Self-driving Paths for Automobile Urban Parking

Abstract

This text discusses autonomous vehicles for parking in smart cities. Three path generation methods are implemented to avoid obstacles: geometric shapes, polynomial parameterization, and genetic algorithm. Tracking uses a pure pursuit algorithm. The results in a simulated 3D environment show the effectiveness of the algorithms. For example, for parallel front-in parking, the method based on genetic algorithm was the fastest (17.59 s) and shortest (1.768 m). For reverse perpendicular parking, the method with geometric shapes performed best (16.39 s and 1.645 m). This provides a basis for further investigations into autonomous parking.