Sensor-based assessment of the in-situ quality of human computer interaction in the cars : final research report.

Kim, SeungJun; Dey, Anind K. · 2016 · ROSA P / Carnegie-Mellon University. Technologies for Safe and Efficient Transportation University Transportation Center

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This research addresses the critical challenge of determining when drivers are interruptible during naturalistic driving. Because human attention is a finite resource, interruptions split focus between the primary driving task and secondary interactions, potentially increasing mental workload and decreasing performance. While previous studies on interruption have focused on static, screen-based tasks, little research has examined interruption timing in dynamic environments like driving, where diverting attention can be dangerous. The study aims to identify low and high cognitive load states in real-time to determine opportune moments for driver interruption. To achieve this, the researchers conducted a field experiment with 25 drivers, though only 15 were included in the final analysis due to data quality issues. Participants drove their own vehicles during two sessions: one on a preferred route and another on a fixed GPS-guided route. The study utilized a comprehensive sensor suite, including an On-Board Diagnostics (OBD) device for vehicle status, four YEI motion sensors on the driver’s wrists, head, and foot, and a BioHarness chest belt for physiological data. Two smartphones recorded traffic video and synchronized the data streams. The researchers derived 152 features from these sensors, including vehicle speed, road curvature, body motion, and heart rate, aggregated into one-second segments. The analysis categorized driving into five states: Steering Only, One-Hand Drive with No Peripheral Interaction, Driving Interactions (central tasks), Peripheral Interactions (PI, such as using the radio), and No-Hand Drive. Statistical analyses revealed that drivers significantly regulated speed during peripheral activities or when both hands were off the wheel. Physiological states during PI and No-Hand Drive were most similar to resting states, indicating higher interruptibility. In contrast, Steering Only and Driving Interactions represented less interruptible states. One-Hand Drive without peripheral interaction remained ambiguous, as physiological states resembled interruptible states, but driving conditions differed significantly. Using a random forest classifier with stratified 10-fold cross-validation, the researchers achieved 94.9% accuracy in distinguishing between interruptible and less interruptible states every second. The study concludes that sensor data can effectively detect driver interruptibility in real-time. These findings support the development of intelligent in-car systems that mediate interruptions based on driver workload. Future work includes validating these models across broader populations, designing interruption mediation strategies, and establishing guidelines for intelligent interruption systems to enhance safety and user experience.

Key finding

Sensor data can discriminate driver interruptibility every second with 94.9% accuracy.

Methodology

naturalistic

Sample size: 15

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• distraction detection algorithms
• manual
• temporal
• in vehicle coaching
• control interfaces
• visual

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: behavioral performance data, physiological data, observational prevalence