| 摘要: | 以輪式探測機器人為基礎的環境探測系統具有移動、採集識別和傳輸數據的特性，能在人類不易進入的現場採集環境數據、影像、自我導航，再將數據傳輸給閘道器節點與遠端電腦，然後進行資料整理、分析、歸納與判斷；探測機器人是智慧型機器人的一種，以此類機器人為基礎的環境監測系統的開發需要微處理機、智慧型控制、影像辨識與無線傳輸等技術的結合；因此，整個系統性能的提升，除了機構的最佳化設計外，上述技術是重要的影響因素。輪式探測機器人在位置伺服驅動進行前，需有正確快速的影像辨識結果。此研究計畫係針對光線不足、物體遮擋或攝影角度差異造成的辨識錯誤與重複，加入一類神經網路辨識法以提升辨識效果，架構中，在影像軟體加入兼具穩定性與可塑性的適應共振類神經網路，以因應環境的變化；另外此研究亦將針對因摩擦力、藕合間隙、馬達參數及負載變動等因素造成的擾動非線性現象，設計一智慧型馬達位置伺服控制器，架構中，將滑動面及自我調變的平移寬度植入模糊推理引擎，以克服系統的非線性與擾動，有效提升伺服系統的精度及穩定性。接著，為使探測機器人能將蒐集到的數據傳回閘道器節點與遠端電腦，或是探測機器人能夠接收遠端電腦的命令(例如追蹤新的標的物)，在探測機器人與閘道器節點間以ZigBee無線方式通訊，而閘道器節點與遠端電腦，則以WLAN方式連接，以發展成為功能完整的無線環境探測系統。此研究提出兩年期的計畫，第一年計畫：針對輪式探測機器人的巡弋過程，利用適應共振類神經網路與模糊推理機制分別設計影像辨識輔助程式與定位伺服控制器，實現探測機器人的伺服控制，預計完成工作： (1) 輪式探測機器人平台架構分析。 (2) 針對影像與馬達定位，推導下列兩種法則。(a)適應共振類神經網路；(b)適應模糊演算法。 (3) 撰寫影像辨識與位置伺服控制程式，進行除錯測試。第二年計畫：以第一年發展的具有影像之輪式探測機器人為基礎，在機器人上加入雨滴、霍爾、火源、溫溼度、光照度感測器及無線電收發器，然後在機器人與閘道器節點間以ZigBee無線方式通訊，閘道器節點與遠端電腦端以WLAN方式連接，以無線方式獲取環境數據，預計完成工作： (1) 使用ZigBee/IEEE 802.15.4協定，建置無線感測系統。 (2) 發展閘道器節點、遠端電腦與機器人端通訊程式。 (3) 測試與除錯。
Wheeled-exploring-robot based environmental monitor system possesses some salient features, such as movement, sensing, gathering, identification and data communication, etc. All gathered data must be transmitted to gateway node and remote computer. Then the actions of arrangement, analyses and decision are carried on the computer. Wheeled-exploring-robot is a kind of intelligent robot and the development of the wheeled-exploring-robot based monitor system needs the combining technologies of microprocessor, intelligent control, image identification and wireless communication. Besides the optimum mechanism of robot, above technologies are important factors to promote the performance of the monitor system. Before position servo operation, the image identification is necessary to provide the precise position information. So the first part of this project is focused on the design of an image identification auxiliary algorithm to cope with the conditions of insufficient light, masked object or detection angle difference. The adaptive resonance neural network (ARNN) is added into traditional identification algorithm to promote the precision. The second part of this project is the design of an intelligent position servo controller to deal with the conditions of friction, coupling gap, variation of motor parameters and load, etc. In the control architecture, the sliding surface and adaptive translation width are embedded into fuzzy inference engine to cope with the nonlinearity, disturbance and to increase the precision and stability. The third part of this project is focused on the wireless data communication to form an integrated environmental monitor system. The adopted communication protocol between the exploring-robot and gateway node is ZigBee, while the general WLAN is utilized to communicate between the gateway node and remote computer. This project is divided into two subprograms and the execution duration is two years. The first-year subprogram is the design of an image identification auxiliary algorithm and an intelligent position servo controller for exploring-robot. Following works will be implemented in this stage: (1) Structure analyses of the wheeled-exploring-robot. (2) Derivations of ARNN and adaptive fuzzy algorithms for image identification and position servo control, respectively. (3) Programming the algorithms and debug. The second-year subprogram is the development of wireless data communication to form an integrated environmental monitor system. Firstly, the raindrop, hall, fire hazard, temperature/humidity, light sensors and transmitter/receiver are embedded into above exploring-robot. Then, following works will be finished in this stage: (1) Establishment of the wireless sensor network. The ZigBee/IEEE 802.15.4 is utilized between the exploring-robot and gateway node, while the general WLAN is utilized to communicate between the gateway node and remote computer. (2) Programming the communication programs of gateway node, remote computer and exploring-robot. (3) Experimental test and debug. |
