Decentralized control for cooperative load transport

Abstract

Coordinated control of multiple robots has always been a popular topic in the robotics control field. The reason behind it is that a cooperative control strategy can easily overcome the limited capability of a single robot. For example, using various small robots to transport a payload versus one giant robot can avoid purchasing the expensive giant robot and keep the maintenance effort low. In this thesis, we examine the task of controlling multiple robots to transport a rigid load in a decentralized manner. The main concept is to design a controller which can command a swarm of mobile manipulators to collaboratively carry a payload that exceeds the capability of a single robot. Based on this target, we propose a decentralized controller inspired by admittance control in robotics control literature. The proposed controller addresses robot’s motions and forces without sharing other participated robots’ kinematic and dynamic information. It ensures that each robot can converge to the desired force setpoint and simultaneously achieve the same predefined velocity. This control strategy is evaluated on two physical robot hardware implementations. A total of five different experiments are conducted and the corresponding results are included to demonstrate the effectiveness of the proposed scheme.