「第6回機械工学セミナー」のご案内

第６回機械工学セミナー
Minimal Information for Estimation and Control:
A Semidefinite Programming Approach
講演の内容は主に、状態推定に関するものです。
人工衛星の姿勢推定への実応用例も紹介予定です。
講師：田中崇資氏
マサチューセッツ工科大学
(Massachusetts Institute of Technology (MIT))
日時：２０１５年４月１５日（水）９:４５～１０:４５
場所：５５２講義室（５号館５階）
使用言語は日本語です。学生の皆さんも自由にご参加ください。
学部学科学年を問いません。
問合せ先：機械工学科橋本（[email protected]）・田原
講演要旨
Information is not free. There are various situations where acquiring necessary
information for real-time decision making is a costly process. In this talk, we
discuss a methodology to design an “information-efficient” sensor, whose
information acquisition rate is minimal while allowing decision makers to make
real-time decisions with acceptable accuracy. Specifically, we formulate a
problem of designing an information-efficient sensor for real-time estimation in
the linear-quadratic-Gaussian (LQG) setting. Introducing a novel variable
elimination technique, we show that the desired sensing policy can be
efficiently found by semidefinite programming (SDP). Since it can be shown
that our problem formulation is equivalent to the Gaussian sequential
rate-distortion (SRD) problem, which is a natural generalization of the standard
rate-distortion problem to multistage settings, our methodology also provides
an efficient numerical algorithm to the existing SRD literature. To demonstrate
practicality of this result, we consider a numerical case study of real-time
satellite attitude estimation using minimal telecommunication. We also
discuss an SDP-based solution to the information-regularized LQG control
problems. Open problems and future perspectives will be also discussed.
御略歴
Takashi Tanaka received his B.S. degree from Tokyo University in 2006, M.S.
and Ph.D. degrees from University of Illinois at Urbana-Champaign in 2009 and
2012, all in Aerospace Engineering. Currently, he is a postdoctoral fellow at the
Laboratory for Information and Decision Systems (LIDS) at Massachusetts
Institute of Technology. His research interests are in the joint area of control,
optimization, game theory and information theory.

Download Report