You can learn a lot from a dummy, but the auto industry's standardized—and federally mandated—crash test dummies are left wanting. Biological engineers often find it difficult to use them to model body blows coming from certain directions or to predict trauma to areas such as the lumbar spine and abdomen. To make a more accurate, responsive model of human injury, nearly two dozen automakers and research institutes have set out together to build a digital complement: an elaborate, 3-D computer model depicting bone, tissue and internal organs from head to toe. To date, the group, known as the Global Human Body Models Consortium (GHBMC), has created a 173-pound adult model and continues to make updates for a broader range of body types and scenarios. The effort comes at a time when U.S. traffic safety administrators report an 8.1 percent increase in traffic fatalities for the first six months of 2015, compared with 2014 data from the same period—the largest increase year over year since 1977.

All Shapes and Sizes

Most automobile safety features were designed with the average adult male in mind, which has left children in particular not as well-protected. From an engineering standpoint, the prepubescent body has different mechanical properties, so one goal of the consortium is to make anthropomorphic details in the model account for kid-sized bodies, as well as other variations in sex and age. A similar model for 10-year-olds is already helping postdoctoral researchers Anil Kalra and Ming Shen of Wayne State University develop new bumpers that make young pedestrians struck by vehicles more likely to land on their hands and feet than on their head.

Injury Prediction

In the event of a crash, General Motors vehicles with the OnStar telematics system already collect collision data, including force and direction of impact, to calculate the likelihood of severe injury to passengers. The system then sends this prediction to emergency responders. J. T. Wang, an engineer at GM and a lead technical adviser to the GHBMC, speculates that the virtual-body model may eventually run fast enough to create real-time simulations that enable vehicles with such systems to give a more specific picture of the crash scene. “Even before the emergency team arrives at the site, we may be able to predict the kind of injuries,” Wang says. This information could also provide paramedics with crucial clues about the underlying conditions afflicting crash victims who are found unconscious.

Driverless Cars

The anticipated widespread elimination of human drivers (and hence human error) will not likely eliminate injuries. As such, Jingwen Hu, a research scientist at the University of Michigan's Transportation Research Institute, suspects the GHBMC models will be instrumental in fine-tuning the safety features of autonomous cars, which have different operating mechanisms than cars today and may sport unconventional seating configurations. For instance, in a recent paper funded by Ford Motor Company, Hu found that the braking of vehicles with driver-assist technology just before a potential collision tends to alter a front-seat rider's posture. As a result, the skull tilts forward—which could increase the risk of head injuries. The models could also soon help engineers reevaluate the operation of seat belts and air bags in this new context.