# Kiwi Drive??

I noticed the link from the recent TechnoKat thread that led me to a video of 857’s kiwi drive from a year or two back.

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

This thing is increible!!! I’ve watched the video several times over, and have spent some time trying to figure out how you guys controled it. I just don’t see how it could be done with simple joysticks. PLEASE explain to me how you acomplished this incredible task.

This is by far the coolest drive system I’ve ever seen!

The best person to contact would be the guy that wrote the program. That person is Adam, the guy who you got the video froms website. I don’t know all the details, however I can tell you this. It is very simple. Adam had tried to work on a program and it ended up being huge. By using the joysticks in the manner we did, we were able to cut the program down greatly.

By simply placing the joysticks 120 degrees apart, the joysticks do all the math. They know to drive two of the motors at 100% and the other one at 0 percent, to go in a straight line, assuming the 0 percent wheel is perpendicular to the path of motion. We do this by making the x axis mean nothing and the y axis mean everything. I wish I had a picture to try and explain better, but I am sure Adam will/can.

Sorry if I confused you even more

Kiwi Drive was their robot name…

The idea isn’t brand new. If you want to do web research, the drive is called the Killough Omnidirectional (holonomic) drive system. However, this system isn’t really feasible without 3 matching motors, or your wheels will have different torques and speeds. It’s a pretty ingenious design though. You may think “Omniwheels? That kills your pushing power.” However, if you look at it no matter what direction you push against the bot, you are pushing directly against 2 omniwheels.

4 Wheel holonomic drive

This one wouldn’t require 4 identical motors. Just as long as you use 2 identical motors in one direction and 2 in the other direction.

*Originally posted by Gope *
**I noticed the link from the recent TechnoKat thread that led me to a video of 857’s kiwi drive from a year or two back.

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

This thing is increible!!! I’ve watched the video several times over, and have spent some time trying to figure out how you guys controled it. I just don’t see how it could be done with simple joysticks. PLEASE explain to me how you acomplished this incredible task.

This is by far the coolest drive system I’ve ever seen! **

I knew someone on the team and saw them at the western michigan regional. They basically had the joysticks set up how they had the motors set up. The handles had been removed I believe and they had a tri-handle thing made of metal that connected them all. very cool, kinda hard to steer though. vectors are fun

I can’t load the video for some reason. Could someone post a picture of the Kiwi drive?

BTW, is there a search option in the picture gallery?

Adam had tried to work on a program and it ended up being huge.

I have developed an omnidirectional robot for another project, and I found that the code initially looks difficult, but may easily be simplified. The euqations start out very complicated with trigonometric functions and vectors. A several hours of work and it can be simplified to about 10 lines of code.

One can simply plug in one joystick and use the following code (initialization sequence skipped):

Code removed - Contact Author for Full Listing

with the 4 wheel holmonic drive how will it affect performance if you have 2 different sets of motors i.e. the chips and the drills?

*Originally posted by abe D *
**with the 4 wheel holmonic drive how will it affect performance if you have 2 different sets of motors i.e. the chips and the drills? **

Well, you can gear the motors so that the speeds are approximately the same. It shouldn’t affect the performance at all. The only time you’ll see problems is when you try to push something using that angled side between the omniwheels. One motor will have more torque than the other and your robot might move crooked. In fact matching the speed/torques perfectly isn’t a limitation in this design. If you wish, you can have it move forward faster than it goes sideways.

Moving forward/backward, left, right and diagonal and turning should be perfectly fine. The coding would be a bit tricky, but an experienced programmer can figure out how to create a 2 joystick drive that utilizes all the holonomic functions. It wouldn’t be that difficult. I can’t think of any way to control all three coordinates, x direction, y direction, and angle (turning) without using 2 joysticks. You could possibly do it with one joystick if you use the triggers or a separate mechanism to control the turning.

The biggest problem is making an Omniwheel with lots of traction. These new TechnoKat omniwheels look perfect… You don’t have to worry about someone overpowering your robot because you have the other omniwheels pushing against any side forces.

There are a few things I’d change in the design on that page. The drill motors have timing (they spin faster running CCW than CW), so in the current setup the robot will run crooked when going forward. This can be easily solved by putting both drill motors on the front half of the robot (put the drill motor on the right side in front of the sprocket rather than behind it).

The 3 and 4 wheel versions are on totally seperate levels of complexity. The 3 wheel has to have the same motor because different motors even geared the same would offer different torque. In the 3 wheel design the motors never really drive “straight” so even if the wheel is turning at 10 ft/s because of the angle the bot actually only moves a fraction of the speed.

In a 4 motor design each set of opposite wheels works on a seperate independent axis. So if one set of the wheels was mvoign at 6ft/s and the other at 8ft/s then the bot would actually move at about 10ft at at angle. I’m not usre how the different torques would effect the turing fo a 4 wheel version though.

I’m looking forward to seeing the drivetrains this year and even more so next year. It seems a 2 speed is now almost standard. Hopefully next year more teams will run kiwis and the new technokat mouse drive.

*Originally posted by sevisehda *
**It seems a 2 speed is now almost standard **
Not quite yet; there are plenty of veteran So Cal teams running only single speed and just with the drills too. The advantages of a two speed are not worth the additional time, effort, weight, and complexity (but that’s an entirely different thread I guess).

*Originally posted by sanddrag *
**Not quite yet; there are plenty of veteran So Cal teams running only single speed and just with the drills too. The advantages of a two speed are not worth the additional time, effort, weight, and complexity (but that’s an entirely different thread I guess). **

Heh, we put our gear shifter together in about 30min one day :yikes:

Just takes a servo, and some aluminum

Look in the white pages, patrick wrote an artical about it.

Greg

does anybody know if that is feesable with a rc controller, and only 2 esc’s?