Sumo Bots C (NY Trial)

[gh]

So, now that we know Sumo Bots hasn't changed significantly, what ideas have you guys been simmering on for the last year? You may recall me as the driver for the three years for Stuy, which has sent the following designs to competition:

• 2007 (regionals) - A sheet aluminum piece of junk that was duct taped to the floor and had an aluminum bar as a lifting arm. And we lost. It brought shame to its builders' families.
• 2007 (states) - A Lexan and aluminum wedge bot with a green top. That thing had great motors and wheels (instant acceleration!), but the electronics were iffy, and failed during a match against our rival team. Everyone got pwned by the black pyramid thing built by Union-Endicott.
• 2008 (regionals and states) - A plastic pyramid bot with four black wheels and a stack of fans to suck down against the floor. We got sabotaged during regionals impound (!!!). At states, we beat Union-Endicott (:o), and then lost to a bot that was taped to the floor (:o). Delicious irony.

I've been the driver for most of that time, (but I only built the 2008 bot), so I've been through a lot with Sumo, and I have a lot of ideas about robot design. Some of these are not really that obvious at first. Anyways, I figured I'd share some of the things I learned this year, which I'll be using on my next design:

• Using more than two wheels: A lot of people use only two wheels and let the scoop rest on the surface of the field. This is not really a good idea. The field is never going to be totally smooth, so the scoop will want to bounce about the axis of the wheels. Also, a lot of your mass will be resting on the scoop rather than on the wheels, reducing your traction. Having more than two wheels allow the mass to rest on the wheels.
• Use a low, sharp scoop: Our 2007 bot used a 40 degree Lexan scoop sharpened by hand with a file and sandpaper. The bot bounced too much, so it was pretty bad at getting under other bots. The 2008 bot had a similar scoop, but it had four wheels, and it was angled 30 degrees. It was able to get under pretty much every robot I faced. However, my kinematics calculations were based on erroneous motor data, so I was not able to actually push once I got under.
• Custom made wheels: The 51:1 550-size gearmotors we used in the 2008 states bot were phenomenal. However, the two motors and two wheels lined up exceed the 30 cm width limit added last year. So, I will be making my own wheels, that do not extend out from the shaft of the motor, but instead wrap over the gearhead. This is a good idea if you want to a set of BaneBots motors but are afraid that they are too large.
• Reducing the moment of inertia: keeping all of the mass near the robot's center of rotation will allow it to attain higher angular acceleration, and thus maneuver better.

I also have some more ideas about tires and center of gravity versus center of rotation, but I'll probably keep those to myself until they are tested.

So, what are your ideas? Taking a trial event seriously is an easy way to win medals, guys... a good thing in NY, where they count towards your team rank. ;)

[Sheogorath]

I'll be doing it again for the third year. I'm pretty disappointed that they didn't change any of the size rules so that our new bot could be different as well as others. But I'm glad they added more time for changing crystals, me and my partner were timing ourselves last year before regionals and states so that we would have time as our receiver is small. I have a few ideas to change, but mostly I'm going to start much earlier, last year we got a late start because of a parts ordering error. That way we will have plenty of time to perfect it. Although I still don't quite like how the whole thing is set up, because last year at states there were teams in the finals who we could have beaten hands down but because we got beaten by like one of the best in the first round we got ranked at like the halfway point. And in the first round we were going against a team and their front wedge got caught on the black tape, as in their entire bot was off of the wood except for like an incredibly small part was caught, not even on the visible wood, and the judges wouldn't call it, and we got stuck so we just had to wait until time ran out, we won because we were lighter even though we should have won fair and square. At regionals we also ran into a problem where the opponent got stuck on the tape putting them at an advantage, i.e. it's harder to push them, so they should try and fix something with that.

[gh]

Haha, that was a big problem at my section during states. We had a bunch of matches where one obviously pushed the other bot out, but both get stuck on the corner, and the winning robot would lose because it was heavier. In Japanese robot sumo, robots lose as soon as they hit the outside of the competition area. The competition area would be raised off of the floor like it was at states last year. Also, in Japanese sumo, the area is round, not square, so you don't have corners to get stuck at. You're allowed a section of the circle to place your robot, which gives a lot of strategic freedom (allowing robots to try to get behind each other instead of getting into head on matches). There was one match where I actually pushed the other robot onto the other guy's shoe (he was standing really close), resulting in both our bots getting stuck.

