PDA

View Full Version : Minimum Energy?



Sienna
07-01-2008, 06:16 PM
Ok, as the converse to the safety limit thread, there needs to be a minimum energy thread.

We are talking about some form of mechanically activated switch I assume for the 'targets'. (I believe right?)

We need to establish a minimum energy needed to hit anywhere on the target to get it to activate. Define it too low, and you will get a lot of false positives from the simple act of walking (which is a jolting experience for rigidly connected assemblies!) Define it too high, and you will not 'feel' all the little tiny tank guns.

I don't remember if this has been answered, but who is designing the targets? Are they going to be another 'range' asset that we have to buy, or are we all going to have to do our own?

JonHylands
07-01-2008, 06:28 PM
Right now, we're planning on having builders do their own targets. We will provide examples, and the specifications that the targets have to meet.

DresnerRobotics
07-01-2008, 06:34 PM
Yup, all of this is still somewhat up in the air, we need to build some prototypes of the scoring/hit circuits and then experiment with how much force is needed to trigger various types of microswitches. We'll be posting our results and final decisions in the months to come.

This is one of the lesser things people should worry about for now, I'd say most plates will be full implementing a solid control scheme and getting their mechs walking.

MSpurk
07-15-2008, 09:23 PM
Hey Tiberious, on a related note has the dimensions of the target been decided yet? I need to make certain my quad is large enough to accomodate 4 targets. Can you at least provide a ballpark?

Thanks,
-Matt-

DresnerRobotics
07-15-2008, 10:00 PM
3"x3" if I had to guess, but again its a guess

ScuD
07-16-2008, 02:05 AM
Actually I had an idea about the targets last night, though i'm not sure if it's feasible.

We'd want to register a hit throughout a plate. Say you take a plate, and attach it to a microswitch.
That requires some amount of force, but also you have to take into account that there can be some binding, when the BB hits the edge of the plate at the side of the microswitch lever, for example.

Now, say you take your armour plate, and attach (glue, or whatever) a cheap-ass piezo speaker to it (you know, the little brass plate ones?)

Now, when something hits the plate, a voltage spike is generated.
Pick up this spike with an opamp, filter/amplify and set a threshold so you can adjust sensitivity.

The entire circuit wouldn't cost more than a microswitch (except for maybe the pcb) but could be a lot more sensitive, or less sensitive, and is not influenced by the angle of attack (to some extent, offcourse)


As a bonus, if you're already using opamps you can use a left-over one as an oscillator for the IR target LED.

ScuD
07-16-2008, 03:54 AM
Come to think of it, it could probably even be just a FET that's triggered by the piezo.
I'll try this approach and let you guys know what I come up with.

ScuD
07-24-2008, 02:42 AM
Okay, after 3 days of searching I gave up looking for a stupid simple little piezo buzzer to test the principle... I can't believe I don't have any of those things laying around anymore.

Anyway, here's (http://robotporn.de/wp-content/uploads/2007/01/triggerlution2.jpg) what I meant. It's one of those electric drum pads viewed from the back, which is basically perfectly what we'd want, detecting an impact, in this case not from a stick but from a BB.

Adrenalynn
07-24-2008, 02:49 AM
Try computer speakers, the onboard motherboard speakers. Most of them are piezo elements these days. If you have a bunch of old junk motherboards laying around like I do, they're a dime a dozen.

ScuD
07-24-2008, 03:25 AM
I'll check it out, thanks for the tip !

I took apart nearly everything I had lying around but only found really flat speakers, as in voice-coil ones.

I did find one in a smoke detector, but it was one with vaporized aluminum, impossible to solder a wire to...

It's moments like these when I hate throwing away stuff

ScuD
08-16-2008, 06:01 AM
Ok, so finally I found one of the piezo speakers I needed to test the theory.

