Meaningful human control and variable autonomy in human-robot teams for firefighting.

Area of Science:

• Robotics
• Human-Robot Interaction
• Artificial Intelligence Ethics

Background:

• Human-robot collaboration is increasing in complex tasks like firefighting.
• Ensuring meaningful human control (MHC) over autonomous robots is crucial.
• Variable autonomy approaches aim to balance robot autonomy with human oversight through accountability, responsibility, and transparency.

Purpose of the Study:

• To operationalize and measure meaningful human control (MHC) in dynamic human-robot teaming for firefighting.
• To address the lack of systematic metrics for designers of variable autonomy systems.
• To explore quantitative methods for verifying MHC in robot-assisted firefighting.

Main Methods:

• A qualitative focus group (n=5 experts) explored quantitative operationalizations of MHC.
• Investigated dynamic task allocation using variable autonomy, with robots identifying moral sensitivity.
• Analyzed focus group data using reflexive thematic analysis.

Main Results:

• Quantifying traceability is key to measuring MHC.
• Situation awareness and performance can objectively measure aspects of traceability.
• Team/robot outcomes verify MHC, but understanding the reasons behind outcomes is essential for determining the level of MHC.

Conclusions:

• Proposed an evaluation method to verify MHC in variable autonomy human-robot firefighting teams.
• The method quantifies traceability subjectively and objectively via human responses during/after simulations.
• Semi-structured interviews identify underlying reasons for outcomes and suggest improvements for variable autonomy approaches.