Artificial hand control by myoelectric signals — reduced DFT approach

Abstract

Smart hand prosthesis control, based on myoelectric signals, strongly depends on signal processing algorithms. This kind of application field forces specific requirements on computational complexity (for dexterity of prosthesis), processing speed (for fast reaction) and size (for portability). The paper presents a concept of reduction of DFT information extracted from EMG signals. Signals are acquisited from 8 channels with 1 kHz sampling frequency. Specialized digital hardware is proposed, capable of parallel processing of series of signals. The design was implemented in VHDL, verified and synthesized for FPGA. In-house developed floating point arithmetic was applied. Satisfying processing speed was obtained for implementation technique that enable embedding in prosthesis.