基於雙目視覺的物件辨識之拾取機器人

Object Recognition Picking Robot Based on Binocular Vision

徐勝彥(SHENG-YEN HSU)

指導教授 ： 廖裕評

中原大學/電機資訊學院/電機工程學系/碩士(2022年)

https://doi.org/10.6840/cycu202201286

摘要

隨著全球的疫情擴散，每天出門必須戴著口罩，酒精隨身攜帶著，不管碰了什麼、拿了什麼物品後，都會拿起酒精消毒雙手。研究報告指出，病毒雖然在手上存活時間不久，但是在這期間假如帶有病毒的手去揉眼睛、摸鼻子、吃東西甚至拿起物品給他人，這時就會將病毒擴散出去。舉例高接觸風險工作者，像是機場清潔員、防疫旅館清潔員、醫院清潔員等，可能只是為了拿取染疫者地板垃圾或是遺留的物品，而感染到新冠病毒COVID19。因此本研究採用YOLOv5物件辨識模型進行物件辨識，透過YOLOv5的檢測速度快和定位精準等優勢，結合雙目視覺來假設人的雙眼，利用辨識結果之目標框得知左右鏡頭的物件座標後，便能使用測距公式來算出物件距離。
本論文中拾取機器人不僅具有物件測距的功能，還搭配機器人作業系統(ROS)能發布物件測距訊息給拾取機器人，也具備導航功能進行室內環境的巡邏。利用同時定位與地圖構建(SLAM)技術使拾取機器人能夠在室內環境中自主導航。並結合手臂將物件拾取，希望可以大幅降低高接觸風險工作者染疫的機率。

關鍵字

YOLOv5 ； 雙目視覺 ； ROS 機器人系統 ； COVID19 ； 機器手臂

並列摘要

With the global spread of the COVID-19 pandemic, people are advised to wear a face mask when they go outside their home and use soap and water or alcohol hand sanitiser to clean their hands regularly. Previous studies showed that the virus is detectable for up to two to three days on plastic and stainless steel. The airport workers and cleaners are at high risk of exposure with COVID-19 to pick up trash for environmental cleaning. Therefore, this paper aims to study objects detection and localization methods based on binocular vision for picking robots. In the proposed system, we utilize the binocular vision to detect objects with the You Only Look Once version 5 (YOLOv5) algorithm running on a Raspberry Pi 4B and calculating object distances.
The picking robot we proposed not only has the function of object estimation, but also can publish object estimation information to the robot with the Robot Operating System(ROS). The proposed system also has the navigation function to patrol in indoor environment. The use of real-time localization and mapping (SLAM) technology enables robots to navigate autonomously in indoor environments. Moreover, the proposed robot system can use the robot arm to pick up objects and help human to clean the environment.

並列關鍵字

YOLOv5 ； Binocular vision ； Robot Operating System ； COVID19 ； robotic arm

參考文獻

[1] C. Ai, D. Geng, Z. Qi, L. Zheng and Z. Feng, "Research on AGV Navigation System Based on Binocular Vision," 2021 IEEE International Conference on Real-time Computing and Robotics (RCAR), 2021, pp. 851-856, doi: 10.1109/RCAR52367.2021.9517359.

[2] Y. Ma, H. Yu, J. Liu and Y. Zhai, "Measurement of ice thickness based on binocular vision camera," 2017 IEEE International Conference on Mechatronics and Automation (ICMA), 2017, pp. 162-166, doi: 10.1109/ICMA.2017.8015806.

[3] Z. Yaping and S. Zhihang, "Research on Binocular Forest Fire Source Location and Ranging System," 2020 IEEE International Conference on Power, Intelligent Computing and Systems (ICPICS), 2020, pp. 199-202, doi: 10.1109/ICPICS50287.2020.9201975.

[4] B. Jia, S. Yang and T. Yu, "Research on Three Picking Arm Avoidance Algorithms for Agaricus Mushroom Picking Robot," 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications( AEECA), 2020, pp. 325-328, doi: 10.1109/AEECA49918.2020.9213575.

[5] X. Zhang et al., "Study on the design and control system for wolfberry harvesting robot," 2016 Chinese Control and Decision Conference (CCDC), 2016, pp. 5984-5985, doi: 10.1109/CCDC.2016.7532069.