超车
灵活性(工程)
计算机科学
巡航控制
MATLAB语言
博弈论
工具箱
序贯博弈
模拟
纳什均衡
过程(计算)
控制(管理)
数学优化
工程类
人工智能
运输工程
微观经济学
经济
程序设计语言
操作系统
统计
数学
作者
Yuewen Yu,Shikun Liu,Peter J. Jin,Xia Luo,Mengxue Wang
标识
DOI:10.1177/0361198120940990
摘要
The lane-changing decision-making process is challenging but critical to ensure safe and smooth maneuvers for autonomous vehicles (AVs). Conventional Gipps-type algorithms lack the flexibility for practical use under a mixed autonomous vehicle and human-driven vehicle (AV-HV) environment. Algorithms based on utility ignore the reactions of surrounding vehicles to the lane-changing vehicle. Game theory is a good way to solve the shortcomings of current algorithms, but most models based on game theory simplify the game with surrounding vehicles to the game with the following vehicle in the target lane, which means that the lane-changing decision under a mixed environment is not realized. This paper proposes a lane-changing decision-making model which is suitable for an AV to change lanes under a mixed environment based on a multi-player dynamic game theory. The overtaking expectation parameter (OEP) is introduced to estimate the utility of the following vehicle, OEP can be calculated by the proposed non-lane-based full velocity difference model with the consideration of lateral move and aggressiveness. This paper further proposes a hybrid splitting method algorithm to obtain the Nash equilibrium solution in the multi-player game to obtain the optimal strategy of lane-changing decision for AVs. An adaptive cruise control simulation environment is developed with MATLAB’s Simulink toolbox using Next Generation Simulation (NGSIM) data as the background traffic flow. The classic bicycle model is used in the control of involved HVs. Simulation results show the efficiency of the proposed multi-player dynamic game-based algorithm for lane-changing decision making by AVs under a mixed AV-HV environment.
科研通智能强力驱动
Strongly Powered by AbleSci AI