2008, 04:08 AM
Have just got the components!!!

I was so eager but now I'm a bit disconcerted due to my lack of practical knowledge. Don't know where to start...going to print the manuals and read them thoroughly. And don't mention the programming, I am a bit apprehensive about how to do that...have not yet started learning mikrobasic...my free time is now being spent typing the conceptual design part...foufff I'm always complaining

I have already assembled the parallax motor kit. Need to get a lead acid battery to test it.

I have had a close look at the maxsonar...is it possible to solder wires directly on it?

As for the sabertooth can it handle the included encoders in the parallax kit?

Adrenalynn
09-18-2008, 05:22 AM
Sure you can solder to it. Try to limit yourself to 350deg C or less for 5 seconds or less. Let it cool a bit betwixt each solder joint.

The sabertooth doesn't know squat about any motor encoder. That's the job of your microprocessor...

revoltage
09-18-2008, 06:57 AM
Thanks Jodie,

How about a 12V drill motor for cutting the lawn?

darkback2
09-18-2008, 10:19 AM
I doubt it would be powerful enough...especially with wet grass. The mower motor is probably going to be your biggest power draw...though maybe I'm wrong, I've never really measured the amount of torque necessary to cut grass.

revoltage
09-18-2008, 11:30 AM
The Robocut robot uses a NiMH 4.1 Ah hour battery to power the two drive motors, a dc motor (i don't know which one) and the sensors. And from what I read from the website is that the robot can mow the grass for more than 2 hrs. I'm quite perplexed how he does this:

http://www.robotshop.se/micro/robocut_us.html

Adrenalynn
09-18-2008, 11:56 AM
Again - this is your thesis/final for your four year degree in mechtronics, right?

When you go get a job with that degree, are you going to ask free-forum dwellers to do all the design work you're getting paid for?

We're all here and happy to help, but I just keep getting this niggling feeling that we're "cheating on your homework" for you...

Quantum
09-18-2008, 04:14 PM
I dont know if the drill will work. It probably would. But why not just get a weed whacker. Its probably easier to incorporate into the project. Its a motor with a whip on it made to cut grass on a aluminum pipe. Ditch the pipe and build you robot around it. It probably runs of 12v and you have the battery in the hand held part of it so you get your motor and batterys. And a already made motor controller. There got to some sort of electronics. And there is also a charger in there for the battery as well. Even thou you use a wall adapter the charger is in the the weed whacker.

So you actually could scavange the motor, batteries, some sort of battery charger, and some sort of motor controller most likly on and off. You would need to build the sfty youself since it usually a s imple trigger which you would use.

My local Sears hardware/parts place has a section of returned things that they cant sell as new. Or repeairs that people left behind and they repair and sell them. Check it out the closer you get to winter the better the deals.

darkback2
09-18-2008, 06:20 PM
HMMMM hybrid!

Why not power the whole thing using a gas weed wacker engine. Then you could make a transmition with a power take off for the mower blade. THe problem with a weed wacker though is that the blade gets chewed up as you mow.

DB

darkback2
09-18-2008, 07:05 PM
lawnmower (http://forums.trossenrobotics.com/showthread.php?t=2417) Thank Matt for this one.

DB

revoltage
09-19-2008, 09:02 AM
Again - this is your thesis/final for your four year degree in mechtronics, right?

When you go get a job with that degree, are you going to ask free-forum dwellers to do all the design work you're getting paid for?

We're all here and happy to help, but I just keep getting this niggling feeling that we're "cheating on your homework" for you...Yep that's complete truth....but I'm trying to improve. Thanks for the advice.

For the maxsonar sensors I'm going to solder pin headers to it for cleaner/easier connections.

For the mowing motors I'll try with the dc first and see the results. Implementing a weed whacker would add too much weight to the robot

revoltage
09-22-2008, 09:34 AM
Anyone knows the underlying electronics principle of the Sabertooth 2x25 driver? Surely it does not use MOSFET? Does it use SCRs?

revoltage
09-23-2008, 10:32 PM
Hmm I think I got the answer: it's a dual h bridge using mosfet

Adrenalynn
09-23-2008, 10:47 PM
The R/C 10A Sabertooth is specifically utilizing the TPC-8017H: http://www.datasheetcatalog.com/datasheets_pdf/T/P/C/8/TPC8018-H.shtml

And it has an ATMEGA48V for its processor.

