Anger & Self Driving Cars
Angry Drivers’ Performance in Self-Driving Cars
Thesis | Quantitative Research | Automated Driving | Human Factors | Auditory Displays
Self driving cars. Are we nearly there?
Automation technology is becoming better
Manufacturers have already started to incorporate self driving capabilities at some level
At this stage we’re moving towards full automation but we are at conditional automation (Level 3) (SAE International, 2018)
The curse of semi-automation
The curse of semi-automation Level 3 autonomous vehicles should support transitions that are smooth and safe
This necessitates the use of highly salient warnings for transitions between manual and autonomous modes
The driver will essentially have to take-over control of the vehicle when required.
Emotions & Automation
Emotion can influence information processing, especially if the emotion induces high arousal (Brave & Nass, 2003)
There is a lot of evidence that anger can degrade driving performance (Jallais et al., 2014; Jeon, Walker, & Gable, 2014)
Driver abilities such as perception, decision making, focus and attention can be affected by emotion and affect driver performance (Eyben et al., 2010)
Angry drivers + takeover in self-driving cars = dangerous driving?
Auditory Alert Design
Participants were exposed to a total of 9 auditory alerts
Higher frequencies can make an auditory warning sound more urgent (Edworthy et al., 1991)
Faster repetitions of auditory warnings can make it sound more urgent (E. J. Hellier, Edworthy, & Dennis, 1993).
Research Objectives
Measure the effects of anger on takeover performance and safety in semi-autonomous vehicles.
Measure the effects of warning urgency on takeover performance in semi-autonomous vehicles.
Methodology
Angry drivers (Blue Line) drove worse than neutral drives (Red Line)
Scenario Participants had emotion induced before drive (Control Group had no emotion induced)
3 Lane Highway simulated drive in semi autonomous vehicle
Participants faced transitions of control (takeover) while in the vehicle where they were required to switch lanes to avoid obstacles .
Participants took over control after hearing an auditory alarm(takeover request) sent 7 seconds (lead time) before their vehicle would crash into an obstacle.
The Nervteh driving simulator running Oktal’s SCANeR studio was used to run experiments.
The Tobii Pro 2 eye tracker was used to measure eye gaze.
Example of a driver taking over control from the self-driving car
Dependent Measures
Reaction time to the takeover warnings
Speeding and deceleration
Lane change duration
Steering wheel angle
How often a driver looks at the rearview mirrors before turning (glance duration & frequency)
Results
There were no significant differences in reaction time between angry and neutral drivers.
Urgent sounding auditory alerts reduce reaction times*.
Eye tracking showed no significant differences between angry and neutral drivers.
Angry drivers drove faster than neutral drivers*.
Neutral drivers slowed down more and turned more than angry drivers*.
Angry drivers took longer to change lanes*.
All the data points toward angry drivers showing behavior that could be unsafe for takeover.
Note: See appendix for related graphs. * means findings were significant.
What does this mean for Self Driving Car Design?
Automation does not live in a silo. It needs to work with unpredictable and often emotion-driven human behavior.
Mitigating the effects of emotions can be an important intervention that promotes safe driving.
Highly urgent sounding alerts can mean faster reactions from drivers, which could mean improved safety.
There needs to be more empirical research into predicting human behavior to urgent alerts, especially angry drivers to improve safety. See my journal article on modeling driver behaviors.
Being still in a transitory phase in vehicle automation, we must design for humans in vehicles that account for the dynamic highly interactive environment that vehicles will be in.
References
Brave, S., & Nass, C. (2003). Emotion in human-computer interaction. The human-computer interaction handbook: fundamentals, evolving technologies emerging applications, 81-96.
Eyben, F., Wöllmer, M., Poitschke, T., Schuller, B., Blaschke, C., Färber, B., & Nguyen-Thien, N. (2010). Emotion on the Road—Necessity, Acceptance, and Feasibility of Affective Computing in the Car. Advances in Human-Computer Interaction, 2010, 1-17. doi:10.1155/2010/263593
Edworthy, J., Loxley, S., & Dennis, I. (1991). Improving auditory warning design: Relationship between warning sound parameters and perceived urgency. Human factors, 33(2), 205-231.
Hellier, E. J., Edworthy, J., & Dennis, I. A. N. (1993). Improving auditory warning design: Quantifying and predicting the effects of different warning parameters on perceived urgency. Human factors, 35(4), 693-706.
Jallais, C., Gabaude, C., & Paire-Ficout, L. (2014). When emotions disturb the localization of road elements: Effects of anger and sadness. J Transportation research part F: traffic psychology, 23, 125-132.
Jeon, M., Walker, B. N., & Gable, T. M. (2014). Anger Effects on Driver Situation Awareness and Driving Performance. 23(1), 71-89. doi:10.1162/pres_a_00169
Jeon, M., Yim, J.-B., & Walker, B. N. (2011, 2011). An angry driver is not the same as a fearful driver.
SAE International. (2018). Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles. In: SAE International.
