Integrated Path Planning and Tracking Control of Autonomous Vehicle for Collision Avoidance based on Model Predictive Control and Potential Field

Abstract

This paper attempts to implement an integrated path planning and tracking controller that generates optimal control inputs guaranteeing a collision-free path. This integrated scheme is achieved by unifying model predictive control (MPC) with a potential field for assigning collision risk. The target vehicle is an electrified autonomous vehicle that is capable of directly controlling traction and braking torques of vehicle. Wheel torque and steering input of vehicle are optimized by receding horizon optimization (RHO) and give us stable and comfort reference trajectories. In the optimization process, control inputs, tracking errors, and collision risk are to be minimized in a single objective function. Collision risk is taken into account by modeling proper potential fields that allow the controller to re-plan the given desired path for avoiding a collision. Simulation is conducted using a high-fidelity vehicle plant model and the control scheme demonstrates promising results by verifying optimal and stable path guaranteeing collision-free.