Dr. Mengmou Li

  • Title: Convergence Rate Analysis of the Mirror Descent Method via Integral Quadratic Constraints
  • Date: 2023.1.30 (Mon) 14:45-16:15
  • Place: 7F-Forum, 14th bldg, Yagami-Campus (tentative)
  • Abstract: This talk focuses on convergence analysis for the mirror descent (MD) method, a well-known algorithm in convex optimization. An analysis framework via integral quadratic constraints (IQCs) is constructed to analyze the convergence rate of the MD method with strongly convex objective functions in both continuous time and discrete time. Finding convergence rates of the MD algorithms can be formulated into feasibility problems of linear matrix inequalities (LMIs) in both schemes. In particular, in continuous time, it is shown that the Bregman divergence function, which is commonly used as a Lyapunov function for this algorithm, is a special case of the class of Lyapunov functions associated with the Popov criterion, when the latter is applied to an appropriate reformulation of the algorithm. Thus, applying the Popov criterion and its combination with other IQCs, can lead to convergence rate bounds with reduced conservatism.
  • Bio: Mengmou Li is currently a postdoc researcher in Tokyo Institute of Technology. He received his B.S. degree in Physics from Zhejiang University, China, in 2016, and the Ph.D. degree in Electrical and Electronic Engineering from the University of Hong Kong, in 2020. From 2021 to 2022 he was a research associate with the Control Group, University of Cambridge. His research interests include robust control, optimization, synchronization, and power systems.
  • Organizer: Masaki Inoue



  • 題目:Cyber-Physical-HUMAN Systems ~ロボティクスにおける歴史から最前線まで~
  • 日時:2022年8月3日(水)15:00~16:30 
  • 場所:オンライン(参加を希望される場合は井上 (minoue at appi.keio.ac.jp) までご連絡ください)
  • 概要:本講演ではこれまでのロボティクス分野におけるCPHS研究をTask Architectureと人間モデルの観点から概観し、最新の研究トレンドを明らかにする。
  • 略歴:Takeshi Hatanaka received the Ph.D. degree in applied mathematics and physics from Kyoto University in 2007. He then held faculty positions at Tokyo Institute of Technology and Osaka University. Since April 2020, he is an associate professor at Tokyo Institute of Technology. He is the coauthor of “Passivity-Based Control and Estimation in Networked Robotics” (Springer, 2015), coauthor of “Control of Multi-agent Systems” (Corona Publishing Co., 2015) and the editor of “Economically-enabled Energy Management: Interplay between Control Engineering and Economics” (Springer Nature, 2020). His research interests include cyber-physical & human systems and networked robotics. He received the Kimura Award (2017), Pioneer Award (2014), Outstanding Book Award (2016), Control Division Conference Award (2018), Takeda Prize (2020), and Outstanding Paper Awards (2009, 2015, 2020) all from The Society of Instrumental and Control Engineers (SICE). He also received 3rd IFAC CPHS Best Research Paper Award (2020) and 10th Asian Control Conference Best Paper Prize Award (2015). He is serving/served as an AE for IEEE TSCT and SICE JCMSI, and is a member of the Conference Editorial Board of IEEE CSS. He is a senior member of IEEE.
  • 世話人:井上

