利用聲場特徵與SVM實現可結合輪式機器人之避障與導航

Abstract

本論文提出一套在室內環境中輪式機器人的導航方法與可結合式子母機器人系統。傳統機器人導航系統使用紅外線感測作為導航依據，當基地端與移動式機器人間有障礙物阻擋時，紅外線感測器導航系統將會失效，而本系統利用聲場特徵與Support Vector Machine(SVM)分類，進行機器人障礙物偵測與導航，解決障礙物遮蔽影響。於移動平台接近基地端時，根據感測器資訊做子母機器人兩者上下的結合，使基地端成為能夠自由行動之機器人。在本論文中，亦透過嵌入式網路技術，建構以802.11為通訊協定之通訊介面，以傳遞基地端與移動平台間感測器的訊息。此外，並使用全向輪建構移動平台，使平台能夠有較高的運動靈活度，以利於結合運動的進行。

The thesis proposes a navigation method of the combinational wheels robot system in a complex indoor environment. The navigation system of traditional robot uses the infrared information to decide the moving method. But if the robot is sheltered by an obstacle from the base station, the infrared will not work effective. In order to solve this problem, the mobile robot can detect the obstacle and start navigation by using sound field characteristic and support vector machine(SVM) classifier in our system. When the mobile robot approach the base station, the robot and base station combine between the higher and lower levels according to the sensors information, and transfer the base station to a mobile station. Also in this thesis, the base station communicates with the mobile robot by the wireless LAN based on the embedded Ethernet technology. And we also propose the omni-directional platform offer a higher mobility in order to progress the combination of base station and mobile robot.