Trust in Autonomous Vehicles FINAL - UMTRI · –Difficulty managing transitions –What did you do...
Transcript of Trust in Autonomous Vehicles FINAL - UMTRI · –Difficulty managing transitions –What did you do...
Trust in Autonomous Vehicles: A Simulator Study
Lindsay H. Ryan, PhDInstitute for Social Research
University of Michigan
The Future of Automotive Safety Conference, UMTRI Automotive Futures Group 2018
Acknowledgements
• Mobility Transformation Center
Funding
• Project Team: PI Lisa Molnar, Co-
Investigators Anuj Pradhan, David Eby,
Renée St. Louis, and Jennifer Zakrajsek
• Additional assistance from Nicole
Zanier and Dana Demchak
Aging Drivers in
the United States
Intersection of Age and Technology
Implications for Trust in Automated
Vehicles
Does experience with other tech matter?
Do beliefs about other drivers matter?
Does actual experience using automated tech matter?
Pilot Study
• Three main components– Level 3 Autonomous Car Simulator– Structured Interview– Self-administered Questionnaire
• Three groups: novice drivers (16 – 19), older drivers (65-75), and comparison group (age 25 – 45)
• Primary focus was on transfer of automated-to-manual control and vice-versa
UMTRI High-fidelity Advanced Driving Simulator
Simulation Design1. Eye tracking calibration2. 10 minute practice, manual and automated3. 20 minute main drive– Begin in manual mode– Instructed to engage automated mode when
comfortable– 8 points during drive prompt automated-to-manual
control change– Timing to re-engage automated mode was up to the
driver
Structured InterviewContent– Expectations about autonomous vehicles before and
after the simulator experience– Comfort with different driving situations in simulator– Difficulty managing transitions – What did you do during automated mode?– Did you trust the automated driving?– Any concerns if this type of tech was available in the
real world?– Are you interested in having a vehicle with this
technology?
Driver Age GroupsDemographics and Driving
Characteristics16-17 25-45 65-75 All
Number of participants 24 24 24 72
Mean age (SD) 16.4 (.50) 33.5 (5.9) 68.8 (3.6) 39.6 (22.2)
% Female 50.0 50.0 45.8 48.6
Avg. days driven per week 5.3 6.1 6.1 5.8
Avg. miles driven per week 79.5 158.7 133.4 123.9
Someone available to give you rides (%) 100.0 79.2 79.2 86.1
Someone depends on you to drive them
(%) 33.3 33.3 33.3 33.3
Indicators of Trust
• Simulator Behavioral Indicator: % of manually driven simulator scenarios– # automated-to-manual
transitions missed due to manual driving
• Self-report Indicator: “Did you trust the automated driving? Why or Why not?”– Coded: 0 = No, 1 = Situation
specific, 2 = Yes
Individual Driver
Characteristics
Comfort with other drivers: “Are you usually comfortable with other people behind the wheel?”
General control preferences: “Generally speaking, do you like to have control over things?”
Experience with other technology: Count of various technologies used
Themes that Emerged from the Interviews
Rural, low traffic, straight driving were the contexts where most drivers were comfortable with the automated mode
Older Drivers Interested in Automated Technology to Extend mobility
Beliefs about Other Drivers and Technology Matter
• In favor of automated vehicle technology
Worried about distracted
drivers
• Not in favor of automated technology
Worried drivers won’t pay
attention during Automated
Mode
Impact of the News/Media
Predictors of Trust in Automated Technology
• Drivers who reported ↑comfort with other drivers behind the wheel also reported ↑ trust in automated drive technology (B = .25, p < .05)
“Did you trust the automated driving? Why or Why not?”
• Coded trust in automated driving was only significant predictor (B = -.15, p < .01)
• Partial validation of this behavioral indicator
% of manually driven
simulator scenarios
Future Directions
• New validated measures of trust in automated vehicles are needed
• Need to test these associations in real-world settings with diverse driver samples– E.g. Samples with less
education and less experience with other technology