How to be Helpful? Supportive Behaviors and Personalization for Human-Robot Collaboration

Area of Science:

• Robotics
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Human-Robot Collaboration (HRC) has advanced, enabling robots in unstructured environments.
• Current HRC systems lack proactive support, autonomous task execution, and effective human assistance.
• Challenges include limited robot manipulation, incomplete task knowledge, and partial environmental awareness.

Purpose of the Study:

• To present a novel collaborative system for Human-Robot Collaboration (HRC).
• To enable robots to proactively assist human workers despite environmental and task uncertainties.
• To improve synchronization and mutual understanding in shared tasks.

Main Methods:

• Developed a framework using a shared, hierarchical task model for human-robot synchronization.
• Derived a Partially Observable Markov Model (POMP) from the high-level task representation.
• Employed an online Monte-Carlo solver for real-time, short-horizon robot planning.

Main Results:

• The system demonstrates interactive replanning and dynamic error recovery capabilities.
• It can infer hidden user preferences, enhancing collaborative efficiency.
• Robustly supported human workers in a realistic furniture construction task.

Conclusions:

• The developed HRC system effectively assists human workers in complex tasks.
• Shared hierarchical models and adaptive planning are key to proactive collaboration.
• This framework addresses limitations in current HRC systems, paving the way for more integrated human-robot teams.