You *really* don't want to have to remove and solder one of those MOSFETs on that board. Reworking that package is NOT fun... ;)

revoltage
09-23-2008, 11:07 PM
That's something. I didn't know the driver had a processor.

Adrenalynn
09-23-2008, 11:10 PM
That's how it handles that PWM (and/or Serial, depending upon the model).

revoltage
09-24-2008, 01:03 PM
Here's a thought:

For the mower motor I buy a 12V rechargeable drilling machine, hack it and used the circuit and battery only for the cutting system only. Then I used a 12V 7Ah battery for the control system and the drive system. I use an optocoupler between the microcontroller and the mower motor so that it does not fry. Ok??

4mem8
09-24-2008, 01:43 PM
Just out of curiosity revoltage what dia is the cutting blade going to be, I would assume it to be in the 300mm range, if so this will cause a problem with one of those drill motors, They are not designed to turn a blade of that size, also the weight of the blade will play havoc on the bearings in that type of motor, Think of the shaft size on the motor, Are you running it through a gearing system? to a heavier shaft?

revoltage
09-24-2008, 09:09 PM
Yep some 30 cm. I was thinking about using the complete drill system; just make a blade and connect it where drills are usually placed??

revoltage
09-28-2008, 02:56 AM
I'm going to get a battery tomorrow; lead acid one. From the Sabertooth manual they used a potentiometer for a first test; will a 10K do???

Also there's some problem people complain about the ground on the Sabertooth; what is it exactly??

I'm sorry I'm pestering you with questions but all I'm trying to do is not mess my components up. I want to be sure I can carry on with them till March 2009 because I don't have any other money to buy blown up stuffs

revoltage
09-30-2008, 09:20 AM
I have just bought a lead acid 12V 7ah

Here are the specs printed on the battery:

Standby use: 13.5 - 13.8 V
Cycle use: 14.4 -15.0 V
Initial current less than 2.1A

Will it be able to give 25 A whenever the motor stall??

Adrenalynn
09-30-2008, 11:28 AM
Do you see what it says right above those numbers?

Constant Voltage Charge?

It's the "Charge" part that is important there.

In use, that battery will deliver 300+A into a stalled motor, quickly reducing the windings in the motor to smoke.

revoltage
09-30-2008, 11:36 AM
Ah ok, it's a technician at my uni who told me this, that's why i was confused. Thanks :)

revoltage
09-30-2008, 10:21 PM
Do you know batteries that can be recharged with solar cells? I'm thinking of the possibility of implementing solar panels to the mower

Adrenalynn
09-30-2008, 10:23 PM
Any battery can be recharged with a solar cell, given enough square footage and enough days, and an adequate solar charger.

revoltage
09-30-2008, 10:41 PM
Thanks again it was at uni that I was told that I need to look for other battery sources that can adapt to solar, i.e. a "special battery for solar charging". Now you see that's exactly why I need this forum to try my best. I get straight great answers here. We are not even among the best 2000 universities in the world after all. Pfff

Adrenalynn
09-30-2008, 11:15 PM
Surely there's got to be someone there not giving you crummy advice?

Here: http://www.batterymart.com/p-12v-15w-batteryminder-solar-battery-charger.html electrically, perfectly sized. Your 'bot will need more than a meter of free space on top for it though... Or have a solar charging station. Notice how the specs (14.2vDC charging, 1A output) match up with the 13.5-15v < 2.1A on your battery?

revoltage
09-30-2008, 11:35 PM
And if I told you my lecturer himself gave the idea (a mechanics one). We are very far behind in developing robotics technology. Electronics lecturers are sometimes reluctant to give you advise when they are not supervising your project...so...think I'm going against my uni too much for today lol

Thanks

revoltage
10-01-2008, 02:46 AM
Can I connect the motors directly to the battery to see it rotate?

Adrenalynn
10-01-2008, 03:16 AM
Sure. Make sure you secure the motors before doing so, and make sure the battery is charged-up.

revoltage
10-01-2008, 05:08 AM
In one of the motors( with the whole kit already mounted) I'm hearing a knocking sound?? It's not inside the gear box but just outside it; between the motor shaft and the axle from the kit.

jes1510
10-01-2008, 01:51 PM
...
In use, that battery will deliver 300+A into a stalled motor, quickly reducing the windings in the motor to smoke.

