Use of soft computing techniques for the control of industrial systems: application on an industrial robot

Abstract

This thesis presents advancements in industrial robotics using soft computing techniques, focus- ing on kinematic and dynamic modeling, sensorless collision detection and optimal trajectory planning. Chapter 3 provides a comprehensive study of robot manipulators, covering funda- mental principles of kinematic and dynamic modeling. A detailed case study on the Fanuc M-710iC/70 industrial robot examines its kinematic structure, dynamic model, and control as- pects, offering practical insights into robotic motion and system behavior. Chapter 4 addresses the limitations of traditional model-based collision detection methods and introduces a novel sensorless approach using a fuzzy momentum observer. By dynamically adjusting observer parameters through fuzzy logic, this method enhances detection accuracy, improving both sen- sitivity and robustness. Extensive simulations validate its effectiveness in detecting collisions with high precision. Chapter 5 focuses on optimal trajectory planning for industrial robots to minimize energy consumption while ensuring smooth motion. A deep learning-based energy model, utilizing a Long Short-Term Memory (LSTM) network, accurately predicts energy con- sumption. Additionally, a Genetic Algorithm (GA) optimizes robot trajectories by considering execution time, jerk, and energy efficiency. The integration of deep learning and evolutionary optimization enables the generation of energy-efficient trajectories, enhancing industrial robot performance. Overall, this research contributes to improving industrial robots by enhancing modeling accuracy, safety, and energy efficiency, making them more adaptive and intelligent in real-world applications.