Efficient Local Navigation Approach for Autonomous Driving Vehicles

This paper presents an efficient and practical approach for a car navigation system (CVM-Car) based on the velocity space optimization paradigm. The method calculates control commands to keep in lane while avoiding obstacles detected by proximity sensors. has follow road path consisting of sequence lanelets. is lower-level reactive that combines pure pursuit obtain reference curvature algorithm keeps vehicle center lane' s free can partially block it. CVM-Car formulates local obstacle avoidance as constrained problem car. In addition dynamics constraints included method, car-shape non-holonomic restrictions are considered space. been applied autonomous prototype.