本文選題：單目視覺(jué)SLAM + 三維地圖��； 參考：《哈爾濱工業(yè)大學(xué)》2017年碩士論文

【摘要】：吸塵器的發(fā)明,改善了人們的勞動(dòng)條件,提高了人們的清潔效率并降低了了人們的勞動(dòng)強(qiáng)度,使清潔環(huán)境的工作變得簡(jiǎn)單而有效率。但傳統(tǒng)的吸塵器體積大、需要專門人員操作,智能化程度不足。隨著科技的不斷進(jìn)步,掃地機(jī)器人結(jié)合了傳統(tǒng)吸塵器與自主移動(dòng)機(jī)器人的優(yōu)點(diǎn)出現(xiàn)在市場(chǎng)上,并從最初的隨機(jī)碰撞式掃地機(jī)器人逐漸向著路徑規(guī)劃式掃地機(jī)器人演變�；谝曈X(jué)SLAM技術(shù)的快速發(fā)展,掃地機(jī)器人成了視覺(jué)SLAM技術(shù)從實(shí)驗(yàn)室走進(jìn)市場(chǎng)的突破口之一。但現(xiàn)在市場(chǎng)上的路徑規(guī)劃式智能掃地機(jī)器人的價(jià)格一直居高不下,而且很多都是國(guó)外產(chǎn)品,價(jià)格便宜的掃地機(jī)器人又不具備足夠的智能。為了降低掃地機(jī)器人的成本,提高隨機(jī)碰撞式掃地機(jī)器人的智能化程度并對(duì)當(dāng)前流行的視覺(jué)SLAM算法進(jìn)行應(yīng)用研究,本文對(duì)基于SLAM的掃地機(jī)器人控制系統(tǒng)進(jìn)行了研究。根據(jù)掃地機(jī)器人的使用環(huán)境與使用要求,選擇并改進(jìn)了相關(guān)算法實(shí)現(xiàn)了掃地機(jī)器人環(huán)境建圖功能和路徑規(guī)劃功能。針對(duì)LSD-SLAM算法獲得的環(huán)境地圖與路徑規(guī)劃算法所使用環(huán)境地圖不一致的問(wèn)題,采用octomap方法和三維投影變換原理實(shí)現(xiàn)了地圖類型轉(zhuǎn)換。在柵格地圖的基礎(chǔ)上,針對(duì)內(nèi)螺旋算法容易陷入死區(qū)的問(wèn)題,使用A*算法對(duì)其做了改進(jìn)以實(shí)現(xiàn)掃地機(jī)器人全覆蓋路徑規(guī)劃。為驗(yàn)證算法的實(shí)際效果,本文進(jìn)行了掃地機(jī)器人控制系統(tǒng)總體方案設(shè)計(jì),說(shuō)明了實(shí)驗(yàn)平臺(tái)各模塊構(gòu)成及功能,根據(jù)功能和使用要求選擇了合適的硬件設(shè)備,并進(jìn)行了相關(guān)計(jì)算,完成了掃地機(jī)器人控制系統(tǒng)構(gòu)建。本文完成了單目攝像頭的標(biāo)定,進(jìn)行了掃地機(jī)器人環(huán)境建圖實(shí)驗(yàn)與路徑規(guī)劃實(shí)驗(yàn)。經(jīng)驗(yàn)證,所采用的LSD-SLAM算法獲得的點(diǎn)云圖最大相對(duì)誤差為1.36cm。在柵格邊長(zhǎng)為14cm的二維柵格地圖基礎(chǔ)上進(jìn)行了點(diǎn)到點(diǎn)路徑規(guī)劃實(shí)驗(yàn)和全覆蓋路徑規(guī)劃實(shí)驗(yàn),點(diǎn)到點(diǎn)路徑規(guī)劃可以使掃地機(jī)器人避開(kāi)障礙物障到達(dá)指定位置,全覆蓋路徑規(guī)劃所得路徑的重復(fù)率為2.7%,覆蓋率為100%。
[Abstract]:The invention of vacuum cleaner improves people's working conditions, improves people's cleaning efficiency, reduces people's labor intensity, and makes the work of clean environment simple and efficient. However, the traditional vacuum cleaner is large in size and needs special personnel to operate, so the degree of intelligence is insufficient. With the development of science and technology, floor sweeping robot combines the advantages of traditional vacuum cleaner and autonomous mobile robot in the market, and evolves from the initial random impact floor sweeping robot to the path planning floor sweeping robot. Based on the rapid development of visual SLAM technology, floor sweeping robot has become one of the breakthrough points of visual SLAM technology from laboratory to market. However, the price of path-planning intelligent floor sweeping robots on the market has been high all the time, and many of them are foreign products, and the cheap ones do not have enough intelligence. In order to reduce the cost of floor sweeping robot, improve the intelligent degree of random colliding floor sweeping robot, and apply the popular visual SLAM algorithm, this paper studies the control system of floor sweeping robot based on SLAM. According to the use environment and requirements of the floor sweeping robot, the related algorithms are selected and improved to realize the mapping function and path planning function of the floor sweeping robot environment. In order to solve the problem that the environmental map obtained by LSD-SLAM algorithm is inconsistent with the environmental map used by path planning algorithm, the octomap method and the three-dimensional projection transformation principle are used to realize the map type conversion. Based on the raster map, aiming at the problem that the inner spiral algorithm is easy to fall into the dead zone, the A* algorithm is improved to realize the full coverage path planning of the floor sweeping robot. In order to verify the actual effect of the algorithm, this paper designs the overall scheme of the floor sweeping robot control system, explains the composition and function of each module of the experimental platform, selects the appropriate hardware equipment according to the function and usage requirements, and carries out the relevant calculation. The control system of floor sweeping robot is constructed. In this paper, the calibration of monocular camera is completed, and the environmental mapping experiment and path planning experiment of floor sweeping robot are carried out. It is verified that the maximum relative error of the point cloud image obtained by the LSD-SLAM algorithm is 1.36 cm. Point-to-point path planning experiment and full coverage path planning experiment are carried out on the basis of two-dimensional raster map with grid side length 14cm. Point-to-point path planning can enable the robot to avoid obstacles to reach the designated position. The path repetition rate of full coverage path planning is 2. 7%, and the coverage rate is 100%.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP242

