| 研究課題/領域番号 |
15K12623
|
|---|
| 研究機関 | 国立研究開発法人産業技術総合研究所 |
|---|
研究代表者 |
GOWRISHANK AR.G 国立研究開発法人産業技術総合研究所, 知能システム研究部門, 国際客員研究員 (10570244)
|
|---|
| 研究分担者 |
宮脇 陽一 電気通信大学, 大学院情報理工学研究科, 教授 (80373372)
|
|---|
| 研究期間 (年度) |
2015-04-01 – 2018-03-31
|
| キーワード | natural learning / brain machine interface / embodiment / machine learning |
|---|
| 研究実績の概要 |
In this year, we tried to setup an extension of the embodiment experiments performed with a rubber hand with a whole robotic hand in Montpellier in France.
The idea was to see if the hand and finger movement can be used to improve the embodiment effect and to what extent.
We also have now developed two versions of the robot finger and feedback device that were planned in the project. We connected this finger to subjects such that they have six fingers in their hand. As the EEG FNA algorithm is progressing, we connected the finger via subject muscle activity (EMG)- We developed a way to isolate the unused (null space) in muscle activity, that we connected to the finger such that the subjects can move all the six fingers (five real +1 robot) independently. We are now perfroming two experiments to show that haptic feedback and agency are critical for embodiment of an extra limb into the body schema.
The FNA algorithm idea was developed and we have started the first fMRI experiments to setup the main experiment. The big issue is in regard to verifying the isolated FNA areas, and we have decided to use neurofeedback learning for verification. This however involves long training time and MRI, due to which the project has been delayed.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
3: やや遅れている
理由
The project is progressing smoothly and we have accomplished the objectives of finger, feedback device development, as well the planned embodiment experiments. The FNA algorithm experiments were delayed due to the constraints with MRI availability and long training period required for neurofeedback learning. We have thus extended the project to this year. As the EEG FNA algorithm were delayed, we created the first experiments by connecting with the subject muscle activity or EMG (instead of EEG) instead. We are using this now to check for embodiment and natural learning issues while we work on the EEG FNA method.
|
| 今後の研究の推進方策 |
This year we plan to finish the experiments the robot finger and feedback device to clarify a)the requirements of embodiment of an extra new limb, b) try the FNA idea to connect the finger with EMG and induce ‘natural learning’.
We plan to finish the fMRI experiment which will give us a possible answer to difficult question of decoding and verifying the FNA in the brain.If this works, we will have the first suh algorithm. We will then strive to connect the EEG to the finger and feedback device.
|
| 次年度使用額が生じた理由 |
The travel cost was used by the PI to go to a collaborating laboratory in Montpellier, France at the start of the fiscal year. This lab has a top robot five fingered hand. We developed experiments with the robot hand to investigate embodiment issues and how they change when the embodied limbs (robot and) is actuated. These experiment served as a preliminary for the robot finger experiment design we developed in the later part of the year.
The article cost was used to buy tools for the robot finger development, and a camera and headphones that are being used (and will be used) in the experiments this year.
|
| 次年度使用額の使用計画 |
We plan to use funding this year for mainly three issues-a) On the UEC side, for the payment for the MRI slots and subject fees.b) on the AIST side, for travel for one conference for the PI and travel between Tsukuba and AIST for experiments.c)A large part will go for the publication cost-we expect atlast 3 open access journal papers from the works to follow this year.
|
