Manipulandum system for arm reaching movements. By measuring motions under various external force conditions, control mechanisms for voluntary movements can be clarified.
2-joint-6-muscle arm robot. By using the redundant actuators and the sensory information about the joints and the actuators, the controller can create movements appropriate to unpredictable environments.
Research activities:
Our main aim is to understand highly harmonic and autonomous biological-information systems in order to propose new designing principles for building innovative systems. Today, state-of-the-art robots display a high level of performance in a predictable and regulated environment. However, they display very poor performance in real-world: an unpredictable and unregulated environment. Why? Because the controller of today’s robots can only process learned information, it requires all the necessary information in advance, and without that it will fail to control its own body. Therefore, when the robots encounter new information in an unknown environment they don’t have the ability to process it and consequently cannot adapt to a changing environment. In contrast, biological systems such as human beings can create information necessary to interpret external stimuli and to control actuators in real-time, by appropriately recognizing and judging unpredictable changes in the real-world.
To understand how living organisms create information, we study biological mechanisms of object recognition, speech recognition, learning and memory, and voluntary-movement control, with various methods such as psychophysics, neurophysiology, computational simulation, developing robot control system.
Research topics:
- Mechanisms of learning and memory in olfaction
- Speaker independent speech recognition by evaluating global spectrum shape
- Three-dimensional visual recognition by integrating motion vision and form vision
- Real-time control mechanisms for voluntary movements
![[Homepage of This Laboratory]](../../image1/a0316.gif){kind=small}
