基於機器人作業系統之自主履帶車跨樓層巡邏系統開發

Abstract

本論文主要是開發基於ROS架構之履帶車，在已知地圖下可以執行室內巡邏導航與自主攀爬樓梯的任務，並且在遠端電腦監控和操作。履帶車使用模糊控制追蹤自訂路徑，並透過AMCL演算法和Hokuyo雷射測距儀得到定位資訊，修正里程計的誤差。另一方面，使用ZED立體攝影機得到深度影像，以及透過一系列的模式，機器人能夠自主攀爬樓梯。在機器人執行任務的時候，透過Qt圖形介面和VPN網路通訊，使用者可以於遠端即時監控並操控機器人。機器人收到指令後，會根據目標樓層和當前樓層判斷最佳路線，自動攀爬到目標樓梯執行巡邏任務。最後經由實驗證實了本論文開發之履帶車能夠實現跨樓層之巡邏導航功能。

In this thesis, we develop a ROS-based autonomous tracked vehicle (ATV), which is able to navigate in indoor environment with map and climb stairs autonomously. Users can monitor and control the ATV on the remote computer. The ATV utilized fuzzy control to track the planned path and corrects odometry through the AMCL algorithm and the Hokuyo laser scanner. On the other hand, we utilize ZED stereo camera to capture depth image, designing a series of action modes to climb stairs autonomously. While the ATV performed tasks, user can control and monitor the robot through Qt-based user interface and VPN communication. After receiving the command, the ATV would determine the best route according to the target floor and the current floor, and then automatically climbs to the target floor to carry out the patrol task. Experimental results verify the effectiveness of the developed ATV for climbing stairs and navigation in the patrol operation under indoor environments.