Training a path planning algorithm for automatic parking of a motor vehicle, path planning for automatic parking of a motor vehicle, and automatic parking of a motor vehicle.

Abstract

To train a path planning algorithm, the following steps are performed for a variety of training scenarios (18, 19, 20): A first path planning agent (16) generates a simulated first path segment from a vehicle start state to a simulated intermediate vehicle state based on a first reinforcement learning strategy, and a second path planning agent (16) generates a simulated second path segment from the simulated intermediate vehicle state to a simulated final vehicle state based on a second reinforcement learning strategy. The respective values ​​of at least one reward or punishment function are determined, and the reinforcement learning strategies are updated accordingly.At least one reward or punishment function depends on a deviation of the simulated intermediate vehicle state from a vehicle anchor state and on a deviation of the simulated final vehicle state from a respective vehicle target state.

Images