Dynamic collision avoidance among multiple mobile robots: A model predictive control approach

This paper deals with dynamic collision avoidance of multiple mobile robots by model predictive control (MPC). Each robot has the same control law based on an MPC problem which has constraints on avoidance of other robots. A numerical example shows that a robot can reach to a goal point from a start point while avoiding collision with another robot that follows its start and goal points.

1. Introduction

Collision avoidance is one of the fundamental requirements in transportation systems. In particular, several types of mobile robots are available in factory automation systems. If mobile robots are regarded as dynamic obstacles to another robot, then it is natural to consider that the mobile robot can avoid the other robots under some methods of dynamic obstacle avoidance. Then a region including a mobile robot can be a dynamic obstacle to other robots. The techniques has been established so far that polytope obstacles can be constraints in a problem of mixed-integer programming [1], [2]. Thus it is possible in principle that the robot can avoid the other robots if each robot has an independent controller based on a problem of model predictive control (MPC) [3] with constraints on obstacle avoidance. For the purpose above, this paper gives a preliminary solution to a problem that the two mobile robots moving from each start point to each goal point avoid one another.