I soldered a length of wire to it, slammed it in a breadboard and hooked up ye olde gameboy-oscilloscope (I really need to hassle the guys at work to get those Lecroys up for sale)

A light tap seems to register fine, but then we'd need something more than a light tap, we don't want false triggers.

So, I looked around for some FET's, but all I have laying around is either a powerFET rating up to 40A continuous, or a tiny double FET in a package I can't solder without using a PCB. No good for small tests.

Then I had an idea. A LED has a threshold voltage, so it would lower sensitivity, and if hooked up without a series resistor, it might even light up (keep in mind, these piezo speakers only source a few µA)

So I tried that, and bingo, a tap won't light up the Led, but a rather hard one would.

After that I took the speaker and just used a piece of scotch tape to tape it to the inside of a little plastic box. Now I could test what I really wanted to see.
http://users.pandora.be/svendecock/robot/sens1.jpg

Eureka! Tap ANYWHERE on the box, even on the sides, and the LED lights up, a hit has been registered!

Practical stuff: KISS. Hooking up a led to see the actual hit has simplified a lot of things. No FET, no filters, no thresholds to set, although that last one may be beneficial.
Next step would be to check if it would work just as good with an optocoupler, maybe a small RC filter to debounce the signal.

Anyway, thought I'd like to post the info, seems like a HELL of a lot easier to make the sensor plates. Make some "armour looking" plates in any shape you like, slap a sensor onto the back with some superglue, presto. No mocking around with microswitches to make sure every hit is registered, no 'floating' plates needed.

Oh, and sorry for the ultimately crappy pic, my little brother stole my camera for the day. I'll try to get better pics next weekend when I'm back from Ireland.

JonHylands
08-16-2008, 07:03 AM
That's really awesome, ScuD... Nice work. Definitely keep us in the loop. If this works, it would make the targets themselves a lot simpler...

4mem8
08-16-2008, 07:47 AM
Scud: Neat idea, appears to work well, I must try it out.

A-Bot
08-16-2008, 07:54 AM
Awesome. That's pretty darn simple. :)

ScuD
08-16-2008, 08:24 AM
4mem8: your mech is a perfect example of where this would be most usefull.
All that work going in to the aesthetics of the bot, looking kick-a$$; but then you'd need to cover it up with microswitches, getting the cast parts on top of those, etc etc.

This way you'd just glue the "sensor" on the inside of the armor, and voila.

I'm looking into making the simplest, cheapest (read low part number) way to get a decent signal out of these things, so question to the designers of the sensor boards: what kind of input signals would you guys be working with? Is a simple TTL pulse ok, or would you prefer an open-collector signal (that way the boards can be universal, either microswitches or the piezo sensors can be used)

/edit: well the local belgian version of ratshack is closed (holiday yesterday) so I didn't get a chance to find any fets, but I whipped up a quick testcircuit with an opamp as a comparator with capacitive feedback, seems to work fine. Adjustable sensitivity, cheap and easy to source parts, only it takes 3 parts + the piezo element... me no like..

BauerMECH
08-17-2008, 02:14 PM
4mem8: your mech is a perfect example of where this would be most usefull.
All that work going in to the aesthetics of the bot, looking kick-a$$; but then you'd need to cover it up with microswitches, getting the cast parts on top of those, etc etc.

This way you'd just glue the "sensor" on the inside of the armor, and voila.

I'm looking into making the simplest, cheapest (read low part number) way to get a decent signal out of these things, so question to the designers of the sensor boards: what kind of input signals would you guys be working with? Is a simple TTL pulse ok, or would you prefer an open-collector signal (that way the boards can be universal, either microswitches or the piezo sensors can be used)

/edit: well the local belgian version of ratshack is closed (holiday yesterday) so I didn't get a chance to find any fets, but I whipped up a quick testcircuit with an opamp as a comparator with capacitive feedback, seems to work fine. Adjustable sensitivity, cheap and easy to source parts, only it takes 3 parts + the piezo element... me no like..

