Recursive Bayesian Human Intent Recognition in Shared-Control Robotics.

Area of Science:

• Robotics
• Human-Computer Interaction
• Artificial Intelligence

Background:

• Effective human-robot collaboration in shared control necessitates understanding human user intentions.
• Current methods often lack robust mechanisms for real-time intent recognition.

Purpose of the Study:

• To develop a mathematical formulation for human intent recognition in assistive teleoperation under shared autonomy.
• To improve the performance and accuracy of intent inference in human-robot shared control systems.

Main Methods:

• A recursive Bayesian filtering approach was employed to probabilistically reason about user intent.
• Multiple non-verbal observations, including contextual data and goal-directed human actions with adjustable rationality, were modeled and fused.
• Human inference on robot motion was examined.

Main Results:

• The proposed approach demonstrated superior performance compared to existing solutions.
• Probabilistic fusion of multiple observations significantly improved intent inference accuracy.
• Human subjects study validated the approach's effectiveness across various scenarios and tasks.

Conclusions:

• The developed mathematical formulation effectively models and fuses multiple observations for accurate human intent recognition.
• The approach enhances shared-control operation by improving intent inference and overall performance.
• This work contributes to more seamless and effective human-robot collaboration.