Impedance Control Method for Tea-Picking Robotic Dexterous Hand Based on WOA-KAN.

Area of Science:

• Robotics
• Control Systems
• Artificial Intelligence

Background:

• Robotic tea-picking requires precise mechanical control for harvesting delicate tea buds.
• Existing impedance control methods often struggle with dynamic changes and parameter optimization.

Purpose of the Study:

• To develop an adaptive impedance control method for robotic dexterous hands in tea-picking.
• To enhance the accuracy and real-time adjustability of impedance parameters during tea harvesting.

Main Methods:

• Implemented an adaptive impedance control framework using a Kolmogorov-Arnold Network (KAN) with cubic B-spline activation functions.
• Optimized KAN's B-splines using the Whale Optimization Algorithm (WOA) for improved nonlinear fitting and global optimization.
• Integrated tactile sensors for real-time force tracking during tea bud interaction.

Main Results:

• The proposed WOA-KAN method significantly reduced overshoot by 14.2% and steady-state error by 99.89% compared to fixed-parameter control under dynamic conditions.
• Experimental validation showed a maximum overshoot of approximately 6% at a 50Hz control frequency during tea-picking.
• Demonstrated dynamic mapping and real-time impedance parameter adjustment for accurate tea bud contact force-tracking.

Conclusions:

• The adaptive impedance control method effectively improves the precision and robustness of robotic tea-picking.
• The integration of WOA and KAN offers a powerful approach for optimizing control parameters in complex robotic tasks.
• The developed algorithm shows significant potential for practical application in automated agricultural harvesting.