本論文內容是關於隨從機器人之行為設計與室內導航之研究,過往的隨從機器人總是跟隨著使用者後方或是與使用者保持相當近的距離,因此本文設計一個新的隨從機器人行為,此行為是藉由機器人沿牆邊同時跟使用者保持一段距離,而被跟隨的使用者可以利用手姿給予命令,去改變機器人的行為動作,而這樣的行為及利用手姿作為命令,可以讓機器人更像隨從。本論文導航部分利用模糊邏輯感測融合估測計算沿牆的方向並在室內未知環境中進行導航。為了使機器人具備辨識人的姿態能力,裝載Kinect 感測器去獲取人資訊,而藉由馬式距離能有效的統計計算使用者的手姿與樣本動作的匹配程度。
This study proposes a new robot behavior where the servant robot runs alongside walls while it follows the human at a distance certain because the common robot follows the human at the closer distance or follows at the back. The human shall give command using gestures at the same time. Consequently, by statistically matching observed gestures with pre-defined templates, the robot changes its behavior depending on the command received. This behavior can make the robot appear more like a servant. The navigation uses fuzzy logic estimation of wall direction for robot navigation in an indoor environment. And the human gesture recognition is based on statistical matching of Kinect skeletal data.