A Switching Model Predictive Control for Collision-Free Path-Tracking of Mobile Robots
Chrystian Pool Edmundo Yuca Huanca, Gian Paolo Incremona, Patrizio Colaneri · 2024
This paper addresses the problem of collision-free path-tracking of a team of mobile robots, which move in the environment in the presence of obstacles. A model predictive control approach is presented, relying on the switched-system formalism to capture each robot model. Specifically, a minimal description of their dynamics, consisting of two modes (rototraslation around a fixed pivot and rotation on spot) capable of efficiently capturing most of the possible motions on the plane, is considered to reduce the curse of dimensionality. Moreover, in order to enable robots to roam around while avoiding collisions and maintaining low computational complexity, obstacle avoidance constraints are relaxed to linear form and included in the optimization problem. The proposal is finally assessed both in simulation and experimentally on the Robotarium remote arena, in comparison with an intrusion-based obstacle avoidance strategy.