Development Of An Adaptive Weighted Prediction-Based Model Predictive Control Using Disturbance Estimation For Longitudinal Autonomous Driving

Abstract

This paper presents an adaptive weighted prediction-based model predictive control using disturbance estimation for longitudinal autonomous driving. To reduce the negative effect of uncertainties on the control performance of model predictive control, the weighting function for linear prediction was designed based on an exponential function with disturbance estimation. The main role of the weighting function proposed in this study is to fade the effect of uncertainties as the prediction step increases. The disturbance of the longitudinal control model for autonomous driving was estimated using a sliding mode observer. The estimated disturbance was used to determine the main parameter of the weighting function. Performance evaluations were conducted using commercial software for reasonable evaluations of the control algorithm. Evaluation results show that the weighted prediction-based control algorithm can provide accurate tracking of desired states compared to the constant predictionbased control algorithm.