Modelling autonomous perpendicular parking procedure for car-like robot Avrora Unior in Gazebo simulator

摘要

This paper describes an implementation of autonomous perpendicular parking and parking spot search algorithms. A procedure of the perpendicular parking is based on geometric constraints of a parking spot and a vehicle. The procedure of perpendicular parking includes driving until a first free parking spot is found, setting a vehicle into an initial pose, performing a rear-drive parking with steering until a vehicle reaches a goal position. For the parking spot search algorithm LIDAR sensor data were used. Based on LIDAR data clusterization parked vehicles detection was performed. HDBScan, OPTICS and Gaussian mixture methods were compared for the parking spot search application effectiveness within the Gazebo simulator.