Computer model for child pedestrians will help automakers design safer cars
December 19, 2016
Every year, around 5,000 pedestrians in the U.S. are killed in car accidents, and more than a 100,000 more are injured.
And although total traffic-accident fatalities are declining, the percentage involving pedestrians is increasing, perhaps as advanced safety features designed to protect the vehicle’s occupants help reduce fatalities among those inside the car.
A new computational model developed at Virginia Tech will help automakers design cars that can protect pedestrians, as well. The research, published recently in the journal Accident Analysis and Prevention, allows safety engineers to predict the consequences of an impact between a car and a pedestrian and test the effectiveness of potential safety features.
Costin Untaroiu, who led the project, is a research associate professor in the Virginia Tech-Wake Forest School of Biomedical Engineering and Sciences and the Department of Biomedical Engineering and Mechanics in the College of Engineering. The team also includes researchers at the Wake Forest University School of Medicine in Winston-Salem, North Carolina.
The pedestrian model is based on anthropometric data for a 6-year-old child. Motor vehicle accidents are a leading cause of death for children; according to the National Highway Transportation and Safety Administration. Approximately one-fifth of children killed in car accidents in 2013 (the last year for which data are available) were pedestrians.
Regulations for cars sold in the U.S. are starting to include requirements for pedestrian safety, which have been part of European regulations for several years. As a result, automakers, suppliers, and researchers are working together to develop and assess safety features, such as external airbags or softer hood materials, that might reduce the risk to a pedestrian in the event of an accident.
Computational models will play a significant role in that effort.
Physical experiments with crash test dummies are time-consuming to set up and conduct, and don’t always accurately represent a human victim. Observational data collected from real-world accidents is informative, but not necessarily representative.
“The problem with real data is that every data point is different,” Untaroiu said. “Was the person running, what was the gait like, how were his legs positioned — this is unknown data.”
Computational models can give a more complete picture of an accident, and yield results more quickly than physical experiments. More powerful computers are making processing times shorter and shorter.
“When I was a Ph.D. student, one numerical simulation took two weeks,” Untaroiu said. “Now we can do the same simulation in one day. In another five years, it might take one hour.”
Creating a computational model of a child is more complex than simply shrinking a model of an adult. Children have different proportions — for example, their heads are larger relative to their bodies than adults’ — and their bones, which do not fuse completely until adulthood, have different material properties.
To create a realistic model of a 6-year-old, the researchers used anatomical data supplied by the Wake Forest University School of Medicine and published data on the material characteristics of juvenile bones.
Using this physical data, the researchers built a computational model that calculates the motion of more than half a million points, or nodes, on the body during the milliseconds following a simulated crash. The results help predict potential injuries, focusing on large bones of the lower extremities — the most likely to be impacted — and the complex joint at the knee.
This model will be used by a global consortium of automakers collaborating to develop universal computational models that can be used across the industry to test designs for pedestrian safety. Untaroiu has previously developed a computational model of a vehicle occupant used by this group.
The team is also developing three computational models for a variety of adult pedestrians. By running tests with computational models spanning a range of physical properties, automakers can ensure that safety features will protect the broadest possible range of pedestrians.
Untaroiu may also add details to the child pedestrian model, such as internal organs, that could help make it even more accurate.
“You can always improve a model,” he said.
The Virginia Tech-Wake Forest School of Biomedical Engineering and Sciences, housed in the Department of Biomedical Engineering and Mmechanics, is partially supported by the Institute for Critical Technology and Applied Science.