| 松田 大樹 | 1751109: M, 1回目発表 | 大規模システム管理 | 笠原 正治 |
|
title: Road reliability model considering secondary disaster
abstract:When a large-scale disaster occurs, evacuees have to evacuate to refuges speedily and safely.To tackle this problem, there has been proposed an automatic, speedy and reliable evacuation guiding scheme based on implicit interactions among evacuees and their mobile devices, by using the geographical information about map and road blockage probability. However, this scheme does not consider the disaster risks caused by secondary disaster, e.g., aftershocks.For example, it is reported that aftershocks whose magnitude exceeded seven occurred twice within one hour after the main shock in the 2011 Great East Japan Earthquake. In this presentation, we propose a road reliability model, which decays with time according to the road blockage probability and the existing aftershock probability model. language of the presentation: Japanese | |||
| 脇本 宏平 | 1751126: M, 1回目発表 | 知能コミュニケーション | 中村 哲 |
|
title: Concept Learning of Action by Structuralization of Tasks using Multimodal LDA
abstract: Robots that support human life are required in nursing places and homes. Such robots need to perform various actions according to the situation, but generation of such actions is difficult. Acquisition of concept of action is important for flexible behavior of robot. In this research, I propose a learning system which acquires action concept using multi - modal information and structuring tasks. language of the presentation: Japanese | |||
| 椛島 基嵩 | 1751032: M, 1回目発表 | 生体医用画像 | 佐藤 嘉伸 |
|
title: Automated Initialization of 2D-3D Registration of Rib Cage from Chest Radiography using Convolutional Neural Network
abstract: The difficulties in automation of 2D-3D registration are two-folds: initialization and failure detection. In this presentation, we focus on automation of the initial pose estimation using a convolutional neural network (CNN). We describe our initial attempts in 2D-3D registration of the rib bones. We learned the CNN using digitally reconstructed radiographs generated from the CT volumes, which outputs the likelihood map of the rib bones in the chest radiography. We discuss the accuracy of the proposed method by experiments using simulated images and radiography. language of the presentation: Japanese 発表題目: 胸部X線画像を用いた畳み込みニューラルネットワークによる肋骨2D-3D位置合わせの自動初期化 発表概要:2D-3D位置合わせの全自動化において，初期姿勢の推定および位置合わせの失敗検出の自動化が問題となる．そこで，初期姿勢の自動推定に焦点を当て，畳み込みニューラルネットワークの適用を提案する．本発表では，初期的検討として，CTボリュームを用いて生成した擬似X線画像を学習した畳み込みニューラルネットワークを用いて，胸部X線画像から肋骨領域の尤度マップの推定，および尤度マップを用いた初期姿勢推定について報告する．さらに，提案手法を用いて，生成画像および実画像を用いた初期姿勢の自動推定精度について定量評価実験を行ったのでそれについて報告する． | |||
| 井上 裕太 | 1751010: M, 1回目発表 | 光メディアインタフェース | 向川 康博 |
|
title:Measurement of scattered light from shadow region by Time of Flight camera
abstract:State-of-the-art technology in computer vision enables material classification from distance distortion in measurement using Time of Flight (ToF) camera. This method requires a database of the combination of shape and material. Thus, even if the material is same, the classification can be incorrect for different shape. In this study, we aim at material classification using ToF camera from the scattered light properties that is independent from the shape. To measure the scattered light on object, we use shadow region that has no direct reflection. A preliminary experiment was conducted to extract the scattered light on an object by measuring its shadow region. In the future, we will analyze this scattered light to model the scattering properties. language of the presentation: Japanese | |||
| 上田 朝己 | 1751016: M, 1回目発表 | 光メディアインタフェース | 向川 康博 |
|
title: Interpretation of separated components by Epipolar imaging
abstract: Epipolar imaging is proposed as a method of light component separation that uses the geometric relationship between a projector and a camera. The components this method separates differ from the conventional High-Frequency illumination method. In this research, I conducted the comparative experiments of these methods and interpreted the components that Epipolar imaging separates. For the experiments, I arranged an experimental environment that is compatible with both methods. It became clear that the refracted light and the reflection on mirror surface, that kept high frequency property, was classified into different components by High-Frequency illumination and Epipolar imaging. language of the presentation: Japanese 発表題目: エピポーライメージングによって分離される成分の解釈 発表概要: プロジェクタとカメラの幾何関係を利用した光の成分分離手法としてエピポーライメージングが提案されている。従来の高周波照明法とこの手法とでは分離される成分が同一ではない。本研究では、高周波照明法との比較実験を行うことでエピポーライメージングが分離する成分を解釈する。そのために、これらの手法を両立できる実験環境を構築した。屈折や鏡面への映り込みなどの高周波成分が残るものについては、高周波照明法とエピポーライメージングでは異なる成分に分類されることが明らかとなった。 | |||
