Towards Mixed-Initiative Human-Robot Interaction: Assessment of Discriminative Physiological and Behavioral Features...

Area of Science:

• Human-Robot Interaction
• Cognitive Science
• Robotics

Background:

• Optimizing operational performance in human-robot interactions requires dynamic task allocation.
• Mixed-initiative systems allow flexible task and authority distribution between humans and robots.
• Monitoring human performance is crucial for effective mixed-initiative system implementation.

Purpose of the Study:

• To investigate the impact of robot automation levels on human performance in a cooperative fire-fighting task.
• To identify behavioral, cardiac, and ocular patterns associated with different performance levels.
• To assess the feasibility of using physiological and behavioral data to predict mission performance and adapt automation.

Main Methods:

• Experimental scenario: participants cooperated with a robot in fire-fighting missions with varying hazards.
• Independent variable: two levels of robot automation (manual vs. autonomous).
• Data collection: cardiac activity, eye-tracking, and user interface actions.

Main Results:

• Participant performance varied, allowing classification into high and low-scoring groups with distinct patterns.
• Higher automation benefited low-scoring participants but hindered high-scoring ones, and vice versa.
• Physiological and behavioral features accurately predicted mission performance, achieving 74% balanced accuracy using combined data.

Conclusions:

• Adaptive human-robot interaction (HRI) systems can leverage real-time analysis of physiological and behavioral markers.
• Dynamic adjustment of automation levels based on these markers can optimize mission performance.
• This approach offers a pathway to more effective and responsive human-robot collaboration.