Human-Centric Autonomous Driving Based on a Two-Stage Machine Learning Algorithm

This paper presents a human-centric autonomous driving system, which is based on a two-stage machine learning algorithm. In particular, driving perception and human features are integrated to develop human-centric autonomous vehicles. Hence, we propose two-stage machine learning algorithms to identify the driver features such as age, location, sense, etc. We consider both online and offline learning to construct a two-stage distribution model and determine the relationship between the driver features and its cluster. The simulation results show the performance of the proposed two-stage learning algorithms in terms of a driver’s feature training performance.