PDA

View Full Version : Target/Scoring System



DresnerRobotics
06-26-2008, 05:06 PM
"We're going to use this wireless module (http://www.sparkfun.com/commerce/product_info.php?products_id=152), one per robot, plus a master. The robot (slave) modules will have a secondary board (which I will design & build) that has an onboard ATmega168, and it will interface with up to four taget modules. Each target module is basically a micro-switch that shorts out two wires when it is hit. It will be up to the robot builder to build the target modules for their robot. We will come up with standards for the size of the target.

The slave module will require a 3-wire connection to the robot - power (5 volts), ground, and a digital input pin that will be strobed each time a target is hit. This is so that the robot can send that information back to the controlling PC, and the person controlling the robot will know how many hits they have. Each slave module will have a unique ID.

Once per second, the master will interrogate each slave to find out how many hits they have, and the results will be posted on a large screen that the audience can see.

In terms of number of targets per robot, that is up in the air, but my initial thoughts are:

# of targets = # of legs

So a biped would have two targets (front and rear), a tripod would have three, and a quad would have four. One of the targets has to be clearly visible and hittable when the robot is facing an opponent.

I'm going to build two version of the slave module - one as mentioned above for non-Bioloid mechs, and one that is a Bioloid bus device for Bioloid mechs (so the three-wire connection will be a standard Bioloid bus connector).

The slave modules will be available at cost, which will be somewhere around $50.

Creating this thread to continue discussion of our Target & Scoring system.

The general idea is we would like to create standardized slave units that will mount on each bot, and interface to 1-4 target plates on the mech, and relay the hit input to a master unit that will in turn interface to a scoreboard.

MYKL
06-26-2008, 06:23 PM
You mentioned 1 plate for a bipedal mech. You also stated that the target should face the opponant. What happens when the bi-ped is making a strategic retreat?

A biped would need to have a front and back target in order to be a fun target... ^_^

DresnerRobotics
06-26-2008, 06:35 PM
So a biped would have two targets (front and rear), a tripod would have three, and a quad would have four. One of the targets has to be clearly visible and 'hittable' when the robot is facing an opponent.

We thought of that ;)

JonHylands
06-26-2008, 06:36 PM
We modified the spec yesterday to require two targets for a biped. We will decide exactly where the targets have to be for each configuration, but for a biped you will need one on the front, and one on the rear.

DresnerRobotics
06-26-2008, 06:39 PM
At this point, number of target plates = number of legs. Figured I would point that out.

dcalkins
07-01-2008, 04:31 AM
Actually, for added realism, you could have a pass-through system:

When a target area (let's say, for example, it's a biped with two arms - so four quadrants) get's enough hits, it can be literaly disabled with NO permanent damage.

Pass through system is very simple: It has, say, 6 header-pin sets per quad (six in and six out). So I plug my quad (arm or leg) into the pass-through, and the pass through onto the CPU.

When enough hits are registered, the pass-through system disconnects the DC. This is a very simple relay system. All Servos pull from the same point, so you can leave the signal and ground wires alone on the pass-through, but all positive DC points (the red wire on a standard servo cable) come to a single point. Should that quad get turned off, all servos lose power.

VERY real. Obvviously, this would suck if you got hit. By the same token, a good programmer could write SW that allows a bot to stand on one leg (hopping at this point would be cool, but it's more realistic to continue standing on one leg and shoot while you can - hoping for the best.)

Perhaps a visual indicator that your leg is about to go belly-up (in a very literal sense) for 3 seconds (call it shock), so the operator can switch to single-leg mode.

Again, a very real and easy-to-implement system. It would make for more... interesting game play. Dunno if that equates to "better" game play, but certainly more realistic (and as someone who has had his right foot literally broken off his leg (at the bone level, held on by tendons and skin), I can attest that it is quite possible to ambulate on one leg.)

Aw hell.... I'm getting all excited about this event.

I DON'T DO THAT SORT OF THING.

