Indirect Force Control of a Cable-Driven Parallel Robot: Tension Estimation using Artificial Neural Network trained...

Area of Science:

• Robotics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Cable-driven parallel robots (CDPRs) rely on accurate cable tension measurements for end-effector force control.
• Inaccurate tension readings due to pulley friction and unmodeled cable properties hinder precise force control in CDPRs.

Purpose of the Study:

• To develop an artificial neural network (ANN)-based indirect force estimation method for CDPRs.
• To compensate for unmodeled effects, such as pulley friction, to improve cable tension accuracy.
• To enhance the force control performance of CDPRs using the estimated cable tensions.

Main Methods:

• An artificial neural network (ANN) was developed to model and compensate for black-box uncertainties like pulley friction.
• The ANN was trained using experimental datasets obtained from CDPR operations.
• A proportional (P) controller was designed using the ANN-estimated cable tensions for force tracking.
• The effectiveness of the ANN model was validated by comparing estimated forces with directly measured end-effector forces.

Main Results:

• The proposed ANN-based method accurately estimated cable tensions at the end-effector by compensating for unmodeled effects.
• Force control implementation using compensated tensions demonstrated improved CDPR performance in wrench space.
• Experimental validation confirmed the ANN model's ability to improve force control accuracy.

Conclusions:

• The ANN-based indirect force estimation method effectively compensates for pulley friction and unmodeled effects in CDPRs.
• This friction-compensation technique significantly enhances the accuracy of cable force control in CDPR applications.
• The proposed method offers a viable solution for overcoming force-control challenges in CDPRs.