| 摘要: | 本論文與另一位同學之論文[1]合作完成一個靈活的機器人頭，並賦予機器人頭聽覺、觸覺與視覺，是一個娛樂性與互動性兼具的機器人頭。為實現靈活的頭部機構，我們採用微型伺服機及連桿式結構來控制可動點，如眼睛、眼皮、眉毛與嘴巴部位的變化，讓機器人擁有人類般的表情變化，也在機器人底部設計一個轉動平台，讓攝影機可以隨使用者轉動，並與[1]所設計之頸部機構結合，完成機器人頭。在系統方面機器人以個人電腦平台開發， 使用Kinect的影像資訊實現機器人視覺，而影像處理目的包括人臉辨識與手勢辨識，讓系統根據使用者手勢給予正確的回應，人臉辨識結果可讓機器人自動追蹤人臉，接著本系統運用人臉區域的深度資訊來建立手勢的深度範圍，並結合膚色偵測得到手勢前景，最後擷取手勢特徵，如手指位置、長度和數量，以及手掌心位置和半徑，以指揮機器人做出該有的動作。本論文在機器人功能方面設計拍照與網路相簿流程，以伺服器架設網頁相簿並設有密碼鎖，可讓使用者方便取得相簿也保有個人隱私。最後，本機器人可以手勢執行四個主要功能： 1)照相與網路相簿功能; 2)音樂歌唱功能[1]; 3)語音辨識與說話功能[1]；4)觸摸與拍打回應功能[1]。運用簡單的手勢流程可讓使用者快速地熟悉功能，提升使用者與機器人的互動體驗。;This study proposes a nimble robot head with auditory, tactual and vision senses. Accomplished with another researcher [1], the robot head has the functions of entertainments and interaction to peoples. To implement the robot nimbleness, the mechanism of head employs a linkage structure and servo motors to establish the movement points. The movement points are eyelid, eye, eyebrow, and mouth. Controlling the movement points can present different facial expressions. The robot has the turntable mechanism in the bottom, such that the camera, which is set on the table, can follow the user. The head mechanism is combined with neck mechanism designed in [1]. As for the system aspect, this robot is developed on PC base. Using Kinect sensor which captures color and depth information of the user is helpful to complete the vision of robot. The robot vision can recognize the face and gestures of the user, and make the robot give correct responses correspondingly. According to the recognized face position, the robot can follow and see the face of the user. In order to recognize the correct gesture, we defined a region of interest (ROI). In ROI, the image process including skin color detection and position/ length of fingers, position/ radius of palm finding can recognize the gestures of the user, so that the robot head can understand the commands from the user and do the correct corresponding motions. The photo taking function is as follows. After taking photos, the user can find the photos from the cloud album conveniently and secretly based on the code on the LCD. In summary, according to the gestures, the robot can perform four main functions for the robot system: (1) photo taking, (2) song singing [1], (3) speech recognition and dialogue [1] and (4) response for the touching and beating [1]. With the simple gestures, the user can easily operate the function rapidly, and it also improves the interaction of robot. |