小野雅裕 博士

  • 題目:自律的ロボットによる太陽系探査の現在と未来
  • 日時:2022年7月7日(木)13:00~14:30 
  • 場所:矢上キャンパス 創想館2階 14-204(セミナールーム4)
  • 概要:本講演は英語のスライドを使用し日本語にて行われます。質問は両言語でお受けします。 This talk will be given in Japanese with English slides. Q&A will be in both languages. Abstract: 太陽系探査の黎明から60年が経ち、輝かしい発見が積み上がる一方、簡単にアクセスできる未踏の探査目標は減りつつある。未来の探索ミッションを遂行するロボットは困難かつ未知な環境下で活動したり(例: 木星・土星の氷衛星の地底の海)、極めて野心的な目標を達成することを求められたりする(例: 月面の1000キロ以上の走行)。そのようなミッションを実現するための鍵となる技術のひとつが、自律化である。たとえば2021年2月に着陸したNASAの火星ローバー・パーサヴィアランスはこれまでのローバーで最も高度な自動運転機能が搭載されており、過去の火星に存在したかもしれない生命の証拠をさがす挑戦的なミッションに大きく貢献している。また、JPLは現在EELS (Exobiology Extant Life Surveyor)と呼ばれるヘビ型のロボットを現在開発している。その目的は将来、土星の衛星エンセラドスの氷の裂け目を降下し地底の海に地球外生命を探すことを可能にすることである。環境の不確定性と地球との通信遅延により、EELSを地球から手動で操作することは非現実的である。この講演ではJPLにおける惑星探査ロボットの自立化の研究開発の一端を紹介し、未来のミッションにおける技術ニーズの知見を共有する。
  • 略歴:NASAジェット推進研究所 Robotic Surface Mobility Group グループリーダー。Mars 2020ローバーミッションの一員として、パーサヴィアランスローバーの運用に関わる。過去にはその自動走行アルゴリズムの開発や着陸地点選定に携わった。また、EELSプロジェクトの主任研究員 (PI)を務め、約50人のチームを率いる。研究テーマは宇宙探査用ロボットの自律化で、とりわけ機械学習の認知、データ解釈や意思決定への応用に興味がある。2005年東京大学航空宇宙工学科学士、2007年マサチューセッツ工科大学修士、2012年同大学博士卒。2012年より慶應義塾大学理工学部助教。2013年より現職。阪神ファン。ミーちゃんのパパ。
  • 世話人:足立修一
  • 参加を希望される方は,必ず足立(adachi[at]appi.keio.ac.jp)と斉藤秘書(elfe[at]appi.keio.ac.jp)に,事前に(7月4日までに)e-mail で,所属,氏名,メールアドレスなどを連絡してください。
  • パンフレット

Prof. Yorie Nakahira (Carnegie Mellon University)

  • Title: Myopically verifiable probabilistic certificate for long-term safety and its autonomous driving application
  • Date: 2022.6.22 (Wed) 15:00-16:00
  • Room: Seminar Room 4 (14-204; セミナールーム4)
  • Abstract: In this talk, we will first introduce our recent work that focused on barrier function-based approaches for the safe control problem in stochastic systems. With the presence of stochastic uncertainties, a myopic controller that ensures safe probability in infinitesimal time intervals may allow the accumulation of unsafe probability over time and result in a small long-term safe probability. Meanwhile, increasing the outlook time horizon may lead to significant computation burdens and delayed reactions, which also compromises safety. To tackle this challenge, we define a new notion of forward invariance on ‘probability space’ as opposed to the safe regions on state space. This new notion allows the long-term safe probability to be framed into a forward invariance condition, which can be efficiently evaluated. We build upon this safety condition to propose a controller that works myopically yet can guarantee long-term safe probability or fast recovery probability. The proposed controller ensures the safe probability does not decrease over time and allows the designers to directly specify safe probability. This framework can also be adapted to characterize the speed and probability of forward convergent behaviors, which can be of use to finite-time Lyapunov analysis in stochastic systems. Building upon the above framework, we will then present an adaptive safe control method that can adapt to changing environments, tolerate large uncertainties, and exploit predictions in autonomous driving. The use of long-term safe probability provides a sufficient outlook time horizon to capture future predictions of the environment and planned vehicle maneuvers and to avoid unsafe regions of attractions. The resulting control action systematically mediates behaviors based on uncertainties and can find safer actions even with large uncertainties. This feature allows the system to quickly respond to changes and risks, even before an accurate estimate of the changed parameters can be constructed. The safe probability can be continuously learned and refined. Using more precise probability avoids over-conservatism, which is a common drawback of the deterministic worst-case approaches. The proposed techniques can also be efficiently computed in real-time using onboard hardware and modularly integrated into existing processes such as predictive model controllers.
  • Bio: Yorie Nakahira is an Assistant Professor in the Department of Electrical and Computer Engineering at Carnegie Mellon University. She received B.E. in Control and Systems Engineering from Tokyo Institute of Technology and Ph.D. in Control and Dynamical Systems from California Institute of Technology. Her research interests include the fundamental theory of optimization, control, and learning and its application to neuroscience, cell biology, smart grid, cloud computing, finance, and autonomous driving. Her group website can be found here: https://www.cmu.edu/ece/learning-control/index.html
  • HostYutaka Hori

