On the Need for Driver Attention Support Systems

On the Need for Driver Attention Support Systems
Victor , T.
Appeared in Journal of Traffic Medicine (2000) Vol. 28, No. 2S
Volvo Technological Development Corporation, Human Systems Integration, 6900, PVH32,
S-405 08 , Göteborg, Sweden. trent@vtd.volvo.se , http://www.tech.volvo.se/human.html
Introduction: Driver inattention is the most prevalent primary cause of collisions, accounting
for an estimated 25-56%. Among the Inattention causes, Distraction and Looked-but-did-notsee
are more frequently reported factors in crashes than Sleepy/fell asleep (e.g. Wang et al,
1996). Important crash types involving inattention have are rear-end, intersection, lane
change/merge, road departure, and single vehicle crashes. Changes in visual scanning
patterns, gaze fixations (number and length), and percentage eye closure are promising
occular-based indicators of attention and alertness, and can potentially be integrated in future
in-vehicle attention support systems. The present research aims at improving driver attention
with feedback and providing vehicle systems with real-time knowledge of driver visual
behavior.
Methods: Volvo developed a novel computer-vision based head pose- and gazetracker with
the Australian National University. This system, integrated in real-time with vehicle
performance data from the CAN bus, and driving support systems (e.g. lanetracking) provides
the opportunity for the development of in-vehicle attention support systems.
Results: Preliminary road tests show that the system robustly tracks head pose, gaze, and eye
closure in real-time, in real vehicle environments. A literature review and focus group
discussions investigated the potential need for attention support, and identified situations in
which inattention occurs. There was concern that new interior distractors (e.g. cellphones,
navigation systems, collision warning systems) divert attention and cause high workload,
however classical sources remain as large contributors, e.g. driver status, loose objects,
interactions with other passengers, and instrumentation. Real-time vehicle- and visual
performance feedback could enable driver self-observation and management of attentive
behavior in addition to threshold warnings. This type of attention support systems should have
a very positive impact on road safety.