Researcing the Fault Tolerance of Robotic System Designed via Use of Neural Network Decision Making Component of Image Processing

This paper proposes and investigates an approach to optimizing the fault tolerance of robotic system designed via use of the neural network component of information processing. The approach suggests creating a special architecture of the neural network decision-making component as part of robotic system. Inside this architecture there are some automated processes that monitor and correct any negative variations in the parameters of computing elements, caused by their partial or full failures due to external and internal destabilizing impacts. The object of this experimental research into the method was the computer model of a robotic system where a neural network decision-making component enabled function to be performed: classification of the object on the image based on the received input information from the primary sensor system. The research has proved the approach to be efficient to ensure the maximum fault-tolerance of neural network component of information processing in robotic system of various applications including the task of image processing.