I don't like you guys any more.

Brain hurts me. Lazy brain make davey-days easy. No like hard work.

Adrenalynn
07-01-2008, 04:35 AM
Tyb's enthusiasm is contagious, IMHO. Blame him.

You could also use a small microcontroller on your pass-through to make the limb start to "misbehave" as it takes damage... Wow. That is cool!

Wingzero01w
07-01-2008, 04:42 AM
Tyb's enthusiasm is contagious, IMHO. Blame him.

You could also use a small microcontroller on your pass-through to make the limb start to "misbehave" as it takes damage... Wow. That is cool!

Kind of like having it spaz out and disobeying original operation?

And yes Tyberius' enthusiasm is a very contagious effect, no cure in sight :)

Adrenalynn
07-01-2008, 04:47 AM
Kind of like having it spaz out and disobeying original operation?

And yes Tyberius' enthusiasm is a very contagious effect, no cure in sight :)

Yup - drop speed, drop torque, throw away packets, all the way to doing the opposite of what you commanded.

That kind of gradual failure is more in line with battle mechtronics, from what I've observed.

DresnerRobotics
07-01-2008, 04:47 AM
Actually, for added realism, you could have a pass-through system:

When a target area (let's say, for example, it's a biped with two arms - so four quadrants) get's enough hits, it can be literaly disabled with NO permanent damage.

Pass through system is very simple: It has, say, 6 header-pin sets per quad (six in and six out). So I plug my quad (arm or leg) into the pass-through, and the pass through onto the CPU.

When enough hits are registered, the pass-through system disconnects the DC. This is a very simple relay system. All Servos pull from the same point, so you can leave the signal and ground wires alone on the pass-through, but all positive DC points (the red wire on a standard servo cable) come to a single point. Should that quad get turned off, all servos lose power.

VERY real. Obvviously, this would suck if you got hit. By the same token, a good programmer could write SW that allows a bot to stand on one leg (hopping at this point would be cool, but it's more realistic to continue standing on one leg and shoot while you can - hoping for the best.)

Perhaps a visual indicator that your leg is about to go belly-up (in a very literal sense) for 3 seconds (call it shock), so the operator can switch to single-leg mode.

Again, a very real and easy-to-implement system. It would make for more... interesting game play. Dunno if that equates to "better" game play, but certainly more realistic (and as someone who has had his right foot literally broken off his leg (at the bone level, held on by tendons and skin), I can attest that it is quite possible to ambulate on one leg.)

Aw hell.... I'm getting all excited about this event.

I DON'T DO THAT SORT OF THING.

I don't like you guys any more.

Brain hurts me. Lazy brain make davey-days easy. No like hard work.

Bravo Dave, amazing idea. Definitely something I'd like to look into, and stays true to "Mechwarrior" style gameplay.

Wingzero01w
07-01-2008, 04:49 AM
Yup - drop speed, drop torque, throw away packets, all the way to doing the opposite of what you commanded.

That kind of gradual failure is more in line with battle mechtronics, from what I've observed.

Question is, how does your average novice program something like that? Sounds pretty complicated.

DresnerRobotics
07-01-2008, 04:53 AM
Question is, how does your average novice program something like that? Sounds pretty complicated.

Yeah this would be great, but would require standardization. It's why we're trying to keep it simple and Jon is designing a standardized module for scoring... not nearly as dynamic as the system David proposed but a bit easier to implement... at least for year 1.

Adrenalynn
07-01-2008, 04:53 AM
It's going to be dependant upon your servo choice - a straight PWM analog servo would be different from a 5990 digital would be different than an AX12 or other module.

Effectively we're just introducing a feedback loop with some simulated "noise".

That said, the programming is actually pretty trivial for an experienced coder with a processor costing literally pennies. I think we were reasoning that it's a system that would be provided by MW itself in order to keep honest people honest, and lower that barrier to entry, right?

