Neuro-Fuzzy Models and Algorithms for Improving the Dynamic Characteristics of MIMO Control Systems for a Manipulator Robot

Abstract. The theory of Multi-Input Multi-Output (MIMO) control systems is an effective approach when dealing with manipulator robots (manipulators) and other objects with a state vector characterized by the mutual influence of components. In the case of a manipulator, the torques that determine static and dynamic loads on the shafts of the actuator subsystem depend nonlinearly on the values of generalized variables. At the same time, the dynamic model describing this dependence often includes several uncertainties related to a specific task being executed, operational environment conditions, robot design, and changes in system parameters during operation. In this regard, it is topical to apply knowledge processing technologies that expand the operational capabilities of control systems under uncertainty. In particular, approximation problems of nonlinear functions of several variables (which include intelligent MIMO control of manipulators) are effectively solved by systems based on neuro-fuzzy approaches; among the latter, the Adaptive-Network-Based Fuzzy Inference System (ANFIS) technology has become the most widespread. This paper considers a modified ANFIS structure for approximating the solution of the manipulator’s inverse dynamics problem as well as its software and algorithmic implementation. Experimental studies are conducted to compare the proposed modification with an approximator based on a set of conventional ANFIS structures and the corresponding control systems. As shown, the proposed approach significantly speeds up the fuzzy inference procedure and, consequently, improves the tracking accuracy of a reference trajectory by a MIMO control system.

Keywords: manipulator robots, intelligent control, neuro-fuzzy control, ANFIS.

PDF (English)

Cite this paper

Smirnov, A.V. and Bykovtsev, Yu.A., Neuro-Fuzzy Models and Algorithms for Improving the Dynamic Characteristics of MIMO Control Systems for a Manipulator Robot. Control Sciences 4, 65–75 (2026).