ゼミナール発表

日時: 9月30日(火)3限 (13:30-15:00)


会場: L1

司会: 樫原 茂
亀本 大貴 1351033: M, 2回目発表 杉本 謙二, 岡田 実, 松原 崇充

title: Electricity market model based on persuasive dialogue systems

abstract: Recently, the increasing demand for energy and growing environmental concerns have created the need for a more efficient power grid that will accommodate distributed and renewable energy resources, storage, and real-time demand response technologies.We consider an abstract market model for designing demand response to match power supply,and propose distributed demand response algorithms to achieve the equilibria.

 
佐原 壮海 1351053: M, 2回目発表 山口 英, 岡田 実, 藤川 和利, 門林 雄基, 奥田 剛
title: Speed-Aware Rate Adaptation in Vehicular Environment
abstract: In order to select an optimum transmission rate in wireless LAN, various rate adaptation techniques have been proposed. These schemes were basically designed for static environments, such as indoor wireless communication. However, they have not been sufficiently researched as to whether rate selection is appropriately performed according to a rapidly-changing communication environment such as a vehicular environment. Therefore, the goal of this study is to propose a speed-aware rate adaptation adapted for vehicular environment. In this presentation, we discuss the problems of the existing rate adaptation schemes based on the simulation results and describe an approach for the problems.
language of the presentation: Japanese
発表題目: 移動速度を考慮したレートアダプテーションの提案と評価
発表概要: 無線LANにおいて適切な伝送レートを選択するためにレートアダプテーション技術が提案されてきたが、それらは主に室内など比較的移動性の低い環境を対象としてきた。今後は車車間や路車間のような移動環境においても無線LAN技術が適用され普及していく。しかし、既存のレートアダプテーションが移動時の通信環境の変動に合わせて適切にレート選択が行われているかは十分に調査されていない。本研究では、車のように速度変化の大きい環境においても、無線区間の通信品質に追従して、適切な伝送レートを選択可能なレートアダプテーションの提案と評価を行う。本発表ではこれまでに得られたシミュレーション結果をもとに、移動環境における既存のレートアダ プテーションの問題点を明らかにし、その改善手法の方針について発表する。
 
泉川 大樹 1351008: M, 2回目発表 山口 英, 藤川 和利, 小笠原 司, 門林 雄基, 猪俣 敦夫, 奥田 剛
title: A method for using Unmanned Aerial Vehicles to collect information on persons in need of aid during disaster
abstract: It is difficult to grasp situations of damage when natural disaster (i.e., an earthquake and/or Tsunami occurs, because roads and communication infrastructure may be unusable. In this study, we propose and evaluate a method for collecting the location information of persons in need of special aid on occasions of evacuation from disaster situations. In the presentation, we discuss the need of the basic evaluation for obtaining the information from UAV. We then describe the approach for achieving it.
language of the presentation: Japanese
発表題目: UAV を用いた要援護者情報取得手法の提案と評価
発表概要: 地震や津波などの災害時には、道路の寸断や通信インフラの断絶等により被災状況を確認することが困難である。そこで、本研究では、特に避難時に特別な支援が必要な要援護者を対象に、UAV(Unmanned Aerial Vehicle)を用いて彼らの位置情報を収集する仕組みについて提案及び評価する。本発表では、まず、UAVを用いる際の基礎評価の必要性について述べ、本手法の実現方法について説明する。
 

会場: L2

司会: 小野 直亮
渡辺 仁 1351114: M, 2回目発表 池田 和司, 佐藤 嘉伸, 久保 孝富

title: Nose position estimation of mice from motion clips and detection of characteristic action based on infinite Gaussian mixture model

abstract: In animal behavior analysis, the demand for automated analysis from motion clips is increasing from aspects of load and reproducibility of experimenters. In this study, we use mice as target animal. Firstly, it is known that mice search their around by their nose, but an estimation method which has high accuracy has not been established. So, we proposed a novel method which is based on simple region segmentation. In experimental evaluation, our method improved the accuracy of conventional method. The other problem is that there are some cases where pre-definition of action patterns become difficult because action patterns of mice have high variability depending on experimental conditions. To solve these problem, we apply infinite Gaussian mixture model which can estimate action patterns from data. In experiment, we confirmed that some characteristic action patters can be detected by this method.

