response to a critical situation during automated driving: can we take drivers out of the loop?
TRANSCRIPT
Response to a critical situation during automated driving: can we take
drivers out of the loop?*
Tyron Louw, Natasha Merat, Georgios Kountouriotis, Ruth Madigan
Institute for Transport Studies, University of Leeds, Leeds, UK
// Introduction & Aims
• Endsley, M. R., and Kiris, E. O. (1995). The out-of-the-loop performance problem and level of control in
automation. Human Factors, 37(2), 381-394.
• Lee, J. D., Regan, M. A., and Young, K. L. (2008). Defining driver distraction. In M. A. Regan, J. D. Lee,
and K. L. Young (Eds.), Driver Distraction: Theory, Effects, and Mitigation(pp. 31–40). Boca Raton, FL:
CRC Press.
• Li, S. Y. W., Magrabi, F. and Coiera, E. (2012). A systematic review of the psychological literature on
interruption and its patient safety implications. Journal of the American Medical Informatics
Association: JAMIA, 19(1), 6–12.
• 30 Participants (39.2yrs ± 14.45)
• Driving experience: 20.17yrs ± 15.26
• Repeated measures, 3 X 3 mixed design
• Vehicle automation is likely to induce mind-
wandering, or stimulus-independent thoughts,
which can interfere with processing external
stimuli, such as roadway hazards (Li et al., 2012).
• This out-of-the-loop (OOTL) state presents an
issue to safety should the driver be called upon to
resume manual control (Endsley & Kiris, 1995;
Lee, 2013). But how does one study this given that
inducing the OOTL state is difficult?
// Results
// Methods
• There is also no objective measure of safety
and quality of the transition of control from
automation to manual driving
1. Can we induce the OOTL state by limiting
system and environmental information?
2. How do drivers make decisions and react in
the face of uncertain automation?
3. Are there other means of evaluating the
transition to manual driving?
Light FogHeavy Fog
30s
Automation On
Lead vehicle actionUncertainty Alert
EVENT START
Egovehicle
Lead vehicle
Screen Manipulation On
90s ≈30s
EVENT ENDScreen Manipulation Off
Non-critical Critical
1 2 3 4 5 6 7
≈ 20 mins
≈150s
// References
Critical Event = Lead vehicle braked at TTC of 5s.
Collision would occur unless driver intervened.
Steering Wheel
Colour
Automation
status
Grey Unavailable
Flashing green Available
Green Engaged
Flashing yellow Uncertain
Red Disengaged
• HMI: FCW & Automation Status • Inducing the OOTL state in automation:
• Schematic representation of each
discrete event
AutomationAutomation
Manual Manual
Critical
Event 1
Critical
Event 2
Critical
Event 1
Critical
Event 2Critical
Event 1
Critical
Event 2
Critical
Event 1
Critical
Event 2
Wit
hin
-Subje
cts
Facto
rs:
Dri
ve &
Event
Wit
hin
-Subje
cts
Facto
rs:
Dri
ve &
Event
Light Fog Heavy Fog
Between-subjects factor: Condition
Automation
Status
Hidden
Light Fog
Screen
Occlusion
Automation
Status
Hidden
Heavy Fog
Screen
Occlusion
Limited Visual Roadway
Information
Auditory and
haptic cues
still present
• Were drivers OOTL?
Percentage Road Centre during
screen occlusion
‘Peeks’ at hidden automation status
during screen occlusion
• What did they do?
In 16% of non-critical cases drivers disengaged
automation compared to 100% in critical cases
Automation Manual
Critical Event 1 Critical Event 2 Critical Event 1 Critical Event 2
Light Fog 7 (2) 10 (1) 11 (1) 12 (0)
Heavy Fog 9 (7) 9 (3) 11 (2) 11 (1)
Lane Changes and collision counts (in brackets)
• How did they do it?
• Automation (vs. Manual) =
↑ Lateral Acceleration (p=.005)
↑ Deceleration (p=.001)
↓ Time headway (p=.011)
*Paper to appear in the Proceedings of the Driver Distraction and Inattention Conference, Sydney 2015