Path Planning for Automobile Urban Parking Through Curve Parametrization and Genetic Algorithm Optimization

Abstract

Parking a car is a difficult task and may be frustrating and stressful for the driver, while commonly causes a traffic jam. One way to mitigate such negative effects is to provide vehicles with self-driving capabilities. As a cornerstone of an automobile ability to move autonomously stands path planning, which despite many works in the literature, is still considered an open problem, especially with regard to nonholonomic vehicles. Based on this scenario, this work presents a path planning algorithm to parallel park an automobile based on polynomial parametrization and genetic algorithm optimization. The aim is to define a law of motion to lead the vehicle from an initial pose near a parking space to a final pose within the latter smoothly, with no interruption and avoiding any obstacles in the way. Simulation results in 3D physics simulation environment are presented to validate the feasibility of the proposed algorithm, which lay the foundation to broader studies.