PDA

View Full Version : Mech Warfare 2009 Competitor Recap Thread



lnxfergy
06-16-2009, 01:00 AM
Ok guys, as we rap up our first round of combat, and while it's still all fresh in our heads, lets recap what we liked, disliked, and what ideas we might have for improvements. Also, what worked/didn't work on your individual mechs. I'll go first:

Event stuff
1) My biggest gripe was our scoring system. The guy who build that piece of crap should be shot. Oh wait. That was me! :sad: Anyways, the entire system wasn't tested enough, since I didn't have a mech when I was designing it -- that problem has been solved. The scoring system still needs some work:


In future years, I'll be bringing one of my computers to run it on, since neither of Andrew's seem to be happy with anything.
The wireless communication part works pretty good, we still need to increase the reliability of the packet structure some... I've been playing with new checksumming methods in my head the whole flight home.
The scoring panels themselves seem to be the most trouble. By the time they are stiff enough to not go off when walking, they really don't detect our low FPS guns too well. Andrew and I discussed possibly switching to using force sensors on the plates rather than piezo vibration sensors, since that is what we really want to measure. I've got quite a few here, that I'll be prototyping stuff with later this week (while I wait for my bots to make it home...). Personally, I'd also like to see us standardize these in some way that everyone is just getting finished panels from Andrew or I, and that way they are all completely the same -- from discussions at the games, I think that most competitors were onboard with this idea.

Overall, I'll be sure we have any updates to the scoring system done very early (hopefully end of the summer, but certainly before December, so that all competitors have at least 6 months to integrate any changes into thier bots.

2) The Arena -- I'd like to see more cover. There was an exhibition match between Squidword and Issy on Saturday night that was lots of fun, cause I tried to shoot Che, so he rearranged the boxes to block him and Squidword from Issy. The result when we got started was that the bots were able to snipe at each other some... and then round the corner and fight more (Squidword fell over before the fight more really happened much). We also need to make sure we have a bigger "kid window" cause we had lots of kids who wanted to see the action. Anybody know a sponser who wants to donate like 12 feet of 4' tall lexan for a front wall? :veryhappy: Also, just so nobody forgets, we need a real door next year, that velcros shut... with the back corner being opened over and over again, we kept pulling the whole side wall open, which meant lots of BBs ending up in the hallway.

3) The Video Team - Adrenalynn and her awesome helper shot tons of amazing video of our matches. A big round of applause to both! We had a quick discussion on Sunday about possibly getting an aerial camera view (mainly for the crowd to see)... ala GoodYear Blimp style... I believe the discussion quickly turned into us wanting a real blimp in the sky over mechwars... hmmm.. just a note here to remember the idea for later.

4) The Roomba - was invaluable in cleaning up. Thanks to Connor for bringing him!

5) Match Length - of course an area of big debate. A few observations on my part. Gary and I had one run that was about 18 minutes long.. but we stopped to reload at one point. I found that on Issy, a single tank gun, with a single Tic-Tac hopper won't last you that long.. I tended to run out after 8-12 minutes, and only getting in about 10-15 actual hits... my targeting system stunk though. Hoppers like Manny's would probably carry quite a few more shots. I think a perfect match length might be more like 15HP or 12 minutes... especially if our mechs can find a bit more speed next year (and if I can get my targeting to work better). I have a feeling battery life is an issue, but servo heat might be even be more of an issue for our bipeds.

6) Match Schedule - next year, when all of our bots will be ready for combat, we should definately have a posted schedule right at our booth of when matches will be. It will help us to keep track of time, help the audience not to miss the matches, and also make sure I don't have to keep saying "soon" everytime a kid asks "when are the bots gonna shoot each other!"

