440 omni flex chassis---center of rotation DNE center of veh

[trautman]

We did precision tests on a rotating 440 omni flex chassis yesterday, and had some interesting results:
1. We mounted a laser tracker target (+/- .001 in accuracy) on a corner on the flex.
2. We commanded the vehicle to execute a circle.
3. We continuously collected the target data.
4. The omni 440 executed a nearly perfect circle (i.e., the end point was within .05 inches of the begin point---obviously this is very good).
5. HOWEVER: when we fit a circle to the data (and the circle fit with about .04 RMS error, so it was a tight fit), the center of that circle WAS 1.25 INCHES TO THE LEFT OF THE CENTERLINE OF THE VEHICLE. FWIW, we conducted this test twice because the result seemed so strange, and got the same answer both times.
6. Although this can be accounted for in higher level planners, are you guys aware of this? This implies that if you were to naively execute a trajectory assuming the center of rotation was the same as the center of the vehicle, you would very quickly accumulate substantial error, possibly resulting in collisions.

7. Is there a patch for this in your controller code? We could send you the transform from the center of the vehicle to the center of rotation if that would help. Or if at least you could point us in the right direction in the low level controller to correct this deficiency.

Thanks!

[phussey]

Please be sure that the flex OMNI configuration is loaded. The wheel base is different than the rigid OMNI and that difference will affect the Mecanum transform; likely in the way you are experiencing.

The wheel base and track width for the FLEX OMNI should be:

Wheel Base Length (m): 0.693
Wheel Track Width (m): 0.693

And the RIGID OMNI:

Wheel Base Length (m): 0.572
Wheel Track Width (m): 0.693

The slight difference will have an impact on the transform.

[trautman]

got it! I think that's probably it...