Impedance Variation and Learning Strategies in Human-Robot Interaction.

Area of Science:

• Robotics
• Control Systems
• Human-Robot Interaction

Background:

• Physical human-robot interaction (HRI) requires robots to adapt their dynamic properties.
• Impedance and admittance control are key for managing physical interactions.
• Learning and online adjustment of these properties are crucial for safe and effective HRI.

Purpose of the Study:

• To systematically review methodologies for varying and learning robot impedance/admittance over the past two decades.
• To categorize and compare assumptions and mathematical formulations for online impedance adjustment in physical HRI.
• To provide an overview of current challenges and research trends in physical HRI.

Main Methods:

• Systematic review of studies focusing on variation and learning of impedance elements.
• Categorization of strategies based on objectives, approaches, and signal requirements (position, force, EMG).
• Review of methods including linear/nonlinear analyses and Gaussian approximation algorithms (GMM, DMP).

Main Results:

• Identified and categorized diverse strategies for impedance/admittance control in physical HRI.
• Compared different approaches regarding their underlying assumptions, mathematical formulations, and data requirements.
• Highlighted the application of optimal control, Lyapunov-based stability, GMM, and DMP strategies.

Conclusions:

• The field has seen significant advancements in adaptive impedance control for HRI.
• Understanding the trade-offs between different methods is crucial for selecting appropriate strategies.
• Future research should address current challenges to further enhance physical HRI capabilities.