Out of curiosity how did you get that information? I have often wondered the same thing and honestly haven't found that info for the various battery packs that I have.

Quantum
10-01-2008, 03:02 PM
Generally a car battery when shorted out can deliver up to 300 amps. But this will destroy the battery and is extremely dangerous. 300 amps creates alot of heat which will create alot of gas from the battery and will start to warp the battery plates. When you crank your car the battery actually creates that many amps but its for a short amount of time.

Adrenalynn
10-01-2008, 04:24 PM
Jes, I've experienced it first hand. :) And my own curiosity has led to experimenting with a large ammeter, a power resistor and a coffee can full of sand. ;)

Quantum - [note: the battery referenced in this thread is a sealed lead-acid battery, not a little R/C pack] I suspect that adage regarding a car battery is largely outdated as far as the maximum current it will deliver. A small car battery, yes. The definition of "Cold Cranking Amps" that you see printed on a battery means the amount of amperage the battery can deliver instantaneously at 10v. Under severe cold weather conditions, the ability of a battery to deliver full voltage is sorely diminished - hence "Cold Cranking Amps" - what the battery can deliver when it's so cold it won't put out more than 10vDC and you still need to fire-up that big V8.

The battery in my truck can deliver 1275 Cold Cranking Amps allegedly - it's the same battery they put in police cars and ambulances here. I don't have anything large enough to test that, although I've left the bed, spot, and interior lights on for 14+ hours and the battery was still at 13v under no load. I've popped the fuse in a 900A ammeter from a car battery listing 850CCA [tested at full charge and room temp].

I agree that deep cycling them down like that is extremely dangerous (explosion risk as well as the risk of flying slag from whatever shorted it - it's more powerful than an average arc welder), and also not good for the long-term health of the battery. The internal contacts and plates don't want to handle that kind of heat, nor do the electrolytes (boil-off).

The point to my original reply was that the listed current on that battery was max charge current, not max current draw. A 7AH battery won't have any trouble delivering 7A for an hour, and it won't have any trouble delivering 21A for 20mins (other than that you'll be below the voltage that your device will run at and you may sulfate the battery running it down that far).

revoltage
10-04-2008, 06:54 AM
Anyone got a clue how much torque the Denso motors can provide? I have contacted Denso but they said they can't provide me with this info. I have contacted Parallax and again they told me the same. Any approximate value please??

Adrenalynn
10-04-2008, 11:47 AM
Yup. Look at the chart you thanked me for.

revoltage
10-04-2008, 12:37 PM
Yep had the graph is on my pc...hmmm... 33 Nm??

revoltage
10-05-2008, 03:03 AM
Hmm the torque is max 13 Nm I think. Attached is a wheel design for improved traction; will make it with 6 mm polypropylene sheets.

darkback2
10-05-2008, 03:06 AM
Its hard to tell exactly what you have there, I think I would have to see an actual wheel. Is 6mm thick enough to support the robot's weight? Seams like it would if it were steel, but poly?

Db

revoltage
10-08-2008, 12:01 AM
I have tested the motors with the Sabertooth and a potentiometer; it's working great. Next step will be writing program and testing it with pwm...

revoltage
10-08-2008, 01:35 AM
How can I design the bump sensors? I mean I know the relatively simple electric schematics but what about the mechanical design so that the microswitch can return to it NO position after it hits the object? Using springs?

Adrenalynn
10-08-2008, 03:15 AM
Have you found a microswitch that isn't broken that doesn't have spring return?

revoltage
10-08-2008, 05:08 AM
Err will it have enough force to push the aluminium or plastic bumper?

jes1510
10-08-2008, 11:54 AM
You could always put the bumper on springs so that it moves a bit like a Roomba. Then you could use IR emitters and detectors to sense when the bumper is pushed back. Or a heavy pushbutton would work too.

revoltage
10-15-2008, 01:27 AM
Thanks jes1510.

Do I have to send PWM signal to the Sabertooth or directly a 5v output from the PIC? It's the Atmel chip in the driver that's intriguing me?? Why send PWM if the Atmel is there to send pwm to the motor?

