Control of Vehicle Trajectory Tracking Under Extreme Conditions Based on ILQR

Vehicle maneuvers resembling drifting offer a significantly expanded range of motion, particularly valuable in extreme collision avoidance scenarios. However, existing control algorithms fall short in fulfilling the demands of trajectory tracking tasks under such extreme conditions. This paper introduces a novel ILQR controller capable of guiding vehicles through maneuvers akin to drifting, particularly in highly demanding collision avoidance scenarios. To address stability constraint issues within the controller, we integrate the Projected Newton method with ILQR theory to resolve state constraint problems within the QP framework. Additionally, input constraint issues are addressed through the utilization of hyperbolic cosine functions. Ultimately, experimental results demonstrate the vehicle's proficiency in accurately tracking collision avoidance trajectories.