Collision avoidance with differential drive robots using MPC-ORCA

Glauber R. Leite; Arthur da C. Vangasse; Ícaro B. Q. Araújo; Heitor J. Savino

doi:10.20906/sbai.v1i1.2774

Glauber R. Leite Instituto de Computação, Universidade Federal de Alagoas, AL
Arthur da C. Vangasse Instituto de Computação, Universidade Federal de Alagoas, AL
Ícaro B. Q. Araújo Instituto de Computação, Universidade Federal de Alagoas, AL
Heitor J. Savino Instituto de Computação, Universidade Federal de Alagoas, AL

DOI: https://doi.org/10.20906/sbai.v1i1.2774

Keywords: Model Predictive Control, Autonomous Mobile Robots, Distributed Control

Abstract

This work describes a distributed solution using Model Predictive Control (MPC), including the collision avoidance algorithm ORCA applied to mobile robots following individual trajectories. Differential drive robots are used, defined by their position on the plane and controlled by velocity commands. Based on an explicit system model and velocity constraints designated by the collision avoidance algorithm, the MPC-ORCA computes optimal control actions to minimize a cost function over a prediction horizon. The methodology can efficiently handle multivariable control systems using state-space model representation and convex quadratic programming (QP). Simulation results show that the combined strategy MPC-ORCA provided smooth and collision-free trajectories in a changing environment. It also requires no trajectory replanning neither direct communication between agents.