I'm in agreement w/ everyone else too. This way those of us who've worked hard making our robots look neat can keep them true to the theme. I was a little worried about the targets before, but this little setup is perfect! Excellent work!

If our bi-peds robots fall/trip, does that going to count against us? Perhaps a tilt sensor in line could eliminate that problem if we’re not keeping track of downs?

sam
08-17-2008, 02:25 PM
Looks great!

I just don't see the size of this thing? Looks pretty big for a mech, no?

Do you have some dimensions you could give us? I assume we could just recreate it at the size we need?

Adrenalynn
08-17-2008, 02:32 PM
You can get piezo elements measured down in the thousandths of an inch for things like strain-measurement on critical components.

A piezo is basically a crystal structure. When flexed, compressed, or distorted it generates a minute amount of electricity (actually polarizes the structure), which ScuD is proposing is then measured.

It would have to be large enough to compress, shock, or deform when any area of the target is hit.

dcalkins
08-18-2008, 02:45 AM
Regardless of how thin, the front plate would still have to be 3" tall by 3" wide.

If you're making a mech for looks, that's fine. But keep in mind it needs to have space for two-four plates on it.

(If all you care about is looks, enter it in bet of show...)

BauerMECH
08-18-2008, 10:43 AM
Don’t get me wrong, functionality is naturally a better idea to pursue than appearance. But, getting the best of both worlds would be ideal. Turning the surface area of the robot’s shell into the “plate” would also make it scalable with the size of the bot.
Oops, I just registered the “regardless of how thin” line you wrote Dave… In other words, we could fab “plates” out of foil and tape that could contour the shells anyway. Has it been determined whether the 3x3 area could be segmented (within reason) to total 3² or does it need to remain totally in tact (ie. one 1”x3” and one 2”x3” w/ ¼” gap between the two)?
Food for thought: air-soft bb’s aren’t too big of a deal, but if we’re calculating damage from weapons in the hardcore class, I would think it would be rather easy to cause damage to a switched plate that would result in a constant closed state (ie. BB denting the plate over where the switch resides). Even if the piezo takes a direct hit it will still reset to low after impact.
Another question – regardless of switch plates or piezo shells, are we going to require the target area be a bright color as to allow autonomous color recognition targeting systems or are we aiming on human and/or other methods of parsing? Maybe a better question to ask would be how many of you out there are thinking they’ll have a color tracking targeting system ready in time to require a need for contrast. I’m game either way.

ScuD
08-18-2008, 01:01 PM
Regardless of how thin, the front plate would still have to be 3" tall by 3" wide.



Point taken,
I have to say I came up with this idea purely for practical reasons. In my opinion the setup would be easier, adjustable, and more reliable than switches.

The fact it can be incorporated in variously shaped parts is just an 'added bonus' really.

By no means am I intending to circumvent any of the rules with this idea, just getting it out there to show there are other ways than switches.

Adrenalynn
08-18-2008, 01:23 PM
Personally I think it's an outstanding suggestion, ScuD. Anyone that thinks common microswitches are all that reliable under stress has never put a common microswitch under stress, IMHO. ;)

dcalkins
08-18-2008, 01:27 PM
Point taken,
I have to say I came up with this idea purely for practical reasons. In my opinion the setup would be easier, adjustable, and more reliable than switches.

The fact it can be incorporated in variously shaped parts is just an 'added bonus' really.

By no means am I intending to circumvent any of the rules with this idea, just getting it out there to show there are other ways than switches.

Oh, I agree totally! I'm really in awe of your solution. I had been playing with buttons (incl. phone keypads), but yours is a really great idea.

My concern would be that some people might want to scale it down strictly for appearance... And I am very* in favor of good looking robots - it's why I started Best of Show and the 5 art bot classes.... But form has to follow function in this particular event.

I played cowboys and indians too many times with the "I shot you" "Did not!" arguments... We need a nice, reliable, automated hit counter... Your solution looks like it will be it.

