Enter The Simulation
Our unique methodology involved using a 300◦ high fidelity driving simulator composed of a five channel projection environment and a fully functioning Ford Focus Cab. We explored the possibility that traditional driving simulators can be used as a proxy for self-driving car research. Our experiment exposed users to two autonomous driving simulations, one safe and one risky. We found that in both simulation conditions there were greater levels of physiological stress, which suggests driving simulators can be utilized for self-driving car research. If a real stress response can be produced from a simulation of autonomous driving, then it would pave the way for future research as a cheaper alternative that offers lower risk and liability. The implications of this possibility extend not only to academia, but also for advertising and development of autonomous technology. If autonomous cars can become more trustworthy to consumers, it would only expedite their ubiquitous presence.
Abstract
The goal of this study was to examine whether a traditional driving simulator could be used as a proxy for self-driving vehicle research. Twenty-eight participants completed one manual driving task and two automated driving tasks using a high fidelity driving simulator. The simulator was programmed to drive the participants through a series of simulated environments with a varying amount of risk-taking behavior such as speeding and drifting out of the lane. The effectiveness of the system as an autonomous vehicle simulator was measured in terms of participant psychophysiological stress. Results show that participants do have an increased psychophysiological stress response when the vehicle drives autonomously as compared to when the users are in control. These findings suggest that users perceive a degree of risk while the simulator imitates autonomous driving, allowing traditional high-fidelity simulators to still be used as tools for this modern research topic.
Methods
- Nineteen individuals (13 females, 19.51±1.31 years)- all had good mental & physical health and drove an average of 4±2.5 times per week.
- Driving Habits Questionnaire- Contained 34 items, such as: "How would you rate the quality of your driving?"
- Automation Trust Survey contained 12 items such as: "The automated driving system is reliable."
- ProComp Infiniti multichannel physio data logger was used to measure: Trapezius EMG, Palmer Skin Conductance & Blood Pulse Volume.
- Driving Simulator (DriveSafety RS 600)-Contains a fully functioning Ford Focus cab and a 300◦ five channel projection environment.
- Driving Conditions entailed:
- Resting Baseline- sitting without activity
- Driving Baseline- simple manual driving
- Automated Risky- vehicle drives in a risky manner (counterbalanced)
- Automated Safe- vehicle drives in a safe manner (counterbalanced)
- A 16-item Simulator Sickness Questionnaire was given after each driving condition.
Results
- The result of paired samples t-test showed that based on the Automation Trust Survey, participants trusted the autonomous vehicle less while in the risky driving mode (2.00±0.90) than in the safe driving mode (3.66±0.55), t(18)=8.28, p<0.001, r2=0.79. (Figure 1)
- There was no significant difference in surface EMG from the trapezius muscle, however participants showed significantly higher skin conductance while in the risky driving mode (4.42±3.59) than in the safe driving mode (4.09±3.46), t(18)=4.09, p=0.001, r2=0.482. (Figure 2)
- Additionally, participants showed significantly higher skin conductance while in the safe driving mode (4.09±3.46) than when driving themselves (3.40±2.96), t(18)=2.95, p=0.009, r2=0.33. (Figure 2)
- A bivariate correlation between feelings of familiarity with the autonomous vehicle and skin conductance showed that familiarity significantly predicted skin conductance during safe driving (r(17)=0.500, p=0.029) and risky driving (r(17)=0.542, p=0.017). (Figure 3)
- Early analysis of heart rate variability data suggests that the sympathetic branch of the autonomic nervous system was more active while participants were in the risky driving mode than in any other condition. (Figure 4)
Conclusions
- Results showed that the use of autonomous vehicles may increase health risks for conditions sensitive to stress levels.
- While using an autonomous vehicle, participants showed an increased stress response compared to when they were in control of the vehicle.
- When participants felt the vehicle was driving in an unsafe manner they showed an even larger stress response. In this case, more feelings of familiarity with the autonomous vehicle were associated with a larger stress response.
- This may suggest that associating an autonomous vehicle with a traditional vehicle accentuates the oddity of it being able to drive itself, which in turn lead to more stress.
- Those in the transportation industry should explore design methods that emphasize how autonomous vehicles are unique from traditional vehicles.
- Those in the health care industry should make an effort to understand how autonomous vehicles will impact their patients as the technology becomes more ubiquitous.