Nonlinear Model Predictive Control for Aggressive Cornering Maneuver Considering Effect of Large Steering Angle

Herein, we describe a newly designed model predictive control algorithm. When we drive under condition of high speed and high steering, the front wheels of the vehicle experience a large lateral force. This lateral force causes longitudinal deceleration, which naturally reduces the vehicle speed. The model predictive control method used for the high level guidance of autonomous vehicles relies on a kinematic model with three states (x, y, and theta), and this model does not take into account the effect of steering on the longitudinal acceleration. We developed a model predictive controller for extreme maneuvering of autonomous driving vehicles, in which the influence of steering on the longitudinal acceleration is considered during cornering. We verified the improvement in terms of lap time reduction and the ability to track the reference trajectory more accurately.