Predicting Return-to-Manual Performance in Lower- and Higher-Degree Automation

Area of Science:

• Human-Computer Interaction
• Cognitive Engineering
• Automation and Control Systems

Background:

• Operator states like workload, fatigue, and trust in automation are crucial for safe system operation.
• Limited research exists on how within-person variability in these states predicts return-to-manual (RTM) performance after automation failure.
• Understanding these relationships is vital for designing adaptive work systems that account for performance degradation and operator strain.

Purpose of the Study:

• To investigate operator state variables (workload, fatigue, trust in automation, task engagement) as predictors of RTM performance.
• To determine if the degree of automation (DOA) moderates the relationship between operator states and RTM performance.

Main Methods:

• Participants performed a simulated air traffic control task with either higher- or lower-degree automation (DOA) for conflict detection.
• RTM performance was assessed when automation failed to resolve conflicts, requiring manual intervention.
• Operator states (workload, fatigue, trust, engagement) were measured periodically via self-report.

Main Results:

• Lower DOA led to faster RTM performance compared to higher DOA; DOA did not moderate operator state effects.
• Increased workload and fatigue were associated with poorer RTM accuracy (fewer conflicts resolved).
• Higher trust in automation correlated with improved RTM accuracy.

Conclusions:

• Operator states, specifically workload, fatigue, and trust, are significant predictors of RTM performance.
• While operator states predict performance, further research is needed due to inconsistencies across studies.
• Adaptive work systems can leverage these findings to mitigate performance decrements during automation failures.