Adaptive robust control of tea-picking-manipulator's position tracking based on dead zone compensation with modified...

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Traditional neural networks struggle with input saturation and accuracy degradation in manipulator control.
• Modeling dead zone nonlinearity is crucial for precise control of robotic systems like tea pickers.

Purpose of the Study:

• To develop an accurate control model for tea picking robots by addressing nonlinearity and dead zone issues.
• To improve tracking precision and accuracy in manipulator control systems.

Main Methods:

• Designed an adaptive compensator using modified radial basis function neural networks (m-RBF) and an adaptive law.
• Implemented the control scheme in Simulink for simulation and verified with a six-axis manipulator tea picking experiment.

Main Results:

• The m-RBF effectively approximated dead zone nonlinearity, showing excellent and stable tracking accuracy in simulations.
• The proposed control scheme achieved a 95.3 score in tea picking experiments, outperforming traditional PID control by nearly two times.
• Demonstrated faster learning rates and avoidance of local minima with m-RBF.

Conclusions:

• The m-RBF-based control scheme offers superior control accuracy, robustness, and self-adaptation.
• This method is highly suitable for real-time control applications, particularly in robotic systems like tea picking robots.