"How does it drive?!?!" Team 857's bot

[Adam Krajewski]

A picture is worth a thousand words and the video is well worth the lengthy download (~100MB), so check them out here. We have a second video showing our ball collection method, but I've run out of space on this server. I plan to DivX both videos in the next couple of days as to reduce the videos to a reasonable size.

I will try to explain the basics drivetrain here before anyone needs to ask. It is an omnidirectional holonomic platform. In English, it can drive in any direction (x, y translation) WHILE rotating on its axis, giving it three complete degrees of freedom. By using absolutely hideous equations (especially when dealing with PBASIC), you can drive each of the three drill motor (thank you FIRST for giving us three this year, it couldn't have worked out more perfectly) powered omni-casters, located 120 degrees apart, so that you go in the desired direction/rotation. For example, to drive forward at full speed, you drive the left and right motors at nearly full speed while the front wheel stays stationary, rolling sideways on the omni-caster.

Adam Krajewski on behalf of MI Roboworks, team 221 and Superior Roboworks, team 857

[Adrian Wong]

Hey, looks good.

Our team was going to go with the same design, except we used paired spherical wheels on the drive system. We ended up abandoning the design just prior to construction because we realized it didn't work for our gameplan very well.

Luckily though, we still have all our CAD drawings and specs, so we plan on building it in the off-season to see how the thing runs.

[Adam Krajewski]

The original prototype, which was started almost two years ago, also used paired spherical wheels. While more impressive to look at and even harder to figure out how it works, their weight and bulk made them an illogical choice when it came down to deciding on a drivetrain for competition.

Adam

[Mike Soukup]

Nice job 857. I've wanted to make a robot with that system and work on the controls for it just for fun. It looks like it would be a blast to program & drive around. Congrats!

Mike

[Kris Verdeyen]

Beautiful machine, 857.

What do you give your driver to work with? A trackball? Please tell me that you have the gyro on board there to let you do absolute steering - that would be the icing on the very very cool cake.

[ReijiH]

I remember a team from last year (or was it year before?) that had a control system which looked like a glove.....something like that would be cool with that kind of drive train.

[Scott England]

Nice, i'll admit, i looked at the pics of that bot in the gallery first and was extremely skeptical about how well that bot would be able to drive. The video definitely put me in my place, that bot dances more smoothly than any one i've ever seen in my 4 previous years of experience. Whether it works well in the competition or not remains to be seen, but it should at least be a hell of alot of fun to watch.

congrats to the rookie team from up north,

though hey, how are all the extreme veteran teams doing? Aside from team 60 i haven't seen many pics in the gallery showing what some of the perennials powerhouses are doing. I'm very anxious to see what 45,47,64,67,71,122,and 192 are doing (those are by no means all the great teams, just some of the ones that have built bots so impressive that i remember their team number)
Less than a week to go, not really any time to steal ideas. Good luck in the final push everyone
~Scott

[Ken Leung]

Unless I am wrong, you guys are the first one successfully implement this kind of drive train on a robot. Great Job! My old team have been trying to develope similar drive train, but it has always been prototypes before competition start. I can't believe I am actually going to see this drive train in action! Woohoo!

It should be interesting to see how your robot is going to dance around other robots, getting to any where you want, however path you want to use. Just do a quick spin, and you can go around any robot trying to push you.

The folks around the country just never stop amazes me how much magic trick they can pull out of their pocket.

Speaking of which, I am interested in calling out to the veteran teams and ask them about their robot. So I am going to start another thread for it.

[Wetzel]

NICE! And fair server speed too...averaged 84 and 62 dl'ing them at the same time.

Congratulations on what appears to be a great bot!

<Edit> Realized that you will be easy to push sideways...if they can catch you. What speeds do you get with the basket on it? </edit>

[Adrian Wong]

That's one of the problems that turned us away from the platform. It wasn't until last week (aaaah!) that we completely shifted designs. It's kind of a shame that we put 4 weeks of work into it.

The thing that played a large role in us abandoning the project was just that problem: side loads. Due to the holonomic nature of the platform, if the robot is pushed perpendicular to the the direction you're moving in, it can easily shift. The wheels were designed with that in mind (to be able to roll freely), but with another intention in mind (to be able to move holonomically).

Since we never actually managed to build our design, we haven't had the chance to try this out in real life. How well does the robot react to being pushed at 90 degree angles to its motion? (Just to validate a few concerns on our team. )

[Kris Verdeyen]

Originally posted by Adrian Wong
How well does the robot react to being pushed at 90 degree angles to its motion? (Just to validate a few concerns on our team.

Seems to me that it would behave in the same way as it would being pushed in any direction. The robot doesn't care which way it's facing - if it were to be pushed against, the driver (or software) could just compensate by increasing the force in that dimension - then it just becomes a question of traction.

That video is incredible, btw. And I still want to see the controls!

[Adrian Wong]

Well, the argument during our design sessions went something like this:

A "normal" two wheel or four wheel drive system is non-holonomic. It goes in one dimension (forward/backward) with the ability to change its direction while in motion. It is also possible to do an in-place turn, but there's lots of scrubbing going on with the wheels.

However, the scrubbing is also a feature. If a robot were to be pushed at 90 degrees to its motion (left/right in an front/back movement), it would be difficult to push due to the scrubbing of the wheels.

Yet, the omni-drive design intrinsically has the capability to roll freely in *any* direction. So, it would be susceptible to attacks from the side.

[Justin 188]

*watches the video*
*chases after his brain*

OMG that is beautiful

Good job guys!!

[Adam Krajewski]

We've got a second video uploaded showing how we load balls into the bot. Also, thanks to Xobes, the webpage is Netscape compatible (At least on Linux). Mary reformatted the first video so it's a reasonable size (8MB), too.

As for the controls, we kicked around many ideas. We wanted a controller that would allow the driver to use the entire range of motion the drivetrain is capable of. The original Killough specs calls for rotation sensors to determine the velocity of each wheel. We found a way around it. We did not use the gyro; we found no need for it. The control system we eventually stumbled upon on doesn't require feedback control, is incredibly simple, and is our little secret for the time being.

Adam

[Adam Krajewski]

I don't see a problem with being pushed. It doesn't exactly slide freely; you're still pushing against the drive motors on at least two sides, if not three. Also, the omnidirectional nature allows you to push against another robot no mater which side is being pushed against. Traction is the biggest issue in pushing. We aren't too worried about being over-powered. We aren't pulling goals, our basket empties very quickly, and we are incredibly maneuverable, so we should be able to slide out of any sticky situations.

Adam