Cooperative Distributed Model Predictive Formation Control of Non-Holonomic Robotic Agents

Abstract

Due to the ongoing advancements in the field of wireless communication and the increasing availability of fast and cost-efficient embedded computing units there is a growing number of applications that can be solved more performantly and reliably by the cooperation of several simple robots instead of one highly complex robot. Such a by design simple mobile robot is for example the differential-drive mobile robot which has already established itself in many everyday tasks. However since its dynamics is subject to a non-holonomic constraint the optimal control of even a single robot is not a trivial task. In particular model predictive control with standard quadratic stage costs does not necessarily lead to an asymptotically stable closed loop. For this reason the present article aims to develop a theoretically-founded distributed formation controller for differentially-driven mobile robots which is substantiated in real-life experiments. Moreover the gained insights are transferred to other non-holonomic systems. Notably it is conjectured how a kinematic car could be regulated based on the aforementioned findings