Initiation and Stabilization of Drifting Motion of a Self-driving Vehicle with a Reinforcement Learning Agent

Abstract

Performing special driving techniques like drifting can be challenging even for professional human drivers. However, such maneuvers can be essential for avoiding accidents in critical road scenarios like evasive maneuvers. This paper reports novel research results whose main goal is to develop a self-driving agent for drift motion control based on vehicle simulation in MATLAB/Simulink. The state representation of the vehicle includes the longitudinal and lateral velocities with the yaw rate. The agent action space consists of two actuators: the throttle position and the roadwheel angle. The goal of the agent is twofold: first, it needs to jump into a drifting state; second, it has to keep the vehicle in drift. The simulation results show that the proposed RL agent is capable of learning to approach a predetermined drift equilibrium from cornering and staying in this drift situation as well. For the training, the solution excluded using any prior data. It only works with information gained from the simulation model, which is a remarkable difference from the actual state-of-the-art RL-based solutions.