基於無線感測網路之機器人定位系統

Abstract

本論文之目的在設計一套基於ZigBee無線感測網路之機器人定位方法。在一個佈建有ZigBee無線感測網路的環境中，機器人經由接收佈置在工作環境中的ZigBee無線感測器網路節點所傳送的訊號強度估算機器人的位置。ZigBee定位誤差是隨機變化而且是在一個有限範圍的值；而基於軸編碼器與陀螺儀之里程計(odometer)在長時間運行下會累積誤差，為了改善定位精確度，本論文提出一個基於模糊邏輯的資料融合方法，整合機器人里程計及ZigBee無線感測網路定位資訊，以提供機器人室內定位之解決方案。經由實驗測試，本論文所發展之ZigBee定位系統平均誤差為1.54m。論文中以實驗室的機器人進行導航實驗，在走廊行走距離10m來回兩趟一共40m，X軸方向平均誤差為8.53cm、最大誤差為14.45cm，Y軸方向平均誤差為39.19cm、最大誤差為56.63cm。

A self-localization system provides robot position(x,y) and orientation(Ө) to the robot. This information is essential for robot navigation. This thesis aims to develop a localization system for a mobile robot using wireless sensor network. Through the received signal strength (RSS) from ZigBee modules deployed in the environment, a localization system is designed and implement for robot navigation. The error of ZigBee localization system is somewhat random and in a limited range. On the other hand, the odometer suffers from accumulated an error. To further improve the localization accuracy, this thesis proposes a method to combine the ZigBee localization system with an odometer. A data fusion system is developed to use the advantages of both the ZigBee system and the odometer. The position error of the realized ZigBee system is 1.54m in an indoor environment. However, for the combined system, experimental results show that the maximum error is 14.45 cm and 56.63 cm in X and Y axis respectively for a 40 m travel in a corridor.