Driving performance analysis of the ACAS FOT data and recommendations for a driving workload manager.

Eoh, Hong; Green, Paul A.; Schweitzer, Jason; Hegedus, Ed · 2006 · ROSA P / University of Michigan. Transportation Research Institute

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 report analyzes driving performance data from the Advanced Collision Avoidance System (ACAS) Field Operational Test (FOT) to establish thresholds for a driving workload manager. The research was motivated by the need to prevent driver overload and distraction as vehicles incorporate more complex in-vehicle systems. The primary objective was to identify specific driving maneuvers and distracted states using vehicle telemetry, thereby determining when secondary tasks should be prohibited. The study addressed four key questions: the distribution of driver inputs and vehicle outputs across road types, the impact of secondary tasks on performance, the efficacy of linear thresholds for maneuver detection, and the predictive value of steering and throttle entropy for distraction. The analysis utilized naturalistic driving data from 96 drivers who accumulated over 100,000 miles in instrumented vehicles. The dataset included 2,914 video clips of driver behavior and corresponding engineering variables, such as steering wheel angle, throttle position, heading, and speed. Researchers categorized driving environments into four main groups: ramps, interstates/freeways, arterials/minor arterials, and collectors/local roads. They compared driving performance during attentive driving against instances where drivers performed zero, one, or two secondary tasks (e.g., phone use, grooming). The study employed nonparametric density estimation and bivariate normal ellipses to map the joint distributions of input and output variables, aiming to distinguish between normal driving and high-demand maneuvering situations. The results indicated that driving performance distributions vary significantly by road type. Interstates and freeways exhibited high throttle variability but low steering variability, whereas local and collector roads showed greater steering variability. Regarding distraction, the presence of zero or one secondary task did not significantly alter driving performance distributions compared to attentive driving. However, performing two simultaneous tasks significantly restricted steering movement, indicating that drivers self-regulate by avoiding complex maneuvers while multitasking. In terms of maneuver detection, linear thresholds for steering and throttle showed mixed success; while some thresholds effectively identified maneuvers like lane changes or turns with high odds ratios, others performed poorly. Furthermore, steering entropy proved to be a predictor of distracted driving, whereas throttle entropy showed no significant differences between distracted and normal states. The study concludes that a workload manager must account for road type to accurately assess driving demand, as performance baselines differ across environments. While simple linear thresholds can identify certain maneuvers, their reliability varies, suggesting that more complex, nonparametric models may be necessary for robust detection. The finding that drivers naturally restrict steering during dual-task performance supports the feasibility of workload managers that inhibit secondary tasks during high-demand maneuvers. These recommendations provide a foundational framework for developing systems that dynamically adjust in-vehicle task availability based on real-time driving conditions.

Key finding

Performing two secondary tasks significantly restricts steering movement variability compared to performing zero or one task, indicating drivers limit lateral control demands during high workload situations.

Methodology

naturalistic

Sample size: 96

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.

• steering pattern
• manual
• mental demand
• temporal
• lane positioning
• distraction detection algorithms

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, observational prevalence
• Theoretical Contribution: computational model