An algorithm is described, which will be part of an autonomous parking system, and its goal is to build a map that makes possible to record the obstacles in the environment of the car from ultrasonic sensor measurement data. The algorithm also tries to find a free parking lot (in case of parallel and diagonal parking) on the map. The sensor data come from the ultrasonic sensors mounted on the car chassis to carefully selected places. The system gets the distance information on the obstacles in the environment. Using the data about the motion of the car, the place and orientation of the sensor (according to the car) and the measured distance, the system determines the position of the sensed point, and adds it to the map. The secondary goal is to be able to determine a suitable parking place using the map, if it is possible, i.e. if such a place exists in the explored area. The parking place geometry ensures that the car can perform the parking manoeuvre without any collision. To achieve this, the Hough-transformation is used, which determines the most dominant line of the map. In the parking case this line will be the sideline of the parking cars. Going along this line one can find a parking place if it exists. The biggest problem caused by the Hough transformation is the computational demand growing experimental in time: the more points are on the map, the number of line coordinates grows squarely. To overcome this problem, the Hough transformation is not executed in every step, only if a certain amount of new points are on the map, but that still was not fast enough.
Map construction and parking lot identification algorithm for autonomous parking system
Kartierung und Parkflächenerkennungsalgorithmus für ein automatisches Einparksystem
2007
4 Seiten, 1 Bild, 1 Quelle
Conference paper
English
Autonomous parking control device and autonomous parking system
European Patent Office | 2022
|AUTONOMOUS PARKING CONTROL DEVICE AND AUTONOMOUS PARKING SYSTEM
European Patent Office | 2020
|European Patent Office | 2024
|European Patent Office | 2021
|