Navigation stack with a very asymmetric diff drive robot

edit

Hello all,

We are using ros navigation for a rectangular robot base with the rotation center (O) placed close to one side (sorry for the foolish sketch). M are powered wheels, c castor wheels

   ________________
  |                |
  | M            c |
  | O    <--       |
  | M            c |
  |________________|

Setting properly the footprint around O (footprint: [[0.18, 0.26], [-0.64, 0.26], [-0.64, -0.26], [0.18, -0.26]]) and base_link also in O, makes the nav stack behave quite bad, getting stuck very easily close to obstacles. The speeds issued by the local planner get very, very low, almost zero. One could said that robot becomes very "scary". By other hand, if I fake the footprint and base_link to place the rotation center in the physical center of the robot, navigation improves a lot, and works as expected (as long as you don't try in-place rotations with and obstacle in the back part sides).

Any similar experience? Any clue about how to better understand what's happening?

Thank you very much. Jorge