Issydunnyet
Issy was pretty fast two weeks ago.. then I put all the stinking mech gear on him and had to slow him way down to handle the load. A few things that really worked:


Bioloid servos and brackets. I put Issy together in an evening, and had him under XBEE control in about 4 days time. I would never have pulled off such a feat using any other servo system. The bracket system also came in handy when I reconfigured Issy to a 2DOF quad at noon on Saturday.
KISS - 2 DOF was far more reliable than 3DOF. My battery life almost doubled, and none of my servos broke 65C on Sunday (they had been in the 100C on Saturday)
Checksums! Quite a few people had trouble with glitching... Issy really didn't. A simple packet structure and checksum is essential to reliable operation in a crazy environment like RoboGames.
Trendnet camera. The jury is in. You just can't beat the cost/reliability/performance combo that the Trendnet Wifi cam has. I found that I could run the cam off a 1400mAh 6V pack for about 40-60 minutes (basically, two matches plus startup/shutdown time). The Jury, however, would like to know of any new cameras that come out that are smaller and lighter....
Having a lazer on your bot can help with targeting. But it also just looks cool. And, the audience knows what you are shooting at, since it's crazy hard to see the BBs flying.

And a few things that stunk:


My tank gun kept jamming. I used a hacked up knock-off of Gdubb's hoppers, I think I missed some of the subtleties of the design. That said, the range/fire-rate/weight/cost combo of these guns is phenomenal, if you can get your hoppers working. Accurate targeting will win the day.
It's really tough not to run into buildings, and it takes quite a bit of driving practice to get used to your bot's walking (to train yourself to compensate for camera lag/dropout). A finished MechDAR would probably help, but I'm weary of more payload or wireless data transmission. Maybe a yellow-dotted-line down the center of our "roadways" would add both realism, and a slight aid in driving?

I'll be posting a full build tutorial on Issy once he gets home (so I can disassemble him and take photos of each step).

-Fergs

Edit: Holy crud... didn't realize I was that long winded until after I hit Submit... Also... please let each of our competitors put in their full thoughts here at the beginning of the thread before this turns into a bloated discussion.

mannyr7
06-16-2009, 05:16 AM
Event stuff
1). I built my scoring panels using Gary's design and judging from the led they seem to work fine. When I finally got Clyde walking, the panel wouldn't register a hit from footfalls, but the slightest tap from center to end would light it up. I am all in favor of standardization though and think force sensors on the plates might be the way to go.

2) The Arena -- I'd like to see more realism. Assuming I drive again next year, it might be possible for me to build and haul a proper arena that is more immersive for the competitors and view-friendly for the spectators. I'll be sure to integrate a real door. As for navigation, I envision a 10'x10' linolium sheet with painted on asphalt and scale road lines. It would not be too much trouble to create building fronts that could be transported to the event and hot glued to a standarized cardboard box.

3) The Video Team - I agree, Adrenalynn and her helper did a fantastic job. I loved the multiple camera angles. Maybe we can alternate in HUD views as well?

4) Match Length - I never got to fire a shot in battle, but my hopper design allowed about 175 rounds per gun. As for my camera, the Trendnet gives me 25 minutes with 4x AA 2500 NiMH's.

Clyde
Where to start!?!


Before Andrew beats me over the head with my Nicads, it's time to upgrade my battery technology.
Always carry an extra controller board!
I should make up a troubleshooting checklist so I won't forget to check out the servo ports for failure next time.
I am pushing the 645MG's too hard at 7.2V. Need to upgrade to 5990TG's all around.
Insulate electrical contacts that could get accidently shorted.

-Manny

gdubb2
06-16-2009, 11:10 AM
Event stuff:
I really don't have any issues with the event as a whole. True, the targeting didn't work very well, but we had to start somewhere.


Arena: A better audience viewing area would help. The netting worked well, but the solid bottom was hard to see through :happy:.


Video Team: Was great, Jodie and team did an exceptional job. I liked the projected display.


Match length: I would like them a little shorter, I think the audience gets bored easily, and with slower robots, that adds to the match dragging out.


Bheka:


1: loose the Airlink camera, It's a piece of crap. :sad:


2: Speed.. got to have more speed. Having Issy dance around you like you don't exist is embarassing. :o


3: Different controller, The Hitec controller is just too limiting. (have to learn how to program now though)


4: Better guns (or gun)


5: Recap... complete redesign, Faster.. smoother..gyro stabilized..(dreaming now)

Adrenalynn
06-16-2009, 12:51 PM
Thanks for all the kind words! My "assistant" was Steve Haynes. A damned good mechanical engineer in his own right, btw.

