Resumen
The review describes the main principles as well as advantages and disadvantages of the modern brain-computer interfaces applied in robotic devices. The invasive and non-invasive devices based on the origin of a signal, invasiveness and location of probes are discussed in the paper. The description of some electrical (EMG, EEG, etc.) and chemical (fMRI, fNIRS, etc.) methods to detect neural activation are concerned. Spatial resolution of electrical neural interfaces is rather low, therefore one of their main disadvantage is the difficulty in detecting the exact region of activation. The main disadvantage of chemical neural interfaces is long reaction time. Unfortunately, none of the non-invasive methods today allows inventing an effective neural interface for interactive control of robotic devices. Modern invasive methods are rather harmful; therefore, they are unacceptable in studies with humans for ethical reasons. In this respect, the most promising is the use of the combined brain-computer non-invasive interfaces, combining sensors of both electrical and chemical activity of the nervous system. In combined neural interfaces the disadvantages of one method are compensated by the advantages of another one. The main area of practical use of neural interfaces in robotics in the foreseeable future will be devices for the rehabilitation of persons with disabilities. The use of neurointerfaces for other robotic devices will have only scientific significance until the advent of new safe invasive neurointerfaces, since non-invasive neurointerfaces do not have significant advantages over traditional control systems for healthy people.