ScuD
08-18-2008, 01:38 PM
Well it still does need work, I'm in Ireland right now, but right before I left I tested it by shooting an airsoft bb on the plastic box at point blank range.
The shot is registered all right, but there's much longer resonance than with a hard snap of the finger.
Best way would be triggering a monostable circuit with the piezo, but that would mean loads of circuitry (costs!!) for a sensor that ought to be quite simple.

When I'm back home I'm going to check if it would be possible to attach it to a microcontroller through a filter / limiting circuit and have a simple soft debounce.
Most micro's have high impedance inputs anyway, so it might work.

Either way, I'll post any updates here. Btw, who are the designers of the hit register board? I'd like some info on what they're planning to use so I could adapt to that.

DresnerRobotics
08-18-2008, 03:05 PM
Looks like Piezo will be the way to go. I'm sending a couple of these: http://www.trossenrobotics.com/store/p/3233-Vibration-Sensor.aspx off to Jon to test with the wireless scoring unit he is developing.

Great suggestion Scud, + rep for you! FYI- Jon is the primary developer of the scoring unit.

JonHylands
08-18-2008, 03:05 PM
I'm designing the hit register board, so you can talk to me about it. I'm using an ATmega168, and I can use either analog or digital inputs. Please try and make the hardware as simple as possible if you can make this happen, because we can do things like debounce and such in software...

sthmck
08-18-2008, 03:23 PM
Jon your board is going to be able to hook up directly to ax-12 bus right?

DresnerRobotics
08-18-2008, 03:41 PM
The idea is that the target/scoring system is going to be largely independent of the rest of the robot, with the exception of having an I/O port to pull optional hit data from.

JonHylands
08-18-2008, 03:52 PM
There will be two versions of my board - one will be a bioloid bus device, and the other will use a simple serial output to provide hit data.

MSpurk
08-18-2008, 07:35 PM
Just to muddy the water a little. Jon, you are planning on creating the hit counter, and that will transmit to the "Master". We will not be responsible for having any wireless serial or anything on our bots, correct?

-Matt-

JonHylands
08-18-2008, 08:19 PM
That's correct. I will be creating the slave units for the mechs, and the master unit the slave units will all talk to. All you will need to do is mount it on your mech, and provide it with a source of power.

sthmck
08-18-2008, 08:24 PM
what is the power source requirements? I am guessing it will need to be regulated?

JonHylands
08-18-2008, 08:38 PM
No, not at all. It will probably need to be above 6.2 volts, but that depends on which regulator I end up using. Most of my boards that are small SMT boards (like this one will be) use an LM3480 low dropout regulator.

If you're using the bioloid variety, of course you don't have to worry about it - you plug it into the bus, and it works.

4mem8
08-19-2008, 12:18 AM
I intend to make carbon/kevlar plates for my Ed 209, these will be strong enough to repel bb's and protect A mech, I will try to incorporate the 3x3 plate in the chest area and back.
JohnHylands: When do you think the slave units would be ready for us to buy?
Adrenalynn: When do you think the camera systems will be available??
Just trying to tie items up and think ahead for building purposes. Thanks guys and gals.

JonHylands
08-19-2008, 07:06 AM
I've been kinda waiting for someone *cough* to send me some information on some wireless units she designed... I have the main part of the board designed, which is easy enough, but I don't know which RF boards I'm going to use, how big they are, and what kind of mounts they require...

ScuD
08-19-2008, 07:59 AM
I'll be sure to try getting a design working with the simplest design possible, Jon.

I'll make sure you get the info asap, but it won't be before the weekend I'm afraid.

Adrenalynn
08-19-2008, 11:08 AM
They're in the mail! :)

ScuD
10-02-2008, 01:35 PM
Well, it took a while, have a lot of stuff going on ATM which has forced me to spend a little less time with the hobby... but hey, I've got something working.

