Research on optimal driving behavior decision method based on multi-objective optimization

Abstract

Aiming at the coupling of traffic laws and traffic participants in the current smart car decision-making process, and the lack of systematic processing of traffic laws, this paper proposes an optimal driving behavior decision-making method based on multi-objective optimization. This paper uses a multi-objective optimization method to establish macro path following indicators, collision risk indicators, traffic efficiency evaluation indicators, and driving burden evaluation indicators to solve the problem of convergence to local solutions and divergence when the existing methods are solved. In addition, the evaluation of the macro path followability is established to ensure that the selected area meets the macro path. In the collision risk assessment, the method based on the hidden Markov model is used to identify the driving intention of the target car, and establish lane occupancy characteristics to determine the risk of collision and the possibility of movement based on the statistical characteristics of the driving habits of object cars. After verification and comparison with existing methods, the optimal driving behavior decision method proposed in this paper is effective and performs well.

Study of medium and long-term free flow capacity and queue discharge rates on roads

“…Interestingly, the distribution of lane flow does not typically have a minimum or maximum value, instead, its value fluctuates in tandem with traffic demand [ 6 ]. A multitude of scholars has delved into the study of driving behavior and its consequential impact on traffic, including parameters of driving behavior under specific circumstances [ 7 ] and lane parameters [ 8 ]. Studies conducted on free-flow speed and fundamental patterns under non-congested conditions have demonstrated that road characteristics, including road capacity, heavily rely on driver behavior and are likely to evolve over time [ 9 ].…”