Adrenalynn
10-15-2008, 03:38 AM
Depends - is it the "R/C" model or not? The R/C model runs from PWM only. The others are able to take serial as input.

Don't forget that AVR is doing a bunch of other stuff. Mixing, linear-vs-exponential throttle, Lithium voltage sensing, etc.

revoltage
10-15-2008, 03:41 AM
It's the 2X25A Sabertooth I bought from Trossen

Adrenalynn
10-15-2008, 04:34 AM
Consider RTFMing. And before you ask: Read The...errr... Fine... Manual.

revoltage
10-15-2008, 07:18 AM
Yep I had read it, and it said that it can use PWM. It was the Atmel processor that was intriguing me. Anyways thanks :)

Adrenalynn
10-15-2008, 11:57 AM
Sorry - I guess we're reading different manuals.

http://dimensionengineering.com/datasheets/Sabertooth2x25.pdf


Many operating modes:

With analog, R/C and serial input modes, as well as dozens of operating options, the Sabertooth
has the flexibility to be used over and over, even as your projects grow more sophisticated. Yet it
is simple enough to use for your first robot project.

...

Mode 1: Analog Input

Analog input mode takes one or two analog inputs and uses those to set the speed and direction
of the motor. The valid input range is 0v to 5v. This makes the Sabertooth easy control using a
potentiometer, the PWM output of a microcontroller (with an RC filter) or an analog circuit.
Major uses include joystick or foot-pedal controlled vehicles, speed and direction control for
pumps and machines, and analog feedback loops.


[...]


Mode 3: Simplified serial.

Simplified serial mode uses TTL level RS-232 serial data to set the speed and direction of the
motor. This is used to interface the Sabertooth to a PC or microcontroller. If using a PC, a level
converter such as a MAX232 chip must be used. The baud rate is set via DIP switches.
Commands are single-byte. There is also a Slave Select mode which allows the use of multiple
Sabertooth 2x25 from a single microcontroller serial port.

Mode 4: Packetized serial

Packetized serial mode uses TTL level RS-232 serial data to set the speed and direction of the
motor. There is a short packet format consisting of an address byte, a command byte, a data byte
and a 7 bit checksum. Packetized serial automatically detects the transmitted baud rate based on
the first character sent, which must be 170. Address bytes are set via dip switches. Up to 8
Sabertooth motor drivers may be ganged together on a single serial line. This makes packetized
serial the preferred method to interface multiple Sabertooths to a PC or laptop. Because
Sabertooth uses the same protocol as our SyRen single motor drivers, both can use used together
from the same serial master.


Mode 3: Simplified Serial Mode

Simplified serial uses TTL level single-byte serial commands to set the motor speed and
direction. This makes it easy to interface to microcontrollers and PCs, without having to
implement a packet-based communications protocol. Simplified serial is a one-direction only
interface. The transmit line from the host is connected to S1. The host’s receive line is not
connected to the Sabertooth. Because of this, multiple drivers can be connected to the same serial
transmitter. If using a true RS-232 device like a PC’s serial port, it is necessary to use a level
converter to shift the –10V to 10V rs-232 levels to the 0v-5v TTL levels the Sabertooth is
expecting. This is usually done with a Max232 type chip. If using a TTL serial device like a
microcontroller, the TX line of the microcontroller may be connected directly to S1.
Because Sabertooth controls two motors with one 8 byte character, when operating in Simplified
Serial mode, each motor has 7 bits of resolution. Sending a character between 1 and 127 will
control motor 1. 1 is full reverse, 64 is stop and 127 is full forward. Sending a character between
128 and 255 will control motor 2. 128 is full reverse, 192 is stop and 255 is full forward.
Character 0 (hex 0x00) is a special case. Sending this character will shut down both motors.

Baud Rate Selection

Simplified Serial operates with an 8N1 protocol – 8 data bytes, no parity bits and one stop bit.
The baud rate is selected by switches 4 and 5 from the following 4 options
2400 Baud: 01x00x 9600 Baud: 01x10x
19200 Baud:


What baud rate to use is dependent on what your host can provide and the update speed
necessary. 9600 baud or 19200 baud is recommended as the best starting points. If
communication is unreliable, decrease the baud rate. If communications are reliable, you may
increase the baud rate. The maximum update speed on the Sabertooth is approximately 2000
commands per second. Sending characters faster than this will not cause problems, but it will not
increase the responsiveness of the controller either.
The baud rate may be changed with power on by changing the DIP switch settings. There is no
need to reset or cycle power after a baud rate change.
There are 2 operating options for Simplified Serial. These are selected by the position of Switch
6.


