Applying Theory to Practice

[benedict]

This is my first post here, but Colin and I have been trading email for a while now, as well as posts on another forum.

A while back I built a mini sumo robot out of a Mark III kit, Shallow Blue. Things went well, so I eventually bought a second Mark III which my wife took charge of. She's doing software development for it and I'm doing the mechanical development. At one of its first appearances in the dohyo it appeared to have a bad tendency to pop a wheelie when it ran ito other robots, namely a Parallax that was built by another club member.

I've since fixed the underlying problem (line sensors were scraping on the dohyo surface, raising the wedge off the ground), but it sparked a nice discussion on mechanics, center of gravity, balance, torque, etc. Colin reminded me of the Mechanics section on the web site, so I went through all the math he presents there. I went through it again when designing ballast for Shallow Blue and the second Mark III, Black Dolphin.

After doing the math on paper a couple of times I went ahead and put it all together in an Excel spreadsheet. This has been particularly handy because the ballast blocks I made are adjustable. Adjusting the ballast and running the new numbers through the spreadsheet let me see the effects of changing how much weight is on the wedge versus the wheels.

I only went as far as the math presented by Colin, and used only values available from simple measurements and from manufacturer data sheets. One important area of consideration I don't cover in the spreadsheet is acceleration from a stand-still, and how the location of the center of gravity above and below the wheel axles affects whelie-popping behavior. But to do this would require acceleration curves for the drive motors and a better way to locate the center of gravity in a robot. Projects for another day...

Anyway, the spreadsheet is available here:

http://vix.dyndns.org/~benedict/robotics/tools/Two-Wheel-Bot-Characterization.xls or as google doc here (EDIT: CAD 02MAY07)

It has numbers in it already for the ballast full-forward setting on Shallow Blue. The numbers aren't typical. That robot runs on a 7.4V battery, so motor torque and speed values are higher than normal. It's got urethane tires, so the coefficient of friction is higher than normal as well. If you're trying to model a different robot, it's important to go through all of the numbers and make sure they're all representative.

Again, a big thanks to you, Colin, for putting all that up on the web site. It's certainly helped me. I hope others are using it as well. Good stuff.

Have fun!

Tom

[slurp]

Glad to have helped

Much more work to do here... I think there's a little more I can add to the maths but it'll soon get academic especially when you start getting a feel for what works.

Must get my finger out and finish a few more pages!

best regards,
Colin

[slurp]

For those that may have missed it... Tom's updated his spread sheet with a few minor tweaks.

Cheers Tom!

Best regards,
colin

[benedict]

You bet!

At some point I'd like to add more to it, but for now it covers the bases.

I do have a first application outside of mini sumo: I'm building a line follower that I'd like to see do two meters per second. It's going to be very low to the ground, so I don't think I'll have the balance issues my mini-sumo sometimes suffers from, but I'm going to have serious traction and torque issues at those speeds.

For the first round of testing and tuning I'm going to assume no right-angle turns. But before June of next year I'll need to sort that out as well. So whatever speed I convince the thing to manage, it'll have to do an abrupt right angle turn, which I should be able to model as a dead-stop.

With any luck this'll let me skip a lot of build/rebuild/rebuild/rebuild. I'll let you know how it goes.

Tom