Autonomous Valet Parking System Architecture

Technology is rapidly changing and the automotive world is constantly improving. One of the challenges for automotive industry is to develop the completely autonomous car. Even if proven feasible, the autonomous driving car is still an important research subject due to systems complexity, costs and legislation. The paper describes an experimental car that Renault is actively developing in partnership with research laboratories and outlines its usage in terms of Autonomous Valet Parking (AVP). During development, a common Renault Zoe vehicle was equipped with sensors, electro-mechanical actuators and a system of data processing. The perception system corresponds to a fusion between an ultrasonic system and a camera. It provides the localization of the vehicle, the navigable area around the car and a map of already visited areas. This information is used afterwards as input for the car control system in order to ensure it's integrity during navigation. The Autonomous Valet Parking system is able to take control of the car and to drive it from the drop-off zone to the parking and from the parking to the pick-up zone. In the following it will be described the hardware and the software system architecture and the main principles of algorithms.