Prospects for the use of hybrid spiking neural networks in biomechatronic systems

An important stage in the development of biomechanical systems is the interpretation of pulses obtained by EEG and EMG methods. The main difficulties are in the classification of signals and the determination of patterns of nerve impulses that actuate the limbs of the human body. If these problems are successfully solved, the classified patterns can be used to control biomechanical systems, for example, such as bionic prostheses. Modern methods of artificial neural networks open up the prospect for realizing an effective classification of the received impulses. However, the effectiveness depends on the chosen model of the artificial neuron and its computational complexity. This article will consider the most popular hybrid neural networks, and it will be shown that they have such advantages as resource intensity, energy efficiency and high level of accuracy of calculations and at the same time are biorealistically enough.