PDA

View Full Version : RTEAM Robotics Club Mech Warfare 2015



artans
03-30-2015, 11:03 PM
5864

We're located in Tucson, AZ.

RTeam Robotics Website
(http://rteamrobotics.weebly.com/)
This thread to post information on things RTEAM robotics is doing with Mech Warfare. Other members in our club might post in this thread.

This is a video of exhibition matches we did during a club meeting on 3/21/2015.

https://www.youtube.com/watch?v=_34-x_FxYH8

Will post more videos and information when we can.

Scoring system we used during exhibition match:
- 2013 Scoring Transponder RTeam Variant
- Piezoelectric Speaker Scoring Panels

We took the 2013 Scoring Transponder and added a FTDI 6 pin connector to it. We then loaded the Arduino Bootloader and modified the original code to work in Arduino IDE.

We also created our own scoring panels, the new scoring panels uses a piezoelectric speaker to detect hits. There is also some additional circuitry added to condition the signal. The panel is designed to be compatible with the old scoring transponder and work with the old scoring panels. In the video, Nomad is using 3 old scoring panels and 1 new piezoelectric panel. Apophis is using 4 of the new piezoelectric panels.

Connor
03-31-2015, 12:03 AM
We also created our own scoring panels, the new scoring panels uses a piezoelectric speaker to detect hits. There is also some additional circuitry added to condition the signal. The panel is designed to be compatible with the old scoring transponder and work with the old scoring panels. In the video, Nomad is using 3 old scoring panels and 1 new piezoelectric panel. Apophis is using 4 of the new piezoelectric panels.

I remember we tried using piezoelectric sensors @ the very first Mechwarfare. They didn't work very well, and were not reliable.. and in some cases.. would trigger when the Mech would walk. That's why we switched from them to Force Sensing Resistors (FSR). Now the issue could have been the mounting method.. or some other issue.. but, from our experiments, the FSR's worked way better.

jwatte
03-31-2015, 11:32 AM
Perhaps a mounting interface based on sorbathane would make the piezos less sensitive to walking?

Btw, if you want a site that supports registration of teams/bots/competition for events, I developed this:

http://mwreg.watte.net/

It currently doesn't do any pictures, which I feel is a significant draw-back, but that can be added in time.

Give me a ping if you want an event listed!

giantflaw
03-31-2015, 12:04 PM
Some of the FSR plate disadvantages that I think we all don't like are: 1.) Expensive 2.) Damage from handling of the four flex connections to the circuit card. 3.) Dead zones between the four FSR sensors. 4.) Not always registering valid hits 5.) Mounting compliance to the body affects hit detection reliabilty. 6.) Less sensative to valid angled hits

The reason we have started using the piezo target panels in Tucson is because there are now six mechs in Tucson and we only have enough FSR target panels for 3 robots and the other guys wanted their own panels. Also because of the above disadvanges of the FSR target plates we developed our own that can be easily and cheaply built. We made them compatible with the existing target transpoder and connector. The target panels do not register false hits, only valid hits are immune to walking gait.

RTEAM piezoelectric target plates are fully compatible with the existing target plate system and I think offer 6 advantages over the FSR plates. 1.) They are much more sensitive than the FSRs 2.) They register target plate hits reliably every time and never resgister any robot vibration or shock to the robot. The target plates are oblivious to walking gait. 3.) They are much cheaper than FSR target plates and easliy made with analog components. 4.) They can be de-tuned to be less sensitive and mimic an FSR low sensitivity target plate. 5.) There are no dead zones in the target panel like there are on FSR target panels. 6.) The target panel can be any shape or size unlike the FSR target panels. 7.) The target panels are tougher than the FSR plates and are not easily damaged

These piezo target panels filter and shape a proper hit pulse while rejecting false hits, vibration, shock and walking gaits. Gun fire vibration from the onboard gun does not register false hits either.

byi
03-31-2015, 12:47 PM
RTEAM piezoelectric target plates are fully compatible with the existing target plate system and I think offer 6 advantages over the FSR plates. 1.) They are much more sensitive than the FSRs 2.) They register target plate hits reliably every time and never resgister any robot vibration or shock to the robot. The target plates are oblivious to walking gait. 3.) They are much cheaper than FSR target plates and easliy made with analog components. 4.) They can be de-tuned to be less sensitive and mimic an FSR low sensitivity target plate. 5.) There are no dead zones in the target panel like there are on FSR target panels. 6.) The target panel can be any shape or size unlike the FSR target panels. 7.) The target panels are tougher than the FSR plates and are not easily damaged

Well, these sound great. Is there any documentation or plans for documentation?

jwatte
03-31-2015, 05:15 PM
That sounds 100% awesome!