| 中西 由希子 | 1751081: M, 1回目発表 | 光メディアインタフェース | 向川 康博 |
|
title: Quantitative evaluation of floating-expression on spherical display
abstract: In this study, we aim to quantitatively evaluate human perception about floating-expression on spherical display.3D vision using binocular parallax has been popular but it can not be directly applied to spherical display.Therefore, visual expression is proposed that gives floating-perception using motion parallax and adding shadow.Then, the plan of the evaluation experiment is described using the paired comparison method about the quantification of floating-feeling. language of the presentation: Japanese 発表題目: 球体ディスプレイにおける浮遊感の定量評価 発表概要: 本研究では、球体ディスプレイにおける浮遊感表現において、人間の知覚を定量評価することを目的とする。3D映像といった両眼視差を用いた立体表現が盛んになっているが、球体ディスプレイには直接適用できない。そこで、運動視差や影付加などを用いた浮遊感知覚を与える映像表現を提案する。そして、浮遊感の定量化にむけて一対比較法を用いた評価実験の計画を述べる。 | |||
| 大須賀 彩希 | 1751021: M, 1回目発表 | 情報セキュリティ工学 | 林 優一 |
|
title: Study on Noninvasive Frequency Injection Attack against TRNG
abstract: Recently, true random number generators (TRNGs) have been studied as the target of side channel attacks. TRNGs are a vital part of many cryptographic protocols since they are used to generate the streams of random bits used as secret keys, random masks, initialization vectors or padding values. If attackers can control the behavior of TRNGs, they can degrade the security of the cryptographic systems. In this study, we will discuss a possibility that degradation of security in TRNG caused by a noninvasive frequency-injection attack. Moreover, we will consider countermeasures against this threat from the viewpoint of EMC. language of the presentation: Japanese 発表題目: *** この部分を発表題目に *** 発表概要: *** この部分を発表概要に *** | |||
| 鍛治 秀伍 | 1751027: M, 1回目発表 | 情報セキュリティ工学 | 林 優一 |
|
title: Arbitrary Command Injection Threats against Electronic Devices and Its Countermasure
abstract: Intentional electromagnetic interference (IEMI) causes degradation of security in electronic devices. Previous studies mainly focused on the degradation of availability of equipment caused by high power electromagnetic (HPEM) interference; there is not enough discussion about confidentiality and integrity. In this study, I will consider security degradation caused using EM injection with much lower power than HPEM. Concretely, we will induce electromagnetic fields in electronic devices by malicious hardware and yield arbitrary commands inside devices. We will also develop countermeasures against this kind of the threat. language of the presentation: Japanese | |||
| 郡 義弘 | 1751040: M, 1回目発表 | 情報セキュリティ工学 | 林 優一 |
|
title: Development of Electrical Level Countermeasure against Side-Channel Attacks
abstract: Side-channel attacks are one of the most powerful attacks against cryptographic modules. So far, many countermeasures were discussed in algorithm, protocol, and circuit level. However, these countermeasures were different for each cryptographic algorithm. In this study, we will develop a side-channel countermeasure which can be used commonly in many algorithms at the hardware level. language of the presentation: Japanese | |||
| 猪川 真琴 | 1751003: M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title: Generation of Robot Motion Producing Favorable Impression and Relationship Between Background and Evaluation of Subjects
abstract: In the future, it is expected that the proportion of the service robots will increase in the robot field, and the opportunity for humans to come into contact with robots probably increases. So, we believe that robots need to be acceptable to many people. Therefore, in this research, we focus on the motion, which is one of the factors affecting human impression. We generate the robot motion that produces the favorable impression and we investigate the likeability of the motion. Based on the investigation, we clarify the relationship between the subjects' background and their evaluations. language of the presentation: Japanese | |||
