基於3D特徵地圖之分散式架構巡邏履帶式機器人

Abstract

本論文主要是設計並開發履帶式機器人之軟體、硬體架構，能於已知的室內環境中進行巡邏導航的功能，並能夠完成攀爬樓梯的任務。硬體方面為全新製作的履帶式機器人，除了在動力方面有所提升外，透過碳纖維的車身以及鋁合金的材質提供足夠強健的車體結構，並增加履帶的寬度強化與地面之間的磨擦力，能夠更穩健的攀爬樓梯；軟體方面，透過ROS的架構來開發履帶機器人的功能，基於此架構我們可以將一個複雜的程序規劃成多個部分來處理，像是感測器資料的接收、不同的演算法、機器人的控制等，將功能模組化，使用上有更大的彈性，並於功能上整合路徑規劃和影像定位，在機器人行走過程中，可以即時的監看Xtion所拍攝到的畫面，實現巡邏導航的功能，最後，經由實驗來驗證自製的履帶機器人有足夠能力完成跨樓層巡邏的任務。

This paper aims to design and develop both software and hardware architectures of a tracked robot such that it is capable of navigating in an indoor environment as well as climbing stairs. Regarding hardware developments, a brand-new tracked robot is designed to improve the dynamics of motors and increase the frictional forces of the tracks. Moreover, a more reliable structure is made by using carbon-fiber as well as aluminum alloy. As for software developments, Robot Operating System (ROS) is employed to provide a computing interface for several algorithms. As such, since a complicated program can be operated in parallel by different modules including receiving sensor measurements, stairs-climbing algorithms, and motion control, the overall design flexibilities can be increased. Integrated with path planning and vision-based localization algorithm, the robot is able to navigate in an indoor environment, and users can monitor the images captured by an Xtion RGB-D camera in real time as well. To verify the robustness of the proposed tracked robot, various experiments are conducted, and experimental results show that the robot is well-developed in terms of navigation and roaming between floors.