Adrenalynn
07-01-2008, 04:55 AM
Yeah this would be great, but would require standardization. It's why we're trying to keep it simple and Jon is designing a standardized module for scoring... not nearly as dynamic as the system David proposed but a bit easier to implement... at least for year 1.

I'll bet we could run these past Jon and get this going without much trouble. David's suggestion really does simplify things in the longer run - as well as add a base for adding more fun stuff as I put forward...

DresnerRobotics
07-01-2008, 04:56 AM
It's going to be dependant upon your servo choice - a straight PWM analog servo would be different from a 5990 digital would be different than an AX12 or other module.

Effectively we're just introducing a feedback loop with some simulated "noise".

That said, the programming is actually pretty trivial for an experienced coder with a processor costing literally pennies. I think we were reasoning that it's a system that would be provided by MW itself in order to keep honest people honest, and lower that barrier to entry, right?

Unfortunately it still is a whole other degree of complexity, and Jon is already gracious enough to offer to design the current proposed system. Too many variables with it, while its a system that I think is awesome- I'm not sure if its feasible to implement within the first year. Definitely something to keep in mind for the future.

Adrenalynn
07-01-2008, 04:58 AM
I can deal with that. It's also something that could be adopted/implemented voluntarily for the "thrill of the game" and the "honor" there-of.

dcalkins
07-01-2008, 05:05 AM
Question is, how does your average novice program something like that? Sounds pretty complicated.

I loathe crossing threads, but to quote myself:


MUCH easier system: All registrants must wire their CPU's with servo extensions. Very simple procedure. If the pass-through system isn't done, no worries, you just have some extra cabling. If it IS done, it can be added in a few minutes time, by disconnecting the extensions, and placing the pass-through system at that juncture. If the pass-through doesn't fail-safe: again - you just re-connect the extensions.

In this fashion, the pass-though can be networked or non-networked. It becomes a wiring matter, rather than a programming one. It also alleviates the inevitable problem of cover plates protecting the CPU... You can leave the cover plate on, and just pop the pass-though on-top of the cover plate.

N'est-ce pas?

Or in short, you don't program anything. Nothing. It's on the event Organizer to do that. You just spend an extra $20 on servo extensions.

BigBug will be happy to pay for them.

Wingzero01w
07-01-2008, 05:07 AM
Or we could just use wireless, most of our mechs are going to have some form of wireless connection from the sounds of it.

DresnerRobotics
07-01-2008, 05:08 AM
BigBug will be happy to pay for them.

With the corporate sponsorships being thrown nilly-willy into the wind these days? :D

dcalkins
07-01-2008, 05:10 AM
It's going to be dependant upon your servo choice - a straight PWM analog servo would be different from a 5990 digital would be different than an AX12 or other module.

So you have dual five-pin (AX-12) and 3-pin (PWM) modules. Either way, they both* have a common VDC line. All you do - in the first rev - is disable the VDC line. The other lines are all pass-through. literally continuous lines for either a PWM line or serial lines. But if you pull the DC, you kill the servo.

There is very little actual programming needed with this solution. When the counter hits its max, a timer flashes an LED in front of the camera on the bot for 3-5 seconds (notifying the 'driver' that he's hit, and should act appropriately), and then the DC to that quad is turned off via a $0.12 relay.

KISS.

Wingzero01w
07-01-2008, 05:12 AM
Yeah but we were talking more like spazzms so like if you get hit your arm leg or whatever goes insane for a short moment.

dcalkins
07-01-2008, 05:14 AM
Or we could just use wireless, most of our mechs are going to have some form of wireless connection from the sounds of it.

Wireless would rely on end-user programming. You would have to program your robot that if it recieved a certain packet, it must then turn off a certain quad.

My system relies on nothing more than the already required "hit system" to flip a relay. no programming, no end-user changes, nothing.

And the error tolerance is also much lower. The cheater tolerance is nil.

dcalkins
07-01-2008, 05:15 AM
Yeah but we were talking more like spazzms so like if you get hit your arm leg or whatever goes insane for a short moment.


