Predictive Guidance and Control Framework for (Semi-)Autonomous Vehicles in Public Traffic

In this paper, a predictive trajectory guidance and control framework is proposed that enables the safe operation of autonomous and semiautonomous vehicles considering the constraints of operating in dynamic public traffic. The core module of the framework is a nonlinear model predictive guidance module that uses a computationally expedient curvilinear frame for the description of the road and of the motion of the vehicle and other objects. The module enforces constraints generated from information about obstacles/other vehicles/objects, public traffic rules for speed limits and lane boundaries, and the limits of the vehicle's dynamics. The module can be configured in two basic modes. The first is a tracking mode, where the control inputs computed by the model predictive guidance module act as references for traditional lower level control systems. The second is a planning mode, where the traffic-optimal state trajectories computed by the model predictive control are reinterpreted for planning the optimal path and speed, which in turn can be tracked by an elaborate speed and path tracking controller. The performance of most aspects of the proposed scheme is illustrated by considering various simulations of the control framework applied to a high-fidelity vehicle dynamics model of the (semi-)autonomous vehicle in typical public driving events, such as intersections, passing, emergency braking, and collision avoidance. The feasibility of the proposed control framework for real-time application is highlighted with the discussions of the computational execution times observed for these various scenarios.