Adaptive automation: automatically (dis)engaging automation during visually distracted driving

Area of Science:

• Human-computer interaction
• Automotive engineering
• Cognitive psychology

Background:

• Automated driving systems aim to mitigate human error but face implementation challenges in real traffic.
• Driver distraction is a critical factor affecting driving safety and performance.
• Adaptive control allocation is a promising approach for integrating automated systems with human drivers.

Purpose of the Study:

• To investigate the effectiveness of an adaptive steering control system that allocates control between a human driver and an automated pilot based on real-time driver distraction.
• To compare driving performance, workload, and acceptance of three conditions: manual driving, automated backup steering, and forced manual steering during distraction.

Main Methods:

• Thirty-one participants completed a simulated driving task at a fixed speed (70 km/h).
• Three driving conditions were tested: Manual (driver-only control), Backup (automated steering during distraction), and Forced (manual steering during distraction).
• Driver distraction was induced by a secondary visual task (changing a CD), and steering control was adaptively managed using an eye tracker.

Main Results:

• The automated backup steering condition significantly reduced mean and maximum absolute lateral errors compared to manual driving.
• The backup condition also resulted in the lowest self-reported workload ratings.
• The backup condition received higher acceptance ratings than the forced manual condition, which exhibited higher lateral errors.

Conclusions:

• An adaptive automated backup steering system is well-accepted by drivers and significantly enhances driving performance and reduces workload.
• This adaptive approach offers a viable solution for managing driver distraction in automated driving scenarios.
• Future research should explore higher simulator fidelity and advanced eye-tracking technology for more robust findings.