본문 바로 가기

로고

국내 최대 기계 및 로봇 연구정보
통합검색 화살표
  • MBD for ANSYS
  • 우수 논문

    뷰페이지 제목 게시판 내용

    Sensing-Based Distributed State Estimation for Cooperative Multiagent Systems  
    Cheolhyeon Kwon(Purdue University)
    United States | IEEE Transactions on Automatic Control
    2018-08-27 | 바로가기
    Sensors, Estimation, Monitoring
    Cited by 4

    ■  View full text

    IEEE Transactions on Automatic Control 

    Date of Publication: 27 August 2018

    https://doi.org/10.1109/TAC.2018.2867341

     

     

    ■  Researchers

    Cheolhyeon Kwon, Inseok Hwang

    School of Aeronautics and Astronautics, Purdue University

     

     

    ■  Abstract

    Distributed estimation has proven to be suitable for many multiagent system (MAS) applications, yet it relies heavily on information exchange via a costly and vulnerable communication network. This paper proposes a sensing-based distributed estimation algorithm that enables a local monitoring agent to expand its estimation capabilities beyond its sensing range without needing communication overhead. The key to expanding the limited sensing range is to incorporate the MAS's cooperative control protocol, allowing the monitoring agent to infer the state of out-of-range agents from the behavior of in-range agents that may interact with them. Then, the state estimation for out-of-range agents is performed through a Bayesian approach that considers the correlation of state estimates between in-range and out-of-range agents. This approach of only taking sensor measurements of local monitoring agents without interagent communications can successfully compensate for the existing communication-based distributed estimation methods. The performance of the proposed sensing-based distributed estimation algorithm is theoretically verified and demonstrated with numerical simulations of a multivehicle formation flight example.

     

     

    ■  Conclusion

    This paper has considered distributed estimation for cooperative MASs, which has been mostly studied using communication-based information exchange between individual agents. In seeking to address the cost and vulnerability issues inherent in the communication network, a sensing-based estimation approach was considered where each agent processes only local observations using on-board sensors. Specifically, we have proposed a sensing-based distributed state estimation algorithm to address the limited sensing range of a local monitoring agent and extend its estimation scope beyond the sensing range without needing communication. Leveraging the cooperative control protocol information over the sensor observations, the proposed algorithm recursively predicts and updates the out-of-range agent state estimates by examining their possible interactions with the observed behavior of the in-range agents. Further theoretical analysis has derived the conditions for the sensing network topology of the MAS to assure the stability of the proposed algorithm, and numerical simulation has demonstrated the performance of the algorithm in comparison to other methods. Our approach has been shown to provide the advantage of augmenting distributed estimation capability by explicitly considering the dynamical interactions between agents instead of through a communication network. The augmented sensing-based estimation could be used as either redundant back-up of communication-based estimation methods or a stand-alone distributed state estimation algorithm in communication-denied environments, leading to more reliable awareness of the MAS. Future work includes the following:

     

    scalability analysis for large-scale MAS estimation and further approximation for computationally more efficient implementation;

     

    sensing-based estimation subject to time-varying network topology that likely happens in mobile MAS applications;

     

    the development of a hybrid scheme where the existing communication-based and the proposed sensing-based approaches are intelligently fused in a way that maximizes the estimation performance while minimizing the communication overhead; and

     

    integration of the proposed estimation algorithm into the cooperative control protocol to facilitate the MAS coordination.

     

     

    * 관련 자료 

    [신진연구자 인터뷰] 제어 이론 기반의 고신뢰 사이버 물리 시스템 개발과 적용

    전체댓글0

    댓글 입력란
    프로필 이미지
    0/250자

    서브 사이드

    서브 우측상단1