On-line sensing and action planning in an autonomous and distributed multi-arm system

Abstract

In this paper, we describe a method for pushing an object on a slanted plane by multiple robot arms in an autonomous and cooperative manner. In order to decide its own action, each arm obtains accurate estimation of poses of the object by fusing information obtained from its own force/torque(F/T) and vision sensors. Furthermore, we present a new planning algorithm for changing pushing points so that the system would have more flexibility to various situations. The effectiveness of the proposed method is investigated by some numerical simulations.