Human Factors Evaluation of Level 2 and Level 3 Automated Driving ...
Human Factors Evaluation of Level 2 and Level 3 Automated Driving Concepts M. Blanco, J. Atwood, H. M. Vasquez, T. E. Trimble, V. L. Fitchett, J. Radlbeck, G. M. Fitch, & S. M. Russell
Introduction
Experiment 1
Experiment 2
• Automated vehicles are those in which at least one aspect of a safety-critical control function (e.g., steering, throttle, braking) is sustained without driver input • NHTSA has defined automation along 5 levels describing a continuum of vehicle control
Purpose • Investigate what HMI characteristics are most effective at issuing TOR of L2 automated vehicles • Characterize transition time
Purpose • Investigate how to prompt operators to monitor road when operating L2 automated vehicles • Investigate effectiveness of prompts over time
Approach • Participants drove L2 automated vehicle on test track for 90 minutes while performing visual non-driving tasks • Presented with 18 alerts that varied in modality and severity • Visual, Visual + Haptic • Cautionary, Staged, Imminent • Unexpected lane drift occurred at end • Participants received visual + haptic imminent alert
Approach • Participants drove L2 automated vehicle on test track for three 60-minute sessions while performing visual non-driving tasks • Experienced 1 of 3 attention prompts (2-s, 7-s, or no prompts) • Prompt stages progressed from visual to haptic to auditory • Experienced normal operation, unexpected lane drift without alert, or unexpected lane drift with imminent visual + haptic alert
Vehicle Control
Level 4 Full Self-Driving Automation Level 3 Limited Self-Driving Automation Level 2 Combined Function Automation Level 1 Function-specific Automation User Control
Level 0 No-Automation
Conclusions This study shows that an appropriate interaction with L2 and L3 automated vehicles is possible providing effective HMIs are utilized. • Most effective HMIs were those that involved non-visual alerts in addition to visual alerts • Able to mitigate primary task reversals • Non-visual elements allowed participants to perceive alerts when visually distracted • May also be because auditory and haptic alerts convey higher urgency than visual alerts • Visual alerts are less annoying • Need HMIs that balance conspicuity, urgency, and annoyance Operators trust the capabilities of the automation • Risk occurs when prioritizing non-driving activities over vehicle operation Future research is needed to better understand HMIs that balance conspicuity, urgency, and annoyance.
NHTSA’s Five Levels of Automation
Purpose • Explore whether operators can interact with L2 and L3 automated vehicles • Explore how this interaction is affected by the human-machine interface (HMI)
Method Three experiments were conducted on closed tracks simulating highway driving • The two Level 2 experiments were conducted at GM’s Milford Proving Ground • The Level 3 experiment was conducted on the Virginia Smart Road
Experiment 3 Purpose • Investigate what HMI characteristics are most effective at issuing TOR of L3 automated vehicles • Characterize transition time
Aerial Views of the GM Milford Proving Ground and the Virginia Smart Road Test Track at VTTI
From top: • 2009 Chevrolet Malibu with prototype L2 automated system used in Experiment 1 • 2010 Cadillac SRX with prototype L2 automated system used in Experiment 2 • 2012 Lexus RX450h Google Car with prototype L3 automated system used in Experiment 3
Approach • Participants drove L3 automated vehicle on test track for a total of 90 minutes while performing visual non-driving tasks • Received one alert per 30-minute session • Staged • Imminent – No External Threat • Imminent – External Threat (i.e., box on road)
Acknowledgements Caption(s) for picture(s)
Summary of the Three Experiments
We wish to thank our project sponsors, NHTSA and the ITS Joint Program Office. We also wish to thank our project partners, General Motors and Google, as well as our stakeholder committee. Blanco, M., Atwood, J., Vasquez, H. M., Trimble, T. E., Fitchett, V. L., Radlbeck, J., Fitch, G. M., Russell, S. M., Green, C. A., Cullinane, B., & Morgan, J. F. (In Press). Human factors evaluation of level 2 and level 3 automated driving concepts: Final report. Washington, DC: National Highway Traffic Safety Administration.
