Soft Shear Sensing of Robotic Twisting Tasks Using Reduced-Order Conductivity Modeling

Area of Science:

• Robotics
• Sensor Technology
• Materials Science

Background:

• Dexterous robotic manipulation relies on rich tactile feedback from artificial fingertips.
• Shear sensing is crucial for tasks like twisting and dragging, but research in soft sensors using electrical impedance tomography (EIT) is limited.
• EIT technology offers a promising avenue for developing advanced tactile sensors.

Purpose of the Study:

• To investigate soft shear predictions using EIT for robotic manipulation.
• To develop and analyze reduced-order models for relating screwdriver twisting tasks to conductivity maps in EIT sensors.
• To enable high-speed, closed-loop robotic control through improved tactile sensing.

Main Methods:

• Proposed and investigated five reduced-order models for EIT-based shear sensing.
• Analyzed EIT signals generated during screwdriver twisting tasks.
• Correlated reduced-order model parameters with physical measurements like torque and diameter.

Main Results:

• Achieved high correlations (0.96 for torque, 0.97 for diameter) between reduced-order parameters and physical measurements.
• Demonstrated that insights can be deduced from noisy EIT signals using the proposed models.
• Showcased the potential for precalculating Finite Element Method (FEM) model signals, unlike traditional methods.

Conclusions:

• The developed reduced-order models effectively predict shear-based twisting in robotic fingertips using EIT.
• This approach offers a pathway towards real-time, high-speed closed-loop robotic manipulation systems.
• The findings advance the field of soft robotics and tactile sensing for complex manipulation tasks.