Option 1: Standard Simplified Serial
Mode

Serial data is sent to input S1. The baud rate is selected
with switches 4 and 5. Commands are sent as single bytes.
Sending a value of 1-127 will command motor 1 Sending a
value of 128-255 will command motor 2. Sending a value
of 0 will shut down both motors.


Option 2: Simplified Serial with Slave
Select

This mode is used when it is desirable to have multiple
Sabertooth motor drivers running from the same serial
transmitter, but you do not wish to use packetized serial. A
digital signal (0v or 5v) is fed to the S2 input. This is
controlled by the host microcontroller. If the signal on S2 is
logic high (5v) when the serial command is sent, then the driver will change to the new speed. If
the signal on S2 is not high when the command is sent, then command will be ignored. Pseudocode
demonstrating this is shown below. After sending the signal, allow about 50 us before
commanding the Slave Select line to a logic LOW to allow time for processing. A hookup
diagram and example pseudo-code are shown in Figures 6.2 and 6.3.


Mode 4: Packetized Serial Mode

Packetized Serial uses TTL level multi-byte serial commands to set the motor speed and
direction. Packetized serial is a one-direction only interface. The transmit line from the host is
connected to S1. The host’s receive line is not connected to the Sabertooth. Because of this,
multiple Sabertooth 2x25 motor drivers can be connected to the same serial transmitter. It is also
possible to use SyRen and Sabertooth motor drivers together from the same serial source, as well
as any other serial device, as long as it will not act on the packets sent to the Sabertooth. If using
a true RS-232 device like a PC’s serial port, it is necessary to use a level converter to shift the –
10V to 10V rs-232 levels to the 0v-5v TTL. Packetized serial uses an address byte to select the
target device. The baud rate is selected automatically by sending the bauding character (170 in
decimal, AA in hex) before any commands are sent.

Packet Overview

The packet format for the Sabertooth consists of an address byte, a command byte, a data byte
and a seven bit checksum. Address bytes have value greater than 128, and all subsequent bytes
have values 127 or lower. This allows multiple types of devices to share the same serial line.
An example packet and pseudo-code to generate it are shown in Figures 7.1 and 7.2

Packet
Address: 130
Command : 0
Data: 64
Checksum: 66
Void DriveForward(char address, char speed)
{
Putc(address);
Putc(0);
Putc(speed);
Putc((address + 0 + speed) & 0b01111111);
}
Figure 7.1: Example 50&#37; forward Figure 7.2: Pseudocode to generate 7.1

Baud Rate Selection:

Packetized Serial operates with an 8N1 protocol – 8 data bytes, no parity bits and one stop bit.
The baud rate is automatically calculated by the first character sent. This character must be (170
in decimal) (binary 10101010) and must be sent before any serial communications are done. It is
not possible to change the baud rate once the bauding character has been sent. The valid baud
rates are 2400, 9600, 19200 and 38400 baud. Until the bauding character is sent, the driver will
accept no commands and the green status1 light will stay lit. Please note that Sabertooth may
take up to a second to start up after power is applied, depending on the power source being used.
Sending the bauding character during this time period may cause undesirable results. When
using Packetized Serial mode, please allow a two-second delay between applying power and
sending the bauding character to the drivers.

Address Byte Configuration:

Address bytes are set by switches 4, 5 and 6. Addresses start at 128 and go to 135. The switch
settings for the addresses are shown in the chart below


Commands:

The command byte is the second byte of the packet. There are four possible commands in
packetized serial mode. Each is followed by one byte of data

0: Drive forward motor 1 (decimal 0, binary 0b00000000, hex 0h00)

This is used to command motor 1 to drive forward. Valid data is 0-127 for off to full forward
drive. If a command of 0 is given, the Sabertooth will go into power save mode for motor 1 after
approximately 4 seconds.