language of the presentation: Japanese

 
福間 良平 0961019: D, 中間発表 池田 和司☆3, 佐藤 嘉伸, 久保 孝富
title: MEG-based neuroprosthetic hand control by patients with paralysis
abstract: [Objective] Brain Machine Interface (BMI) has been shown to improve the quality of daily life for severely paralyzed patients. However, it remains unclear how the motor information is preserved in the paralyzed patients, how the properties of the preserved motor information affects the efficacy of BMI, and the relation of neurological symptoms and the preservation of motor information. [Methods] One amputee and five patients suffering from brachial plexus root avulsion, who all have phantom limb sensations, attempted to grasp or open their affected hands while magnetoencephalogram (MEG) signals were recorded in an open-loop condition. As a control, MEG signals from non-paralyzed, right-handed subjects executing actual movements on their right hand were also measured. The motor information was evaluated from the slow component of the MEG signals. Moreover, the ability to control the BMI was evaluated using a neuroprosthetic hand controlled by the slow component of MEG signals in a closed-loop condition. The ability to control BMI, as well as the motor information, were compared among patients and between the patients and non-paralyzed subjects. [Results] Even while the severely paralyzed patients attempted to move their affected hands in the open-loop condition, the characteristic activation on the contralateral sensorimotor cortex was found to be similar to the non-paralyzed subjects performing hand movements. The preserved motor information was correlated with the subjective movability of the phantom limb and the ability to control the BMI during the closed-loop condition. [Interpretation] It was found that the motor information of the paralyzed patient was dependent on the subjective difficulty of moving their phantom limb. Moreover, the preserved information affected the ability to control the BMI. The proposed BMI system is thus suggested to be suitable for evaluating the ability to control neuroprosthetic hands and as a training tool to improve control ability.
language of the presentation: Japanese
発表題目: 麻痺患者によるMEGを用いた神経義手制御
発表概要: [Objective] Brain Machine Interface(BMI)は重度麻痺患者の日常生活の質を改善できることが示されている.しかし,BMIの対象である重度麻痺患者において,どのように運動情報が残存し,BMIの適応性に影響するか,また残存した運動情報と関連した神経学的症状は明らかでない. [Methods] 幻肢のある5人の腕神経叢引き抜き損傷患者と1人の切断肢患者が,患肢の離握手を企図した際の脳磁界信号 (MEG)を計測した.また,対照として健常右利き被験者が右手で同じ運動を行った際のMEGを計測し比較した.さらに頭頂部を中心とした84チャンネルから計測したMEG信号の低周波成分を用いてリアルタイムに神経義手を制御し,患者毎の義手制御能を評価した.また,同MEG信号から運動情報量を評価し,健常者と患者および患者間で比較した. [Results] 患者が患肢の運動を企図した際には健常者と似た脳磁界信号が計測され,その主な信号源は健常者同様に患肢の対側の感覚運動野であった.また,残存している運動情報量は主観的な幻肢運動の容易さと相関していた.さらに,リアルタイムでの義手制御能は残存した情報量と相関していた. [Interpretation] 麻痺患者の運動情報が患者の主観的な幻肢の運動の困難さと関係していることが示された.また,残存した情報量はBMIの操作能と正に相関することが示された.本研究で提案したBMIが,侵襲的BMIの非侵襲的な適応判定や,適応性を上げるためのトレーニングに応用できることが示唆された.
 
石原 佑彌 1351007: M, 2回目発表 小笠原 司, 佐藤 嘉伸, 高松 淳, 池田 篤俊
title: Robotic Motor Training System for Decreasing Tremor during Postural Control
abstract: It is said that the adjustment of the joint stiffness by co-contraction of the antagonist muscle and agonist muscle is important for human posture control. However, it has been recognized that co-contraction is generated tremor at around 4-12 Hz by the interaction between antagonist muscle and agonist muscle. And it is hinder steady posture control. The purpose of this research is development of robotic motor training system which can decrease effectively tremor during posture control. In this presentation, I explain about exercise physiology mechanisms by generated tremor, training theory may have possibility of decreased tremor, outline about robotic motor training system and verification experiment.
language of the presentation: Japanese
発表題目: 姿勢制御中の震えを軽減するロボティック筋トレシステム
発表概要: ヒトの姿勢制御においては主動筋と拮抗筋の共収縮による関節剛性の調整が重要であると言われている. しかし,共収縮時には筋の相互作用により4-12Hz程度の振動が発生することが知られており,安定した姿勢制御の妨げとなっている.本研究では,トレーニングにおいて姿勢制御における振動を効率的に軽減させるロボティック筋トレシステムの開発を行う.本発表では,振動の発生する運動生理学的なメカニズムを紹介し,振動を軽減させうる可能性のあるトレーニング理論を紹介する.また,ロボティック筋トレシステムの概要と実験計画について述べる.