View Full Version : [Project] Meet S.A.M.i

06-05-2008, 06:46 PM
My first Bot! Nothing like starting big! As I have mentioned in the past I am building an autonomous mower, well two autonomous mowers to be exact. Here is the beginning of the big project S.A.M.i which is the zero-turn bot. (S.A.M.i stands for Smart Autonomous Mower by the way)
Life started for S.A.M.i as a cart pusher (it will take two to make one mower).

After stripping these down, I think I have some good parts to work with.
They came with motor controllers, built in RC units which will need to be modified or replaced, and 36v chargers. They had lead acid batteries in them, but they are mostly toast. The axles are limited slip differential just like on the family car. The motor is a 36v 2.5A motor that is direct into the axle.
My plan is to mount two motor/axles side by side on the original chassis which will allow for the zero turn skid style steer. Which is now done.



Ok, it looks kind of funny now, but hopefully as the project progresses it will look less goofy.
I have to attach “anti-spin” plates to the inboard hub of each axle to eliminate the "slip" from the limited slip axle. That will be the next step. Then I can “re-hub” the outside hubs for larger tires.


After that it is on to building the rails for hanging the mowing deck – complete with a gas vertical shaft engine that will spin two blades. After that it is just a matter of working out the bugs with the RC and learning a lot about programming.

Hopefully in the end it will look like an actual zero turn mower – minus the seat of course. Then I will be able to add the "brains" and sensors to get S.A.M.i "thinking" for him/her self.

Since I started the project, I have found out that it may be July before I get any quality time back in the workshop. I will keep updating though as progress is made.

06-05-2008, 07:17 PM
That's awesome!

Where does the 2.5A come from? Is that really stalled, or average operating, or ...? It seems like a really low number for something designed to move that much weight that quickly.

I have some 36v, .75HP electric scooter motors that draw 60A nominal, 180A stalled...

06-05-2008, 07:23 PM
I will look again, but I am pretty sure that is what is stated on the motor. They wouldn't lie would they:veryhappy:
I had the same thought, but didn't do the math.

06-05-2008, 10:42 PM
How does one acquire cart pushers! That does seem like something useful.

And big bots are cool!

06-06-2008, 02:47 AM
It could be no-load current too.

Sienna, I see 'em out there at auction. It's the freight that kills ya!

Droid Works
06-06-2008, 04:26 AM
Very resourceful, Awesome!

06-06-2008, 12:59 PM
Thank you for the kind words.

Update the sticker says 3.5a. Once hooked up I can determine for myself.

I haven’t chosen a “brain” or an operating system yet. Everybody feel free to persuade me to use their favorite!

06-06-2008, 01:03 PM
I'll be strung-up for my recommendations...

PicoITX running Windows XP. There. I said it.

If you were an experienced software engineer, I'd say "PicoITX running Linux". But dang this C# stuff with aftermarket libraries is pretty darned awesome.

06-06-2008, 01:20 PM
I'll be strung-up for my recommendations...
Not at all! Even though I prefer going with embedded processors as required for given tasks, I don't have any problem using a PC style motherboard where it would be appropriate. :happy:

PicoITX running Windows XP. There. I said it.
Ack, -cough-, Phhht, -choke- :P

If you were an experienced software engineer, I'd say "PicoITX running Linux". But dang this C# stuff with aftermarket libraries is pretty darned awesome.
I'm not an experienced software engineer, and I run and use Linux. :veryhappy: All this talk about C#.. Mono and Portable.Net might be worth looking into.. Hmmmm..


06-06-2008, 01:23 PM
Downside is loading the aftermarket libraries just isn't likely to happen in many cases. I haven't seen anyone get Roborealm working on Linux yet?

06-06-2008, 02:08 PM
So, something like this (http://www.logicsupply.com/products/artigo) is a good idea?

(Link courtesy of Droid Works from another thread)

I have to admit that the XP sounds good to get started. Just how steep is the Linux learning curve?

06-06-2008, 02:30 PM
Yup. I've recommended LogicSupply to a few people now, and so far all have been more than happy with 'em.

It's not the linux learning curve that's high (imho), it's the tools learning curve. Windows has much more drag-and-drop and simpler IDE type tools. As well as things like RoboRealm (roborealm.com) that are ready to go out-of-the-box.

06-06-2008, 02:33 PM
Top of my list so far!

06-10-2008, 06:22 PM
I found some time over the weekend to get a little done on S.A.M.i
Here is what he looks like now
Still a long way to go!

06-10-2008, 07:07 PM
That looks tremendous! A lot of potential there. Is the background the yard that you intend to mow? If so you need to deal with a steep grade and a lot of interesting obstacles.

