自主分散式救援機器人與其合作策略之實現

Abstract

自主式救援機器人主要目的在於進行搜救任務，相較於傳統的搜救機器人，不需透過操作員的搖控，透過機器人本身對於環境的認知，就可以自主執行搜救的任務，同時經由無線網路，使監控端可以隨時觀測機器人的狀態。在機器人的行為控制方面，透過感測器對環境的感測，而設計個別行為，經由比較優先權的機制來產生最後的行為輸出，而無線網路方面，建立ZigBee網狀拓撲網路，結合遠端電腦的人機介面，達到網路化監控的功能。 本論文分別依照機器人對環境的認知來設計合作搜尋策略，在已知環境方面，設計虛擬區域劃分策略，由合作策略產生群體行為，機器人自主能力的個別行為，結合ZigBee無線通訊，設計基於行為模式之多機器人合作搜尋架構，經由實驗結果，合作策略的使用將提升整體搜尋效率38%。 在未知環境方面，另外設計動態區域劃分策略，經由實驗結果，合作策略的使用將提升整體搜尋效率30%。

The proposed autonomous rescue robot transmits information of its environment within a certain disaster area through the wireless network including the image and voice communication. Furthermore, they may also achieve the mission to bring materials to the victim. Therefore, a critical issue of the rescue robot is how they cooperate to search for victim more efficiently and effectively. In this thesis, the behavior-based motion cooperation is proposed for both the single robot and the multiple robots, separately. Then, the mesh topology network is adopted with the ZigBee wireless system and their integration can be further applied to the resultant final behavior output based on the assigned priority.

With both known and unknown environments, different behavior-based cooperative search strategies are also proposed for the multiple rescue robots. For the known environment, the strategy with virtually dividing approach is adopted and the group behavior is determined by both the cooperative strategy and the individually autonomy behavior. With the ZigBee communication, the framework of the behavior-based multi-robot cooperation is achieved and experimental results indicate that the searching efficiency is reliably improved 38%. For the unknown environment, the dynamic area dividing strategy is proposed and experimental results indicate that the searching efficiency is also improved 30%.