Please use this identifier to cite or link to this item:
Outdoor Localization and Reactive Control of an Autonomous Mobile Robot
|關鍵字:||Autonomous Mobile Robot;自動導航車;Dead Reckoning;Inertial Navigation System;Global Position System;Differential Global Position System;Kalman Filter;Prediction Dynamic index;方位推估器;慣性導航系統;全球定位系統;差分式全球定位系統;卡爾曼濾波器;動態防碰撞指標||出版社:||電機工程學系||摘要:||
The thesis develops methodologies and techniques for design and implementation of 3D outdoor self-localization system and distributed fuzzy reactive behavior of an autonomous mobile robot, and studies their application to vehicle navigation. Experimental setups for self-localization and behavior control of the robot consist of an odometer, two gyros, a Differential Global Position System (DGPS), and sixteen ultrasonic sensors with ring structure. By using the Kalman filter, the proposed 3D outdoor self-localization system merges measurements obtained from the DGPS and a laboratory-based 3D dead-recokening device to estimate absolute 3D position and orientation of the robot. The proposed system is extendly used to vehicle navigation. The design of distributed fuzzy reactive behaviors is divided into two steps. The first step uses the fuzzy control theory to establish and achieve five basic behaviors which consist of goal-seeking, dynamic avoid obstacle, left-wall-following, right-wall-following, Corridor-following behavior. In the second step, a new two-layered fuzzy fusion approach is presented to deal with multiple enviroment data so as to fuse aforementioned behaviors to generate a suitablely emergent behavior, which makes the robot move smoothly in outdoor, unstructured and dynamic environment.
Computer simulations are described which were performed to illustrate effectiveness and feasibility of the proposed methods. Experimental results show that the proposed methods are capable of given good results for self-localization and reactive control of autonomous mobile robot.
|Appears in Collections:||電機工程學系所|
Show full item record
TAIR Related Article
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.