Bah. Smasms are for TV. 3 seconds of warning via an LED within your camera, and then your quad is out.

DresnerRobotics
07-01-2008, 05:15 AM
So you have dual five-pin (AX-12) and 3-pin (PWM) modules. Either way, they both* have a common VDC line. All you do - in the first rev - is disable the VDC line. The other lines are all pass-through. literally continuous lines for either a PWM line or serial lines. But if you pull the DC, you kill the servo.

There is very little actual programming needed with this solution. When the counter hits its max, a timer flashes an LED in front of the camera on the bot for 3-5 seconds (notifying the 'driver' that he's hit, and should act appropriately), and then the DC to that quad is turned off via a $0.12 relay.

KISS.

How does this work on a serial chained servo like the AX-12s though? You cut DC to a servo and every servo chained off of it if going to also lose power. I'm sure theres a way around it, but it certainly adds a bit more complexity to it.

Your LED in front of the camera idea is killer though, and very simple. Hell I think that should be implemented just as a 'taking damage warning' every single time the mech takes a hit.

Adrenalynn
07-01-2008, 05:17 AM
I agree that your solution is absolutely KISS and works fine across the board. I was pie-in-the-sky'ing to another level of "realism" - which I acknowledge is moderately more difficult, larger BOM, a little more cost, and is semi-platform-dependant.

dcalkins
07-01-2008, 05:18 AM
How does this work on a serial chained servo like the AX-12s though? You cut DC to a servo and every servo chained off of it if going to also lose power. I'm sure theres a way around it, but it certainly adds a bit more complexity to it.

Four serial lines. One for each quad. Unless the Bioloid (I admit, I haven't used one), has to have all servos in one single loop - which seems like a dseign flaw, since in this scenario, any single servo failure would kill the whole system.

Just link a given quad to itself as part of the rules. If a contestant links their arm to their leg... shit, that's a design flaw in need on proving itself 'worthless' on the field of battle.

JonHylands
07-01-2008, 07:39 PM
On the Bioloid, all the servos are on a common bus. The three wires are shared across all the devices. Typically, you daisy-chain all the servos on a given "limb". With MicroRaptor, I have five separate "chains", all originating from 5 of the 6 bus connector plugs I have on my main board.

One chain for each leg, one for the tail, one for the neck/head, and one for the IMU.

If you interrupt one of the chain's power line, it will affect all the servos downstream of where that "break" happens. When you remove power from an AX-12, it immediately goes "soft", which would cause (if it was one of the legs) the robot to collapse instantly.

dcalkins
07-02-2008, 12:09 AM
On the Bioloid, all the servos are on a common bus. The three wires are shared across all the devices. Typically, you daisy-chain all the servos on a given "limb". With MicroRaptor, I have five separate "chains", all originating from 5 of the 6 bus connector plugs I have on my main board.

One chain for each leg, one for the tail, one for the neck/head, and one for the IMU.

If you interrupt one of the chain's power line, it will affect all the servos downstream of where that "break" happens. When you remove power from an AX-12, it immediately goes "soft", which would cause (if it was one of the legs) the robot to collapse instantly.

I played with my bioloid kit today... It's easy to have the proposed kill-sicth interupt one change, w/o affecting others.

At each change in leg/arm, you interupt the chain. That portion of the chain get's it's power from the proposed pass-through board. Put in a diode to prevent back-flow to other chains. It would work just as reasily as with standard servos...

Omegaspecter
03-14-2009, 04:16 PM
One thing I have not noticed mentioned is the size of the targets. What are the dimensions on the target plates?

lnxfergy
03-14-2009, 04:22 PM
One thing I have not noticed mentioned is the size of the targets. What are the dimensions on the target plates?

Target plates should be 3"x3", 2 of them for bipeds, 4 for quads.