Have you looked at commercial robot lawnmovers? We have one by Friendly Robotics (http://www.friendlyrobotics.com/) that we bought years ago (long enough that our model is no longer featured on their front page). I case you haven't, let me just describe how it works to give you some ideas.

There is wire that defines the perimeter of the lawn area which carries a signal that the robot can sense. We just staked it down one afternoon and over a few months the lawn grew over it. Because I've had to dig it up a few times to repair breaks I know that it has migrated naturally right down to the base of the sod. It has an electronic compass for orientation and a circle of bump switches completes its sensors.

On activation it mows around the perimeter by following the wire. Once it detects that it has turned a complete circle it starts mowing the interior. It does this by zigzagging from one wire to another, going backwards and then repeating at a different angles for a fixed amount of time. Here's their movie (http://www.friendlyrobotics.com/video/robomow-mowing_method.mpeg).

We love the stupid thing. Even when it gets stuck and calls for help, the robot lifestyle is still better than doing things youself.

06-10-2008, 07:40 PM
depending on speed, we had one of those invisible fence collars for our cats. They mostly just ran through it though.


06-10-2008, 09:23 PM
Thank you for the kind words.
Most of what is visible in the background is the neighbors back pasture. But I have portions of my yard that are as steep. I don’t expect S.A.M.i to handle the steep grades, that will be the next one. But I have lots of yard (a few acres) that isn’t steep. None of it is flat however, and I understand the commercial mowers like the Robomow doesn’t handle the not flat yards well.

06-12-2008, 12:36 PM
Will you have adjustable blade heights and terrain sensing? (I've always wanted to work on topo sensing...)

The wire perimeter is cool, but how about a laser fence instead? No trenching to worry about.

That's an awesome start, Ooops!

06-12-2008, 04:53 PM
Will you have adjustable blade heights and terrain sensing? (I've always wanted to work on topo sensing...)The wire perimeter is cool, but how about a laser fence instead? No trenching to worry about.

Deck height will need to be adjustable. My plan at this point is to attach it so that it can first be manually adjusted and later adjusted “on the fly” . I haven’t considered what will be he best way to adjust it “on the fly” although I have a few thoughts. The trick will be to do the initial installation so that it can be easily adapted later.

The wire perimeter would seem mostly fool proof, however as I mentioned before, lots of yard to mow, so lots of yard = lots of wire= lots of wire burying. Also, it wouldn’t be easy to use on a friends yard, or to demo/test it on a football field or other large field in need of mowing. I gather the laser fence would be easier on the back, but slow setup in new areas?

So, without enough knowledge or experience to draw any firm conclusions my first thought would be to start with a compass and GPS, and work in a laser beacon … I envision something like a laser level … rotating laser on a tripod. But with the topography being so (for the lack of a better word) hilly, I am not sure how that would work without having more than one, or moving the one, once a particular area is done. Which wouldn’t work well for me having to drag it around.
I also had a rather abstract thought on the spectrometry (wow is that really a word? and is it the right one?) of the sap from the cut grass. I am not sure if the cut grass would reflect light differently, or for how long, but you would think that it is possible that it might and that a sensor could be utilized for telling where the grass has been cut and where it hasn’t. That would sure be an improvement to efficiency.
Once again, any and all suggestions are appreciated!

06-12-2008, 07:19 PM
One idea I had been toying with was vision, although I don't know how well that would work with your hardware configuration. A camera hanging over the front of the mower and pointed straight down would be able to see what's ahead as well as to the left and right. Since the ground is parallel to the camera and there is limited parallax it might not be hard to create an algorithm that could distinguish grass from everything else. Edging would follow the edge of the not-grass area, and mowing the interior could go back and forth across the grass area. You could also track your absolute position for mapping purposes by computing the motion vector by auto-correlation, like an optical mouse.

06-12-2008, 08:57 PM
In my imagination, I see a laser on a stake/pole at the border, and a mirror at each virtual fence corner, also on a stake. The laser is pointed so that it hits and bounces from each mirror in series. You could create non-regular-polygons that way too.

