Hybrid ANN–Fuzzy Logic Controller for Autonomous Mobile Robots in Industrial Environments

Abstract

Autonomous Mobile Robots (AMRs) are becoming essential components of modern industrial automation systems, performing tasks such as material handling, inspection, and intralogistics within dynamic and uncertain environments. Despite significant advancements in sensing and computation, achieving robust, adaptive, and real-time motion control in industrial settings remains a major challenge due to nonlinear robot dynamics, narrow aisle navigation, dynamic obstacles, payload variations, and sensor noise. Conventional controllers such as PID offer simplicity and low computational cost but lack adaptability to environmental uncertainties, while standalone Artificial Neural Network (ANN) controllers require extensive training data and may suffer from stability concerns. Similarly, Fuzzy Logic Controllers (FLCs) effectively manage uncertainty through rule-based reasoning but depend heavily on expert-designed membership functions and lack autonomous learning capability. To address these limitations, this paper proposes a Hybrid Artificial Neural Network–Fuzzy Logic Controller (ANN–FLC) for autonomous mobile robots operating in industrial environments. The proposed framework integrates the nonlinear approximation and adaptive learning capability of ANN with the interpretability and robustness of FLC. The ANN dynamically tunes fuzzy membership functions and rule weights in real time based on tracking errors and sensory feedback, enabling continuous performance optimization. The controller is evaluated through comprehensive simulations and experimental validation in a structured warehouse environment with dynamic obstacles and noise disturbances. Comparative results demonstrate significant improvements in trajectory tracking accuracy, convergence speed, collision reduction, and robustness compared to PID, standalone ANN, and standalone FLC approaches. The findings confirm that the proposed hybrid controller provides a reliable, computationally efficient, and safety-oriented solution for next-generation industrial autonomous mobile robotic systems.