宮下令央 特任講師 (東京大学)

  • 題目: 高速画像処理による視覚情報操作
  • 日時: 2022.5.25 (Wed) 16:00-17:00
  • 場所: セミナールーム4 (14-204)
  • 概要: 計算機の発展により情報処理の高速化が進んでいる。高速な情報処理を用いてアクチュエータを制御することで、人間の動作速度を超える高速ロボットシステムが構築され、生産性の向上や動的対象の操作が実現されてきている。一方で、人間を対象とする応用では、人間の知覚速度を超える高速システムを構築し、没入感の高い拡張現実空間を創造することができる。すなわち、実世界の現象を捉え、情報処理を施した後、実世界に反映する操作を人間が気づかない速度で行うシステムを組み込むことで、実世界の物理に相当する部分を仮想的に再定義することが可能であると考えられる。さらに、物理を仮想的に再定義することは、人間が物理を通して物質を捉えている以上、物質をも拡張することを意味している。本講演では、高速画像処理による高速センシングと高速ディスプレイを用いて、人間の知覚速度を超える高速情報システムを構築し、実世界の物質や物理を視覚的に書き換え、拡張する研究について紹介する。
  • 略歴: 2012年東京大学工学部計数工学科卒、2017年東京大学情報理工学系研究科システム情報学専攻博士課程修了、博士(情報理工学)、2020年より東京大学情報基盤センター特任講師。映像情報メディア学会 丹羽高柳賞 論文賞、井上科学振興財団 井上研究奨励賞、船井情報科学振興財団 研究奨励賞など受賞多数。高速画像処理を用いた高速センシングシステム、リアルタイムXRシステムの開発に従事。
  • ブログ記事
  • 世話人堀豊 & 井上正樹

Mr. Hiroyasu Tsukamoto (California Institute of Technology)

  • Title: Contraction Theory for Designing Safe, Stable, and Robust Learning-based GNC: A Tutorial Overview
  • Date: 2022.5.11 (Wed) 9:00-10:30
  • Place: Zoom
  • Abstract: AI and machine learning technologies have been utilized for achieving safe and stable autonomy of aerospace and robotic systems. Stability and safety are typically research problems of control theory, while conventional black-box AI approaches lack much-needed robustness, scalability, and interpretability, which are indispensable to designing control and autonomy engines for safety-critical robotic missions on land, in water, or in deep space. This talk gives a brief tutorial overview of contraction theory for deriving formal robustness and stability guarantees of various learning-based and data-driven automatic control problems, with some illustrative examples including the recent NASA JPL-Caltech RTD project on learning-based Interstellar Object (ISO) exploration. This talk is based on a tutorial session, Contraction Theory for Machine Learning, which we organized at the 2021 IEEE Conference on Decision and Control.
    For more information, see https://sites.google.com/view/contractiontheory.
  • Bio: Hiroyasu Tsukamoto is an aerospace Ph.D. student at GALCIT, Caltech. His research interest includes deep learning-based robust optimal control, estimation, and motion planning for general nonlinear systems, aerial swarms, and autonomous aerospace systems. (Google Scholar: https://scholar.google.com/citations?user=G9iATfcAAAAJ&hl=ja, LinkedIn: https://www.linkedin.com/in/hiroyasu-tsukamoto-32500a1a5/)
  • Blog
  • Organizers: Yutaka Hori & Masaki Inoue

古賀朱門博士 (University of California, San Diego)