08-04-2008, 09:43 AM
S.A.M.i update:
Well, after weeks and weeks of looking at the "pile" of parts I finally found a few hours to get some work done.
I had left off needing to eliminate the limited slip differentials of the two axles which will provide for tank style steering. Then it would be time to test drive.
I had ordered a big 36v 80amp motor controller and was itching to see S.A.M.i run around the yard on RC. So, after a few more hours than expected (somehow it always takes longer than you think it should originally) the "slip eliminators" were installed and nothing left to do but wire it up and go for a spin. Unfortunately by now it is after dark Saturday night and I will have to postpone the test run until Sunday.
First thing Sunday I wired up all the connections, tested voltage (38v) and plugged the power to the board ... zzzt-zzzt then blue smoke:(
I suspect that the board had a problem right out of the box, I double checked, triple checked, and then checked again that everything was setup correctly. On about the third pass through the "instructions" on the board I noticed that there was a note for voltages above 36v and to move the J7 jumper to the HV pins. ... "ooops!!!"
I don't think that the two volts over the limit should have fried the board. Will wait and see what the board Mfg says. If two volts are that critical then there should be a very large sticker on the controller itself, and the jumper should be in a baggy and not installed from the Mfg.
So, I will wait patiently for another controller and start thinking about how to install the mower deck.

08-04-2008, 12:35 PM
Wow! Letting the magic smoke escape - did you pick your handle or did it pick you? :P
Prototype working, Ooops. I could get you running on the *real* controllers before you could get the repair job back. Incidentally, the real controllers are 160A/channel, and the motors are drawing over a hundred before they reach stall...

08-04-2008, 01:31 PM
I am all about the real controllers for obvious reasons!!!
Sign me up:)
Thank you, thank you, thank you:)

I just figured since I had that one I would test drive it.
In retrospect, it is probably a blessing that it didn't work. I had SAMi on the ground and would have been running him without any type of fail safe. I think next time I will jack him up off the ground and get real comfortable with the RC range before turning him loose. I would guess that with the twin axles and steel construction he weighs 325+Lbs. That would leave a mark if it hit something. And or worse, if it decided to go down the hill, I would have to drag or trailer him back up.

08-04-2008, 01:46 PM
I'm in exactly the same boat. I put OSCAR up on jackstands for testing, if you have a better idea, I'm all ears, otherwise the cheap "SUV" jackstands and a creeper work pretty well for working on him. To mount that heavy axle and drivetrain up I used a transmission jack and smaller jackstands. These 300+lb bots start feeling like you're working on heavy equipment, don't they? :)

For those not in the know, we're talking about parts from surplus shopping-cart pushers. The motor controllers are Curtis 160A single-channel units. They're *very* different from our hobby stuff, but have some remarkably cool features and are made for serious industrial abuse. Just the filter coils are the size around of my wrist and weigh in near a pound and you could house a dozen 45A Banebots motor controllers in the filter's housing alone. The external heat-sinking aluminum is a half inch thick. These are big-bad-maama-jaamas. Took a little reverse-engineering work to figure them out, but they are the sweetest motor controllers I've ever even seen. You know you're dealing with industrial when the motor controllers have outputs for electromagnetic brakes, remote panic braking, key switch inhibitors, manual push speed limiters ... If you're building a really big 'bot, you need really bad-a' hardware. ;)

08-15-2008, 10:05 AM
SAMi update ... Motor controller back and repaired. Thank you motor controller guys!!!
I will be adding a switch and hopefully test running the motors over the weekend. On jack stands this time! Fingers crossed:)

09-12-2008, 10:07 PM
any updates ooops?

09-12-2008, 11:41 PM
I think he's more focused on the TRAMP right now... :)

09-15-2008, 08:04 AM
any updates ooops?

Librab, thank you for asking:) Yes, at the moment I am distracted by TRAMP, along with the 6,832 things on my "to do" list. Hopefully, TRAMP will come to completion soon, and I can get back to SAMI and my Bioloid. I do have the new/repaired motor controller, and will slide it in when the opportunity arises.

12-30-2008, 10:30 PM
At long last a little progress:

YouTube - samy rolls

This was run using RC, which will be the main form of control during the "build phase". I still aspire to autonomous control, but will be patient in implementing.

The RS80D motor controller isn't working on the higher voltage (36v+), so this is run at 24v which at the end of the day isn't all bad, maybe a blessing in disguise. You take a 250+++ lb robot and run it a full power during the initial R&D and a small "oops" could easily turn into a call to the insurance adjuster, or worse.
***Thank you to the guys at R-S, they got the RS80D working again after I had let the magic smoke out previously!***

The "limited slip eliminators" worked quite well. I think it is ready for bigger tires and wheels. The new hubs are "chopped" and ready to go on. Looks like a trip to the wheel store is on the horizon:)

At some point I will have to get some closer pictures of the modifications so far, but I wanted to see if they worked before documenting. Not to mention they are not clean or painted ... don't want to post ugly close-up pictures.