Here's the schematic I'm using:
http://users.telenet.be/svendecock/robot/BBsensor.png

(actually, I'm not using a zener since I didn't have the right one, but since I'm 'abusing' the pic input, I've included it in the schematic)

What the schematic is doing, is simply clamping the output of the piezo transducer to 5.1v, and making sure reverse voltage spikes from the piezo are shorted (well, clamped to -0.6v actually)

The clamped output is attached to a schmitt trigger input on a PIC16f628, which means it will only trigger above 0.8*Vdd, in this case 4volts.
A software debounce is used to clean up the signal a bit, other than that the code consists of a simple 'counter' to drive the LED's (223 byte hex, and that's without cleanup work)

ok, enough talking, here's the vid: Piezo BB sensor

Note that the debounce period was a little short, causing the harder taps to be registered as two hits. Some tuning of the debounce period is required, but I need to get BB's to test it.

Adrenalynn
10-02-2008, 01:43 PM
+Rep on that! Nice simple implementation. Debounce is easily addressed in firmware or hardware - it's just tuning. GREAT job!

ScuD
10-02-2008, 02:07 PM
Thanks Lynn :veryhappy:

One more thing I'm gonna try is the ADC approach - so you'd be able to set the sensitivity within a window. It'll take a little more on the software side though, but the hardware side should be able to handle it as-is.

Debounce time is indeed only a matter of tuning the timer value, I'm quite happy with the simplicity of the system. Turned out better than I'd thought.

Adrenalynn
10-02-2008, 02:14 PM
IMHO, that's a bitchen super-inexpensive and tweakable design. It has wide application outside of just Mech Wars in robotics. It's a great solid-state replacement for a bump-switch that gets rid of the inconsistancy and short lifespan of the microswitch. It's also a limit-switch. I should have done something like that on my competition 'bot last year.

How about turning this into a tutorial? I think it's the PERFECT application for a tutorial.

ScuD
10-02-2008, 02:23 PM
How bout this :

I figure out the analog part of the software, and once I get that working I'll pour it into a tutorial containing both hardware, software, digital and analog version.

I'll supply both source-code (pic assembly) and pseudo-code so people with other controllers can grasp the concept as well.

DresnerRobotics
10-02-2008, 02:47 PM
Scud you're the fricken man, that's awesome! +Rep for sure.

BauerMECH
10-03-2008, 10:00 AM
I love it! :veryhappy: Thabnks ScuD!
For simplicity sake, I wonder if having an easily visible array of LED’s (Like you've got here) would be the way to go versus an additional setup just for transmitting the score – final LED being red indicating “death”. Nor would it be too hard to tie the output into an A/D port and animate our bots into playing a disabling/shut-down sequence or something when hit for the final time.

JonHylands
10-03-2008, 10:09 AM
The issue with that is the targeting system (the wireless slave these sensors attach to) will be doing its own "debouncing" or governing to only allow one hit per second. That information will be available on the slave interface, so if a mech builder wants to incorporate that they can.

DresnerRobotics
10-03-2008, 10:22 AM
I love it! :veryhappy: Thabnks ScuD!
For simplicity sake, I wonder if having an easily visible array of LED’s (Like you've got here) would be the way to go versus an additional setup just for transmitting the score – final LED being red indicating “death”. Nor would it be too hard to tie the output into an A/D port and animate our bots into playing a disabling/shut-down sequence or something when hit for the final time.

With the output pin we'll have available, I'm sure it would be quite easy to add something visual on the bots. Personally I'll be using it to put a "Damage level" indicator on my control program.

The reason for a main scoring unit was so that we could have the score on a large display, so that the audience could see.

Sienna
10-03-2008, 11:24 AM
The biggest concern I have with any "3rd party" scoring system, is how do I as a mech builder assure that walking doesn't set off the "hit" sensor? Because I am making the educated guess that some of these mechs will not walk with twinkletoe steps as it were.

I think using a piezo sensor makes this even more relevant, because now the sensor can detect inadvertant shocks coming from the legs, instead of being limited to the plunge direction of the switches.

