Hybrid Path Planning for Non-Holonomic Autonomous Vehicles: an Experimental Evaluation

Path planning of an autonomous vehicle as a non-holonomic system is an essential part for many automated driving applications. Parking a car into a parking lot and maneuvering it through a narrow corridor would be a common driving scenarios in an urban environment. In this study a hybrid approach for dynamic path planning is presented which deals with the limited degrees of freedom of a non-holonomic system, as well as the real-time constraints and limited computational power. Furthermore, it takes into account ride comfort of the passenger by generating smooth path. The proposed method has been evaluated in simulation environment and is implemented and tested on a Toyota Prius test vehicle with an active steering system.