豊橋技術科学大学

Search

Search

Fukumura, Naohiro

Affiliation Department of Computer Science and Engineering
Title Associate Professor
Fields of Research Computational Neuroscience
Degree Doctor of Engineering (Toyohashi University of Technology)
Academic Societies The Institute of Electronics, Information and Communication Engineers / Japan Neural Network Society / Robotics Society of Japan
E-mail fukumura@cs
Please append ".tut.ac.jp" to the end of the address above.
Laboratory website URL http://www.bmcs.cs.tut.ac.jp
Researcher information URL(researchmap) Researcher information

Research

ヒトなどの生体は視覚などの感覚器官から得た多くの感覚情報を統合して外界の情報を知覚し、それを基に柔軟な運動を行なうことで、外界の状況に対処しています。このような生体の脳の高次機能である感覚-運動情報統合の機能を計算論的神経科学の立場から理解する事を目指し、対象物認知の過程やその認知に基づく手や腕の運動を計測する心理物理実験を行ないます。さらに生理学的な知見も取り入れてそれらの情報処理機能を再現する数理モデルを構築することで、ヒトのもつ柔軟な情報処理の仕組みを探ります。さらにその数理モデルを応用した、柔軟で人に優しいシステムの実現を目指しています。

Theme1:Computational Research about Human Voluntary Arm Movements

Overview
Fig.1 Measurement system for human arm movements

Humans can executes various dexterous movements. This process includes extracting important information for the movement from environment, selecting an adequate movement of limbs and body, and controlling many muscles for the movement. We focus on upper limb movements like a reaching movement or a grasping movement. We measure and analysis human arm and hand movements under various conditions and construct computational models of such human movements to investigate the information process of the upper limb movements.

[publications]
Fukuda,H., Kakutani, N., Fukumura, N., and Uno,Y., “A Neural Network Model for Planning Force and Posture in Three-digit Object Holding,” Proceedings of International Joint Conference on Neural Networks, pp.1308-1313 2007
Fukumura, N., Otane, S., Uno, Y., and Suzuki, R., “A Neural Network Model for Extracting Correlated Information in Sensory Integration,” Proceedings of International Conference on Neural Information Processing, Vol.2, pp.873-876 1998

Keywords

reaching movement, grasping movement, motor planning, motor control, sensory-motor integration

Theme2:Research about Human Motor Learning

Overview
Fig.2 A Learning model of many degrees of freedom system using module structure to control an inverted pendulum by a human arm.

One of the essential reasons why humans are able to execute dexterous movements is a flexible motor learning ability. We can learn new movements under the various conditions. One of the motor learning methods is a reinforcement learning method that humans learn by oneself through a trial and error. Another method is called an imitation learning that humans imitate a movement of other expert of the task. We investigate the principle of the motor learning by measuring and analyzing the change of the movement patterns on the process of such learning schemes. Moreover, we verify such motor learning schemes by implementing to some learning motor tasks by a computer simulation.

[publication]
Ohama, Y., Fukumura, N., and Uno, Y., “A Simplified Forward-Propagation Learning Rule Applied to Adaptive Closed-Loop Control,” Proceedings of International Conference on Artificial Neural Networks),pp.437-443 2005
Fukumura, N., Wakaki, K., and Uno,Y., “A Modular Structure of Auto-encoder for the Integration of Different Kinds of Information”, Proceedings of 1st International Conference on Natural Computation, pp.313-321 2005

Keywords

motor learning, reinforcement learning, imitation learning, neural network

Title of class

Engineering and Science Laboratory / Control Engineering / Bio-physical Information Systems


to Pagetop