Potential field-based path planning for emergency collision avoidance with a clothoid curve in waypoint tracking

Abstract

A novel potential field-based model curve fitting method (PF-MCF) is presented in this paper to handle emergency collision avoidance in waypoint tracking (following waypoints from a leading vehicle). It is reported that the PF has high performance on real-time obstacle avoidance in robotic path planning. However, some inherent limitations exist in the PF, such as no passage between closely spaced obstacles and oscillations in the presence of obstacles, which is rarely discussed in the research field of PF-based autonomous vehicles (AVs), especially in waypoint tracking. To solve these problems, we propose the PF-MCF method to transform the clothoid curve into a quadratic programming form and solve the optimization under reasonable constraints to consider the different waypoints from the PF and the leading vehicle. The proposed PF-MCF method is validated via MATLAB/Simulink and CarSim to avoid multiple obstacle vehicles in waypoint tracking compared with the two latest methods (PF-based model predictive controller and PF-based curve fitting method). The simulation results confirm the efficiency of finding a passage in closely spaced obstacle vehicles and eliminating oscillations (under 0.5 [deg] in tire steering angle) in the presence of obstacles compared to other latest methods.