Xevel
03-31-2015, 05:32 PM
That's cool :) I would be glad to get some more info on your mechanisms (target, feeder,...) too.

giantflaw
04-01-2015, 01:42 AM
Yes, we can post the Eagle files, laser cutter files with some pics and directions. Have to pull it together as different people have done pieces of it in the club.

GhengisDhon
04-18-2015, 08:14 PM
5869

We had our second Mech Warfare battle today. We're planning to hold them roughly on a monthly basis. We have several mech's that are either fully functional, or close to being. Still need a bit of work on our arena....


https://youtu.be/MMrtD-6hsFs

If anyone is ever in the Tucson area, feel free join the mayhem.

jwatte
04-19-2015, 12:11 AM
Regular practice? You will beat down anyone else if you keep that up!

jpieper
04-25-2015, 05:51 AM
Yes, we can post the Eagle files, laser cutter files with some pics and directions. Have to pull it together as different people have done pieces of it in the club.

We'd love to be able to build something similar if you could post anything, even just a bill of materials, but of course the more the better.

GhengisDhon
04-26-2015, 02:34 PM
Here's a quick overview of the construction of the target plates. They're the standard size 3.5" x 3.5 ". Material is 0.063" Delrin (Acetal) from Zoro. (http://www.zoro.com/i/G0518481/) We cut them out at the local Hackerspace using a laser cutter. (Could use aluminum sheet. The first prototype was aluminum and it worked fine.) We use surgical tubing as a damping material. I use 3/8" tubing, but I image other sizes would work fine. The piezos one of us (giantflaw) bought dirt cheap from a surplus house (don't know where).

All the real magic is in the piezo filter card. It takes the high voltage, short duration pulse from the piezo and converts it to a signal readable by the transponder card. Right now we have a potentiometer on the filter card to adjust the sensitivity, but I imagine once we've settled on a good value, we'll put discrete resistors in it. (Filter card was designed by giantflaw.)

The transponder (not shown) is the same as used in previous Robo games, but we've made a few minor tweaks to it - (courtesy of Artans).

All in all, the target plates are dirt cheap (maybe <$5 each?) and have proved reliable in both testing and in our matches. We've talked it over in the club and we have no issues releasing the drawings or eagle cad files for the elx. Or maybe if there's enough interest, since they're cheap enough we could just kit up a bunch of them.

We plan to make a full data package on the target plates and post it on our website, but don't know how quick that will happen.

5882

5880

5881

5883
http://rteamrobotics.weebly.com/

jwatte
04-26-2015, 04:07 PM
Looks nice!
You could presumably use ABS instead of Delrin, like the old panels? Or maybe even thin polycarbonate, like the old covers?
I am somewhat weight conscious... (about my bots, too :-)

jpieper
04-26-2015, 08:08 PM
What piezo speaker are you using?

What type of circuit is on the "magic" filter card? Is it just an op-amp low pass filter, or does it have any DSP?

bloftin
04-28-2015, 08:57 PM
I posted the LED and transponder design to github, it can be found here:

https://github.com/bloftin/rbots

bloftin
04-30-2015, 11:06 AM
also got the target plate design up there too


I posted the LED and transponder design to github, it can be found here:

https://github.com/bloftin/rbots

DresnerRobotics
05-01-2015, 07:41 PM
This is fricken awesome guys! Keep it up. I love that you're doing your own thing with the Transponder system.

jwatte
05-01-2015, 07:58 PM
in the best of worlds, we could use a ESP8266-12 $3 wifi module running 802.11b for both transponder and scoring logic.
It is possible to build custom firmware for these and they have GPIOs.
The serial port could be used to send match and score status to the 'bot. And total weight is less!

GhengisDhon
06-17-2015, 11:46 PM
6046

http://rteamrobotics.weebly.com/

We had another scrimmage this weekend in Tucson. Had three mechs compete. Unfortunately, I took mine apart about a month ago to clean up the harnessing and haven't had time to get it back together (been busy at work). Not to mention, I'm working on a turret-less design. Basically using the body pitch/yaw to aim the gun. I've been set up to go turret-less for some time. Just never given it a go yet. Maybe at our July Scrimmage....

