基于深度强化学习的移动机器人轨迹跟踪和动态避障

吴运雄; 曾碧

doi:10.12052/gdutxb.180029

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

基于深度强化学习的移动机器人轨迹跟踪和动态避障

综合研究 | 更新时间：2026-05-27

- 基于深度强化学习的移动机器人轨迹跟踪和动态避障
- Trajectory Tracking and Dynamic Obstacle Avoidance of Mobile Robot Based on Deep Reinforcement Learning
- 广东工业大学学报 2019年36卷第1期页码：42-50
- 作者机构：
  
  广东工业大学　计算机学院,广东,广州,510006
- 作者简介：
  
  吴运雄(1992–)，男，硕士研究生，主要研究方向为计算机视觉、深度强化学习
  曾碧(1963–)，女，教授，博士，主要研究方向为智能机器人、嵌入式智能计算. E-mail：zb9215@gdut.edu.cn
- 基金信息：
  
  广东省自然科学基金资助项目(2016A030313713);广东省应用型科技研发专项项目(2015B090922012);广东省产学研合作专项项目(2014B090904080)
- DOI：10.12052/gdutxb.180029
  中图分类号： TP242.6
- 网络首发：2019-01-25，
  
  纸质出版：2019
- 稿件说明：
移动端阅览
吴运雄, 曾碧. 基于深度强化学习的移动机器人轨迹跟踪和动态避障[J]. 广东工业大学学报, 2019, 36(1): 42-50.

Wu Yun-xiong, Zeng Bi. Trajectory Tracking and Dynamic Obstacle Avoidance of Mobile Robot Based on Deep Reinforcement Learning[J]. 2019, 36(1): 42-50.
吴运雄, 曾碧. 基于深度强化学习的移动机器人轨迹跟踪和动态避障[J]. 广东工业大学学报, 2019, 36(1): 42-50. DOI： 10.12052/gdutxb.180029.

Wu Yun-xiong, Zeng Bi. Trajectory Tracking and Dynamic Obstacle Avoidance of Mobile Robot Based on Deep Reinforcement Learning[J]. 2019, 36(1): 42-50. DOI： 10.12052/gdutxb.180029.

摘要

针对移动机器人在局部可观测的非线性动态环境下

实现轨迹跟踪和动态避障时容易出错和不稳定的问题

提出了基于深度强化学习的视觉感知与决策方法. 该方法以一种通用的形式将卷积神经网络的感知能力与强化学习的决策能力结合在一起

通过端对端的学习方式实现从环境的视觉感知输入到动作的直接输出控制

将系统环境感知与决策控制直接形成闭环

其中最优决策策略是通过最大化机器人与动力学环境交互的累计奖回报中学习获得. 仿真实验结果证明

该方法可以满足多任务智能感知与决策要求

较好地解决了传统算法存在的容易陷入局部最优、在相近的障碍物群中震荡且不能识别路径、在狭窄通道中摆动以及障碍物附近目标不可达等问题

并且大大提高了机器人轨迹跟踪和动态避障的实时性和适应性.

Abstract

A method of visual perception and decision making based on deep reinforcement learning was proposed

to solve the problem of malfunction and instability in the trajectory tracking and dynamic obstacle avoidance of mobile robot in a partly observable nonlinear dynamic environment. This method was used in a general form to combine the perceptual ability of convolutional neural network (CNN) with the decision-making ability of reinforcement learning. The visual perception input of environment was transformed into the direct output control of actions by the way of end-to-end learning style

so that the system environment perception and decision-making control directly formed a closed loop. The optimal decision-making strategy was acquired from the maximization of interactive cumulative reward between robot and dynamic environment. The results of simulation experiment showed that this method could meet the requirements of multi-task intelligent perception and decision making

and well solve problems of the traditional algorithm such as easily falling into local optimum

vibrating and failing to identify the path among the similar obstacles

wavering in the narrow passage and failing to reach the targets near obstacle. It greatly improved the instantaneity and adaptability of robot trajectory tracking and dynamic obstacle avoidance.

关键词

Keywords

references

ZENG B, LIN Z P, DENG J H. Algorithm research on recognition and improvement for corridor of autonomous mobile robot [J]. Journal of Guangdong University of Technology, 2016, 33(5): 9-14.

MA X D, ZENG B, YE L F. An improved visual odometry/SINS integrated localizationalgor-ithm based on BA [J]. Journal of Guangdong University of Technology, 2017, 34(6): 32-36.

PRENTICE S, ROY N. The Belief roadmap: efficient planning in belief space by factoring the covariance [J]. Robot, 2009, 29(11-2): 1448-1465.

KOREN Y, BORENSTEIN J. Potential field methods and their inherent limitations for mobile robot navigation [J]. IEEE International Conference on Robotics and Automation, 2002, 2(2): 1398-1404.

YANG S X, LUO C. A neural network approach to complete coverage path planning [J]. IEEE Transactions on Systems Man Cybernetics Part B Cybernetics A Publication of the IEEE Systems Man Cybernetics Society, 2004, 34(1): 718-724.

CASTILLO O, TRUJILLO L, MELIN P. Multiple objective genetic algorithms for path-planning optimization in autonomous mobile robots [J]. Soft Computing, 2007, 11(3): 269-279.

CLERC M, KENNEDY J. The particle swarm explosion, stability, and convergence in a multidimensional complex space [J]. IEEE Trans Evolutionary Computation, 2002, 6(1): 58-73.

LILLICRAP T P, HUNT J J, PRITZEL A, et al . Continuous control with deep reinforcement learning [J ] . Computer Science, 2015, 8(6): A187-A195.

SUN Z J, XUE L, XU Y M, et al . Overview of deep learning [J ] . Application Research of Computers, 2012, 29(8): 2806-2810.

VOLODYMYR M, KORAY K, DAVID S. Human-level control through deep reinforcement learning [J]. Nature, 2015, 518(7540): 529-536.

MNIH V, KAVUKCUOGLU K, SILVER D, et al . Playing atari with deep reinforcement learning [J ] . Computer Science, 2013, 56(1): 201-220.

SCHULMAN J, LEVINE S, MORITZ P, et al . Trust region policy optimization [J ] . Computer Science, 2015, 24(1): 1889-1897.

VAN H V, GUEZ A, SILVER D. Deep reinforcement learning with double q-learning [J]. Computer Science, 2015, 34(2): 2094-2100.

GLASCHER J, DAW N, DAYAN P, et al . States versus rewards:dissociable neural prediction error signals underlying model-based and model-free reinforcement learning [J ] . Neuron, 2010, 66(4): 585-595.

SYLVAIN G, SILVER D. Monte-Carlo search and rapid action value estimation in computer go [J]. Artificial Intelligence, 2011, 175(11): 1856-1875.

MNIH V, PUIGDOMENECH A, MEHDI M. Asynchronous methods for deep reinforcement learning [J]. Journal of Machine Learning Research, 2016, 33(6): 1928-1937.

MNIH V, KAVUKCUOGLU K, SILVER D, et al . Human-level control through deep reinforcement learning [J ] . Nature, 2015, 518(7540): 529-533.

LEVINE S, FINN C, DARRELL T, et al . End to-end training of deep visuomotor policies [J ] . Journal of Machine Learning Research, 2016, 17(1): 1334-1373.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于方向约束强化学习的左心室内膜分割

面向多用户动态频谱接入的改进双深度Q网络方法研究

离散制造智能工厂场景的AGV路径规划方法

人工噪声辅助无人机太赫兹安全通信的鲁棒设计

多套餐共享储能模式下考虑功率互济的社区联盟优化运行策略