The AVR transponder board is 2.4x2.4" (mounting holes are in all four corners, 0.15" in from the edge, sized for a #4 machine screw)

I'm putting final touches on the transponder firmware and completing the PC scoring app this weekend. Tyberius is working on mechanical mounting recommendations for the target plate this weekend. There should a mostly finished version posted later this week.... and orders for boards will commence very soon.

-Fergs

Zacattack
03-15-2009, 12:28 PM
hmmm just wondering, if u gave us all the weight, size, and requirements of the target plates then we can buidl the mechs with them in mind? just a thought

lnxfergy
03-15-2009, 12:38 PM
hmmm just wondering, if u gave us all the weight, size, and requirements of the target plates then we can buidl the mechs with them in mind? just a thought

Size of everything is in the previous post... weight is nearly negligible... as for other requirements read the rules... (for things like, your target plates should be in same plane as your camera/guns, if you have a panning torso)

-Fergs

Zacattack
03-15-2009, 12:41 PM
ok thanks, i didnt know how high they needed to be mounted thansk

lnxfergy
03-15-2009, 02:12 PM
ok thanks, i didnt know how high they needed to be mounted thansk

The actual design of the target plates is still be explored. Tyberius should be doing some testing this weekend to finalize the design of the plates.

-Fergs

Zacattack
03-15-2009, 05:30 PM
oh ok, hmm it would be neat if we were allowed to put our colors or a logo on the target plates, aka steady hand and sharpy!

DresnerRobotics
04-06-2009, 03:44 PM
Just as an FYI- Fergs and I are working out the final details on the target units. I'll start taking orders for them this week most likely, however, due to the time constraints I am under and the amount of time it takes to build, configure, and test these, priority will go to those competing this year.

Zacattack
04-14-2009, 10:05 AM
ok cool, i know that its probly in these forums somewhere, but im on my ipod since my computer and laptop is at home and am getting a bit frustrated on how long it is taking just to read throught a forums page, but about how much is a unit going to cost?

lnxfergy
04-14-2009, 10:22 AM
ok cool, i know that its probly in these forums somewhere, but im on my ipod since my computer and laptop is at home and am getting a bit frustrated on how long it is taking just to read throught a forums page, but about how much is a unit going to cost?

Board (assembled, programmed, tested) is going to be about $25 (final price will be set by Tyb, but I believe that's what he is shooting for)
XBEE module - $19
Target plates use this sensor - $9 each (2 for bipeds, 4 for quads). http://www.trossenrobotics.com/store/p/3233-Vibration-Sensor.aspx

-Fergs

DresnerRobotics
04-23-2009, 05:57 PM
I am now taking orders for the Scoring system. Please send me a PM for details.

Note: Priority will go to those participating in this year's Mech Warfare competition. If you are not going to participate this year, but still want a unit, you may order one however I will be building the units for those competing first.

societyofrobots
07-12-2009, 12:06 AM
Tyberius is working on mechanical mounting recommendations for the target plate this weekend. There should a mostly finished version posted later this week....

Was this ever done? I searched and can't find it . . .

Adrenalynn
07-12-2009, 12:29 AM
The plates are being standardized for next year. It will be a turn-key requirement.

lnxfergy
07-12-2009, 06:57 AM
Was this ever done? I searched and can't find it . . .

There was a thread entitled "Mech Scoring 2009" or some such that I created, my internet connection here sucks so you'll have to do the searching. It had mounting recommendations. But remember, this was year 1. Please note: 2010 rules have not yet been posted, nor has the renewed scoring system been finished -- in other words, don't go buying lots of parts for the scoring system, as it will be somewhat different from last year -- the scoring panels were not overly reliably.

-Fergs

societyofrobots
07-12-2009, 09:20 AM
Thanks!

http://forums.trossenrobotics.com/showthread.php?t=3300&highlight=2009+scoring

I'm definitely not going to buy anything until things are finalized, I just need some information to help budget weight/space on my bot in the CAD design.