The three competing Mechs were:

Apophis (Driven by Artans)
6045


Nomad (Driven by Giant Flaw)
6042


Thorn (Club Mech driven by Joey)
6043


and '99 (My mech that didn't compete)
6044

GhengisDhon
06-17-2015, 11:49 PM
Thorn vs. Apophis

https://www.youtube.com/watch?v=b4Adt-AUrqk&amp;feature=youtu.be

GhengisDhon
06-17-2015, 11:50 PM
Nomad vs. Thorn

https://www.youtube.com/watch?v=wv-tytn6GXQ&amp;feature=youtu.be

GhengisDhon
06-17-2015, 11:52 PM
Thorn vs Apophis (Switched Drivers)

https://www.youtube.com/watch?v=0MMIryMCQp4&amp;feature=youtu.be

GhengisDhon
06-17-2015, 11:52 PM
Nomad vs. Apophis

https://www.youtube.com/watch?v=c9SQdBusDXI&amp;feature=youtu.be

GhengisDhon
06-17-2015, 11:53 PM
Final Round

https://www.youtube.com/watch?v=5aAPvNRO4Gc&amp;feature=youtu.be

jwatte
06-18-2015, 11:00 AM
Looks like Nomad's the one to beat :-)

Does it work, on average, to step behind some cover and try to change the point of engagement? These matches look a bit like, once found, the bots just stand still and trade shots.

giantflaw
06-18-2015, 11:54 AM
We noticed that to. We are thinking of trying some different things next month. Here are some ideas we may try in July.

1.) Hit once and then run to another location to hit again and then repeat.
2.) Limiting rounds to 40 balls per machine.
3.) Use a spare robot as a stationary neutral arena guard to hit any bot seen. The arena guard could be taken out with 5 hits.
4.) GhengisDhon proposed 2 on 2 with only one shooter on each team. The other bot would be a runner. The two shooters would not shoot at each other, only at the runners. So the shooters would have to take out the runners.
5.) We've been lax in enforcing the (10 second move 1 body length) rule that happens at robogames. We could enforce that better with a referee.
6.) Melee practice
7.) 3 on 3

Next month we hope to have some improvement to our buildings. Nothing as good as robogames but better than just cardboard boxes.

jwatte
06-18-2015, 12:36 PM
5.) We've been lax in enforcing the (10 second move 1 body length) rule that happens at robogames. We could enforce that better with a referee.

That rule allows standing still when engaged in a firefight though.


Use a spare robot as a stationary neutral arena guard to hit any bot seen. The arena guard could be taken out with 5 hits.

If I could dream, I would have a neutral automated bot patrolling the outside, or perhaps stationary in the center, and shooting at anything it sees, This could be temporarily silenced for 30 seconds with, say, two hits. It would need target panels to be a known color (like hot pink.)
I can dream, can't I? :-)

GhengisDhon
06-18-2015, 09:29 PM
Looks like Nomad's the one to beat :-)

Does it work, on average, to step behind some cover and try to change the point of engagement? These matches look a bit like, once found, the bots just stand still and trade shots.

Ha! Is it that obvious from the videos that once you move out from behind the buildings it becomes a long range sniping fest? We've also found out, that once the sniping begins, if you move, you lose. Your opponent is racking up hits on you while you move, and your getting none on him. If you look at the later videos, you will notice we've tried to put obstacles in the way to force more movement. (Not very successful.)

As for Nomad, he's one of the veterans of the group, being to several Robogames. Think he has a bit more practice than the rest of us & his Bot has had more run time on it. So he is the one to beat right now. He did lose one of the earlier scrimmages, though. He had his gun fire rate set to low, and couldn't keep up with the sniping. :happy:

GhengisDhon
06-18-2015, 09:40 PM
If I could dream, I would have a neutral automated bot patrolling the outside, or perhaps stationary in the center, and shooting at anything it sees, This could be temporarily silenced for 30 seconds with, say, two hits. It would need target panels to be a known color (like hot pink.)
I can dream, can't I? :-)

Ya... we've had similar dreams. Ours are that its not an automated neutral bot though, buts its our real bot. One of the other members (bloftin - Ra's driver) has been experimenting with the idea and giantflaw has been playing with the pixy camera recently.

As for the stationary bot, we've been thinking a stationary automated turret with target panels.

So much to do... and work always gets in the way

tician
06-18-2015, 09:54 PM
I seem to remember the Ruleset additions including the possibility of host arena provided tanks (cannon fodder: 2~3 hits disables it; ~20 shots with ~5 second reload?) and turrets placed under audience control. I quite like the idea of matches involving 'capture point' turrets: at start of match turret will shoot anyone/everyone until 2~10 hits disables it for 20~30 seconds; if bot stands on capture point for 5~10 seconds during disabled state, then will capture/'repair' the turret and start it attacking only the opposing team; if (re-)disabled but not (re-)captured/'repaired' by a team, then will go omnicidal again.

