Virtual Sensor for Kinematic Estimation of Flexible Links in Parallel Robots

Area of Science:

• Robotics
• Control Systems Engineering
• Mechanical Engineering

Background:

• Controlling flexible link parallel manipulators presents challenges, particularly in endpoint trajectory tracking.
• Robot arm flexibility and deformation complicate accurate estimation of the Tool Centre Point (TCP) position.
• Existing methods for deformation estimation often rely on costly sensors or computationally intensive integration techniques.

Purpose of the Study:

• To develop a novel virtual sensor for precise estimation of flexible link deformations and their derivatives.
• To address the limitations of existing methods in terms of cost and computational complexity.
• To validate the virtual sensor's performance in a real-world robotic application.

Main Methods:

• Implementation of a virtual sensor approach to estimate link deformation.
• Calculation of deformation derivatives (velocity and acceleration).
• Validation using simulation results and comparison with a flexible multibody dynamics software (ADAMS).

Main Results:

• The virtual sensor achieved less than 2% error in estimating flexible link deformation.
• Derivatives were estimated with less than 6% error for velocity and 13% error for acceleration.
• TCP position estimation in a Delta Robot showed less than 0.03% error compared to the flexible approach in ADAMS.

Conclusions:

• The proposed virtual sensor offers a precise and computationally efficient solution for estimating flexible link deformations.
• This method significantly enhances the accuracy of TCP position estimation in flexible robots.
• The virtual sensor demonstrates high validity and potential for advanced control applications in robotics.