Platoon Separation Strategy Optimization Method Based on Deep Cognition of a Driver’s Behavior at Signalized Intersections

Abstract

Semantic understanding of drivers' behavior features at intersections plays a pivotal role in the proper decision-making of a platoon. This paper presents a flexible framework to automatically extract the driver's driving features from observed temporal sequences of driving raw data and traffic light information. An approach, which contains two key sub-problems, is proposed to select the separated vehicles from the platoon in the vicinity of the intersection. Then, the first sub-problem, accurately capturing the drivers' driving behavior features under the impact of traffic lights, is addressed by using the Bayesian nonparametric approach, which could segment drivers' driving raw data temporal sequences into small analytically interpretable components (called driving primitives) without using prior knowledge. In addition, the extracted driving primitives are used to obtain the vehicle separation strategy (which is also the second sub-problem) by considering safety, efficiency, and energy consumption. Finally, 200 groups of raw data of human-driven vehicles approaching the intersection are used to validate the effectiveness of the proposed primitive-based framework. Experimental results demonstrate that the acceleration indeterminacy of separated vehicles could be decreased 37%-72% by segmenting the captured driving behavior features into 3×15 patterns. Moreover, the vehicle separation strategy could not only increase the efficiency, but also the safety, and the energy consumption could be decreased.