You all saw the video stuff working relatively smoothly, but we saw our own internal flaws. We were "fortunate" enough to be driving back to my place every night, and as you probably noticed - the equipment on the table grew every day. ;)

Changes I propose for next year on the video side:

I'd like to take over the onboard bot video wireless networking. I've been thinking about the architecture and interference on 802.11* and have some suggestions there.

The basic architecture:

A commercial-grade Access Point on the ceiling. The access point has a circular polarized patch antenna pointed down. I'll spec a flat patch antenna (about the size of a target plate and inexpensive) that is circular polarized and pointing up. We'll be using a different polarization from everyone else which gives massive signal rejection. Also, the patches are directional up and down, so side rejection on signals will be in the 60-90dB range. We can throw a few dB of amplification on the 'bot with just the antenna, and probably ~30dB on the ceiling. We'll just overpower any noise. The A/P will come out over fiber (3gbit backplane) to an HP switch with a 33gbit/sec backplane. Each port on the HP is 10/100/1000 ethernet. You guys plug your machines into the HP and it routes to the access point. Having multiple routers stacked on each other, we were probably jamming ourselves as much as anything...
This also resolves some issues with antennas that stick up being in the way, and it addresses two more potential issues: I suspect your routers were shooting over the top of the 'bots, and I also suspect that having that antenna just a few inches above that rubberized concrete floor was contributing to a LOT of microwave ground absorbtion. This solves both those.

Pure video-wise: The street level cameras were CLEAR winners. The overview camera was something everyone also seemed fond of. Better low-light security cameras are a requirement for next year. I'll stick with wired cameras because they didn't interfere and performed so well. I'd like to add a PTZ camera over the top of the arena to follow the action, especially as buildings become denser. There was a suggestion (Fergs' dad maybe?) for foilage (plants working as trees) that I thought was awesome. I was recording and switching four channels of video - that wasn't enough. I need to go at least 8, maybe 16 for next year, and bring the HUDs in to that mix. I'd also like to fix a camera watching the bot operators. Wish I'd thought of it at the time.

Our production values were pretty low by our standards. We'll definitely want to bump 'em up a notch or two next year.

I'd also like Tybs and Fergs to consider a live webcast, at least of bracket matches. Or if not a live webcast, a delayed webcast (produce and upload that evening). I think this is doable.

The projection viewable from both sides of the box was great (Thanks Ooops and Steve for working that out!), combination of forward and reverse projection. Thanks to Manny for the loan of the projector! That said, I think we need to shield it with an awning on top to block the overhead lighting from blowing it out. That was totally MacGyvered this year, we had no _idea_ what the environment would be, or even what projector we'd have (or if we even would). Knowing what we know now, we can do a *much* better job.

Audio: I'm not even sure what to do with that... Open to suggestions. The volume coming from the battle bots was so insane, even a two element shotgun was just blowing out from the side. And there's not really much audio to be had anyway.

Overall, I think the video team performed at an acceptable level. Plenty of room to step this up next year, and increase the reliability (and enjoyment!) of the competitors at the same time. I'm open to any and all suggestions, here or via PM/email.

We tested the non-802.11 2.4 and 5.6 cameras and they were (as predicted) entirely worthless. Anyone even THINKING they're going to go that route needs to just hush-up, suck it up, and buy a decent IP camera. NOTHING else will work. We were testing with commercial grade VideoComm rx/tx pairs with real antennas and more weight than the bots could hope to carry - and those didn't work either, even with an $800+ price tag. In fact, I propose the event organizers, Tybs and Fergs, ban those entirely as they increase the overall noise floor and would contribute to jamming the legit IP cameras and XBee control systems.


