Real-Time Simulation For Control Of Soft Robots With Self-Collisions Using Model Order Reduction For Contact Forces

Abstract — In rigid robotics, self-collision are usually avoided since it leads to a failure in the robot control and can also cause damage. In soft robotics, the situation is very different, and self-collisions may even be a desirable property, for example to gain artificial stiffness or to provide a natural limitation to the workspace. However, the modeling and simulation of self-collision is very costly as it requires first a collision detection algorithm to detect where collisions occur, and most importantly, it requires solving a constrained problem to avoid interpenetrations. When the number of contact points is large, this computation slows down the simulation dramatically. In this paper, we apply a numerical method to alleviate the contact response computation by reducing the contact space in a low-
dimensional positive space obtained from experiments. We show good accuracy while speeding up dramatically the simulation. We apply the method in simulation on a cable-actuated finger and on a continuum manipulator performing exploration. We also show that the reduced contact method proposed can be used for inverse modeling. The method can therefore be used for control or design.

Comments are closed.