Soooo, is there a contest for the BIGGEST bot?

12-30-2008, 10:35 PM
I'd say that was pretty darned successful for a first test of a 'bot approaching 300lbs! I have one of those motor/gearboxes (thanks!) and just one of those things is near 85 back-breaking pounds!

12-30-2008, 10:48 PM
Thank you:)
You notice I stuck to the driveway area for the current test, my fear is having to push that thing uphill if something fails:)

12-30-2008, 11:10 PM
Isn't that why you had kids?

12-31-2008, 05:28 AM
haha very funny addy.

all in all, looks like an amazing project

12-31-2008, 04:50 PM
Awesome ooops, your work is coming on great, What size tires will you eventually fit?

12-31-2008, 05:15 PM
Nice project and it's coming along! :wink:

HAve a happy year!

01-02-2009, 12:53 AM
Awesome ooops, your work is coming on great, What size tires will you eventually fit?

Not sure ... I will be tire shopping tomorrow:)

01-03-2009, 09:02 PM
Well, I got the biggest knobby tires that I thought would fit S.A.M.i
Unfortunately, since I had to swap hubs, I didn't have an exact measurement for the wheel offset and was left to guess at what would most likely fit. Let's just say I was close but no cigar:( I did go for a tire/wheel combo, and may be able to keep the tires (22x11.00x8) and swap the wheels for some with a much shorter offset?
Just for comparison - what is on there now is a 4.10/3.50x6 - for those that don't "speak tire" (aka tyre) twice as tall, 3 times as wide
Since Northern tool is about an hour away, I can only hope to get back there soon???
Needless to say I am bumming!!!

01-06-2009, 12:47 AM
Or swap out the hubs' lugs for longer units, and throw in a spacer. Unless I read that backwards.

01-06-2009, 08:05 AM
Or swap out the hubs' lugs for longer units, and throw in a spacer. Unless I read that backwards.

I very seriously considered this, but opted to go for a bit smaller tire - 20x7x8 as pictured.
still this is almost twice as tall and twice as wide. So, not a bad compromise, now I am just waiting on the brown truck to show up.

01-09-2009, 09:54 AM
Finally I had the chance to install the new tires! Wow, it looks better and it is a lot faster now. Pretty much doubled the gear ratio to the ground so even at the 24v it is fast. I am thinking with the addition of the mowing deck, and gas engine that it might slow down a bit. Hopefully given the chance this weekend I can get some new pictures and video.

01-11-2009, 07:46 PM
Too cold to shoot video outside this weekend. So here are some stills to show off the ground clearance.
Also, there is a closer view of the anti- anti-slip device.
One step closer:)

01-11-2009, 09:58 PM
Those are more like boom-lift tires! ROFL!

You need to put some Rancho's on there and jack that puppy up! It needs a gun-rack! Bwahahahahaha!

01-12-2009, 03:42 PM
WOW!!!! Do u plan on cutting grass in the snow haha. Can't wait for the video.

01-12-2009, 05:19 PM
Now you're cooking with gas! She's shaping up nicely. Is the mowing deck going to use a gas motor?

01-13-2009, 10:26 AM
Thanks for the votes of confidence.

As for the ranchos, maybe after I get the big round fog lights:)lol

Plan on using an electric start gas. No reason to pull a cord on such a highly technological piece of machinery:)
As soon as I get a break in weather, and can get outside to weld, I hope to get the mower deck going, then can shift to the engine. All that following some test runs in it's current condition.

01-13-2009, 01:23 PM
Looks like this is shaping up to a pretty fun project. I had a thought concerning automated deck height for your mower. My thought was a lightweight roller that was free floating on rods. Placed in front of your mower deck it would move vertically based on any changes in terrain. If the roller moves beyond the tolerances you set, the deck height could be adjusted. The "roller" could be as simple as some 3" diameter PVC. Attach rail from a linear actuator(sp?) to your vertical rod and set a gear on the stationary piece and you'll get a consistant reading of the height based from gear revolutions. Hope this helps.

01-15-2009, 09:24 AM
Scumm, welcome to the forum.
I like your thoughts:) I had originally thought of a proximity switch to control height, and then IR and sonar, or a mix of all but they all have such a specific area that they "look at" that I am unconvinced that even with a group of strategically placed sensors and software averaging, that a mechanical (dumb system) wouldn't be a better alternative.
Thanks for the suggestion, it will get serious consideration since the deck is the next step.
BTW: weight isn't really an issue:)

01-16-2009, 10:39 AM
My bud tells me that a lot of lawnmowers have a floating deck. This could be a simple solution as well.