On paved terrain and in defined urban environments UGVs are nowadays able to drive partially autonomously as the DARPA Urban Challenge (2007) and the VisLab Intercontinental Autonomous Challenge (2010) have shown. But even in this surrounding UGVs still sometimes require support from the human operator. In non urban environment this is even more difficult as the surface itself is often a challenge and it might happen that the paths which have been driven by the first vehicle cannot be passed by those following later due to changes of the ground. This presentation will present a method to control convoys of UGVs in unpaved terrain using the concept of adaptive robot control combined with a layer based approach for behaviors that are capable to control these UGVs. To allow rapid development and the capability to rerun and repeat different mission settings, a large simulation environment has been created and compared with results obtained from experiments. The algorithms for the control of UGVs that are handled are starting with A* and Mixed Integer Programming for Path Planning coming to sets of Fuzzy Systems for the control of the stability of UGV formation before some algorithms for path following and collision avoidance are discussed.

講演者	Prof. Roberto Tempo IEIIT-CNR, Politecnico di Torino
講演題目	"Design of Uncertain Complex Systems:　　　　　　A Randomization Viewpoint"

概要

In recent years, we have seen a growing interest in probabilistic and randomized methods for design of uncertain complex systems. In this lecture, we provide a broad perspective of this research area and discuss several randomized algorithms. In particular, we present sequential algorithms for convex problems and we discuss non-sequential algorithms for non- convex problems. Furthermore, we demonstrate the utility of this approach for analyzing specific applications, such as the computation of PageRank in the search engine Google and the design of autonomous unmanned aerial vehicles. 問い合せ先 藤田政之 機械制御システム専攻 電話/FAX: (03)5734-3813 E-mail: fujita@ctrl.titech.ac.jp 2010年度 第2回システム制御セミナー 日時　平成22年8月3日（火）　14:00 〜 15:00 会場　東京工業大学大岡山キャンパス南5号館312号室 講演者 Prof. Tyrone Vincent Center for Automation, Robotics, and Distributed Intelligence, Colorado School of Mines, USA 講演題目 "An LFT approach to identification to interconnected systems" 概要 System identification of a complex, multi-dimensional system from input/output data often requires the imposition of a-priori information in order for the identification to be practically feasible. One powerful type of a-priori information is that of interconnection structure - that is, a known decomposition of a system into simpler subsystems with sparse input/output connections. However, existing system identification tools do not allow for the user to easily incorporate this information. Recently, use of a linear fractional transformation (LFT) has been proposed as a canonical data structure for system identification, which has the advantage of allowing easy specification of structural information. This talk will review how this data structure can be used as both an analytical and computational tool of a variety of structured system identification problems, covering linear and non-linear systems and both parametric and non-parametric models. In particular, the LFT framework allows for a unified treatment of issues such as identifiability, persistence of excitation, robustness, and experiment design in interconnected systems. 問い合せ先 藤田政之 機械制御システム専攻 電話/FAX: (03)5734-3813 E-mail: fujita@ctrl.titech.ac.jp 2010年度 第1回システム制御セミナー 日時　平成22年7月12日（月）　10:30 〜 11:30 会場　東京工業大学大岡山キャンパス南5号館312号室 講演者 岩崎 徹也 先生 カリフォルニア大学ロサンゼルス校 講演題目 動物ロコモーションの推進原理と最適周期運動 概要 歩く、飛ぶ、泳ぐなど生き物のロコモーションは、 体の周期運動が環境と相互作用することにより実現される。その推進原理を理解し、神経制御機構を解明すれば、ロボティクスや 医療技術などへのさまざまな応用が期待できる。我々はパターン発生を目的とした新しい制御理論を構築することを目指し、 生物制御系のモデリングと解析を行ってきた。本発表では、まず生物のロコモーション制御系の全体像について具体例(蛭の泳ぎ)を 用いて説明し、神経オシレータ、筋肉駆動ダイナミクス、流体力学などの要素がフィードバック系を構成していることを示す。 その後、最適周期運動に焦点を絞り、与えられた振幅のもとで推進力を最大化した結果生じる動きとして蛇のうねりなどを説明できる ことを明らかにする。また、より一般的な最適周期運動発生問題を考え、その解がシステム行列の一般化固有ベクトルで与えられる ことを示し、くらげ、蛭、エイなどの泳ぎの原理の解明に応用する。 問い合せ先 藤田政之 機械制御システム専攻 電話/FAX: (03)5734-3813 E-mail: fujita@ctrl.titech.ac.jp