摘要：Abstract"Human-vehicle-road" collaborative optimization is an important issue in current research. And the topic of how to build an appropriate driver model integrating future traffic information has not been totally solved. In this paper, Gaussian Mixture Model (GMM) was adopted to learn different driving modes. And a linear acceleration model, was identified by least square method which was weighted by probability density of the GMM. Therefore, the most energy-saving acceleration model could be evaluated by a comprehensive score. Based on this model, the planning and control of driving speed in single traffic light scene was explored. Simulation results show that the planning strategy proposed in this paper can save fuel up to 15.81% compared with the aggressive driver; and 14.54% reduced compared with the fixed speed planning algorithm based on vehicle to infrastructure (V2I) communication, with traffic light restriction satisfied.
关键词：KeywordsDriving BehaviorGaussian Mixture ModelTraffic-lights OptimizationEco-drivingActive Disturbance Rejection Control (ADRC)
