系统仿真学报 ›› 2024, Vol. 36 ›› Issue (5): 1199-1210.doi: 10.16182/j.issn1004731x.joss.22-1546

• 研究论文 • 上一篇    下一篇

基于深度学习的机器人局部路径规划方法

刘泽森(), 毕盛(), 郭传鈜, 王延葵, 董敏   

  1. 华南理工大学 计算机科学与工程学院,广东 广州 510006
  • 收稿日期:2022-12-29 修回日期:2023-03-06 出版日期:2024-05-15 发布日期:2024-05-21
  • 通讯作者: 毕盛 E-mail:ftyg@live.com;picy@scut.edu.cn
  • 第一作者简介:刘泽森(2001-),男,本科生,研究方向为移动机器人系统。E-mail: ftyg@live.com
  • 基金资助:
    广东省科技计划(2020A0505100015);高校教师特色创新研究项目(2022DZXX03);华南理工大学“百步梯攀登计划”(j2tw202202079)

Deep Learning Based Local Path Planning Method for Moving Robots

Liu Zesen(), Bi Sheng(), Guo Chuanhong, Wang Yankui, Dong Min   

  1. School of Computer Science and Engineering, South China University of Technology, Guangzhou 510006, China
  • Received:2022-12-29 Revised:2023-03-06 Online:2024-05-15 Published:2024-05-21
  • Contact: Bi Sheng E-mail:ftyg@live.com;picy@scut.edu.cn

摘要:

为了将视觉信息融入到机器人导航过程中,提高机器人对各类障碍物的识别率,减少危险事件的发生,设计了基于二维CNN及LSTM的局部路径规划网络。提出了基于深度学习的局部路径规划方案利用机器人视觉信息及全局路径信息推理产生机器人在当前时刻完成避障导航任务所需转向角度搭建了用于对规划器核心神经网络进行训练和验证的室内场景提出了以路径总长度、平均曲率变化率及机器人与障碍物之间的距离为性能指标的路径评估方案。实验表明:该方案在仿真环境及真实场景中均体现了较优秀的局部路径生成能力。

关键词: 机器人导航, 路径规划, 实时避障, 深度学习

Abstract:

In order to integrate visual information into the robot navigation process, improve the robot's recognition rate of various types of obstacles, and reduce the occurrence of dangerous events, a local path planning network based on two-dimensional CNN and LSTM is designed, and a local path planning approach based on deep learning is proposed. The network uses the image from camera and the global path to generate the current steering angle required for obstacle avoidance and navigation. A simulated indoor scene is built for training and validating the network. A path evaluation method that uses the total length and the average curvature change rate of path and the distance between robot and obstacle as metrics is also proposed. Experiments show that the proposed approach has good local path generation capability in both simulated and real scenes.

Key words: robot navigation, path planning, real-time obstacle avoidance, deep learning

中图分类号: