利用通訊密度設計群組仿生機器人之擴散與派遣移動演算法

Abstract

群組機器人系統擁有可共同合作之特性，在許多應用上帶來了極大的優點，因此近年來受到了重視並且被廣泛地研究探討。而在群組機器人系統中，為了達成「合作」此一特性，即使在移動狀態之下，機器人仍舊必須維持住一個良好的通訊網路。而為了達成此一目的，一定量的資訊通常是必須的，但這在實現上很有可能會受到硬體的限制而造成問題。在本論文中，我們提出了一個群組機器人擴散演算法，使得群組機器人可以在移動狀態之下仍舊維持良好的通訊網路狀態。而此演算法的特點為，只利用極簡單的一項資訊：通訊密度，也就是個別機器人與其相鄰機器人們的通訊連接數，便可達成系統的需求。該項資訊在實際應用中是極簡單即可獲得的一項資訊，不需要一些複雜的設備，例如全球定位系統、攝影機或是雷達等。由此我們達成了低資訊量需求的一大優點。除此之外，在搜尋目標物的任務中，對於一個在環境中釋放機器人去偵測目標物的基地台，我們也根據了此移動演算法設計了一個機器人派遣法則。利用此派遣法則，基地台可以彈性地釋放出更多的機器人向外支援，並讓整體搜尋涵蓋面積隨時間線性增加，達成資源運用與搜尋速度並重的優點。最後在本論文中，我們討論了此移動演算法和派遣法則的模擬結果。同時，包括涵蓋面積，網路分裂率，花費時間與機器人停止率等數個特定參數，關於其模擬結果的數據統計值也做了詳細的探討，並且得出了一些極有益的參數關係特徵。

In recent years, the research topic on multi-robot systems has received great attention since the cooperation characteristics of robots provides lots of advantages. To achieve the cooperation characteristics and information sharing, robots must maintain a good communication network while moving in order to share information. But in real implementation, information quantity could be a problem due to the hardware limitation. In this thesis, a multi-robot dispersion algorithm is proposed for robots to move under maintaining a good communication network. In particular, this algorithm utilizes only simple information, which is the communication density, or in other words, the number of communication links. This information could be easily obtained without complex equipments such as GPS, camera or radar. Hence the advantage of low information quantity is achieved. Moreover, for a base station that releases robots to search for targets, a dispatch rule is proposed based on the dispersion algorithm. With this dispatch rule the base station could flexibly resupply robots to enlarge the coverage area linearly with time. Extensive simulations of the dispersion algorithm and dispatch rule are well studied. Statistics on certain parameters, namely coverage area, partition rate, spending time and stop percentage, are discussed and good characteristics are found in them.