Research Concept Vehicle - Automatic Parking
The project goal is to develop and implement a system with processor, bus and sensors capable of controlling the RCV so that it can start in front of the garage door, or any other point on the parking lot for that matter, and self-drive to a parking lot achieving a desired final pose while avoiding obstacles on the way there.
According to the requirements, an algorithm has been designed such that it has two states, one is a path tracking state and the other is an obstacle avoidance state. For the path tracking state, the RCV follows the dynamically adjusting path created by a fuzzy logic controller based on the real-time GPS location and heading data. When an obstacle is detected by our own sonar ranging system securing the whole perimeter of the RCV, the state is switched from path tracking to obstacle avoidance. Another fuzzy logic algorithm is applied here to get around the obstacles based on the distance signals collected by 5 ultrasonic sensors. When the risk is no longer posed, the path tracking state will once again take over and lead the vehicle to the target parking lot.
We have successfully tested the path tracking state with the RCV and achieved the real-world autonomous parking outside of the Transport Labs. Our method has been verified to be effective for the obstacle avoidance task using the Prescan simulation environment. Due to a time limitation this state has not been tested on the RCV.
Frank Wang and Yavor Trasiev
RCV - Autonomous Parking.pdf (pdf 2.0 MB)