SIA OpenIR  > 其他
移动机器人自主跟踪与动态避障研究
Alternative TitleResearch on Mobile Robot Autonomous Tracking and Dynamic Obstacle Avoidance
任恒乐1,2
Department其他
Thesis Advisor徐方
Keyword同时定位与建图 社会力模型 行人模型 人工势场法 A*算法
Pages62页
Degree Discipline检测技术与自动化装置
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本文首先对移动机器人的自主跟踪与动态避障技术的发展现状进行了概述,阐明其研究的难点以及未来的趋势;然后,在第二章对全向移动平台的搭建工作进行了简要介绍;接着围绕同时定位与建图、行人检测与识别及路径规划三部分展开研究,详细情况如下:(1)搭建了全向移动平台。设计移动平台机械结构,对其进行运动学分析。由于其没有非完整性约束,具有可向任意方向移动的优点,可使机器人在跟踪的时候尽可能的朝向目标行人,这样减小了跟踪丢失的可能性。这部分为后续的行人识别与跟踪工作奠定硬件基础。(2)利用深度相机基于ORB-SLAM实现了同时定位与地图重建功能。使得当机器人距离目标行人过远或由于干扰跟丢失后,机器人能够利用该部分进行自身的定位及对周围环境的感知建图,从而使机器人能够重新跟上目标行人。(3)本文提出了一种行人模型来进行行人检测与识别。该模型利用方向梯度直方图信息来初步检测当前帧中可能是行人的位置。然后结合深度图中的相应深度获得行人的实际三维坐标。接着利用社会力模型预测上一图像帧中每个待匹配行人的期望位置。另外,对每个行人分段提取颜色直方图来。将以上信息作为行人模型的输入,利用这些信息,行人模型将当前帧与上一帧图像中的行人相匹配来完成行人的检测与识别。通过实验验证了该模型在行人间短期内相互遮挡或者由于行人距离机器人太近而只能采集到人体局部信息时也能准确的识别。(4)对人工势场法加入了方向因子进行改进,将其与A*算法相结合来进行路径规划。通过实验验证了机器人能够实现对目标行人的自主跟随及对障碍物的动态避碰。
Other AbstractFirstly, the development status of autonomous tracking and dynamic obstacle avoidance technology for mobile robots is summarized, and the research difficulties and future trends are clarified. Then, in chapter 2, the construction of omnidirectional mobile platform is briefly introduced. Then, three parts of SLAM, pedestrian detection and recognition and path planning are studied. The details are as follows: (1) An omni-directional mobile platform was built. With the advantage that the robot can move in any direction without nonholonomic constraints, it can make the robot move towards the target pedestrian as far as possible in tracking, thus reducing the possibility of tracking loss. This part lays the hardware foundation for the subsequent pedestrian recognition and tracking work. (2) Using the depth camera based ORB-SLAM achieve simultaneous localization and mapping function reconstruction. When the robot is too far away from the target pedestrian or because of interference and loss, the robot can use this part to locate itself and map the perception of the surrounding environment, so that the robot can catch up with the target pedestrian again. (3) This paper presents a pedestrian model for pedestrian detection and recognition. The model uses the directional gradient histogram information to preliminarily detect the possible pedestrian position in the current frame. Then, the actual three-dimensional coordinates of pedestrians are obtained by combining the corresponding depth in the depth map. Then, the social force model is used to predict the desired position of each pedestrian in the previous image frame. In addition, color histogram is extracted for each pedestrian segment. With the above information as the input of the pedestrian model, the pedestrian model matches the current frame with the pedestrian in the previous image to complete pedestrian detection and recognition. Experiments show that the model can accurately recognize the human body when the pedestrians occlude each other in a short time or because the pedestrians are too close to the robot and can only collect local information of the human body. (4) The direction factor is added to the artificial potential field method to improve it, and it is combined with A* algorithm to carry out path planning. Experiments show that the robot can achieve autonomous tracking of the target pedestrian and dynamic collision avoidance of obstacles.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/25173
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
任恒乐. 移动机器人自主跟踪与动态避障研究[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
移动机器人自主跟踪与动态避障研究.pdf(2390KB)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
[任恒乐]'s Articles
Baidu academic
Similar articles in Baidu academic
[任恒乐]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[任恒乐]'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.