Real-time Autonomous Parking in Unstructured Scenarios with an Indirect Optimal Control Approach
Abstract
This paper presents a framework to generate in real-time autonomous parking maneuvers for generic and unstructured parking scenarios. Our method is based on the solution of minimum-time optimal control problems by means of an indirect approach. In a single optimization, the framework computes parking maneuvers composed of two or more segments of forward and reverse driving. The trajectory planning tasks are solved in real-time, and a fine grid is used to discretize the domain of the optimal control problems, resulting in accurate and collision-free solutions. Moreover, we introduce a novel method to deal with static obstacles in the optimal control problems, using penalty functions defined as regularized three-dimensional clip functions. The results show the effectiveness of our approach in various scenarios with narrow parking spots.
Downloads