Comparison of distributed model predictive control approaches for transporting a load by a formation of mobile robots

Abstract

This paper investigates and discusses two different distributed formation control approaches based on model predictive control (MPC). Specifically, the presented control schemes are used to govern the motion of omnidirectional mobile robots that shall maintain a given formation shape while following a path through a previously unknown environment. The setup of the control schemes accounts for the requirements of transporting an elastic plate purely by normal and friction forces. The intricacy of this task motivates the choice of model predictive control since it allows to explicitly constrain the movements of the robots. The two schemes analyzed in this contribution are fundamentally different in their optimization and communication strategies. The performance of the schemes is carefully examined in various simulations. These include situations in which the robots have to shrink the formation in order to squeeze through narrow passages. An exemplary experimental result involving real robot hardware is also presented.