Anyways, to fix the problem, at least one of these changes would have to be made:

1. Don't make the area raised (then it's kind of boring)
2. Make the playing area non-square (hard to set up for the event sup at tournaments)
3. Have the robots start at opposite sides rather than opposite corners (probably the easiest to do)

As for the tape thing, I didn't encounter that at all this year; Metro region bots didn't have very good scoops. ;) Heck, most of the robots in the state didn't have good scoops. The rest of us have our work cut out for us. :D

[JBoyd-NY]

There are two rules changes to try to fix the problem of bots getting stuck when they are partly off and partly on the raised ring (or getting caught on the tape on the edge of the raised ring):

1. The definition of "out of square" has been changed to state a bot is out of the square when it is no longer touching the tape OR when it is completely off the ring if a raised ring is used. This change was made to allow the event supervisor who is using a raised ring to do so without having to apply tape to the edges of the ring - no tape, no problems with bots getting stuck on it.

2. On flat rings (or raised rings where the event supervisor decides to go with tape on the edges anyway): Rule #10 under competition was added so that once the two bots become entangled (or stuck) so that neither bot moves for a period of 10 seconds, time is called and the teams place their bots back at the starting positions and competition resumes.

Rule 10 was added to avoid teams with light bots winning over obviously superior bots because they got hung up on the edge of the ring and the superior (but heavier) bot was unable to get them completely off the ring. With last year's rules, the team with the lighter bot simply had to hope that they were hung up so badly that time ran out before the better bot could force them out. With the rule change, if a bot gets hung up on the edge so that it can't move but is stuck so bad that their opponent can't move either while trying to force them out, the judges will stop the clock, the bots will be disentangled and placed back at the starting point, and the bout will resume.

One other suggestion: at last year's State competition there were a number of bouts where a bot would get hung up on the edge of the ring. Their opponent would continue trying to push them out, without any luck. During one of these bouts, one team member suggested that his partner, who was controlling the bot, back up and hit their opponent with more force to get it completely off the ring. The student running the bot wouldn't do so because he was afraid that his opponent would get back onto the ring if he backed off. What he failed to see was that all four of his opponent's wheels were in the air - that bot wasn't moving, whether he was pushing it or not. Had he backed off and made a run at the bot from a distance, his added momentum would probably have supplied enough force to finish removing his opponent from the ring.

Teams need to practice, and try all possible scenarios (including having your opponents bot get hung up on the edge of the raised ring), so that you have a plan of attack for whatever situation you run into.

[sachleen]

do you guys build multiple bots and have them fight or what?

[gh]

Tempting, but no. Each team fields one robot. It's not supposed to be destructive either.

I never noticed those rule changes. I think it covers most of my concerns about getting stuck, but I feel that most matches will end up being head-on pushing matches, since both bots start very defensively, backed up against the corners.

Anyways, I was reading a paper about sumo as I was CADing (my obsession grows...). It was about evolving tactics against various opponents using genetic programming. One of these tactics, used against heavier opponents, actually match up to Jim's idea about getting momentum by backing up and ramming. Here is the graphic from the paper, with time-lapse paths of the optimal tactics against various opponents:

The paper is titled "Multi-Phase Sumo Maneuver Learning," but aside from that image, it doesn't really apply to the SO Sumo Bots events all that much. But hey, science is science.

[Sheogorath]

do you guys build multiple bots and have them fight or what?

Well I have done the even for the past two years so we have two sumobots, they changed the size requirement last year so one is too large, but we practice with that along with just stuff that ways about the same and has really good traction to give our bot a workout.

[zscioly]

Gh, what software did you use to make the model here: http://www.stuytech.com/ghf_images/atea ... hoover.png

?

It would really help me if you could tell me, since I want to make a model as well.

[gh]

I used SolidWorks, zscioly. I got a student edition of it for free at a robotics competition. I do designs in CAD since I need to making drawings for parts that need to be cut and I'm bad at drawing.

[zscioly]

Thanks a lot, this is perfect for what I need.