权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Analysis of Brain Information Components and Its Transmission to Humanoids

大脑信息成分分析及其向人形动物的传输

基本信息

批准号：
15300077
负责人：
MATSUYAMA Yasuo
金额：
$ 10.75万
依托单位：
Waseda University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (B)
财政年份：
2003
资助国家：
日本
起止时间：
2003 至 2004
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-15300077/
关键词：
humanoid motion control motion recognition unification of human motion and humanoid machine independence network environment independent component analysis 脳情報ヒューマノイドエージェント一般化された対数

项目摘要

This grant was applied to the unification of the human movement, the animation, and the humanoid over the computer network. The injection of the brain signal to the humanoid is another objective. The following results were obtained.(1)This research group was able to find the method to unify the human body motion, the cartoon character and the humanoid over the network environment. The designed system includes the recognition of human body motions. The system finds the body motion's abstract expression in a language level. The APNNA Best Paper Award for Application Oriented Research was given in 2004 to the research paper on this method,(2)Because of the abstraction to the language level, the human body motion can be transmitted and used as a command to different humanoids and other robots. In other words, the machine independence between humanoids was obtained.(3)It is not yet possible to obtain granular commands from brain signals because the resolution is still low by the contemporary technology. But, this study found that the abstract commands of (2)can be combined with the overwriting urgent signal from the brain. This method was found useful.(4)For the estimation method of active states of the brain, this study developed the f-ICA which includes the conventional ICA method as a special case. The new method is applicable to a wide class of information sources ; not limited to the brain signal. These targets include digital images and DNA segments, for which successful results were obtained.

这笔拨款用于在计算机网络上统一人类运动，动画和人形。将大脑信号注入人形机器人是另一个目标。得到了以下结果：(1)本课题组能够在网络环境中找到将人体运动、卡通人物和人形统一起来的方法。所设计的系统包括对人体运动的识别。该系统在语言层面上寻找身体动作的抽象表达。该方法的研究论文获得了2004年APNNA面向应用研究的最佳论文奖。(2)由于将人体运动抽象到语言层面，可以将其作为指令传递给不同的类人机器人和其他机器人。换句话说，获得了类人之间的机器独立性。(3)目前尚不可能从大脑信号中获得细粒度的指令，因为当代技术的分辨率仍然很低。但是，本研究发现(2)的抽象指令可以与来自大脑的覆盖紧急信号相结合。人们发现这种方法很有用。(4)对于大脑活动状态的估计方法，本研究开发了包含传统ICA方法作为特例的f-ICA方法。新方法适用于种类广泛的信息源；不仅限于大脑信号。这些目标包括数字图像和DNA片段，并获得了成功的结果。