SIA OpenIR  > 其他
协作机器人的障碍物检测与避碰方法研究
Alternative TitleResearch on obstacle detection and collision avoidance of collaborative robot
康杰1,2
Department其他
Thesis Advisor贾凯
Keyword人机协作 骨骼跟踪 高斯混合模型 分层预测框架 反应式运动规划
Pages73页
Degree Discipline控制理论与控制工程
Degree Name硕士
2019-05-17
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract本文首先对人机协作的发展现状和研究背景进行概述,阐明其研究的主要内容以及当前的研究热点;然后,在第二章对人机协作仿真平台进行了介绍,并对课题相关理论进行了研究学习;在前两部分工作基础上,本文针对不同的问题开展了具体的工作,详细情况如下:(1)考虑实际作业场景,利用深度相机获取深度信息并进行点云处理,从而确定人体各关节点的位置并实时跟踪,最后我们在仿真环境中搭建了人体模型,与真实环境中的人体实时同步运动,为轨迹预测提供了很好的基础。(2)针对传统方法的不足,即可以判断关节的运动趋势,但是只对手臂末端进行预测,而对于肘部等不能同时给出良好的预测结果,特别是异常的轨迹,更不能给出良好的判断,因此以上的方法还有很大的局限性,基于这些考虑,本文基于Linux系统和机器人操作系统ROS搭建了协作机器人仿真平台,准确的捕捉手臂各关节的运动轨迹,同时提出了基于minimum-jerk的异常行为预测方法和可同时预测多个关节运动的分层轨迹预测框架,良好的解决了以上算法的局限性,提高了系统的适应能力和鲁棒性,更具有实际意义。(3)利用基于采样的运动规划方法RRT-Connect,从安全和效率角度进行优化,实现实时的反应式运动规划。首先按照预定规划好的轨迹进行运动,假设在机器人运动过程中检测到新的变化,则需要做以下决策,并重新规划路径,以满足安全性、可靠性和高效性。在环境变化的情况下,如果在正在执行的规划路径上预计到碰撞,规划器立即开始重新规划,一旦再次获得另一条无碰撞路径,就执行它;在路径重新规划过程中,除非机器人接近指定安全距离内的障碍物,否则机器人将继续运动。在重新规划过程中,预期出现的障碍物可能会消失,在这种情况下,重新规划取消,机器人继续执行原始路径而不停止。最后,在机器人仿真环境中进行了测试,实验结果表明,该方法有效的缩短了运动规划时间,提高了系统可靠性。
Other AbstractThis paper firstly summarizes the development status and research background of man-machine collaboration, and expounds the main content of its research and current research hotspots. Then, in the second chapter, the man-machine cooperative simulation platform is introduced, and the related theories are studied. On the basis of the previous two parts, this paper has carried out specific work on different issues, with details as follows: (1) Considering the actual working scene, the depth information was obtained by using the depth camera and point cloud processing, so as to determine the position of each human node and track it in real time. Finally, the human model was built in the simulation environment to synchronize with the real time movement of the human body in the real environment, which provided a good foundation for trajectory prediction. (2) Aiming at the shortcomings of the traditional method, which can judge the trend of movement, but only to forecast the end of the arm, and for the elbow and so on can't good prediction results are given at the same time, especially the trajectory anomalies, more can't give good judgment, so the above method has much limitation, based on these considerations, this article is based on Linux system and robot operating system built collaborative robot simulation platform ROS, accurately capture the arm trajectory of each joint, meanwhile, a minimum-jerk method for predicting abnormal behavior and a hierarchical trajectory prediction framework for predicting multiple joint motions were proposed, it solves the limitation of the above algorithm, improves the adaptability and robustness of the system, and has practical significance. (3) RRT-Connect, a sample-based motion planning method, was used to optimize the motion planning from the perspective of safety and efficiency, and real-time reactive motion planning was realized. Firstly, the robot moves according to the predetermined planned trajectory. Assuming that new changes are detected during the robot's movement, the following decisions need to be made and the path needs to be re-planned to meet the requirements of safety, reliability and efficiency. In the case of environmental changes, if a collision is expected on the planning path being executed, the planner immediately starts replanning and executes it once another collision-free path is obtained. During the path replanning process, the robot will continue to move unless it approaches an obstacle within a specified safe distance. The path is executed by slowing down when approaching an obstacle and stopping completely at a safe distance; During the replanning process, the expected obstacles may disappear. In this case, the replanning is cancelled and the robot continues to execute the original path without stopping. Finally, the simulation results show that the proposed method can effectively shorten the motion planning time and improve the system reliability.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/25177
Collection其他
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院大学
Recommended Citation
GB/T 7714
康杰. 协作机器人的障碍物检测与避碰方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2019.
Files in This Item:
File Name/Size DocType Version Access License
协作机器人的障碍物检测与避碰方法研究.p(2812KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[康杰]'s Articles
Baidu academic
Similar articles in Baidu academic
[康杰]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[康杰]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.