請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/64602
標題: | 運用被動稀疏分布標籤之容許錯誤無線射頻辨識同時定位與建地圖演算法 Failure Tolerant RFID SLAM with Sparse Passive Tags |
作者: | Jiun-Fu Chen 陳俊甫 |
指導教授: | 王傑智 |
關鍵字: | 無線射頻辨識技術,同時定位與建地圖,容許錯誤, RFID,SLAM,Failure Tolerant, |
出版年 : | 2012 |
學位: | 碩士 |
摘要: | 錯誤處理對長期或終生運作的機器人學是一個必需要解決的問題。在本篇論文中,我們主要考慮由不一致性( inconsistency )所造成的錯誤,此錯誤主要來自於線性化的誤差。因此, 論文中提出了單位圓表示法( Unit Circle Representation ) 來改進預測步驟並提出保留相關係數膨脹法( Correlation Coefficient Preserved Inflation )來改進更新步驟。
除了容許錯誤外,此論文的第二目標為發展低成本的同時定位與建地圖( SLAM )系統。無線射頻辨識技術( RFID )已經被用在許多定位與建地圖的應用上,但是大部分的應用使用長距離、多個讀取器、並且在稠密分布標籤的環境中進行工作任務。而在這篇論文中,主要針對使用短距離、單一讀取器、並在稀疏分布的環境中使用延伸式卡曼濾波器( Extended Kalman Filter )來同時定位與建地圖。 基於分析與了解此同時定位與建地圖演算法,兩種改進方法被分別運用在預測與更新步驟上。此兩種改進方法可以在只使用里程計( odometry )與短距無線射頻感測器容許與減少不一致性。而提出的方法主要利用線性轉換來減少線性誤差並且在放大可能性的同時保留機器人與標籤之間的關係。最後,模擬與真實資料的結果也展示了提出方法的可行性與性能。 FAILURE handling is one of the essential issues while going forward the long-term or lifelong robotics. In this thesis, we state that the inconsistency issue should be considered. The major source of the inconsistency is given by the inearization errors. Therefore, the Unit Circle Representation is proposed to improve the prediction performance while the Correlation Coefficient Preserved Inflation is utilized to enhance the update performance. The second purpose in this thesis is to develop a low-cost SLAM system. The radio frequency identification (RFID) technology has been studied and used to solve the localization and mapping problems in which most of studies use long-range active or passive RFID systems with multiple readers and dense tags. In this thesis, we aim to use a short-range passive RFID reader to accomplish extended Kalman filter (EKF) based simultaneous localization and mapping (SLAM) with sparse tags in large indoor environments. Based on the understanding of the EKF-based RFID SLAM, the modifications are applied to perform the failure tolerance SLAM with only odometry and short-range RFID data. The proposed methods are capable to compensate the linearization errors by linear transformation and covariance inflation while preserving the relationship between the robot and features. Furthermore, the simulation and experimental results demonstrate the feasibility of the proposed method. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/64602 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-101-1.pdf 目前未授權公開取用 | 13.81 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。