[archiver]

Posted by Dodd Stacy.

Engineer on team #95, Lebanon Robotics Team, from Lebanon High School and CRREL/CREARE.

Posted on 9/7/2000 7:51 AM MST


In Reply to: CRAB!!!!!!! posted by Josh on 9/6/2000 4:24 PM MST:



Josh,

There are a number of different ways to accomplish 'crabbing' in a mobile platform. I agree with Joe's comments as they relate to approaches that involve steering (and powering) the driving wheels in the direction(s) you want them to go. That involves huge challenges in the four areas noted, and succeeding like Team 47 has done in the last two years (a few others, as well) is a truely amazing accomplishment.

However, an approach using 'holonomic' wheels provides equally universal platform maneuverability with no particular challenges in pivot design (there aren't any), wheel position feedback (they don't steer), or programming (well, maybe a little). Holonomic wheels develop traction and can deliver drive power in the direction of wheel rotation, but are able to slip freely (without steering) in the perpendicular direction if something else pushes the platform in that direction. There are lots of ways to do this, and a number of teams have made them in years past. Wheel design and manufacture is a chore, but nothing unusual compared to the mechanisms that many teams use elsewhere in their robots.

We used such a system last year, and it worked really well in a purely manual control mode. We were not able in time to successfully work in the use of the yaw rate sensor as a gyro-compass and transform the controls to an absolute spatial reference (go North) versus the standard robot reference system (go right). So our driver still had to 'put his brain on the robot' to consider its heading in making stick inputs, but that's what all the drivers do.

We used three holonomic wheels, each with an independent drive motor - a left and a right on a common axle line, and a middle wheel at 90 degrees to the other two. The left and right were programmed in conventional 'tank drive' style and operated off a single joystick. The middle wheel simply slips 'sideways' (its idea of sideways) when the left and right are driving the platform fore and aft, and serves as a pivot point when the left and right operate differentially to rotate the platform. The middle (or 'crab') wheel was controlled either off the side-to-side motion of the second stick or off a pot, depending on driver preference. Whenever the crab wheel drives the platform sideways, the left and right wheels slip freely in that direction. All modalities can operate simultaneously to move the bot in any compass direction while also rotating in either direction.

I will agree with Joe on the importance of driver practise with any omnidirectional system - we're not used to driving machines with this kind of mobility! In that regard, there are many different ways you can rig the control inputs without involving fancy programming, depending on what feels good to the driver(s). You can, for example, do fore/aft and crab left/right on the Y and X of one stick, with rotate left/right on another stick or on a pot. You can use the gyro chip to feedback to the rotate routine to maintain the bot's pointing direction constant in the absence of driver command inputs to rotate - I think many teams have done this in their tank drive program to compensate for left/right drive differences.

One final note - you need to insure that any and all of the driving wheels are adequately weighted so that they can deliver power to the ground under all conditions. This may mean some kind of limited suspension so the wheels can follow irregularities in the floor or accomodate some push and shove without losing traction. Good luck.

Dodd