Skip to Content

A Smarter Kind of Crash Test Dummy

Researchers combine real vehicle, scene, and medical data to simulate car crashes on a supercomputer.
December 3, 2015

One person dies in a road traffic accident every 25 seconds, according to estimates from the World Health Organization. Crash test dummies with accelerometers, force sensors, and strain gauges have helped auto manufacturers keep the death toll from rising.

Researchers at Wake Forest University, however, believe that the mannequins may have reached the limits of their usefulness. For the past five years, the researchers have run thousands of virtual crash simulations, each using data drawn from real-world examples, through a supercomputer. “By simulating real-world crashes, we can study the effect of vehicle design parameters, safety features, and occupant factors and propose solutions that would prevent and mitigate occupant injury,” says Ashley Weaver, assistant professor of biomedical engineering at the university and a key member of the research team.

It’s not a new idea: in the 1930s, researchers pioneered similar tests on human cadavers. Later, volunteers such as John Paul Stapp offered to be live subjects in modest impact tests, while research teams routinely used live pigs to test the effects of more serious collisions.

Anthropomorphic test devices, as these ill-fated mannequins are better known in the industry, provide data on around 20 points on the body. Digital simulations such as the one designed by the Wake Forest team, by contrast, allow researchers to examine the effects of a crash to a far greater degree, testing a variety of body shapes and sizes and different body positions at the moment of impact. The Wake Forest model can quantify the risk of bone fractures and damage to soft tissue and organs, injuries unaccounted for by crash test dummies.

The data is proving invaluable to car manufacturers. “Digital crash dummies [allow us] to determine the best methods to modify vehicle chassis, interiors, seats, headrests, safety belts, dashes, and active safety systems, such as airbags, to improve safety very early in the vehicle-design process,” says Bill Veenhuis, an engineer at Nvidia, which provides commercial hardware for crash simulations to more than a dozen car manufacturers.

The work can save money as well as lives. Figuring out safety improvements before the manufacture of sheet metal and other parts reduces costs later in the design phase. “Testing with actual crash dummies then becomes a method to validate the digital crash dummy tests, instead of discovering energies and deformations very late in the vehicle engineering process,” Veenhuis says. U.S. vehicles must meet federal standards on 35 different tests to ensure sufficient protection in front or side collisions. The more accurate the simulation, the greater the chance the car manufacturer will pass the expensive live crash tests the first time.

Weaver’s research, which has been sponsored by Toyota, differs from current commercial applications in that its data is based on detailed vehicle, scene, and medical data drawn from a database of injury research. Using an advanced digital model that contains no fewer than 1.8 million elements that combine to accurately reproduce the human form, from precise bone strength to the structure of organs, and which is capable of predicting injuries to both soft and bony tissues, the team ran simulations until the model accurately mimicked the effects of different crashes on real-world crash victims. Such complex work is only possible thanks to recent developments in computer hardware power and efficiency. “Two decades ago, a vehicle crash analysis involving a digital crash dummy took two weeks to solve,” says Veenhuis. “Today, we can run a typical front impact study overnight.”

Nevertheless, as with any model, the work required several simplifications and assumptions, explains Weaver. For example, the team had to use a generic, one-size-fits-all vehicle in the simulations, as not enough data exists in the public domain of different vehicles. And while the model is able to simulate the effects of a crash on different human body sizes and up to 140 different positions, it is currently unable to adapt the results according to a passenger’s age or health.

Nevertheless, Weaver is confident that the research will ultimately save lives and reduce the severity of injuries to both bones and organs. “My hope is that the research will provide a cost-effective solution for evaluating new and existing automotive safety features,” she says.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.