A Finite State Machine Based Automated Driving Controller and its Stochastic Optimization

In this paper, we develop a finite state machine based automated highway driving controller. The controller is described by feedback control laws in each state and state transition conditions. We test the controller in a traffic simulator and evaluate its performance based on a metric function. Furthermore, we propose a stochastic gradient based optimization approach to achieve optimal calibration of the developed controller. We expect that this controller can serve as a baseline for automated driving algorithm developments.