Thanks to Steve, Ooops, and Manny for their contributions - with a big thanks to Steve who worked tirelessly, cleverly, and still managed to make the ~4hr a day commute every day!

nbdeveloper
06-16-2009, 12:54 PM
Hmmm, live webcast...

http://bbs.chinadaily.com.cn/attachments/month_0801/drooling_homer_MzY5CNUnEpK4.gif

ooops
06-16-2009, 01:49 PM
A spectators two cents:
On the scoring system - Fergs, with you taking possession of that I am confident it will be great. I would suggest that it be displayed where spectators can see it. That would go a long way for generating/keeping interest in the Mech-wars. A further thought along those lines would be some sort of "sound effects" for hits/kills. I know that seems silly, but interest at Robogames seems to be somewhat "sound driven", so if you build it noisy they will come!

On the arena - The arena as it was worked. But very "unfriendly" for spectators. I assume there will be a push in future years to "go lexan". But the mesh at the top of the arena worked pretty well for protection, and allowed a pretty good view. With some creative PVC "plumbing" and that mesh a very lightweight, very portable arena could be fabricated with a door and everything. Trees and bigger buildings would be cool - but only if it can be seen.

On the Video - The video like the event was pretty much a trial by fire sorta thing. I was truly impressed that Mechwars was the only event that I saw with any overhead video. TRC - Blazing the trail - Great job video crew:)
With a year to work on it, seems as though it wouldn't be difficult to incorporate the screen (or better yet screens) into the arena set up. Maybe beg borrow or steal two projectors (or more) to give scoring, arena views, feeds from the bots (HUD) and more.

Hopefully everyone reads this as positively as I typed it. From my point of view Mechwars was a huge success! I will be back next year, but this time to kick some Mech butt!
Great job everyone it was fun just hanging out with you!

mannyr7
06-16-2009, 02:14 PM
Would it be possible to incorporate a text overlay of the competitor/mech's name, their score and the match countdown timer in the center, like modern fighting games? Maybe not as extravagant as this, but you get the general idea.
http://lh3.ggpht.com/_RBehtyGR3tY/SjfufLjvPMI/AAAAAAAAAZI/goldVCLdhj4/scoring.jpg

tom_chang79
06-16-2009, 02:18 PM
On the arena - The arena as it was worked. But very "unfriendly" for spectators. I assume there will be a push in future years to "go lexan". But the mesh at the top of the arena worked pretty well for protection, and allowed a pretty good view. With some creative PVC "plumbing" and that mesh a very lightweight, very portable arena could be fabricated with a door and everything. Trees and bigger buildings would be cool - but only if it can be seen.


My $0.02 would be to implant some cameras where the perspective makes the bots look as if it's 30-foot tall or something like that. But yeah, a clear Lexan would be cool!!!

Another cool thing would be a hit sensors around the arena itself, on the mock up buildings and trees. Once it registers a certain amount of hits (say two or three), maybe we can implant some firecrackers and pyro it! Hmm, the local fire department might not like that idea...

:D

ooops
06-16-2009, 02:27 PM
Would it be possible to incorporate a text overlay of the competitor/mech's name, their score and the match countdown timer in the center, like modern fighting games? Maybe not as extravagant as this, but you get the general idea.



Anything/everything is possible when Adrenalynn is on the job!

darkback2
06-16-2009, 02:33 PM
Ok...a couple of things.

First and formost, thank you to Andrew, and everyone else who made mechwarfare a reality. It was really fun to get to see everyone in the flesh and all, and without all of the hard work Andrew put into it we wouldn't have had much of anythingto do.

Pros: First of all the attitudes of everybody there. This was the first year of the event, and everyone showed up looking to be helpful, and willing to just have a good time.

The arena worked. While future iterations could have more windows, the canvas and mesh both held up really well, and did their job of keeping people watching safe. Lets all remember Andrew payed for the arena out of pocket. Over the summer I'll get out my sewing machine and try to stich something up with pretty much the same design, only I'll try to add more windows, and make the wall shorter...and try to include a velcro flap for a door. Again, the arena worked...

Cons...

WIFI...There was just too much noise for anything else. I proposed we use a really powerful router, and everyone is assigned an IP address from the start. Xbee seamed to work relatively well, though there were some lost packets. Anything else...don't bother. I also propose we have a rule that noone can have on wireless devices if they are not actively competing...well...anything that cuts down on noise.