Introduction
Experiment 1
Experiment 2
• Automated vehicles are those in which at least one aspect of a safety-critical control function (e.g., steering, throttle, braking) is sustained without driver input • NHTSA has defined automation along 5 levels describing a continuum of vehicle control
Purpose • Investigate what HMI characteristics are most effective at issuing TOR of L2 automated vehicles • Characterize transition time
Purpose • Investigate how to prompt operators to monitor road when operating L2 automated vehicles • Investigate effectiveness of prompts over time
Approach • Participants drove L2 automated vehicle on test track for 90 minutes while performing visual non-driving tasks • Presented with 18 alerts that varied in modality and severity • Visual, Visual + Haptic • Cautionary, Staged, Imminent • Unexpected lane drift occurred at end • Participants received visual + haptic imminent alert
Approach • Participants drove L2 automated vehicle on test track for three 60-minute sessions while performing visual non-driving tasks • Experienced 1 of 3 attention prompts (2-s, 7-s, or no prompts) • Prompt stages progressed from visual to haptic to auditory • Experienced normal operation, unexpected lane drift without alert, or unexpected lane drift with imminent visual + haptic alert
Vehicle Control
Level 4 Full Self-Driving Automation Level 3 Limited Self-Driving Automation Level 2 Combined Function Automation Level 1 Function-specific Automation User Control
Level 0 No-Automation
Conclusions This study shows that an appropriate interaction with L2 and L3 automated vehicles is possible providing effective HMIs are utilized. • Most effective HMIs were those that involved non-visual alerts in addition to visual alerts • Able to mitigate primary task reversals • Non-visual elements allowed participants to perceive alerts when visually distracted • May also be because auditory and haptic alerts convey higher urgency than visual alerts • Visual alerts are less annoying • Need HMIs that balance conspicuity, urgency, and annoyance Operators trust the capabilities of the automation • Risk occurs when prioritizing non-driving activities over vehicle operation Future research is needed to better understand HMIs that balance conspicuity, urgency, and annoyance.
NHTSA’s Five Levels of Automation
Purpose • Explore whether operators can interact with L2 and L3 automated vehicles • Explore how this interaction is affected by the human-machine interface (HMI)
Method Three experiments were conducted on closed tracks simulating highway driving • The two Level 2 experiments were conducted at GM’s Milford Proving Ground • The Level 3 experiment was conducted on the Virginia Smart Road
Experiment 3 Purpose • Investigate what HMI characteristics are most effective at issuing TOR of L3 automated vehicles • Characterize transition time
Aerial Views of the GM Milford Proving Ground and the Virginia Smart Road Test Track at VTTI
From top: • 2009 Chevrolet Malibu with prototype L2 automated system used in Experiment 1 • 2010 Cadillac SRX with prototype L2 automated system used in Experiment 2 • 2012 Lexus RX450h Google Car with prototype L3 automated system used in Experiment 3
Approach • Participants drove L3 automated vehicle on test track for a total of 90 minutes while performing visual non-driving tasks • Received one alert per 30-minute session • Staged • Imminent – No External Threat • Imminent – External Threat (i.e., box on road)
Acknowledgements Caption(s) for picture(s)
Summary of the Three Experiments
We wish to thank our project sponsors, NHTSA and the ITS Joint Program Office. We also wish to thank our project partners, General Motors and Google, as well as our stakeholder committee. Blanco, M., Atwood, J., Vasquez, H. M., Trimble, T. E., Fitchett, V. L., Radlbeck, J., Fitch, G. M., Russell, S. M., Green, C. A., Cullinane, B., & Morgan, J. F. (In Press). Human factors evaluation of level 2 and level 3 automated driving concepts: Final report. Washington, DC: National Highway Traffic Safety Administration.
