Decentralised Cooperative Mobile Manipulation with Adaptive Control Parameters

Abstract

In this work a collaborative controller for a team of mobile manipulators is designed for transporting a rigid workpiece to a desired position and orientation. Impedance control is used for the robotic arms to allow some compliance in the physical contact. The mobile platform is localised with a Bayesian filter and controlled kinematically using a velocity reference composed of two parts: a) a velocity towards the goal position and the desired formation, and b) one that keeps the robot gripper in a proper range relative to the platform. According to the accuracy information from the localisation algorithm, the control gains are changed to reduce the motion error and internal forces on the workpiece. The system is decentralised and no communication between the robots is required. Simulation results are presented to illustrate the performance.