Chief Delphi

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)
-   Technical Discussion (http://www.chiefdelphi.com/forums/forumdisplay.php?f=22)
-   -   Omnidirectional drivetrains (http://www.chiefdelphi.com/forums/showthread.php?t=14599)

patrickrd 30-09-2002 16:39

Omnidirectional drivetrains
 
I know there was at minimum one team who applied this technology on their own last year (my apologies for forgetting who you are as it was quite impressive!)... but this type of drive system should be of common knowledge in any robotics competition as it provides 3 complete degrees of freedom: x-translation, y-translation, and rotation, all simultaneously and independently controlled. It's existed in industrial applications for quite some time and in other robotics competitions such as robocup for several years.
Last year in order to both (a) pass my classes and (b) participate in FIRST simultaneously, I used a required project in my robotics class to research how this drive system might be installed on an approximately FIRST-sized robot with approximate FIRST rules and material. Please note I worked on a four wheel version, which deviates from the classic three wheels on most omnidrive bots. I did this since FIRST robots have a history of falling over and three wheels could be detrimentally unstable with a high center of mass and high maneuverability. The major roadblock for *actually* using this design is that you would have to use two different sets of motors such that the wheels rotate with the same angular velocities for a given torque... My design uses four identical drill motors, which has classicly been illegal according to FIRST rules.

Anyway, I've posted the paper online (link is below) which should give any of your teams a good understanding of how it works and what you need to do to get a prototype working of some sort. I'd love to see some teams apply this technology successfully! If you have any questions feel free to email me at prd8@cornell.edu as I do not check these forums regularly.

http://www.people.cornell.edu/pages/...bot/report.htm

Good luck to all teams in 2003!


Elgin Clock 30-09-2002 16:56

We were one of the teams who used this design this year.

READ THIS THREAD

AND LOOK AT THIS PICTURE

It worked very well for us this year to get out of some sticky situations when backed into a corner!!

Madison 30-09-2002 20:23

Well, I think there is still an advantage to what Patrick's proposed over what 237 did. Patrick's omnidirectional system has complete freedom over its range of motion. On your robot, Elgin, you can't use all four wheels to move at, say, a 45 degree angle to the robots axis, can you? Patrick's can power all 4 motors at once to move.

Amber H. 30-09-2002 20:37

237's bot may be halfway there, but it still rocks. It seems like a good strong step in the direction of the illustration in the first post.

me/ likes that picture. My father is an inventor/mechanic. Brings back fond memories.

Is my asumption correct in that the model can spin on a dime, so to speak?

I think 237 came as close as you can get and still be within the rules. It'd be cool though if it could be made to go the next step.

Sweeeet picture Elgin (of the bot I mean). My kids thought I was looking at porn or something because I kept saying "Oh baby" (How embarrassing).

Yan Wang 30-09-2002 21:51

1 Attachment(s)
Hey hey Patrick! Too bad you can't be on our team this year due to your excessive "college work"... But yes, thanks for wishing us to do well. I think Pat got half the inspiration from seeing the multidirectional drivetrain at the Canadian regionals, and the omni wheels from our own team design, and also the plastic (very nicely made ones) at the Cornell RoboCup lab. That was an awesome night we spent there. A picture of our omni wheels is attached below... though that's what it looked like brand new :P Now it's completely locked up and everything from competition and also Brian and I driving it at a BorgWarner presentation on concrete for about 2 hours... oh well, it's running and I'm sure it'll be fine to present with Dean Kamen when he comes here on the 30th of this month with some Segways for us to ride!

ajlapp 01-10-2002 02:02

we premiered the first killough platform last year
 
Last year team 857 used the killough platform as the drivetrain for our robot. This design needs only three wheels to achieve perfect holonomic omni-directional motion. This means complete control over direction and orientation of travel. You can see pics of the design in the robot photo section of this website.......and if you're interested I'd gladly send you the existing cad files of how we achieved this momentus task. No luck on helping you control it though...that's proprietary. :)

keep up the good work.

