Decentralized multi-robot formation control in environments with non-convex and dynamic obstacles based on path planning algorithms

Abstract

In this paper, we propose a new strategy to solve the multi-robot formation problem. Considering a set of holonomic robots, a decentralized algorithm is proposed to guide the robots to achieve a predefined formation while avoiding collisions with non-convex obstacles, dynamic obstacles, and other robots. Local collision avoidance is achieved using a variant of the well-known ORCA (optical reciprocal collision avoidance) algorithm. We modify this algorithm to ensure the continuity of the robots’ controls (velocities). The implementation of an online replanning algorithm, RRT, is essential to guide the robots and prevent them from getting stuck in minima. The resulting method guarantees formation convergence, and several simulations are presented to illustrate its effectiveness.