A Framework for Inclusion of Unmodelled Contact Tasks Dynamics in Industrial Robotics

Area of Science:

• Robotics
• Control Systems
• Machine Learning

Background:

• Unmodeled dynamics in robot end-effectors or loads pose challenges for collision detection and operational context understanding.
• Existing methods often require detailed dynamic parameter identification, limiting their applicability.

Purpose of the Study:

• To develop a method for integrating unmodeled dynamics into robot operation analysis.
• To enhance collision detection and context understanding without dynamic parameter identification.

Main Methods:

• Utilized a modified Dynamic Time Warping algorithm.
• Applied a universal algorithm for kinematic parameter identification.
• Categorized contact tasks based on external contact force profiles.

Main Results:

• Successfully incorporated unmodeled dynamics into robot operation context.
• Demonstrated applicability to arbitrary robot configurations.
• Validated the method through realistic testing on an industrial robot.

Conclusions:

• The proposed method effectively addresses unmodeled dynamics in robot tasks.
• Applicable to diverse tasks like load manipulation, assembly, and drilling.
• Offers a robust solution for improving robot safety and operational awareness.