Sincerely,

Anthony Lapp
Lead Design Engineer
Michigan Tech -- Superior Roboworks

Andy Baker 01-10-2002 09:51

Kiwi drive
 
I must say that 857's drive train was sweeeet. They even let me drive it in their pit.

This robot was probably the smoothest-moving FIRST 'bot ever. It is a shame that these guys from Michigan Tech didn't win some technical awards last year, because they deserved it.

Andy B.

Elgin Clock 01-10-2002 12:20

Quote:

Originally posted by Michael Krass
Well, I think there is still an advantage to what Patrick's proposed over what 237 did. Patrick's omnidirectional system has complete freedom over its range of motion. On your robot, Elgin, you can't use all four wheels to move at, say, a 45 degree angle to the robots axis, can you? Patrick's can power all 4 motors at once to move.
No but we used to be able to do a "death spin" by having all 4 wheels go in the same direction around a circle. Yeah, we stopped that after a bad incident with a broken transmission and chains. lol Not to mention the battery flying off the robot and hitting someone. (that was before we had it totally secured in the frame) oops!!

Quote:

Originally posted by Andy Baker
I must say that 857's drive train was sweeeet. They even let me drive it in their pit.

This robot was probably the smoothest-moving FIRST 'bot ever. It is a shame that these guys from Michigan Tech didn't win some technical awards last year, because they deserved it.

Andy B.

AH HA!!! That is the one!! I have thier video promo that put out after the six weeks, cool as anything I have seen with the "You Spin Me Right Round Baby Right Round!" song in the background!!!

Adrian Wong 01-10-2002 15:58

That's amazing, Patrick. This year, I intend to do odometry and self-localization research on the Killough omnidirectional platform. I even got the two IEEE articles on his research.

I received those articles late last week, and I was formulating some of the guidelines for the project. I was having trouble finding construction plans, so I started drafting my own. I got as far as developing the wheel assemblies (the Killough platform's wheels are slightly more complex in comparison to the "universal" whee) and chassis. I still had to do work on the sensor platform (inertial sensors), but the general idea was there, albeit, unrefined.

It's so great that some other people are doing work on this platform. I talked with some people from 857 last year, when they were designing their omnidrive. Mostly, it was about control issues and the decision between the Killough orthognal wheel assemblies and the universal wheel. Later that year, I heard they resolved their control issues ("proprietary" ;)) and they put up that stellar video.

Hopefully we'll see more omnidirectional bots this year!

Marc P. 08-10-2002 22:38

I believe it was the Kiwi drive system, but I as well have forgotten the team number. It is really an impressive display of technology, and there was a video posted demonstrating it's capabilities. If i can find it I'll post a link here.

http://stuweb.ee.mtu.edu/~alkrajew/FIRST/

That's it, check out the video, it's just plain great.

Don Taylor 08-10-2002 23:20

Our Robot, Metal in Motion (Team 343) also used Omni-directional multi-mode drive train. We had a 6 wheel drive system (4 in the X direction and 2 in the Y direction). The X direction had a 2 speed transmission and the Y a High torque single speed transmission. The primary mode that we operated in was the 'Tank Drive' mode with all 6 wheels driven which gave us a very fast forward and reverse speed and tremendous rotational torque. Also if the driver pulled the trigger on his joystick our robot would switch to the X/Y mode where we could move forward/reverse left/right or crab.

Don Taylor
Metal in Motion
Team 343

Moshingkow 09-10-2002 20:03

Sure 857 had mobility, but the real question is, did they have speed and torque? I would really like to know how they did in tug-o-war matches. That killough drivetrain looks mighty cool, and I think that My team may attempt somthing to that effect, at least before the competiton, just to see what its like, and the coolness of it. Maybe we can use it as a tethered cart for the robot. who knows.

tenkai

Marc P. 09-10-2002 22:43

I've seen it in action, they have incredible speed and maneuverability, but not much in the torque department. It's driven by 3 drill motors, so it has the strength, but not much as far as traction goes. I'm sure it wasn't their intention to haul goals though, I think they were going for the agility to collect balls, which I think they did one hell of a job of!!