Then, I have one more question. If a piezo sensor can be attached to any plastic 'form', does that mean we can do away with the flat vertical surface requirement? For instance, if I make my canopy out of vacu formed lexan, and attach the sensor there, can I use my canopy as my hit sensor, provided that the canopy's 2D projected area is 9 in^2 ?

DresnerRobotics
10-03-2008, 12:15 PM
The biggest concern I have with any "3rd party" scoring system, is how do I as a mech builder assure that walking doesn't set off the "hit" sensor? Because I am making the educated guess that some of these mechs will not walk with twinkletoe steps as it were.

I think using a piezo sensor makes this even more relevant, because now the sensor can detect inadvertant shocks coming from the legs, instead of being limited to the plunge direction of the switches.

Then, I have one more question. If a piezo sensor can be attached to any plastic 'form', does that mean we can do away with the flat vertical surface requirement? For instance, if I make my canopy out of vacu formed lexan, and attach the sensor there, can I use my canopy as my hit sensor, provided that the canopy's 2D projected area is 9 in^2 ?

More than anything we just wanted a '3rd party scoring system' to simplify things for builders and to make sure we have a verified working system. What we don't want to see is people showing up with their own scoring systems that do not perform as expected. I would think we could set a certain threshold on the piezos so that simply walking does not trigger them.

As to your other question, I'm certainly open to this, nothing is yet set in stone. I would think to be fair to others using the standard 3x3" targets we would need verified proof that yours works in the same exact manner (ie: photos/video). To be honest I was thinking about proposing the same.

4mem8
10-03-2008, 12:31 PM
Will this system be able to be applied on my Ed209 mech? Botboard2 and SCC-32 system, or is it only suitable for the Bioloid system?

JonHylands
10-03-2008, 12:33 PM
Yes, there will be two versions of the slave module, one for Bioloid, and one for other mechs.

4mem8
10-03-2008, 12:49 PM
JonHylands" Thanks for the info, Will you let us know when they are ready for purchase.

Sienna
10-03-2008, 12:58 PM
As to your other question, I'm certainly open to this, nothing is yet set in stone. I would think to be fair to others using the standard 3x3" targets we would need verified proof that yours works in the same exact manner (ie: photos/video). To be honest I was thinking about proposing the same.

I think thats perfectly fair. In fact, when Scud releases his tutorial, I will be more then willing to build one, and take my airsoft collection to it, and video tape it.

(would need to get some lexan sheet and a piezo first, but I need to get that anyway)

ScuD
10-03-2008, 03:03 PM
The biggest concern I have with any "3rd party" scoring system, is how do I as a mech builder assure that walking doesn't set off the "hit" sensor? Because I am making the educated guess that some of these mechs will not walk with twinkletoe steps as it were.

I think using a piezo sensor makes this even more relevant, because now the sensor can detect inadvertant shocks coming from the legs, instead of being limited to the plunge direction of the switches.

...


This is actually the reason I haven't released anything yet. At this stage, I'm not too confident on the systems reliability yet regarding sensitivity and adjusting of this sensitivity.
As soon as I'm happy with it, I'm able to clearly set threshold levels and have consistant results, I will put up the tutorial.

In the meantime, anyone willing to try it out for themselves can PM me and I'll give you any information I can to get you started with the sensor as-is.

sthmck
12-10-2008, 08:08 PM
Any new news in this area?

DresnerRobotics
12-10-2008, 08:11 PM
We're using the electric airsoft guns' inertia as the minimum velocity. Not sure what that is exactly, but if you use one of those or something more powerful, you'll be good to go.

sthmck
12-10-2008, 08:59 PM
I meant with the piezo thing.

ahweh
01-15-2009, 03:03 PM
I do not know if this has been fully resolved but these have been selling on ebay recently.
http://www.meas-spec.com/downloads/MiniSense_100.pdf