As for making the city more realistic, I've offered to take pictures of store fronts in Los Angeles NY, and Seattle. I"ll print them out large enough to cover the sides of the box, and try to get them printed on hard cardstock. That way we can velcro them to the front of ordinary cardboard boxes.

As for the streets...I'm not really sure that matters. I guess for now I think we have enough things to worry about...and paintlines could be solved with yellow duct tape.

OK...again, thanks for a good time Andrew...also Adren Steve, and everyone else who helped.

Adrenalynn
06-16-2009, 04:51 PM
I keep hearing "assigned an IP address" [not just from you, but several time] and I'm just wondering what possible difference that could make? There were no IP address conflicts, and by design, couldn't be. It won't help interference or speed in any way that I can see.

What is it that I'm missing?

Adrenalynn
06-16-2009, 05:00 PM
I had a titler going at one point, Manny, with timer and mech names. We stopped producing live as much the second day. In the future, I could see projecting the scoreboard up, absolutely, which had the names and hitpoints - thanks to Fergs' genius

darkback2
06-16-2009, 06:04 PM
I keep hearing "assigned an IP address" [not just from you, but several time] and I'm just wondering what possible difference that could make? There were no IP address conflicts, and by design, couldn't be. It won't help interference or speed in any way that I can see.

What is it that I'm missing?

I mean that so that we can use one router...and given that I communicate with my bots/others look for their cameras based on IP addresses, in that way we would avoid having one more thing that could go wrong as we switch over for a new match.

For example, I use two PCs...one talks to the other over IP...So in my software I have to set the IP address to send data to, as well as the one from which to look for the video data. I'm hoping to switch over to the Trend net camera for next year...but it would be good if all of that was set up ahead of time...so that we wouldn't get the "there is already a device with that IP address error...which I got several times when starting up both using connor's router, and my own.

This might also have to do with the fact that neither router was password protected...so maybe that would be enough, but not having to reset IP addresses before a match would be greatly helpful for me.

Oh...and thanks for bringing all of the video gear, as well as putting so much work into shooting video!

DB

jes1510
06-16-2009, 06:08 PM
If a static IP addy is needed then that is as simple as restricting the DHCP range on the router.

Adrenalynn
06-16-2009, 06:08 PM
My pleasure!

So what you really need is a dynamic domain name server (DDNS). Assigning fixed IPs is almost always a nightmare since someone is going to accidentally [invariably] forget to reassign the IP and you end-up knocking your router off the network. ;)

DDNS addresses that rather robustly.

Testing2000
06-16-2009, 06:25 PM
I totally agree with Adrenalyn on the noise level from the main arena, it was bordering on painful to me at one point but I have hearing issues. Perhaps a request to be as far away as possible tho you may not get a choice. Many thanks to Trossen Robotics for the pass Saturday evening which allowed me to roam the pits of despair and destruction. I also saw a great many youngsters trying to see so perhaps a mesh down to a bit above 18" so the bot cameras have a consistant background. It was a pleasure to meet everyone and share a bit of your enthusiasm.

Perhaps Chicago,
Best regards,

Testing

Connor
06-17-2009, 01:24 AM
My router is capable of assigning the same IP to the same device via MAC address. Also, My router is capable of up too three independent Wini Cards and has Three 10/100 network ports. It uses a 800MHz ARM processor and is what I would call a professional grade Access Point/Router

It's late and I'm about to pass-out, just got in from my flight.. I'll talk more later.

Thanks, Connor

lnxfergy
06-17-2009, 03:11 PM
Ok, thanks to everyone for the great ideas. Keep 'em coming! A few of my thoughts in response to ideas above:

Moer realizm: I totally agree. I don't think we need to put a floor down, the floor is already nice and black, I'll have rolls of white/yellow electrical tape next year to mark roadways down. I'm also envisioning a standard building design that can be folded down and travel-sized. I'm going to be building an arena setup (minus outer walls) in my garage this summer for testing. My plan is that those buildings could be transported to RoboGames (an any other events we have). I'm also a big fan of trees and maybe a small astroturf "park" that makes for an interesting change in ground (it would be something that wouldn't be overly large, as some mechs might not be able to cross it, but it would be really cool).

