Development of An Efficient and Practical Control System in Autonomous Vehicle Competition

In this paper, we develop a self-driving car called Clothoid towards Hyundai autonomous vehicle competition 2019. This work presents a development of the Control System which composed low and high-level control. Global Path Planning based on an HD-map generates the shortest path for the vehicle to reach the destination. Mission Planning and Local Path Planning use Finite State Machine(FSM) with behavior, which to deal with missions and control state of the vehicle, respectively. The lower level, the Local Path Planning uses a real-time hybrid A* algorithm based on probabilistic occupancy grid map to avoid obstacles and generates smooth trajectory while minimizing zerk in change lane scenario. And the Lowest-level Control System implements the Purse Pursuit for lateral control and PID controller for a longitude control. Our algorithms are implemented in the ROS framework; which written by C++ and Python in Linux Operating System. The results show that our technology works effectively in both simulation tool and practical proving ground in K-city, South Korea.