General Framework for the Optimization of the Human-Robot Collaboration Decision-Making Process Through the Ability...

Area of Science:

• Robotics
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Current Human-Robot Collaboration (HRC) methods often define fixed utility functions, limiting adaptability to changing performance metrics.
• Existing frameworks struggle to dynamically adjust task performance criteria within the same system.

Purpose of the Study:

• To propose a novel, flexible decision-making framework for Human-Robot Collaboration (HRC).
• To enable easy modification of performance metrics for diverse HRC scenarios.
• To enhance real-time adaptation of robot actions to human actions.

Main Methods:

• The proposed framework models HRC as a constrained optimization problem.
• The utility function is bifurcated into a task completion constraint and a modifiable reward function for collaboration performance.
• Decision-making is underpinned by game theory principles, specifically Nash Equilibrium and perfect-information extensive form games.

Main Results:

• The framework effectively handles changes in performance metrics across different scenarios.
• It allows for real-time adaptation of robot behavior to human actions.
• Simulations and experimental studies on an assembly task validated the framework's effectiveness.

Conclusions:

• The developed framework offers superior flexibility in HRC by separating task constraints from performance rewards.
• It provides a robust method for optimizing collaborative tasks with varying metrics like completion time and error probability.
• This approach enhances the adaptability and efficiency of human-robot teams.