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 權(quán)美香;樸松昊;李國(guó);;視覺(jué)SLAM綜述[J];智能系統(tǒng)學(xué)報(bào);2016年06期

2 趙航;劉玉梅;卜春光;李艷杰;劉博;;掃地機(jī)器人的發(fā)展現(xiàn)狀及展望[J];信息與電腦(理論版);2016年12期

3 張毅;杜凡宇;羅元;熊艷;;一種融合激光和深度視覺(jué)傳感器的SLAM地圖創(chuàng)建方法[J];計(jì)算機(jī)應(yīng)用研究;2016年10期

4 梁文莉;;競(jìng)爭(zhēng)激烈的掃地機(jī)器人市場(chǎng)[J];機(jī)器人技術(shù)與應(yīng)用;2015年02期

5 王忠立;趙杰;蔡鶴皋;;大規(guī)模環(huán)境下基于圖優(yōu)化SLAM的圖構(gòu)建方法[J];哈爾濱工業(yè)大學(xué)學(xué)報(bào);2015年01期

6 Li-wei LIU;Yang LI;Ming ZHANG;Liang-hao WANG;Dong-xiao LI;;K-nearest neighborhood based integration of time-of-flight cameras and passive stereo for high-accuracy depth maps?[J];Journal of Zhejiang University-Science C(Computers & Electronics);2014年03期

7 徐嵩;孫秀霞;劉樹(shù)光;劉希;楊朋松;;攝像機(jī)畸變標(biāo)定的模型參考逼近方法[J];光學(xué)學(xué)報(bào);2013年07期

8 杜釗君;吳懷宇;;基于激光測(cè)距與雙目視覺(jué)信息融合的移動(dòng)機(jī)器人SLAM研究[J];計(jì)算機(jī)測(cè)量與控制;2013年01期

9 杜光勛;全權(quán);蔡開(kāi)元;;視覺(jué)與慣性傳感器融合的隱式卡爾曼濾波位置估計(jì)算法[J];控制理論與應(yīng)用;2012年07期

10 陳小龍;唐強(qiáng);車軍;劉林;;基于人工視覺(jué)的四旋翼飛行器室內(nèi)定位與控制[J];兵工自動(dòng)化;2012年05期

相關(guān)碩士學(xué)位論文 前2條

1 柴劍;智能掃地機(jī)器人技術(shù)的研究與實(shí)現(xiàn)[D];西安電子科技大學(xué);2014年

2 甘波;移動(dòng)服務(wù)機(jī)器人環(huán)境建模技術(shù)研究[D];沈陽(yáng)工業(yè)大學(xué);2012年

，

本文編號(hào)：1874609