Integrated nonlinear model predictive control for automated driving

This work presents a Nonlinear Model Predictive Control (NMPC) scheme to perform evasive maneuvers and avoid rear-end collisions. Rear-end collisions are among the most common road fatalities. To reduce risk of collision, it is necessary for controller react as quickly possible exploit full vehicle maneuverability (i.e., combined control longitudinal lateral dynamics). The proposed design relies on simultaneous use steering braking actions track desired reference path with preceding vehicle. A planar model was used describe dynamics. In addition, dynamics brake system were included in NMPC prediction model. Furthermore, incorporates constraints ensure stability account actuator limitations. this respect, defined Kamm circle Ideal Brake Torque Distribution (IBD) logic optimal tire force torque distribution. evaluate design, performance compared two ”more classical” MPC designs that rely on: (i) linear bicycle model, (ii) nonlinear these three evaluated simulation (using high-fidelity simulator) via relevant KPIs, such tracking Root Mean Square (RMS) error, controller’s rise/settling time, Distance Collision distance by which collision avoided safely). Different single-lane-change tested behavior controllers presence wind disturbances, friction variation, maneuver aggressiveness.