1: Drive backwards motor 1 (decimal 1, binary 0b00000001, hex 0h01)

This is used to command motor 1 to drive backwards. Valid data is 0-127 for off to full reverse
drive. If a command of 0 is given, Sabertooth will go into power save mode for motor 1 after
approximately 4 seconds.

2: Min voltage (decimal 2, binary 0b00000010, hex 0h02)

This is used to set a custom minimum voltage for the battery feeding the Sabertooth. If the
battery voltage drops below this value, the output will shut down. This value is cleared at startup,
so much be set each run. The value is sent in .2 volt increments with a command of zero
corresponding to 6v, which is the minimum. Valid data is from 0 to 120. The function for
converting volts to command data is
Value = (desired volts-6) x 5

3: Max voltage (decimal 3, binary 0b0000011, hex 0h03)

This is used to set a custom maximum voltage. If you are using a power supply that cannot sink
current such as an ATX supply, the input voltage will rise when the driver is regenerating
(slowing down the motor) Many ATX type supplies will shut down if the output voltage on the
12v supply rises beyond 16v. If the driver detects an input voltage above the set limit, it will put
the motor into a hard brake until the voltage drops below the set point again. This is inefficient,
because the energy is heating the motor instead of recharging a battery, but may be necessary.
The driver comes preset for a maximum voltage of 30V. The range for a custom maximum
voltage is 0v-25v. The formula for setting a custom maximum voltage is
Value = Desired Volts*5.12
If you are using any sort of battery, then this is not a problem and the max voltage should be left
at the startup default.

4: Drive forward motor 2 (decimal 4, binary 0b00000100, hex 0h04)

This is used to command motor 2 to drive forward. Valid data is 0-127 for off to full forward
drive. If a command of 0 is given, the Sabertooth will go into power save mode for motor 2 after
approximately 4 seconds.

5: Drive backwards motor 2 (decimal 5, binary 0b00000101, hex 0h05)

This is used to command motor 2 to drive backwards. Valid data is 0-127 for off to full reverse
drive. If a command of 0 is given, the Sabertooth will go into power save mode after
approximately 4 seconds.

6: Drive motor 1 7 bit (decimal 6, binary 0b00000110, hex 0h06)

This command is used to drive motor 1. Instead of the standard commands 0 and 1, this one
command can be used to drive motor 1 forward or in reverse, at a cost of lower resolution. A
command of 0 will correspond to full reverse, and a command of 127 will command the motor to
drive full forward. A command of 64 will stop the motor.

7: Drive motor 2 7 bit (decimal 7, binary 0b00000111, hex 0h07)

This command is used to drive motor 2. Instead of the standard commands 4 and 5, this one
command can be used to drive motor 1 forward or in reverse, at a cost of lower resolution. A
command of 0 will correspond to full reverse, and a command of 127 will command the motor to
drive full forward. A command of 64 will stop the motor.

Mixed mode commands:

Sabertooth can also be sent mixed drive and turn commands. When using the mixed mode
commands, please note that the Sabertooth requires valid data for both drive and turn before it
will begin to operate. Once data for both has been sent, then each may be updated as needed, it is
not necessary to send both data packets each time you with to update the speed or direction. You
should design your code to either use the independent or the mixed commands. Switching
between the command sets will cause the vehicle to stop until new data is sent for both motors.

8: Drive forward mixed mode (decimal 8, binary 0b00001000, hex 0h08)

This is used to command the vehicle to drive forward in mixed mode. Valid data is 0-127 for off
to full forward drive.

9: Drive backwards mixed mode (decimal 9, binary 0b00001001, hex 0h09)

This is used to command the vehicle to drive backwards in mixed mode. Valid data is 0-127 for
off to full reverse drive.

10: Turn right mixed mode (decimal 10, binary 0b00001010, hex 0h0a)

This is used to command the vehicle to turn right in mixed mode. Valid data is 0-127 for zero to
maximum turning speed.

11: Drive turn left mixed mode (decimal 11, binary 0b00001011, hex 0h0b)

This is used to command the vehicle to turn leftt in mixed mode. Valid data is 0-127 for zero to
maximum turning speed.

