In-Vehicle Driver Distractions & Eye Movements Presented by: Briana Nelson Kraayenbrink
What is driver distraction? (NHTSA, 2000) Any activity (technology or non-technology related) that takes a driver’s attention away from the task of driving Any distraction has the potential to cause or contribute to a crash Driver distraction is one of the most common causes of traffic crashes Approximately 25% of police reported crashes
What is driver distraction? (NHTSA, 2000) Four distinct categories: Visual distraction Auditory distraction Biomechanical distraction Cognitive distraction Many distracting activities can be more than one of these categories
Previous Research Cell Phones Hands-free Navigation Guidance Systems
Visual Distraction Research Antin, Dingus, Hulse, & Wierwille (1990) found that drivers spent 80% of their time looking at the road ahead when not engaged in distracting activities Distracting events reduced amount of time looking at forward headway and increased the glance durations to mirrors, windows, and inside the truck
Visual Distraction Research Glance duration & frequency to an in-vehicle display are dependent upon the display characteristics (e.g., character size, info density, complexity of task) Compared to large displays & controls, small displays and controls can produce a 20% increase in mean glance duration (Rockwell, 1998) More glances are required with manual than speech task Radio/CD task took three 1s glances, dialing phone # required twelve 1s glances (Faerber, Faerber, & Meier-Arendt, 1999)
Visual Distraction Research ETAK study (Wierwille, Dingus & colleagues, 87-88) Compared visual attention demand for the navigation system tasks (with ETAK) and conventional tasks (tuning radio) On road studies in an instrumented 1985 Cadillac DeVille fitted with ETAK navigator
ETAK Studies
ETAK Studies Eye glances to in-vehicle instruments, mirrors, etc. were low-frequency occurrences and accounted for 4-6% of glances Percentage of glances to navigation system ranged from 7-33%. Percentage of glances to forward, left, and right roadway ranged from 62-92%
Visual Distraction Research Design Guidelines Zwahlen, Adams, and DeBald (1998)
Visual Distraction Research Design Guidelines European Statement of Principles on Human Machine Interface JAMA (2000) Guidelines Apply to OEM and after-market The IVIS Demand HF Design Guidelines for ATIS and CVO SAE J2364 (the 15-second rule) Statement of Principles by the Alliance of Auto Manufacturers CAMP Driver Workload Metrics Project
Route Navigation Studies-UMTRI Tech Report UMTRI Tech Report on Visual and Task Demands of Driver Info Systems Purpose Provide background material for the development of SAE J2364 Provide info to enhance the safety and usability of driver interfaces for navigation systems
Route Navigation Studies-UMTRI Tech Report What is the relationship between visual demand and crashes? Time spent looking inside the vehicle is not spent looking at the road for potential crash-inducing hazards.
Route Navigation Studies-UMTRI Tech Report How long does destination input and retrieval take? With manual controls and visual displays, destination entry takes 1 – 2.5 minutes Point of interest entry takes 10% longer than street address or intersection methods Destination retrieval takes about 10s depending on the address and database The worst interface for a task takes twice as long to use as the best
Route Navigation Studies-UMTRI Tech Report How long do drivers prefer to look in a vehicle at an object? From their experience, drivers have a sense of how long is too long and how frequent is too frequent Experimental approach is for drivers to look at nonattention demanding target for as long as they feel safe/comfortable to do so Also addressed by having drivers rate how safe/comfortable/secure they felt after completing various tasks Suggestions have been made that total glance times should not exceed 5-10s for drivers to feel safe
Route Navigation Studies-UMTRI Tech Report How much time do drivers spend looking at in-vehicle controls and displays? Mean glance durations typically do not exceed 1.2 – 1.5s But how many times are they glancing over to look at the display in a certain time window??
SAfety VEhicles using adaptive Interface Technology (SAVE-IT) VNTSC awarded contract to Delphi Electronics & Safety to develop, demonstrate, and evaluate the potential safety benefits of adaptive interface technologies that manage the info from various in-vehicle systems based on real-time monitoring of the roadway conditions and the driver’s capabilities SAVE-IT is designed to mitigate distraction with effective countermeasures and enhance the effectiveness of safety warning systems. Comprehensive research team includes researchers from U of IA, UMTRI, GM, Ford, and Seeing Machines
SAVE-IT Visual Distraction Research Experiment 14 subjects (35-53 y.o.) Delphi driving simulator using FaceLab eye tracking system NASA-TLX used to measure workload and perceived task difficulty IVs: road type, amount of visual distraction (reading task), display eccentricity
SAVE-IT Visual Distraction Research Delphi Driving Simulator
SAVE-IT Visual Distraction Research Results As the visual distraction level increased, the TLX score increased. TLX score also increased as display eccentricity increased
SAVE-IT Visual Distraction Research
SAVE-IT Visual Distraction Research Recommendations for future research Repeat experiments in NADS, naturalistic on-road driving to provide more generalizable results Actually doing this in phase 2 of SAVE-IT
Future Research Recommendations Non-intrusive, automatic, and reliable eye tracking systems should be used to examine driver’s visual behavior and assess the level of distraction. Could even use a few video cameras to determine where individuals are looking
Precision & Accuracy of Video-Based Measurements of Driver’s Gaze Location (Schieber, et. al, 1998)
Future Research Recommendations Time-based eye glance measures may be more appropriate than task-based eye glance measures when developing a system designed to mitigate driver distraction Several eye glance measures are reliable measures of visual distraction that can be used in a real-time system using adaptive interface technologies. Total eyes-off-road glance duration Sum of mean glance duration, number of glances, and frequency of use for a task or device
Comments/Thoughts The technology becoming available in vehicles is progressing faster than research. Entertainment systems: DVD players Collision avoidance systems Speech-based internet and e-mail Laptop/handheld computer Night visions systems What about the combination of more than one driver distraction? What about the novice-expert difference?