Real-Time Decentralized Neural Control via Backstepping for a Robotic Arm Powered by Industrial Servomotors

Area of Science:

• Robotics
• Control Systems
• Artificial Intelligence

Background:

• Trajectory tracking is crucial for robotic arm applications.
• Accurate dynamic modeling is challenging for complex robotic systems.
• Decentralized control offers robustness and scalability.

Purpose of the Study:

• To develop a continuous-time decentralized neural control scheme for a 2-DOF direct drive vertical robotic arm.
• To online identify the robotic arm's dynamics using a novel neural network structure.
• To validate the controller's effectiveness on a custom robotic arm platform.

Main Methods:

• A decentralized recurrent high-order neural network (RHONN) in a series-parallel configuration.
• Filtered error learning law for online dynamic identification.
• Backstepping approach for deriving the local neural controller.

Main Results:

• Successful online identification of the robotic arm's unknown dynamics.
• Effective trajectory tracking achieved by the decentralized neural controller.
• Validation on a custom-built robotic arm platform with industrial servomotors.

Conclusions:

• The proposed decentralized neural control scheme is effective for trajectory tracking of a 2-DOF robotic arm.
• RHONNs provide a robust method for online dynamic identification in robotic systems.
• The control strategy is validated on a real-world robotic platform with unknown parameters.