Constraint-Oriented Obstacle Avoidance Control for Autonomous Vehicles Without Local Trajectory Replanning

Abstract

Obstacle avoidance, as an indispensable part of the autonomous driving process, plays an essential role in safeguarding vehicular safety. The intricacies of the driving environment coupled with the uncertainties in vehicle dynamics render the formulation of an obstacle avoidance strategy a formidable challenge. In this study, a novel obstacle avoidance control is proposed for autonomous vehicles that eschews local trajectory replanning based on the principle of constraint-following. Initially, trajectory tracking is achieved by formulating equality constraints on the vehicle's states, which are based on kinematic relationships between the desired trajectory and the controlled vehicle. By analyzing the geometric relationships between obstacles and the vehicle, the obstacle avoidance inequality constraints of the vehicle position are established. Based on a potential function, we transform the inequality constraints into equality constraints, thereby recasting the obstacle avoidance as a constraint-following control problem. Subsequently, a closed-form constraint force based on the Udwadia-Kalaba (U-K) approach and an adaptive law are put forward. Through Lyapunov minimax analysis, it has been demonstrated that the derived control ensures the constraint-following performance. Finally, the Simulink-CarSim co-simulations are implemented. The results indicate that the proposed control guarantees the vehicle trajectory tracking and collision-free performance.