  • 題目:SLAMのための最適制御と強化学習
  • 日付:2022.4.26 (火)13:00-14:30
  • 場所:セミナールーム1
  • 概要:自動運転車やドローンなどの移動型ロボットが近年目覚ましい発展を遂げている。特に、GPSが遮断された環境下で、移動型ロボットに組み込まれているセンサ(例:カメラ、ライダー)を用いて、不確定かつ複雑な環境を再現する技術が注目を浴びている。そのような課題は、Simultaneous Localization and Mapping (SLAM) という問題に帰着され、災害救助や惑星探索、また拡張現実などの幅広い応用に高い期待がされている。本講演では、SLAMの基礎からはじめ、またその精度を向上させるようなロボットの軌道設計についてお話しする。数学的には、SLAMをカルマンフィルタによって行い、ある情報理論的な評価関数が最小になるようなロボットの入力を求める最適制御問題に取り組む。モデルに基づく制御法から、モデルに依存しない強化学習までの手法を提案し、実データを用いて性能の有効性を立証する。
  • 略歴:2014年慶應義塾大学理工学部物理情報工学科卒、2020年カリフォルニア大学サンディエゴ校機械航空工学科博士課程修了。博士課程在学中にレンセラー工科大学にて訪問研究、またNASAジェット推進研究所、三菱電機研究所にてインターンを経験。2020年から、カリフォルニア大学サンディエゴ校電気コンピュータ工学科にてポスドク研究員。アメリカ制御学会の最優秀論文賞、カリフォルニア大学サンディエゴ校の制御分野における最優秀博士論文賞などを受賞。博士課程では偏微分方程式の制御理論、Extremum Seekingによる最適化と学習、またそれらの様々な応用(3Dプリンタ、リチウムイオン電池、熱エネルギー貯蔵、交通渋滞、気候変動)に従事。現在はロボティクスの最適化と機械学習、特に位置地図同時推定(SLAM)や軌道設計などのプロジェクトに携わる。
  • ブログ記事
  • 世話人:Yutaka HoriMasaki Inoue

井手剛博士 (IBM Thomas J Watson Research Center)

  • 題目:人工知能ブームの現在と最近の研究から
  • 日付:2022.4.5 (火)10:00-12:00
  • 場所:Zoom
  • 概要:言語、画像、音声の分野での深層学習のブレイクスルーをきっかけに、AI (artificial intelligence) が一般メディアの話題になるようになってしばらく経ちます。この講演では最初に、少し前に盛んに言われていたAIに対する期待と、2022年における現実を俯瞰します。その後、AI脅威論の流れで注目を集めてきた「AIの説明可能性」についての所見を述べ、最近の研究成果(Ide et al., AAAI 21; NeurIPS 21) について概要を紹介したいと思います。
  • ブログ記事
  • 世話人:足立修一


Dr. Zoltan Tuza (Imperial College London)

  • Title: Characterization of Biologically Relevant Network Structures from Time Series Data
  • Date: 2019.11.8 (Fri) 16:30-18:00
  • Abstract: High-throughput data acquisition in synthetic biology leads to an abundance of data that need to be processed and aggregated into useful biological models. Building dynamical models based on this wealth of data is of paramount importance to understand and optimize designs of synthetic biology constructs. However, building models manually for each data set is inconvenient and might become infeasible for highly complex synthetic systems. In this talk, we present state-of-the-art system identification techniques and combine them with chemical reaction network theory (CRNT) to generate dynamic models automatically. On the system identification side, Sparse Bayesian Learning offers methods to learn from data the sparsest set of base functions necessary to capture the dynamics of the system into ODE models; on the CRNT side, building on such sparse ODE models, all possible network structures within a given parameter uncertainty region can be computed. Additionally, the system identification process can be complemented with constraints on the parameters to, for example, enforce stability or non-negativity—thus offering relevant physical constraints over the possible network structures. In this way, the wealth of data can be translated into biologically relevant network structures, which then steers the data acquisition, thereby providing a vital step for closed-loop system identification.
  • Organizer: Yutaka Hori

Prof. José María Maestre Torreblanca (University of Seville)

