Implementation of a Parking State Machine on Vision-Based Auto Parking Systems for Perpendicular Parking Scenarios
Saeid Sedighi, Duong-Van Nguyen, Klaus-Dieter Kuhnert · 2019
Recently autonomous parking has got a great attention from researchers from both Academy and Industry and being known as one of the key subsets of autonomous driving. Since autonomous driving is at its hype but with many remaining concerns, autonomous parking is more practical for daily applications and more feasible to be immediately realized mass-production in the automotive industry. Current solutions for autonomous parking require multiple maneuvers and a significant amount of time to park, which lead problematic in many real-life applications, depending on traffic situation around the ego-vehicle. In this paper, we present a parking strategy for vision-based autonomous parking systems in which the ego-vehicle is capable of completing its auto-park by one maneuver, or up to maximum three required maneuvers. Experiments show that the proposed method provides a significant improvement in reducing the number of maneuvers as well as time to park. Indeed, a rate of more than 95% successful parking in heavy traffic flows within a parking area is achieved using this method. We will also prove that the method is lightweight and very efficient for real-time applications.