家用機器人之影像追蹤系統

Abstract

本論文發展一套應用於家用機器人之影像追蹤系統。本系統以使用者之輪廓與膚色為特徵，偵測人臉是否在畫面中，進而執行追蹤之任務。本系統除機器人之二自由度頭部可追蹤人臉，同時能隨時修正本體的姿態，使本體亦能追隨人移動，達到一種互動的功能。本論文並完成家用機器人的影像追蹤控制器，本控制器是基於影像(Image-base)的伺服控制系統，使用影像的特徵為控制命令，整個控制器為一個閉迴路控制系統，使控制器能確實將機器人頭部的攝影機移動到指定的位置。在本論文中並以DSP為核心發展一運動控制卡，應用於頭部與本體的運動控制。本體的運動控制包含路徑追蹤之功能，使用軸編碼器與陀螺儀量測偏向角度，並應用最佳控制中的線性二次調節器(LQR)消除運動誤差，內建路徑追蹤控制器於本體的運動控制卡中，使機器人可隨所設之路徑行走。整合測試之實驗結果顯示本系統具有即時追蹤人臉之功能。

In this thesis, we developed a visual tracking system for a home robot. The system can detect human faces and track a person by controlling a two-degree-of-freedom robot-head. The visual tracking controller can pursue a person by controlling the robot body motion as well as the head orientation. An image processing system has been developed to extract facial features for a COMS PC Camera. An algorithm is proposed to recognize a human head in this thesis by using skin colour and elliptical shape of a human face. A DSP-based motor control card is developed for robot motion control. The visual tracking control system has been integrated on a self-made home robot H2. Experimental results show that the robot can track a person in real-time.