If you’re anything other than a mid-sized adult male or a small, lightweight adult woman, beware. Your senior citizen ribs, massive toddler head, or full-gear, military uniformed body has never been crash tested.
“With 33,000 people in the U.S. dying last year in collisions, there’s a disconnect between crash test results and what underserved people are experiencing in accidents,” says Matt Reed, Head of the Biosciences Group at the University of Michigan Transportation Research Institute (UMTRI).
To back up his talk, Reed and his team of researchers are employing unlikely tools. A $220 dollar scanning device, designed to record depth information and commonly used by videogame creators, is capturing body shapes in real time, allowing UMTRI to build virtual humans for vehicle assessment.
“For decades, car companies have been able to test engines and [inanimate] objects. Our work is making body shape and posture of vehicle occupants more realistic,” says Reed.
In addition to a $50,000 handheld scanner that measures 60 landmarks on a test crash model and the $120,000 laser scanner that captures 500,000 points in 12 seconds, these faster, highly detailed scans are changing the world of crash test modeling.
“In federal rules tests, there are just two dummies tested,” Reed reveals. “And manufacturers are using computer simulations that are effective in predicting what will happen in a dummy test.”
But what about real humans? What about the 75-year-old man with soft ribs and a curved back, whose short-sightedness causes him to sit close to the dash?
Reed says state-of-the-art computational models are tweaking airbags, seatbelt and vehicle designs to accommodate him and the rest of the world’s drivers.
“We’re not just building one model,” Reed tells Wired. “We’re building a representation of a huge range of body shapes and dimensions.”
Zooming in on the dummy models, UMTRI is collecting data on the forces exerted on the body in a crash; acceleration to the head, strain on tissues – some of which are strong enough to break bones.
Unlike crash dummies, which can only be positioned by shoving their hips and heads into limited positions, virtual human models can be morphed into a variety of configurations. Even data-based grasp predictions are possible, when calculating grasp motions for ergonomic applications and modeling virtual shapes.
And unlike crash tests, with their $45,000 base model dummies and $100,000+ fully-loaded-with-technology relatives, (not to mention the test vehicle) virtual testing is economical.
“The cost of a full body scan is a tiny fraction, [relative] to developing a car,” Reed claims.
3-D anthropometry isn’t new. In the late 1990s the U.S. Air Force’s CAESAR project scanned people in three different postures. Reed says the volunteers were not anatomically diverse and included two unsupported seated positions. But both complaints about the collected data pale in comparison to the fact that we haven’t changed testing methodologies for the population in over three decades.
“The last large-scale study of children in vehicles was in 1977. We [now] know about childhood obesity and how the population is changing body size. We hope to redo that 1977 study with modern technology,” he says.
Crash dummies used in the U.S., commonly referred to as the “5th-percentile female,” “50th-percentile male,” and “95th-percentile male,” stand up well to current stature percentiles, but fall further off the mark for the two male models when it comes to mass. Today, they might be renamed “the 33rd- and the 81st-percentile male.”
Already, Reed’s biosciences group has 150 children under observation. The kinetics of a crash involving kids ages 3 through 11 are unique: Their supple necks resist injury unless directly impacted and they suffer far fewer rib fractions than elderly passengers.
And for soldiers, the third underserved demographic his researchers study, Reed says, “The thing that’s unique is that we’re measuring the soldiers in realistic postures and quantifying for the body armor, because of the safety implications.”
UMTRI’s researchers know that real crash tests will always be necessary. But armed with tools to capture the full spectrum of human variability and projecting that data into virtual human models, they plan to better predict and protect the world’s population, saving lives and development dollars in the process.