Principal Investigator: Valentina Graci, PhD, Children's Hospital of Philadelphia

WHAT WAS THE PURPOSE OF THIS PROJECT?

An increasing number of vehicles on today’s roadways have automatic emergency braking (AEB). Federal testing criteria are mainly based on the vehicle’s successful speed reduction to avoid collision with an obstacle. Therefore, AEB pulses can vary between vehicles and manufacturers; accordingly, occupants’ motion when exposed to these different pulses also varies. This project aimed to quantify important pulse characteristics for each AEB system and use that information to categorize pulses for future testing.

HOW WAS THE RESEARCH CONDUCTED?

We extracted publicly available information on 2,279 AEB tests from a range of vehicles from the Insurance Institute for Highway Safety test database. We identified tests in which the vehicle contacted the target (613) and had no contact (1,666). Utilizing machine learning principles – a relatively novel technique now used in autonomous driving research – we identified categories of pulses from a range of pulse characteristics, including deceleration magnitude, speed reduction, ramp time, and duration of the deceleration phase.

WHAT WERE THE FINDINGS?

The categorizations of the AEB pulses show few differences between pulses: When all AEB pulses (contact and no contact) were considered for the machine learning cluster analysis, only three clusters were identified. However, those three clusters show statistically significant differences between their pulse characteristics, such as ramp-time, jerk, and maximum deceleration, suggesting categorical grouping of pulses can be achieved. Additionally, we found that contact between the vehicle and the target decreased dramatically in more recent model vehicles. In AEB tests from 2013, 60% showed contact between the vehicle and the target. In 2019 tests, only 1.7% of vehicles contacted the target. This indicates that AEB technology has improved significantly through the years, so much so that the likelihood of contact between the vehicle and the target is low when AEB is present.

WHAT ARE THE IMPLICATIONS FOR INDUSTRY AND WHAT’S NEXT?

While there was a significant decrease in contact during testing when AEB was present in more recent model year vehicles, it is unclear how or if occupant motion is affected by the differences in AEB pulses. Because that determination can only be made via laboratory testing, an important next step in developing AEB systems and standards should be to examine not only the efficacy of the AEB, but also the way the braking is achieved to improve occupant safety, particularly for children.

Students

Madeline Griffith, University of Pennsylvania; Mikayla Schneider, University of Notre Dame

IAB Mentors

Shawn Sinclair, Consumer Reports; Yi Glaser, General Motors Holdings LLC; Mark LaPlante, Graco Children’s Products Inc.; Jerry Wang, Humanetics Innovative Solutions Inc.; Jason Stammen, National Highway Traffic Safety Administration; Jason Hallman, Toyota USA; Schuyler St. Lawrence, Toyota USA; Uwe Meissner, Technical Advisor