Mecanum Suspensions

[Dan Zollman]

Originally Posted by Tristan Lall

Probably because any three points of contact define a plane (e.g. the floor), and the fourth is redundant (in this case, not necessarily touching the ground). So, they leave their front wheels fixed, and let both of the rear wheels pivot so that they effectively have two planes of contact—either rear wheel plus both front wheels.

If you put it on all four, it works, but then the body of the robot becomes (potentially) unstable, because it's now supported on four moving pivots. It's like body roll in a car, only probably quite a bit more pronounced.

I believe that a mecanum system actually needs every wheel on the ground because the direction is controlled by a different vector from each wheel. If a wheel isn't touching the ground, the robot would drive at the wrong angle because force in one direction is missing, while the bot is programmed to use all four vectors. (Unless the robot was programmed to change its math based on which wheels are touching the ground...)

[Alex.Norton]

While I don't have experience with the Mecanum drive, our team has done a harmonic drive for a couple of years which is pretty much the same thing. When we first put together our drive the chassis would steer off in random directions because the floor (and no floor will be ,ever) wasn't perfectly flat (we had shimmed the wheels to be flat). This is why you use a suspension so that all wheel can stay in contact with the floor and cause good control of the vectors. Our first suspension was with each wheel mounted on a hinge and a spring, this would tilt whenever the robot would accelerate in any direction. So this year we used a "tractor" type of suspension. This didn't add any weight really but kept all wheels in contact always.
http://www.chiefdelphi.com/media/photos/22700

We also added a rotational accelerometer to keep track of any variation in the bots direction of travel and last year when I broke out our original chassis for the first time in months so the programmers could work with it one of the motors wasn't plugged and we didn't notice. Despite the lack of an entire vector the accelerometer allowed the robot to drive straight (if a little drunkenly).

Hope this helps
Alex

[Alan Anderson]

Originally Posted by Alex.Norton

...our team has done a harmonic drive for a couple of years...

I think you mean holonomic.