12: Drive forwards/back 7 bit (decimal 12, binary 0b00001100, hex 0h0c)

This is used to command the vehicle to move forwards or backwards. A command of 0 will
cause maximum reverse, 64 will cause the vehicle to stop, and 127 will command full forward.

13: Turn 7 bit (decimal 13, binary 0b00001101, hex 0h0d)

This is used to command the vehicle turn right or left. A command of 0 will cause maximum left
turn rate, 64 will cause the vehicle to stop turning , and 127 will command maximum right turn
rate.

Checksum:

To prevent data corruption, each packet is terminated with a checksum. If the checksum is not
correct, the data packet will not be acted upon. The checksum is calculated as follows:
Checksum = address byte +command byte +data byte
The checksum should be added with all unsigned 8 bit integers, and then ANDed with the mask
0b01111111 in an 8 bit system.

Example of Packetized Serial:

The following is an example function for commanding two Dimension Engineering motor
drivers using Packetized Serial Mode. Figure 7.3 shows an example hookup and Figure 7.4
shows an example function.


Emergency Stop:

In Packetized Serial mode, the S2 input is configured as an active-low emergency stop. It is
pulled high internally, so if this feature isn’t needed, it can be ignored. If an emergency stop is
desired, all the S2 inputs can be tied together. Pulling the S2 input low will cause the driver to
shut down. This should be tied to an emergency stop button if used in a device that could
endanger humans.

4mem8
10-15-2008, 12:26 PM
Thanks also for this info Adrenalynn, as I have one of these.

Adrenalynn
10-15-2008, 12:40 PM
No problems. Really, though, you should reference the manual (http://dimensionengineering.com/datasheets/Sabertooth2x25.pdf), since it has pretty diagrams and dip-switch positions and stuff. That's just cut-and-paste from that manual.

revoltage
10-15-2008, 10:38 PM
I'm sorry. I wrongly interpreted the manual. Thanks

revoltage
10-18-2008, 01:22 AM
Here a 2D view of the designed wheel. I'm planning to make it with polypropylene. Is the material choice proper or will it still slip and not provide a good traction?

revoltage
10-19-2008, 01:16 AM
Hi,

Anybody know an online store where I can get these components:OA95,BD137?

Adrenalynn
10-19-2008, 03:38 AM
Why not design your own circuit using commonly available components?

Have you tried cross-referencing them?

revoltage
10-19-2008, 10:02 AM
I have got a schematic for the perimeter wire far easier that that used by robocut and they used these components. After some research I found out that RS components sold most of them except the OA95 diode. For some reasons I am not finding its equivalent.

revoltage
10-21-2008, 10:46 AM
Could someone please confirm that the maximum torque of the Denso motors is 13Nm?????

Adrenalynn
10-21-2008, 11:39 AM
Nope. There is no theoretical "maximum torque" on an electric motor. The more voltage you put in with more amperage pushing it, the more torque you get. Right up until the motor fries itself. That's why my graphs are lines and curves and not a point.

darkback2
10-21-2008, 01:06 PM
Here a 2D view of the designed wheel. I'm planning to make it with polypropylene. Is the material choice proper or will it still slip and not provide a good traction?

How thick is your poly prop, and how heavy is this robot? Seams like you may be out snapping wheels if your not carefull. Maybe you could use the wheels that came with the denso kit, and mount the poly prop gear dealies on the side of the wheels. That way most of the weight is beared by the wheel, and the gear can just deal with the whole traction bit..

DB

Adrenalynn
10-21-2008, 01:49 PM
That's why I like bicycle chainrings. Conveinient mounting holes to drill right into the original wheels, and it's for sure they're going to carry the weight. (And they're free/cheap at yard sales, in dumpsters, flea markets, public auctions, ...)

revoltage
10-22-2008, 03:06 AM
The wheels should be 34 cm in diameter and the one from parallax is only 15cm. The poly is 6 mm thick

Adrenalynn
10-22-2008, 03:17 AM
Right, but what he's saying is to using the Parallax wheels as "hubs" for your 34cm wheels.

revoltage
10-22-2008, 06:10 AM
Hmm could try that thanks to both of you :)

revoltage
10-23-2008, 11:47 PM
I need the robot to move at a speed of 8-17 rpm. Is it possible to do this with the denso motors which has a no load speed of 150 rpm via pwm without losing efficiency?