| 榎本 貴仁 | M, 2回目発表 | 知能システム制御 | 杉本 謙二, | 安本 慶一, | 小林 泰介 | ||
|
title:Shared Autonomy for Overlooked Visual Field
abstract: Imitation learning is an effective method towards autonomous driving because it only requires driving data instead of complex dynamical model. However, the driving data would contain noise and/or outliers such as driving mistake, which disturb learning driver skill. To solve this problem, we propose a robust method for noise and/or outliers by using Tsallis statistics. This proposed method decreases the influence of outliers at computation of loss function to be minimized. We conduct a verification experiment and compare the proposed method to the conventional method. As a result, the conventional method fails to acquire driver skill due to outliers, while the proposed method can imitate driver skill without being affected by outliers. language of the presentation:Japanese | |||||||
| 小坂 麻人 | M, 2回目発表 | 知能システム制御 | 杉本 謙二, | 安本 慶一, | 小林 泰介 | ||
|
title: Bayesian Optimization for Force-based Operation in Mobile Robot
abstract: This paper proposes a framework for optimizing robot operability in physical human-robot interaction (pHRI) only based on the operator's subjectivity. The robot controller in pHRI is defined by the dynamics with virtual impedance parameters, although their design has not been established. In particular, the operator's subjectivity is basically ignored. Therefore, in this paper, the impedance parameters are designed by Bayesian optimization based on the operator's subjectivity. We analyze the behavior of the designed parameters. The results showed that the operator optimized the robot's behavior to have steady speed with emphasis on predictability. language of the presentation: Japanese | |||||||
| 西澤 佳祐 | M, 2回目発表 | サイバネティクス・リアリティ工学 | 清川 清, | 安本 慶一, | 荒牧 英治, | 酒田 信親 | 磯山 直也 |
|
title: Copy and Paste Techniques for Web Browsing in Immersive VR
abstract: In the future, it is assumed that when people become familiar with life with HMD, they will be able to browse the web in VR on a daily basis. Therefore, we aim to realize a new web browsing experience using the features of VR environment, such as depth, spatial interaction and wide field of view. As a first step, we focus on the copy and paste operation. The current method of operating a web browser in VR is a simple ray-casting technique, but it is difficult to select with high precision. So it is difficult to perform tasks such as a copy-and-paste. Therefore, we propose a method of copy-and-paste that is suitable for web browsers in VR environment. Specifically, we duplicate a web browser in the VR and operate it as if you were touching the browser window with your hand, so that you can operate it with high precision. We describes the basic design of the proposed system and the implementation of the prototype system in detail, and reports on preliminary experiments including the feeling of use. language of the presentation: Japanese 発表題目: 没入型VRのWebブラウジングに適したコピー&ペースト手法の提案 発表概要: 将来HMDを装着する生活が身近になったとき，VR空間内でWebブラウジングを日常的に行うことが想定される．そこで，我々はVR空間の特徴である奥行き，空間的インタラクション，広視野性を利用し，新たなWebブラウジング体験の実現を目指している．その第一歩としてコピー＆ペースト操作に着目する．現状のVR空間内でのWebブラウザの操作には単純なレイキャスティング手法が用いられているが，高精度の選択がしにくいという問題があり，コピー&ペーストなどの作業が難しい．そこで，VR空間内のWebブラウザに適したコピー&ペーストの手法を提案する．具体的には，VR空間内で手元にWebブラウザを複製し，ブラウザウィンドウに手で触れるように操作を行うことで，高精度な操作をできるようにする．本稿では提案システムの基本設計と試作システムの実装について詳述し，その使用感などの予備実験について報告する． | |||||||
| 長谷川 瑛一 | M, 2回目発表 | サイバネティクス・リアリティ工学 | 清川 清, | 安本 慶一, | 酒田 信親, | 磯山 直也 | |
|
title: The effect of the speed of visual information on the viewer's work speed on smart glasses
abstract: I conducted experiment and confirmed that the faster the speed of the video viewed, the faster the work speed tended to be. Smart glasses enable users to view information before and after work unlike the VDT work environment in which a monitor is viewed only while working. Although it has been shown that the viewer's subjective time and work speed are affected by certain visual information, it is thought that by applying these to the information presentation of smart glasses, it is possible to use the system to improve the efficiency of later work by casually watching videos and the like. In this study, we aim to clarify how the speed of visual information viewed on the smart glasses affects the work to be done later on, and to implement an information presentation system that enables users to start working efficiently. I have conducted experiment and confirmed that the faster the speed of the video viewed, the faster the work speed tended to be. language of the presentation: Japanese | |||||||