Abstract
This research proposes an automated parking system for a car-like robot which uses a roadmap where every node has not only position but also orientation information of the robot. In the proposed roadmap, the orientation of the robot is not expressed as a fixed value but as range of values. The roadmap is constructed by dividing the orientation space in variable resolutions considering the non-holonomic constraints of the robot and collision-avoidance constraints between the robot and surrounding obstacles. The proposed method evaluated both in numerical simulations and in an experiment using a real electric vehicle.
