Modeling and Control of Robotic Manipulators Based on Artificial Neural Networks: A Review

Recently, robotic manipulators have been playing an increasingly critical part in scientific research and industrial applications. However, modeling of robotic manipulators is extremely difficult due to their complicated structures, nonlinear characteristics, and so forth. Based on the unique black-box characteristics and self-learning capability, artificial neural networks (ANNs) are considered effective tools for modeling and controlling robotic manipulators with uncertain dynamics due to the advantages of both convenient hardware implementation and high-speed parallel distributed calculation. This review attempts to summarize the current research on modeling and control of robotic manipulators based on ANNs. Firstly, the various types of robotic manipulators and the development of ANNs are discussed briefly. Then, the ANN-based modeling methods of robotic manipulators are described. Both traditional and intelligent control methods based on ANNs for robotic manipulators are discussed subsequently. Besides, some potential directions, possibly deserving investigation in a variety of different types of modeling as well as control methods by ANNs, are described and discussed as well. The proposed summary is aimed at aiding researchers to effectively comprehend the characteristics of various ANNs and their applications in the modeling and control of robotic manipulators while providing a reference for future directions related to robotic manipulator research.