RPi and 4~6S LiFePO4 in the base with RPi-camera with kogeto lens stationary mounted on a pole/tube in the center. Trossen 120mm slew bearing with diy 'slip ring'/'brushes' for powering two gearboxes and central laser pointer; continuous pan of turret achieved by printed ring gear attached to inner ring of bearing and driven by cheap stepper motor mounted in base; tilt achieved by tube/ring controlling height of barrels at one end of a pivot and driven by another cheap stepper motor mounted in base. If I manage, tomorrow, to make a functional adapter plate for the RPiCamera and kogeto so that I can actually get a usable FOV/resolution, then I might be able to get it to track multicolor 'team/bot tags' (40mm or 80mm square) with simple geometric design to differentiate between targets (sub-divided into 4x4 grid with corner sub-squares and surrounding border one fixed color for reference/tag-finding then have four center sub-squares a second color and eight edge sub-squares a third color for team/bot ID).

jwatte
06-19-2015, 10:55 AM
Agreed -- a roving bot doesn't have to be walking; it could be a tank.

Josh's super-omni-whatever-bot actually started learning how to move and fire at the same time towards the end of Boston Mech Warfare, which made it quite dangerous. I think perhaps evolution of the sport will drive us all to that point...

giantflaw
06-21-2015, 06:02 PM
I was talking to Jeff Thorn who built much of the R-Team MechWarrior robot Thorn. He was not at the scrimmage this month but had several other ideas to try in July's scrimmage to help liven up a battle.
1.) Remove the front target plate. This may end the long distance snipping fests and cause pilots go for side and rear shots. As soon as the target robot turned to face the shooter the shooter would move to find another side/ rear shot again.
2.) Dusk fighting, turn down the lights so the BBs can be seen. Right now in the brighter light only the pilots can see the BBs in the camera video. If the lights are turned down then everyone can see the BBs. This would make better audience appeal and better videos.
3.) Have stationary targets like cans or whatever that are hard to hit. If hit a health point could be gained.
4.) Have target plate hits cause a hit noise on the robot.

tician
06-21-2015, 06:32 PM
An alternative to having the front panel removed would be requiring a different target panel to be hit before the last hit panel (if front or sides) can be triggered again (or just a longer ~10s timeout for that panel?) (have rear panel exempted from that rule to discourage slow bots running away?). E.g. You sneak up and snipe an opponent from the side to get a hit, then you can get another hit against the front panel if the opponent turns to fire on you and they can get a hit on your front panel. Once they are facing each other and both have gotten their front panel hit, they both have move around to get any more hits. Teams would make it easier if they can isolate a single opponent and attack from two sides, but leaves them open to retaliation from opponent's teammate.

jwatte
06-21-2015, 08:58 PM
Have target plate hits cause a hit noise on the robot

Agreed! In Boston, Josh made the scoring system call out the bot being hit on each hit; very useful as second indication when busy watching the video for the control.

jpieper
06-21-2015, 09:27 PM
At most I'd think you would want a longer timeout. It wouldn't be that hard to just ensure that one side of your bot always faces the opponent given that most bots can turn much faster than they can move.

Deimos
06-21-2015, 11:18 PM
I think it could be interesting to remove the 1 hit/s cap on the panels and instead imposing a rate of fire limit on all mechs. My thinking being that getting in close and unloading in bursts would be more effecting (and exciting) than taking time to carefully aim (1 bb/s is pretty slow). That said, I've only been to one event so I don't really have all that much experience to go off of.

ArduTank
06-22-2015, 12:47 AM
Maybe rate caps mixed with higher HP counts? (Say, 100-200HP instead of twelve.) Drop opponent's HP to 0 to win like always.

But, say, you have a 4 rounds/sec cap, with a cooldown period after X seconds/rounds (Thinking weapon heat), to force you to run and hide to get out of your opponent's sights, who has not fired yet.

Less sniping due to cover requirements to not get blasted by multiple bursts of fire.

This could also feed into a classing system, ala X number of guns equals X class, set the rest from there.

Light mechs: high dps, faster, longer cooldown (fewer "heatsinks")

Medium: middle of the road

Heavy: lower DPS, faster cooldown (more "heatsinks") maybe an extra gun or two?? Slowest by far.

jwatte
06-22-2015, 10:27 AM
Rate caps are mechanically really hard to implement -- the airsoft AEGs aren't precision mechanical assemblies, different versions use different driving motors, and the driving speed also depends on the state of charge of the battery. (Not to mention: Some bots use 2S, others use 3S batteries.)

I think there's more life in the idea of a cool-down period. After 5 hits registered, no more hits will be registered for 10 seconds or whatever.