Model Predictive Path Planning Based on Artificial Potential Field and Its Application to Autonomous Lane Change
Pengfei Lin, Woo Young Choi, Seung-Hi Lee, Chung Choo Chung
Abstract
In this paper, we propose a vehicle lane change system using model predictive path planning (MPPP) based on the artificial potential field (APF) for speeding vehicles. It is shown that APF has high performance in real-time obstacle avoidance. However, it remains unpractical for self-driving cars because the point model used for the APF ignores the lateral vehicle dynamics for the lane-keeping system. To resolve the problem, this paper introduces a novel curve-fitting method combined with the APF applied to plan a drivable path for autonomous vehicles in the lane change action. The proposed system was validated through MATLAB/Simulink with the empirical kinematic model. The simulation results indicate that the model predictive path planning algorithm is highly effective in high-speed lane change scenarios to avoid dynamic obstacle vehicles.