Regression Model of Autonomous Lateral Rectification with Driver Perception

Abstract

It is usually hard for autonomous vehicle to perform human-likely in lateral rectification between desired path and ego-vehicle, since those methods choose linear preview point according to speed. This paper presents a regression model of autonomous lateral rectification with driver perception to describe relationship between preview point and speed. It is shown in Driver Perception Model(DPM) that logarithmic relationship between speed and preview point distance. Experimental results show that compared with lateral rectification of Stanley method, regression model of autonomous lateral rectification with driver perception can 1) acquire lower lateral acceleration, which enhances comfort; 2) promote robustness of algorithm on parameter tuning.