We are looking for real world experience based insights or concerns re repeatable predictable precision navigating (including turns) in Auto for:
four 6" AM Mecanum platform, and
a 2 traction wheel and 2 Omni wheels platforms
versus
a standard 6 wheel system.
All will have encoders and a gyro. Are platforms 1) and or 2) equally capable of doing sub 1" precision navigating or do they give anything up compared to a 6 wheel system. Are there wear and tear considerations over a long season re repeatable accuracy ? Do they natively drive in a straight for long distances ?
Posts like …“roller performance (Minimal axial free play and roller spinning friction) is critical to proper drive function” raise concerns to me about the wheels being able to natively track true on their own on the long term or at least requiring a lot of roller maintenance / inspection.
Also, on the topic of calculating distances traveled with Mecanum based wheels, I understand there might be a bias (different distance traveled) for the same encoder counts depending if the platform is traveling forward/reverse versus left and right. Calculating the distance traveled on angles would appear to be complicated by that.
Might as well ask, do 8" mecanums offer any advantage over 6" (other than larger radius for going over obstacles benefits)?
In the offseason for the past two years we’ve done extensive testing of mecanum drives. We have never run one in competition because we were worried about defence, however we are seriously considering them for this game.
Seeing as you specifically asked about autonomous, I’ll comment on that. In comparison to our competition robots which have been 6 wheel tank with traction wheels, we were able to have our mecanum robots drive just as straight, stop just as accurately, and turn just as accurately. We were able to even do some precision strafing.
In my experiences, these drives tend to rotate about their center of mass, regardless of how offset it is. 6WD/4WD traction skid steer have more forces at play, thus the center of rotation is somewhere between the center of mass and the center of area (CoA being the intersection of the diagonals created by track width and wheel base).
All this really means is that the characteristics are different. Mecanum and 2+2 will translate slightly as they rotate if the center of mass is not co-located with the intended center of rotation.
However, in our trials, the robot is consistent in how it turns for the most part. I don’t know that we’ll get precision turning without some fine tuning, but we won’t know until week 6 when the robot is fully loaded in its final form. Not that our autonomous ideas require too much precision turning ( < 1 degree accuracy & consistency ) without translation anyways.