A Driving-By-Targets Algorithm Based on Deep Reinforcement Learning for Intelligent Vehicle
Hongbo Gao, M H Zhao, Zhengtian Wu, Yong Wang, Yu Kang, Yuchao Liu, Shaoyuan Li, Danwei Wang 等 9 位
University of Science and Technology of China Suzhou University of Science and Technology Shanghai Jiao Tong University Nanyang Technological University
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
Due to the complex nonlinear control systems and the variability of driving tasks, intelligent vehicles face significant challenges in control. To address this problem, the theory of driving-by-targets is proposed to formalize basic driving tasks using mathematical models. Then, a driving-by-targets controller based on deep reinforcement learning (DRL) is designed to achieve coupled lateral–longitudinal control. Considering the driving task uncertainty, the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm trains the proposed controller. The state information is directly mapped to the steering angle and acceleration output, enabling control in a continuous action space. In addition, the reward function is constructed to facilitate policy learning. Experiments show the accuracy and real-time performance of the driving-by-targets controller.
逐年被引趋势
暂无年度引用数据
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
计算机 / AIReinforcement Learning in Robotics
Autonomous Vehicle Technology and Safety · Vehicle Dynamics and Control Systems