您的位置:首页  > 论文页面

智能车轨迹预测及动态避障

发表时间:2023-03-30  浏览量:3362  下载量:191
全部作者: 蒋金桥,黄海波
作者单位: 国家工业信息安全发展研究中心
摘 要: 研究针对智能车轨迹预测及动态避障问题进行研究。通过对环境的实时感知,完成环境的栅格化重建,导出分层代价地图,实现智能车的定位和实时环境建图;依托卡尔曼滤波器算法预测移动障碍物行进轨迹,结合人工势场法,确定危险系数,划分智能车危险区域,通过斥力规避障碍物,引导智能车进行动态避障。基于此考虑势场法中易出现的局部极小点问题,进行势场梯度下降的优化工作,确保智能车能以更加平滑的姿态实现动态障碍物的躲避。本文通过ROS平台的Gazebo软件,对智能车路径规划和动态避障问题进行仿真验证。实验结果表明:结合卡尔曼滤波器和势能场法可实现智能车对动态障碍物的规避,具有较强的环境适应性及实用性。
关 键 词: 机械工程其他学科;动态避障;轨迹预测;仿真模拟
Title: Intelligent vehicle trajectory prediction and dynamic obstacle avoidance
Author: JIANG Jinqiao, HUANG Haibo
Organization: China Industrial Control Systems Cyber Emergency Response Team
Abstract: In this paper, intelligent vehicle trajectory prediction and dynamic obstacle avoidance were studied. Through the real-time perception of the environment, the rasterized reconstruction of the environment is completed, the hierarchical cost map is derived and the positioning of the intelligent vehicle and real-time environment mapping is realized. Relying on Kalman filtering algorithm, the trajectory of moving obstacles is predicted. Combined with the artificial potential field method, the risk factor is determined, the dangerous area of the smart car is divided. By avoiding obstacles with repulsive force, the smart car is guided to avoid obstacles dynamically. Based on this, considering the local minimum point problem that is easy to occur in the potential field method, the optimization work of the potential field gradient descent is carried out to ensure that the smart car can avoid dynamic obstacles with a smoother attitude. In this paper, the Gazebo software of the ROS platform is used to simulate and verify the path planning and dynamic obstacle avoidance problems of intelligent vehicles. The experimental results show that with the combination of Kalman filter and potential energy field method, smart vehicles can avoid dynamic obstacles, which has strong environmental adaptability and practicability.
Key words: other subjects of mechanical engineering; dynamic obstacle avoidance; trajectory prediction; simulation
发表期数: 2023年3月第1期
引用格式: 蒋金桥,黄海波. 智能车轨迹预测及动态避障[J]. 中国科技论文在线精品论文,2023,16(1):1-7.
 
10 评论数 0
暂无评论
友情链接