Moshingkow 09-10-2002 23:02

Marc-
Did you see what theyre mini-wheels on the omni wheels are made out of? from the looks of the video, it seems like they were made out of delron, or some sort of hard plastic. Id think if you coated the wheels in a sticky rubbery coating, that it would have MUCH more traction, or better yet, make them out of metal spikes :)
mmm spurs

tenkai

Greg Ross 10-10-2002 00:26

Quote:

Originally posted by Moshingkow
Marc-
Did you see what theyre mini-wheels on the omni wheels are made out of? from the looks of the video, it seems like they were made out of delron, or some sort of hard plastic. Id think if you coated the wheels in a sticky rubbery coating, that it would have MUCH more traction, or better yet, make them out of metal spikes :)
mmm spurs

tenkai

High traction wheels won't work well with Patrick's drive. Remember, when the 'bot is traveling in the direction of one set of wheels, the other two wheels have to be able to slide sideways!

ajlapp 10-10-2002 02:32

traction and the Killough
 
you guys are absolutely right. the setup we used last year didn't account for traction because we were strictly an offensive ball collecting machine.

the small wheels inside the big wheels were made of "uhmw." it's basically a delron or teflon type plastic.

this isn't to say though that the killough can't achieve good traction.......i don't think that we'd ever use the drivetrain if we were planning on going defensive, but i will say that the stuff you'll see from us this year will definately have much more pulling power. including the bot you saw at competition we have four prototypes either working or near working with all different types of wheel assemblies. two of the bots use the omni-wheel design, and two use the original spherical wheel concept.(which by the way may turn out to be the more high traction of the two designs.)

for the record we have a delphi type swerve system also. we love this drivetrain, but chose the killough for it's uniquness and coolness factor, not because of it's performance.

i'd encourage teams to check this drivetrain out for the upcoming season. start now though! as with the swerve or other high complexity drivetrains, control issues will put a damper on your season if you don't have ample time to tweak. not to mention the extremely steep learning curve for beginning drivers.

we were super pumped to be the first to use the killough in competition, so i hope we inspired evryone to venture out into uncharted waters! if you have any questions feel free to ask...we love talking about the bots and how to make improvements. thanks for all the cool replies.

Anthony Lapp
lead design (killough fanatic) -- Superior Roboworks

##i'll try attaching pics of the other bots so people can check out what we've done.......we offered people a chance to drive one of the bots at the regionals we attended, if you didn't get a chance see us this year and we'll make sure you get a chance to give it a ride.

ajlapp 10-10-2002 02:39

try the pic again
 
1 Attachment(s)
pics of the bots

Moshingkow 10-10-2002 07:18

oooh, the wheels were sliding? hmmm what if they were on bearings? that would allow them to slide, and still have the forward traction that I desire...

Marc P. 10-10-2002 07:21

*drools* Yeah, that's definitely the coolest drivetrain I've ever seen...

At nationals I had a good conversation with someone about how the whole system works. I believe it's driven by three crosslinked joysticks (not in software mind you, but literally, three interconnected joysticks, one for each axis, tied together by a metal apparatus) Movement was achieved by moving the center of the apparatus in the direction of which you wished to travel, if I'm not mistaken. Can anyone confirm this?

Clark Gilbert 10-10-2002 22:31

Very Cool
 
I got to drive the top one when a group of TechnoKats went over to check out the "Kiwi" bot (am i right?) in Chicago (?). The controls were somewhat easy to pick up on (3 joysticks linked together) and it drove like nothing else....

Very Cool :D

Greg Ross 11-10-2002 01:17

Re: traction and the Killough
 
Quote:

Originally posted by ajlapp
two of the bots use the omni-wheel design, and two use the original spherical wheel concept.(which by the way may turn out to be the more high traction of the two designs.)
I had a hard time envisioning how AJ's sperical wheel base would work, but I did a little Google Search, and found a couple videos which show a slightly different version in action.

I think you're right AJ. It looks like it has potential for a fair amount of traction.

sanddrag 11-10-2002 10:54

All I can say is OH WOW!!!!!!!


/me digs through childhood toys in search of legos.


All times are GMT -5. The time now is 01:47.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi