Making your robot drive easier

[Jay Lundy]

Originally Posted by Chris Hibner

I think the key to an easy-to-drive drivetrain is having the optimal lateral coefficient of friction (not the lowest). That is much easier said than done, but here is why I feel this way:

1) If you have too much lateral friction, your drivetrain will not turn well (or maybe not turn at all). This is obviously not desirable for many reasons that everyone knows, so I'm not going to explain further.


2) If you have too little lateral friction, the robot can become difficult to drive. Why? Inertia!

I couldn't agree more.

For years we had trouble tripping breakers while turning with a 4WD base due to too much lateral friction. Last summer we started experimenting with the 6WD base, and when we finally got to drive it we found it turned TOO well. Our center wheel is offset 3/16", and when we turned at full speed, the robot would basically turn on the 2 center wheels. The other wheels were still in contact, but they had very little weight on them so there was very little friction preventing the robot from turning. In high gear our robot goes about 15 ft/s, and when we turned the robot flew around at more than 1 rev/s.

We liked the extra manueverability, but it was too difficult to control with a standard 2 joystick tank drive. We ended up using a steering wheel and a throttle. See this thread for more information about how we implemented the steering wheel.

Another problem was going straight. Less lateral friction meant it was easier for the robot to turn even when one side was only slightly more powerful than the other. In our case the flip-flopped drill motors were our biggest problem, but even things like worn down treads or slightly more friction on one side could cause the robot to veer off to one side if you powered the motors at the same voltage. 980 had the same problem as us. We (254/60) solved it using prox sensors which gave us about 54 counts/rev and allowed us to adjust the voltage being sent to one side to keep it going straight. 980 used a gyro (the ADXRS150EB) in basically the same way. We were going to use a gyro too, but Analog Devices ran out of them.

I don't like the 2WD idea because it would be way too manueverable. You need some friction out there on the ends of your robot to provide some torque to slow you down while turning. And if you try to solve it by adding more friction to the skid-plates then you just lose power when you want to go straight. You can try to solve it using software and sensors (and in fact I'm probably going to work on something over the summer to use sensors to have the robot maintain a certain orientation), but it's not going to be easy.

We plan on sticking with the 6WD at least for the next couple of years. Sure there are alternatives out there (treads, swerve, holonomic), but as John said 6WD is a great combination of strength and agility.