- 學位論文
語音辨識搭配影像色彩標籤之家庭服務機器人
Development of a home service robot with speech and image tag recognition
摘要
本論文整體系統主要由全向輪機器人 FESTO Robotino、頭戴式麥克風、網路攝影機以及個人電腦所組成，使用者透過麥克風與電腦溝通，聲控命令機器人前往目的地，機器人接受命令後會先利用影像處理探索周圍的目標物，並使用最鄰近點前進法往目的地移動。 語音辨識與合成是採用Microsoft Speech SDK，使用者可透過麥克風對機器人下達語音命令，而機器人透過喇叭輸出語音與使用者互動。影像處理主要目的在於在各目標物上貼上色彩標籤，當機器人隨意置於空間中時，可以利用搜尋這些色彩標籤，推算出目前最鄰近終點之地標，先往最鄰近點前進之後，再進行搜尋，依序進行最後即可抵達目的地。 機器人往目標前往的路徑上如遇障礙物，機器人會使用沿牆法(Wall Following) 進行閃避障礙物之動作，沿牆法的部份主要是利用前方五顆紅外線感測器之電壓值，將電壓值輸入給模糊控制器，其輸出將控制機器人的全向輪進行閃避障礙物的動作。
並列摘要
In this paper the robot overall system consists of the wheeled platform FESTO Robotino, headset microphone, webcam and personal computer. The user communicates with the computer through a microphone and issues a command to the robot for the destination. After receiving the command the robot will first explore the space around for possible targets by decoding the color tag using image processing technique, and move to the destination through the method of nearest point connection. Microsoft Speech SDK system is utilized for speech recognition and synthesis for robot. The user can make a verbal command to the robot through a microphone speaker. The robot can interact with user through the speaker. The main purpose of image processing is to decode the color labels on the targets. When the robot is randomly placed in the space, it can search for these color labels to calculate the nearest target. It moves toward to the nearest point, then re-searches again the space around and moves sequentially until the final destination is met. If an obstacle is met when robot moves to the target, the robot will take wall following action to avoid the obstacle. Five infra-red sensors in front face of the robot are employed for obstacle detection and for robot wall-following. The sensor voltage is input to the fuzzy controller for controlling the robot's wheels to wall follow the obstacle.