Trajectory planning and control for autonomous vehicles: a “fast” data-aided NMPC approach

A huge research effort is being spent worldwide by automotive companies and academic institutions for developing vehicles with high levels of autonomy, ranging from advanced driving-assisted systems to fully automated vehicles. Nonlinear Model Predictive Control (NMPC) has the potential to become a key technology in this context, thanks to its capability to deal with linear and nonlinear systems, manage physical constraints and satisfy multi-objective performance criteria. However, NMPC is based on the on-line solution of a nonconvex optimization problem and this operation may require a high computational cost, compromising its real-time implementation. In this paper, a "fast" data-aided NMPC approach is developed, aimed at trajectory planning and control for autonomous vehicles. In particular, a Set Membership approximation method is used to derive from data tight bounds on the optimal NMPC control law. These bounds are used to restrict the search domain of the underlying NMPC optimization process, allowing a significant reduction of the computation time. The proposed NMPC trajectory planning and control approach is tested in simulation and compared with other state-of-the-art methods, considering different road scenarios.