Related Experiment Video
Updated: May 2, 2026
06:58
A Structured Rehabilitation Protocol for Improved Multifunctional Prosthetic Control: A Case Study
Published on: November 6, 2015
9.0K
Adaptive neural network motion control of manipulators with experimental evaluations
S Puga-Guzmán1, J Moreno-Valenzuela2, V Santibáñez3
1Instituto Tecnológico de Tijuana, Boulevard. Industrial S/N, 22510 Tijuana, BC, Mexico ; Instituto Politécnico Nacional-CITEDI, Avenida del Parque 1310, Mesa de Otay, 22510 Tijuana, BC, Mexico.
Thescientificworldjournal
|February 28, 2014
Summary
This study introduces an adaptive neural network controller to improve robot arm accuracy. Experimental results demonstrate enhanced tracking performance in complex mechanical systems.
Area of Science:
- Robotics
- Control Systems
- Artificial Intelligence
Background:
- Traditional controllers struggle with complex dynamics and external forces in robotic systems.
- Accurate torque estimation is crucial for precise motion control in robotics.
- Adaptive control strategies offer potential for improved performance in dynamic environments.
Purpose of the Study:
- To propose a novel nonlinear controller combined with adaptive neural network compensation for robotic systems.
- To develop and validate a two-layer neural network for accurate torque estimation.
- To demonstrate the effectiveness of the proposed adaptive control scheme in real-time experiments.
Main Methods:
- A nonlinear proportional-derivative controller was augmented with a two-layer adaptive neural network.
- Adaptation laws for neural network weights were derived to ensure stability and convergence.
- The controller was implemented and tested on a two degrees-of-freedom robot and a one degree-of-freedom arm.
- Performance was evaluated with and without neural network compensation.
Main Results:
- Asymptotic convergence of position and velocity tracking errors was theoretically proven.
- Neural network weights were demonstrated to be uniformly bounded, ensuring system stability.
- Experimental validation confirmed significant improvements in tracking accuracy with the adaptive neural network controller.
- The controller showed effectiveness in systems with varying dynamics, including gravitational forces.
Conclusions:
- The proposed adaptive neural network controller effectively enhances tracking accuracy in robotic systems.
- The controller is robust and adaptable to different mechanical configurations and external disturbances.
- This approach offers a promising solution for high-precision robotic control applications.