  • Title: Centralized, Distributed, and Coalitional Model Predictive Control
  • Date: 2019.9.5 (Thurs)
  • Abstract: Model predictive control (MPC) has become of the most popular control techniques due its flexibility. Issues such as constraints on the control problem variables, delays in the system dynamics, and multiple objectives can be handled explicitly in the MPC framework. The evolution of computer, information and communication technologies has motivated the application of MPC to problems beyond its scope years ago and the development multiple noncentralized MPC approaches. The goal of this talk is to present a coherent and easily accessible overview regarding model predictive control and some of the latest developments regarding its application to large-scale cyber physical systems, including topics such as coalitional control and human in the loop.
  • Bio: J.M. Maestre received the Ph.D. degree in automation and robotics from the University of Seville, where he works as associate professor. He has also worked in LTH at Lund University, TU Delft, and Tokyo Institute of Technology, where he is currently funded by the Japanese Society for the Promotion of Science. Besides his PhD, which was awarded with the extraordinary prize of the University of Seville, he also has master degrees in intelligent buildings and economics. His main research interests are the control of distributed systems and the integration of service robots in the smart home. He has authored and coauthored more than one hundred publications regarding these topics. He is also editor of the books “Service robotics within the Digital Home: Applications and Future Prospects” (Springer, 2011), “Distributed Model Predictive Control Made Easy” (Springer, 2014), and “Domotica para Ingenieros” (Paraninfo, 2015). Finally, he is one of the founders of the technological firms Idener and Eskesso.
  • Blog
  • Organizer: Masaki Inoue

Prof. Ahmet Cetinkaya (National Institute of Informatics)

  • Title: Randomized Communication Protocols for Secure Networked Control Under Jamming Attacks
  • Date: 2019.8.1 (Thurs)
  • Abstract: Recent control architectures in cyber physical systems utilize wireless communication technologies for transmission of measurement and control data packets to remote locations. As the Internet of Things is becoming more popular, the use of wireless technologies in networked control systems is expected to increase even more. These new developments are bringing efficiency to control systems, but they are also expected to introduce vulnerabilities that can be exploited by cyber-attackers. For instance, jamming attackers may be able block the transmission of data packets on a wireless channel by emitting strong interference signals. It has been shown recently that wireless networked control systems that are designed based on classical periodic (or event-driven) sampled-data control approaches may suffer from jamming attacks and face instability even under attacks coming from an energy-constrained attacker. To avoid instability under such control approaches, a restriction on the attack frequency becomes necessary. Specifically, the frequency at which the jamming attacks can be turned on and off is required to be less than the frequency of the communication attempts. In this talk, we will look at a new randomized control & communication approach that allows secure operation under jamming attacks with arbitrarily large frequencies. In this approach, control and measurement data packets are attempted to be transmitted at random time instants with a fixed expected interval between them. We show that this randomized scheme guarantees infinitely many successful transmissions in the long run as long as the intervals of jamming attacks do not cover the entire time domain. In multi-agent systems with marginally-stable integrator agent dynamics, this suffices for achieving consensus. To handle networked control of plants with unstable dynamics, we further look at the long run average number of successful transmissions. We show by employing tail probability bounds that this average number is actually lower-bounded by a constant that depends solely on the attack durations and not on the attack frequency. By using a recent result on the stability of networked control systems, we show that our randomized approach can guarantee almost-sure asymptotic stabilization if the average jamming duration is sufficiently small regardless of how frequent the attacks may occur.
  • Organizer: Masaki Inoue


澤田賢治准教授 (電気通信大学)

  • 題目:制御工学で制御システムのセキュリティは守れるのか?
  • 日時:2018.6.27
  • 場所:セミナールーム1
  • 概要:2010年のイラン核燃料施設のStuxnet感染事件から始まり,2017年のランサムウェア大規模感染は「制御システムが悪意ある攻撃者に狙われていること」が事実であること世界に強烈に印象付けた.本講義では,世界が制御システムのセキュリティ解決に苦労している状況とその理由を解説しつつ,講演者が関わる制御システムのセキュリティ技術動向を解説する予定である.
  • 略歴:2009年大阪大学大学院工学研究科機械工学専攻博士後期課程修了.同年電気通信大学システム工学科助教,22015年同大学 i-パワードエネルギー・システム研究センター准教授となり現在に至る.博士(工学).ハイブリッドシステムや制御系セキュリティに関する研究に従事.2016年より制御システムセキュリティセンター顧問.
  • 世話人:Masaki Inoue