Identification of Intrinsic Friction and Torque Ripple for a Robotic Joint with Integrated Torque Sensors with...

Area of Science:

• Robotics
• Control Systems
• Machine Learning

Background:

• Integrated joint torque sensors in robots measure both interaction forces and intrinsic robot dynamics.
• Distinguishing these forces is crucial for delicate manipulation tasks where interaction forces are small.

Purpose of the Study:

• To develop and validate a method for identifying and predicting robot intrinsic dynamics.
• To improve the accuracy of measuring external interaction forces using integrated sensors.

Main Methods:

• Linear regression with non-linear radial basis functions was used to model intrinsic dynamics.
• The model was validated on a wheel-bearing turning task to measure friction.

Main Results:

• The proposed model significantly reduced the error in measuring bearing torque compared to methods without intrinsic dynamics compensation.
• The RMS error for bearing torque measurement was 11% when using the joint 7 torque sensor.

Conclusions:

• The developed approach effectively compensates for robot intrinsic dynamics.
• This enables more accurate force and torque measurements for robots, enhancing delicate manipulation capabilities.