硕士学½论文
基于
SLAM
的扫地机器人控制系统研究
RESEARCH OF CONTROL SYSTEM OF
SWEEPING ROBOT BASED ON SLAM
陈玉
哈尔滨工业大学
2017
年
6
月
½内图书分类号:
TP242.3
½际图书分类号:621
学校代码:
10213
密级:公开
工程硕士学½论文
基于
SLAM
的扫地机器人控制系统研究
硕 士 研 究 生
: 陈玉
导
师
: 王伟杰副教授
申 请 学 ½
: 工程硕士
学
科
: 机械工程
所 在 单 ½
: 机电工程学院
答 辩 日 期
:
2017
年
6
月
授予学½单½
: 哈尔滨工业大学
Classified Index: TP242.3
U.D.C: 621
Dissertation for the Master Degree in Engineering
RESEARCH OF CONTROL SYSTEM OF
SWEEPING ROBOT BASED ON SLAM
Candidate:
Supervisor:
Academic Degree Applied for:
Specialty:
Affiliation:
Date of Defence:
Degree-Conferring-Institution:
Chen Yu
Associate Prof. Wang Weijie
Master of Engineering
Mechanical Engineering
School of Mechatronics Engineering
June, 2017
Harbin Institute of Technology
哈尔滨工业大学工程硕士学½论文
摘
要
吸尘器的发明,改善了人们的劳动条件,提高了人们的清洁效率并降½了
了人们的劳动强度,½清洁环境的工½变得简单而有效率。½传统的吸尘器½
积大、需要专门人员操½,智½化程度不足。随着科技的不断进步,扫地机器
人结合了传统吸尘器与自主移动机器人的优点出现在市场上,
并从最初的随机
碰撞式扫地机器人逐渐向着路径规划式扫地机器人演变。基于视觉
SLAM
技
术的快速发展,扫地机器人成了视觉
SLAM
技术从实验室走进市场的突破口
之一。½现在市场上的路径规划式智½扫地机器人的价格一直居高不下,而且
很多½是½外产品,价格便宜的扫地机器人又不具备足够的智½。
为了降½扫地机器人的成本,
提高随机碰撞式扫地机器人的智½化程度并
对½前流行的视觉
SLAM
算法进行应用研究,本文对基于
SLAM
的扫地机器
人控制系统进行了研究。
根据扫地机器人的½用环境与½用要求,
选择并改进了相关算法实现了扫
地机器人环境建图功½和路径规划功½。针对
LSD-SLAM
算法获得的环境地
图与路径规划算法所½用环境地图不一致的问题,
采用
octomap
方法和三维投
½变换原理实现了地图类型½换。在栅格地图的基础上,针对内螺旋算法容易
陷入死区的问题,½用
A*算法对其做了改进以实现扫地机器人全覆盖路径规
划。
为验证算法的实际效果,本文进行了扫地机器人控制系统总½方案设计,
说明了实验平台各模块构成及功½,
根据功½和½用要求选择了合适的硬件设
备,并进行了相关计算,完成了扫地机器人控制系统构建。
本文完成了单目摄像头的标定,
进行了扫地机器人环境建图实验与路径规
划实验。经验证,所采用的
LSD-SLAM
算法获得的点云图最大相对误差为
1.36cm。在栅格边长为 14cm
的二维栅格地图基础上进行了点到点路径规划实
验和全覆盖路径规划实验,
点到点路径规划可以½扫地机器人避开障碍物障到
达指定½½,全覆盖路径规划所得路径的重复率为
2.7%,覆盖率为 100%。
关键词:单目视觉
SLAM;三维地图;路径规划
-I-
哈尔滨工业大学工程硕士学½论文
Abstract
The invention of the vacuum cleaner improves people's working conditions
and cleaning efficiency.And it reduces people's labor intensity and makes the work
of environment cleanning simple and efficient. But the traditional vacuum cleaner
still needs human operation and lacks of intelligence. With the continuous progress
of science and technology, sweeping robots combined with the advantages of
vacuum cleaners and robots came into being, from the initial random collision
sweeping robot gradually toward the path planning sweeping robot. Based on the
rapid development of visual SLAM technology, sweeping robots has become one
of the breakthroughs from the laboratory into the market. But now intelligent path
planning sweeping robot’s prices have been high in the market, and many are
foreign products, cheap sweeping robots do not have enough intelligence.
In order to reduce the cost of sweeping robots and improve the intelligent
degree of random collision sweeping robot and study the application of the popular
visual SLAM algorithm, the SLAM-based sweeping robot control system is
studied in this paper.
According to the service environment and operating requirements, we select
and improve the relevant algorithm and realize the sweeping robot’s environment
mapping function and path planning function. In order to solve the problem that
the environmental map obtained by LSD-SLAM algorithm is inconsistent with the
environment map used in the path planning algorithm, the octomap method and the
three-dimensional projection transformation principle are used to realize the map
type conversion. On the basis of the grid map, Aiming at the problem that the
internal spiral algorithm is easy to fall into the dead zone, the A * algorithm is used
to improve it to achieve coverage path planning.
In order to verify the practical effect of the algorithm, we design the overall
scheme of the control system of the sweeping robot and explain the composition
and function of each module of the control system. We select the appropriate
hardware equipments according to the function and the requirement, and complete
the construction of the control system of the sweeping robot. We complete the
calibration of the monocular camera, and carry out the environment modeling
experiment and the path planning experiment.
It is verified that the maximum relative error of the point cloud map obtained
by LSD-SLAM algorithm is 1.36cm.
On the basis of a two-dimensional grid map with a
grid length of 14 cm, We conducted point-to-point and full coverage path planning
-II-
评论