FEATURES OF USING DIGITAL SOLUTIONS FOR BIONIC PROSTHESIS CONTROL TASKS

Abstract

The use of digital solutions for bionic prosthesis control tasks, through the use of deep machine learning capabilities, allows us to solve the problems of identifying and formalizing the movement of individual fingers and adequately carry out the process of identifying human hand movements by reading data from a video camera and processing the received data using OpenCV software function libraries in real time. This approach allows us to reproduce the movements of the fingers of a real human hand with the bionic prosthesis while simultaneously using the MediaPipe and OpenCV platform libraries. Identification of the positions of the fingers of the human hand is carried out according to the developed program, the algorithm of which consists of the stages of recognizing the left and right fingers and their placement in two-dimensional space. The output function of this program starts the operation of the servo drives of the bionic prosthesis. Which, in turn, requires the implementation of such an approach in real hardware, which was presented in the specified work. The bionic prosthesis is controlled by transmitting output data via the UART port to the ATMega328 microcontroller, which in turn controls the servo drives of the bionic prosthesis via the SPI interface. A real printed circuit board has been designed, which allowed, using the developed programs, to simulate the movements of a human hand to work out scenarios of its movements for setting up and validating the operation of the servo drives of bionic prostheses. A structural diagram of the proposed control system for a bionic upper limb prosthesis has been developed. The specified system has been designed and a PCB board has been developed, which allowed experimental studies to work out the adequacy of its operation and validate the movements of the fingers of the bionic prosthesis with a real living object in real time.