Optimal assigner decisions in a hybrid predictive control of an autonomous vehicle in public traffic

The complex public traffic environment requires an autonomous vehicle to have the ability of planning and executing a sequence of different maneuvers, such as maintaining cruising speed, changing speed to follow a vehicle in front or to lead a vehicle in the rear, and changing a lane when necessary and possible. This paper presents a hierarchical hybrid predictive control framework that integrates the discrete optimization problem of maneuver selection with particle motion-based model predictive trajectory guidance for an autonomous road vehicle. To address the challenge of solving the resulting mixed integer nonlinear programming (MINP) problem efficiently for online implementation, a relaxation method is introduced that transforms the MINP problem into a nonlinear programming problem with improved feasibility for online implementation. The performance of the proposed framework is illustrated via simulations of the autonomous vehicle in highway scenarios.