| 7 | 0 | 90 |
| 下载次数 | 被引频次 | 阅读次数 |
针对城市双车道场景下智能车辆变道决策易保守、固定奖励强化学习难以兼顾安全性与通行效率问题，提出一种模糊逻辑深度确定性策略梯度(FL-DDPG)变道决策方法。首先，选取自车速度和目标车道前车相对速度差Δvreal作为风险-收益表征量，构建双输入双输出模糊控制器，通过隶属函数和模糊规则自适应输出安全权重和效率权重；其次，将模糊权重嵌入DDPG奖励函数，并设计连续Actor输出到离散高层变道动作的阈值映射机制，使智能体能够在不同交通风险水平下动态调整变道倾向；最后，在CARLA仿真平台中构建4类典型变道场景，并与传统DDPG方法进行对比验证。结果表明，FL-DDPG平均变道率由37.25%提高至56.25%,提高19.00个百分点，相对增幅为51.01%;平均碰撞率增加2.87个百分点。结果说明所提方法在提升主动变道能力的同时会引入一定探索风险，需结合安全校核模块进一步约束高风险动作。
Abstract：To address the conservative lane-changing decisions of intelligent vehicles in urban two-lane scenarios and the difficulty of balancing safety and traffic efficiency in fixed-reward reinforcement learning, a Fuzzy Logic Deep Deterministic Policy Gradient(FL-DDPG) lane-changing decision method is proposed. First, the ego-vehicle speed v and the relative speed difference Δvreal of the preceding vehicle in the target lane are selected as risk-benefit indicators, and a dual-input and dual-output fuzzy controller is constructed to adaptively output the safety weight ws and efficiency weight we through membership functions and fuzzy rules. Second, the fuzzy weights are embedded into the DDPG reward function, and a threshold mapping mechanism from continuous Actor output to discrete high-level lane-changing actions is designed. Finally, four typical lane-changing scenarios are constructed in CARLA for comparative validation. The average lane-changing rate increases from 37.25% to 56.25%, with a relative increase of 51.01%, while the average collision rate rises by 2.87 percentage points. The results indicate that the proposed method improves active lane-changing capability but introduces certain exploration risks, which should be further constrained by a safety checking module.
[1] 陈虹，郭露露，宫洵，等.智能时代的汽车控制[J].自动化学报，2020,46(7):1313-1332.
[2] 冯付勇，魏超，吕彦直，等.基于GCN和QP的智能车辆换道决策规划[J].北京理工大学学报，2024,44(8):820-827.
[3] Liu X C,Hong L,Lin Y R.Vehicle lane change models:A historical review[J].Applied Sciences,2023,13(22):12366.
[4] Zheng X,Li H,Zhang Q,et al.Intelligent decision-making method for vehicles in emergency conditions based on artificial potential fields and finite state machines[J].Journal of Intelligent and Connected Vehicles,2024,7(1):19-29.
[5] He S,Zeng J,Zhang B,et al.Rule-based safety-critical control design using control barrier functions with application to autonomous lane change[C]//American Control Conference.New York:IEEE,2021:178-185.
[6] 严永俊，彭林，王金湘，等.混驾环境下基于主从博弈的多车协同决策规划[J].中国公路学报，2024,37(3):117-133.
[7] Yu H,Tseng H E,Langari R.A human-like game theory-based controller for automatic lane changing[J].Transportation Research Part C:Emerging Technologies,2018,88:140-158.
[8] 李伟东，马草原，史浩，等.基于分层强化学习的自动驾驶决策控制算法[J].吉林大学学报(工学版),2025,55(5):1798-1805.
[9] Wang P,Chan C Y,De L F A.A reinforcement learning based approach for automated lane change maneuvers[C]//IEEE Intelligent Vehicles Symposium.New York:IEEE,2018:1379-1384.
[10] Ye F,Cheng X,Wang P,et al.Automated lane change strategy using proximal policy optimization-based deep reinforcement learning[C]//IEEE Intelligent Vehicles Symposium.New York:IEEE,2020:1746-1752.
[11] Li D,Liu A.Personalized lane change decision algorithm using deep reinforcement learning approach[J].Applied Intelligence,2023,53(11):13192-13205.
[12] Shen Y,PengG J,Kang C,et al.Collaborative optimisation of lane change decision and trajectory based on double-layer deep reinforcement learning[J].International Journal of Vehicle Design,2023,92(2/4):336-356.
[13] Guo H,Keyvan E M,Xie K.Modeling coupled driving behavior during lane change:A multi-agent transformer reinforcement learning approach[J].Transportation Research Part C:Emerging Technologies,2024,165:104703.
[14] Wang J,Zhang Q,Zhao D.Highway lane change decision-making via attention-based deep reinforcement learning[J].IEEE/CAA Journal of Automatica Sinica,2022,9(3):567-569.
[15] Bayuwindra A,Wonohito L,Trilaksono B R.Design of DDPG-based extended look-ahead for longitudinal and lateral control of vehicle platoon[J].IEEE Access,2023,11:96648-96660.
[16] Han L,Ma X Y,Wang Y R,et al.Safe and efficient DRL driving policies using fuzzy logic for urban lane changing scenarios[J].Journal of Intelligent and Connected Vehicles,2025,8(1):9210054-1-9210054-13.
[17] 潘超，王雪涵，高俊平，等.基于深度学习的计算机视觉技术在交通场景中的应用[J].长春工业大学学报，2022,43(3):251-257.
[18] 刘丽伟，侯德彪，侯阿临，等.基于SimAM-YOLOv4的自动驾驶目标检测算法[J].长春工业大学学报，2022,43(3):244-250.
基本信息:
DOI：10.15923/j.cnki.cn22-1382/t.2026.3.02
中图分类号:TP18;U463.6
引用信息:
[1]汉家欣,王乙壬,韩玲.模糊强化学习驱动的智能车辆变道决策方法研究[J].长春工业大学学报,2026,47(03):202-210+307.DOI:10.15923/j.cnki.cn22-1382/t.2026.3.02.
基金信息:
国家自然科学基金(5237120306)
2026-06-15
2026-06-15