Abstract
The disabled persons such as arm amputees or spinal cord injured patients usually need to use auxiliary apparatus like electric prosthesis or functional electrical stimulation system. It is an important topic to control these devices. The goal of this research is to develop an application specific integrated circuit (ASIC) prototype with real-time pattern recognition for the EMG signal. The system of real-time pattern EMG recognition can be used as the control device to the assisting equipments of the patients.This research presents dedicated ASIC design for an EMG pattern recognition system. The recognition speed by requisition is much slower than the maximum operation frequency of our chip, therefore time-multiplexing and folded architecture are adopted. To avoid rapid growth of logic units due to circuit duplication and implementation of real-time pattern recognition system, our design reduced the hardware functional units in the several operations such as multiplication and addition in algorithm. The cepstral parameters are modified and analyzed from our databases which are determined by our trained-model in the ASIC. These parameters can then be used as the basis for recognition. In addition, the discrimination of different motions made by the user was found by the dynamic timing warping (DTW) method using the cepstral parameters. The specification show that the operation clock rate is 2MHz, the core size is 0.84 X 0.84 mm2 , and the chip size is 1.789 X 1.789 mm2. Now the project goals are achieved and applying the SOC technology to improve the people healthcare and the use of the disability tools in the future.
| Date of Award | 2006 |
|---|---|
| Original language | Chinese (Traditional) |
| Supervisor | Chih-Lung Lin (Supervisor) |