A Systematic Review and Meta-Analysis of Takeover Performance During Conditionally Automated Driving.

Area of Science:

• Human-Computer Interaction
• Automotive Engineering
• Cognitive Psychology

Background:

• Conditionally automated driving systems (ADS) require drivers to resume control in limited operational domains.
• Driver takeover performance is influenced by time constraints, prior task engagement, and information provided during transitions.
• Understanding these factors is crucial for ensuring safe human-machine interaction in automated vehicles.

Purpose of the Study:

• To synthesize experimental research on factors influencing driver takeover performance in conditionally automated driving.
• To identify key variables affecting the safety and efficiency of transitions from automated to manual control.
• To provide evidence-based recommendations for future ADS design and implementation.

Main Methods:

• Systematic literature search identifying 8446 articles, from which 48 articles (51 experiments) were selected for meta-analysis.
• Independent variables coded included time budget, non-driving related task engagement, resource demands, and information support.
• Dependent variables focused on takeover timing and quality measures.

Main Results:

• Engaging in non-driving tasks, especially those with overlapping resource demands, significantly degrades takeover performance.
• Limited evidence suggests shorter time budgets also impair takeover performance.
• Current information support strategies did not demonstrably improve takeover performance.

Conclusions:

• Future research and development should prioritize increasing the time available for drivers to take over control.
• Further investigation into effective information support mechanisms is warranted.
• Implementation of vehicle-to-everything (V2X) services and robust driver monitoring systems are recommended.