| 北村 勇希 | 1751035: M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title: Development of a cane system for falls prevention while walking
Fractures caused by falls become serious problems from large influence on the body function. There is a strength training that improves physical function and walking aids such as canes and carts for falls prevention. However, it takes a long period of time to prevent the falls by strength training, and walking aids have problems such as many accidents. Therefore, many cane robot have been developed. However, they have problems with portability. In this research, we propose a cane system that prevents falls of a user. The proposed cane system detects falls of a user and it actively leads cane tip to a position where the user doesn't fall down. language of the presentation: Japanese | |||
| 木下 勝吏 | 1751036: M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title: Quantitative evaluation of proficiency in rowing with non-contact sensor and assistance of coaching
abstract: Rowing is a sport which propel a boat on water using oars. It requires extreme physical strength and endurance, a high level of skill and technique is essential to enable an effective transfer of power through the rowing sequence. Because it is difficult to coach the crew from riverside, Feedback in rowing is traditionally presented as paper reports or video overlaid with data once a session has been completed. In this study, we evaluate quantitatively proficiency in rowing by unobtrusive method like ultrasonic sensor or IMU. Moreover, we develop an on-board rowing coaching system which provide auditory and visual feedback function. language of the presentation: Japanese | |||
| 佐久間 達也 | 1751046: M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title: Development of universal gripper with optical sensing
abstract: In recent years, as a flexible gripper, the universal gripper has been attracting a great deal of attention due to its very simple structure and the advanced grasping ability of the irregular and delicate objects. In this research, in order to improve the grasping stability and operability, we develop a novel universal gripper with optical sensing function. We select the materials of the universal gripper to make it transparent and suitable for optical sensing by inserting a camera in. We also propose a method to estimate the shape of the grasped object and the grasping force based on the camera images. language of the presentation: Japanese 発表題目: 光学式センシング機能を有するユニバーサルグリッパの開発 発表概要: 近年，構造が単純であり，多種多様な物品に対応可能な柔軟なグリッパとしてユニバーサルグリッパが注目されている． 本研究では物体把持安定性と操作性を向上させるため，光学式センシング機能を有するユニバーサルグリッパを開発する． ユニバーサルグリッパを構成する材料を変更することで，カメラによる光学センシングが可能な構造とし， カメラ画像を用いて対象物との接触部を解析することによって把持物体の形状と把持力の推定を実現する． | |||
| 安田 晃也 | 1751117: M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title: Manipulation for packing various objects
abstract: Short of hands in logistics is a problem to be solved quickly. In logistics, there are simple hand works such as packing objects ordered by e-commerce in the warehouse. However, these works have not yet been automated. The reason for this is that there are various items such as simple items and deformable objects that are sold by e-commerce, and it is difficult to perform accurate manipulation of them. In this study, we propose manipulation of various objects to solve the problem. We propose action procedures necessary to realize accurate manipulation of various objects and give way to solve each procedure. language of the presentation: Japanese | |||
| 佐野 哲也 | 1751052: M, 1回目発表 | ロボティクス | 小笠原 司 |
|
title: Eye movement measurement using a wearable device for communication robots
abstract: Recently, robots for communication with humans are being developed but their eye movement is not natural enough. One of the reasons is because we do not know how the human eyes move during a conversation so we need to measure the eye movement for a long time while humans talk. Therefore, I propose to measure the eye movement for long periods of time with a wearable device and then create an eye model for the eye motion. I will measure the gaze direction, the head angle and the eye saccades to create this model. I plan to use smart glasses as the wearable device to obtain such information. In this presentation, I will explain the details of my approach, the smart glasses design and the current progress. language of the presentation: Japanese | |||