Intelligent control of robot engineering systems in uncertainty conditions

INTELLIGENT CONTROL OF ROBOT ENGINEERING SYSTEMS IN UNCERTAINTY CONDITIONS.
Aloydinov Muhammadjon Gayrat oglu, Tashkent State Technical University, 1st year master [phone], [e-mail]
Annotation: This article should have formulated approaches to the construction of intelligent control systems for robotic complexes that do not change in terms of specific performance characteristics, taking into account the incompleteness of the input data and various uncertainties. To achieve this goal, it is necessary to solve a number of interrelated tasks: analysis of the architecture of intelligent control systems for robotic complexes, development of a generalized algorithm for situational identification of a robotic system, development of a generalized scheme of a control system for a robotic complex, development of intelligent control systems for a manipulation robot, a mobile robotic platform and a flexible automated line.
Keywords: robotic systems (RS), architectures of intelligent systems, control of robotic systems, automatic control, neural network, artificial neural network (ANN), identifier, universal algorithm.
Intelligent control - the use of artificial intelligence methods to control objects of various physical nature. In the field of robotic systems control, artificial intelligence methods are most widely used. This is primarily due to the autonomy of robots and the need for them to solve non-formalized creative tasks in conditions of incomplete information and
various types of uncertainty.
Until recently, this class of tasks remained the prerogative of natural intelligence: the operator of the control object, engineer, scientist, i.e., a person. Modern achievements in the field of automatic control theory, intelligent methods for formalizing semi-structured tasks and controlling complex technical systems make it possible to implement very complex robotic systems, which include mobile robotic platforms, flexible automated lines and android robots.
Robotic systems operate under conditions of incomplete input information, when the fundamental impossibility of measuring a number of parameters imposes significant restrictions on the control program. This leads to the need to develop a database of algorithms that allow, based on indirect signs and measured indicators, to calculate non-measurable parameters.
The uncertainty of the external environment in which the robotic system operates makes it necessary to include various types of compensators, modules for adapting, accumulating and ranking information in the control system.
In the course of the research, methods of the general theory of automatic control, the theory of fuzzy sets, neural networks, system analysis, and the theory of expert assessments were used.
Location of the robotic system in the external environment
For the implementation of intelligent control algorithms, the primary task is the current identification of the situation in which the robotic system is located. To solve this problem, a block diagram of the situational identification system has been developed (Fig. 1).

The block of technical vision and sensory perception is designed to determine changes in the state of the external environment and present a sensory map of the environment for further processing. The sensory map of the environment is an image of the situation in which the robot is at the current time. The time interval for constructing a sensory map is selected based on the specifics of the subject area.
Fig. 1. Block diagram of the situational identification system.
Working memory, by analogy with expert systems, is designed to process information coming from sensors and processed using the existing database of algorithms and the knowledge base (KB) of the robotic system.
The base of algorithms includes algorithms for sensor map preprocessing (digital signal processing, recognition of sound images and images), calculation of non-measurable parameters (functional dependencies on measured parameters), restoration of information completeness (knowledge check for completeness and inconsistency, knowledge adaptation taking into account non-stationary ™ and variable external conditions), mathematical operations, etc.
The knowledge base is a complex hierarchical structure containing a priori information about the external environment, laid down at the training stage, complete and consistent knowledge acquired by the robot in the process of functioning and perception of the external environment. Knowledge in the KB is ranked according to relevance and updated taking into account changes in the specifics of the robot's functioning based on knowledge adaptation algorithms.
The most important block is the situation identifier. It is this block that is responsible for the correct recognition of the image of the situation based on the sensory map. The result information of this block is decisive for the choice of the robotic system control program.
And finally, an intelligent interface, which is necessary for communication with the operator. The operator controls the functioning of the robotic system, as well as monitoring the process to achieve the set goals. As a rule, communication between a robot and an operator should occur using a natural language interface in a limited subset of natural language.
The implementation of algorithms and programs for intelligent control of robotic systems under conditions of uncertainty is associated with a number of significant difficulties.
The complexity of the input information preprocessing algorithms and the structural uncertainty of the behavior model of the robotic system itself determine the redundancy of the structure of the intelligent control system.
To solve the problem of controlling a robot under conditions of uncertainty, the following architecture of the intelligent control system was designed (Fig. 2).
The situational identification system (SIS) should be part of any intelligent control system for a robotic system. The intelligent control device (ICD) contains in its composition a CU and a control program selection unit (CPSU).

The purpose of this block is to generate a control action for the system of electric drives (ED) acting on the mechanical system (MS) of the robot.