• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to project page

2023 Fiscal Year Final Research Report

Efficient Deep Learning and its implementation for Robot-Based Rehabilitation Using Brain Signals

Research Project

  • PDF
Project/Area Number 21K03970
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 20020:Robotics and intelligent system-related
Research InstitutionHosei University

Principal Investigator

Capi Genci  法政大学, 理工学部, 教授 (20389399)

Project Period (FY) 2021-04-01 – 2024-03-31
Keywords深層学習 / BMI / ロ ボ ッ ト 動作 / 転移学習
Outline of Final Research Achievements

In this research work, we developed efficient deep-learning algorithms for Brain Machine Interface (BMI) systems. The two main research topics were: 1) Optimization of electrode channels to improve the recognition rate of CNNs. We evaluated the performance in motor execution (ME) and motor imagery (MI) tasks. For channel optimization, we integrated DL and Deep Reinforcement Learning (DQL) algorithms. The primary objective of this system is to minimize the computational complexity and training time of the DL network without deterioration in the system performance. 2) Training data optimization for high-accuracy EEG classification using Genetic Algorithm (GA). EEG data from motor imagery and real hand/arm motion tasks were considered. The developed optimized systems are implemented in efficient robotic applications. The robots are controlled using EEG/EMG bio-signals. Such applications can be implemented in rehabilitation, human-robot interactions, and assistive robotic systems.

Free Research Field

知能ロボット

Academic Significance and Societal Importance of the Research Achievements

研究結果の科学的および社会的意義は以下の通りである:1. 訓練データの質を向上させることで、BMIシステムに向けたDLベースのネットワークを改善した。2. 訓練データを最適化し、CNNの認識精度の顕著な向上と訓練時間の短縮を実現した。3. Deep Q-learningベースの手法を用いて、BMIシステムにおけるEEGチャネルの最適化を実現した。
これらの結果は、脳信号を使用したCNNの高速訓練にも利用することが可能である。
また、このような柔軟性があり訓練が容易なCNNは、リハビリテーションなどのヒューマン・ロボット・インタラクションに応用することができる。

URL: 

Published: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi