Hybrid MPC and Spline-based Controller for Lane Change Maneuvers in Autonomous Vehicles

This paper presents a novel approach for collision detection and overtaking maneuver. The proposed technique is considered hybrid as it combines the smoothness of spline curves for nominal trajectories and the capabilities of Model Predictive Control (MPC) to respond to unexpected obstacles on the lane. To detect collisions, a temporal spline parameterization procedure is introduced. In the case of a possible collision, two decoupled longitudinal and lateral MPC controllers adjust the vehicle's path without modifying the nominal trajectory, using linear models to ensure short computation times. The complete control system has been implemented on Robot Operating System (ROS) using the hyper-realistic simulator CARLA, with successful results in overtaking scenarios.