An Optimisation-Based Distributed Cooperative Control for Multi-Robot Manipulation with Obstacle Avoidance

Abstract

Multi-robot manipulation systems are usually high-dimensional, kinematically complex and the internal forces are sensitive to robot motion errors due to the physical coupling, especially when the manipulated object is rigid. In this work, a distributed cooperative controller is designed for this scenario. Besides transporting the object, obstacle avoidance and manipulability enhancement are also achieved online by a novel optimisation-based approach. Since the local controller does not require the other robots to send the model or joint-space data, the system is flexible and the communication cost is minimal. experiments show that no internal force is generated when the robots are changing their poses for the additional tasks and the online computation is fast.