A Receding Horizon Trajectory Tracking and Obstacle Avoidance Strategy for Constrained Differential-Drive Robots
Alexis Marino, Cristian Tiriolo, Walter Lúcia, Francesco Tedesco · 2025
This paper presents a control strategy for tracking trajectory and obstacle avoidance in constrained differential-drive robots operating in static but unknown environments. A robust receding horizon tracking controller is proposed, capable of handling state-dependent input constraints that arise when the robot’s dynamics are addressed through a feedback linearization technique. When a potential collision is detected along the reference trajectory, a switching law activates an obstacle avoidance mode. This mode leverages a vision module to build a grid map and applies an A* path planner to generate dynamically updated, obstacle-free waypoints that guide the robot around the obstacle. The same control scheme is used for both tracking and obstacle avoidance, ensuring consistency and efficiency. This framework aims to compute a collision-free, shortest, and safe path while adapting to environmental changes and satisfying velocity constraints. Once the obstacle is cleared, the robot seamlessly resumes the nominal tracking policy. The proposed method is validated using the digital twin Qbot2e differential-drive robot, achieving an average tracking RMSE of 0.06 m, maintaining a minimum obstacle clearance of 0.34 m, and avoiding all collisions across 100 runs.