請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/4765
標題: | 在多機器人同時定位與移動物體追蹤中的適應式估測與量測分享 Adapting Measurement and Belief Sharing in Multi-Robot Simultaneous Localization and Tracking |
作者: | Chun-Kai Chang 張鈞凱 |
指導教授: | 王傑智(Chieh-Chih Wang) |
關鍵字: | 通訊,多機器人,定位,追蹤,合作感知,部分可觀察馬可夫決策過程, Communication,multi-robot,localization,tracking,cooperative,POMDP, |
出版年 : | 2015 |
學位: | 碩士 |
摘要: | 目前的多機器人合作感知的方法,根據機器人間分享資訊的方式,主要可以分為兩類:基於量測分享的合作感知與基於估測分享的合作感知。當通訊品質良好的時候,基於量測分享的方法可以達到理論上最佳的結果,然而基於估測分享的方法則不行。但是基於估測分享的方法在通訊不穩定的情況下,因為一組估測結果已經包含了多筆量測資料,所以相對來說表現比較穩定。啟發於量測分享與估測分享在不同情況下各有優劣,在本論文中,我們提出了適應式量測與估測分享的方法來考慮不同的通訊與感知情況,來整合量測分享與估測分享兩者的優勢,用以達到更好的效能與結果,並且處理通訊不穩定時所帶來的問題。然而要如何決定資訊分享的方式,是一種多機器人部分可觀察馬可夫決策過程的問題。我們藉由最大化地降低預期的不確定性,來決定要分享的量測資料或是估測資訊,透過有效通訊的期望值以及對於未來量測結果的預估,適應式量測與估測分享方法在複雜度上所面臨的問題可以被有效的處理,來即時地處理通訊上所遇到的問題。此外,我們也透過模擬實驗與真實數據實驗,來驗證所提出的適應式方法,透過模擬不同通訊情況以及資料映射的情境,可以發現我們提出的適應式量測與估測分享方法可以達到比只進行量測分享或只進行估測分享的演算法準確的結果。 Existing multi-robot cooperative perception solutions can be mainly classified into two categories, measurement-based and belief-based, according to the information shared among robots. With well-controlled communication, measurement-based approaches are expected to achieve theoretically optimal estimates while belief-based approaches are not. Nevertheless, belief-based approaches perform relatively stable under unstable communication as a belief contains the information of multiple previous measurements. Motivated by the observation that measurement sharing and belief sharing are respectively superior in different conditions, in this thesis an adapting algorithm, communication adaptive multi-robot simultaneous localization and tracking (ComAd MRSLAT), is proposed to combine the advantages of both to tackle the unstable communication conditions. However, the decision process of what kind of information to share is only based on a probability distribution of states, which is estimated according to a set of observations and observation probabilities. Therefore, it could be seen as a multi-robot partially observable Markov decision process (POMDP) problem. The information to share is decided by maximizing the expected uncertainty reduction, based on which the algorithm dynamically alternates between measurement-sharing and belief-sharing without information loss or reuse. With using the expected effective communication and information receiving, the proposed ComAd MR-SLAT can tackle the complexity issue and online decide the sharing strategy to adapt different communication conditions. The proposed ComAd MR-SLAT is evaluated in communication conditions with different packet loss rates, bursty loss lengths, and data association conditions. The proposed ComAd MR-SLAT outperforms both measurement-based and belief-based MR-SLAT in both localization and data association accuracy. In addition, the real data are also collected and evaluated, the experimental results demonstrate the effectiveness of the proposed adapting algorithm and exhibit that the ComAd MR-SLAT is robust in the simulation and real data experiment. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/4765 |
全文授權: | 同意授權(全球公開) |
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-104-1.pdf | 539.17 kB | Adobe PDF | 檢視/開啟 |
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。