Predictive trajectory guidance for (semi-)autonomous vehicles in public traffic

The safe and reliable operation of autonomous and semi-autonomous vehicles in public traffic requires the tight integration of environmental sensing and vehicle dynamics control. In this paper, a predictive control framework is outlined that connects both areas. Specifically, a trajectory guidance module is posed as a nonlinear model predictive controller that computes the optimal future vehicle trajectory using information from environmental sensing for other objects as well as by imposing public traffic rules. It is also sought to minimize the number of vehicle specific parameters needed for the guidance by adopting a particular particle motion description for the vehicle. The computed control input set for the trajectory guidance is passed as a reference for lower-level vehicle dynamics control systems. The definitions of the objective functions and constraints and the adopted vehicle motion model allow for a unified predictive trajectory guidance scheme for fully autonomous and semi-autonomous vehicles in public traffic with multiple dynamic objects. The performance of the proposed scheme is illustrated via simulations of an autonomous and a semi-autonomous vehicle in a few traffic scenarios such as intersections and collision avoidance. Execution time considerations are also analyzed.