As for the enclosure, what we had worked great, I'd just like to see the entire front wall be clear in the bottom for the kiddos). Also, what do other drivers think, I didn't really even have a hard time with the mesh.. the video pixelated enough on me that the mesh looked nearly solid....? I almost think that we could get away with all mesh (or just the 18" of solid like Testing recommended).

Live Webcast: Andrew and I had a wireless link onsite so we could download stuff related to the scoring system. Let's just say, the link wasn't fast enough to view youtube, so I can imagine a live broadcast would be impossible unless we can tie into a hardwire back plane somewhere. A delayed upload might work (we just need to find somewhere with fast internet... my link at the Marriot wasn't so great, although I guess if Jodie is commuting again, she might be able to upload).

Scoring System: If I can get my hands on an LCD (or another projector...) that we could like hang from the rafters or the arena frame, we could put the scoring display right up on there (it displays mech names, scores, timer, and a ticker tape that tells whats going on...). I think we'll make an LED from the scoring system mandatory next year (it helps everyone in debugging, gives the crowd feedback, and is also a back up should the scoring computer crash). I'll also be working with Jodie to see how we can send out audio feedback (explosion noises when a mech is hit).

Wireless: There will always be lost packets. It's how you deal with them that sets you apart. I have a tutorial coming up in a day or two that will cover XBEE usage, and focus also on designing protocols that are noise immune.

-Fergs

societyofrobots
07-04-2009, 09:56 AM
Whats this 'Trend net camera' everyone is talking about? A google search reveals a half dozen of them . . . All the good cameras I can find are too big/heavy for a biped . . . and the 1.2GHz are so cheap, but probably won't work with all that interference. Anyone used a 1.2GHz?

I'm still considering doing this next year, so I'd be interested in finding out what you guys decide to do for targets. It should probably come as a complete kit, where you can throw it on your robot within ~20 min and it'll 'just work' . . . also nice for it to pulse a digital pin when a hit occurs so we can interface it with our own controllers (display hit counts on our own laptops, etc).

lnxfergy
07-04-2009, 10:45 AM
Whats this 'Trend net camera' everyone is talking about? A google search reveals a half dozen of them . . . All the good cameras I can find are too big/heavy for a biped . . . and the 1.2GHz are so cheap, but probably won't work with all that interference. Anyone used a 1.2GHz?

I'm still considering doing this next year, so I'd be interested in finding out what you guys decide to do for targets. It should probably come as a complete kit, where you can throw it on your robot within ~20 min and it'll 'just work' . . . also nice for it to pulse a digital pin when a hit occurs so we can interface it with our own controllers (display hit counts on our own laptops, etc).

See this thread for details on the camera and a link to amazon: http://forums.trossenrobotics.com/showthread.php?t=2745

-Fergs

Adrenalynn
07-04-2009, 11:22 AM
1.2Ghz is law-enforcement band. Check Parts 92 and 95 before even considering it.

dcalkins
07-04-2009, 02:09 PM
Anybody know a sponser who wants to donate like 12 feet of 4' tall lexan for a front wall?

My offer still stands. We own like 4000 square feet of the stuff. Or just roll-the mosquito netting around the lower half as well as the upper. But my saftey-nazi side would prefer you just use my spare lexan. But RoboGames will help pay next year, so poor andrew desn't spend all his kids' college money.

Arena innards: TechShop has a styrofoam CNC. With the prep time and cad files, you could make 3D styrofoam bldgs, and paint them. They can do fairly complex stuff in terms of detailing. Also, we own a large format printer. we can print 44" by 8'. I like Andrew's idea about adding trees.

WiFI: It sucks in the bldg. It'll never get better. Sorry I can't offer more advice (athough anyone using X-Bee has no problems...)

Noise: ComBots are loud. the dB isn't much different anywhere in the bldg. You just have to deal with it.

lnxfergy
07-04-2009, 04:26 PM
Dave,

Wasn't aware of the lexan offer, but I totally agree with your safety considerations to use lexan over netting for that bottom section.

Thanks again for hosting such a great venue! Had a fabulous time.

-Fergs