IIHS-HLDI test drives uncover driver assistance system quirks

November 10, 2016

Hundreds of vehicles make their way through the IIHS Vehicle Research Center each year, and most are crash-tested and evaluated for headlight, front crash prevention and LATCH ratings and then released. Since March, a pool of vehicles outfitted with automated technologies has been traveling back roads and city streets as part of an internal test-drive program.

IIHS and HLDI staff drive select vehicles on their regular commutes and longer trips for several days to weeks and complete questionnaires about their experiences. Models include a 2017 Audi A4, 2017 Audi Q7, 2016 Honda Civic, 2016 Infiniti QX60, 2016 Tesla Model S and 2016 Toyota Prius.

"The main goal is to collect data on what people think of these technologies and what kinds of driving scenarios the systems have trouble handling," says David Kidd, who runs the program with Ian Reagan. Both specialize in human factors research.

"The findings will help us shape future research and test programs," Reagan says. "So far, staff experiences have varied."

Employees reported instances where vehicle automation didn't perform as expected and also cases where they overrode the system in what they considered to be potentially unsafe situations. For example, some drivers reported that adaptive cruise control braked too late for their liking or at times they deemed unnecessary.

On driver trust, side-view assist scored the highest, while lane-keeping assistance scored the lowest. Systems that center vehicles in their lanes were a common source of frustration. Some drivers felt they had to fight the steering wheel when trying to steer outside of the marked travel lane in construction zones or when another vehicle encroached into their lane.

"These systems are immature," Kidd says. "They can't improvise or adapt to normal changes in the driving environment like humans can. Experienced drivers know they might need to drive onto the shoulder to avoid an obstacle, for instance, but lane-keeping assistance can't make the same judgement call. It is programmed to keep the vehicle on the road within the lines."

Cresting hills is a challenge. Because cameras that monitor lane markings point upward away from the pavement as the cars make their ascent, they can't "read" lane markings beyond the crest and can drift as systems hunt for the lane, catching an unaware driver off-guard. In some cases, the lane-keeping systems shut off altogether and the driver had to intervene to keep the vehicle in the intended travel lane.

With adaptive cruise control, drivers reported braking instances after the vehicle ahead left the lane or exited the road. Or sensors didn't detect slow-moving or stopped vehicles or ones that abruptly cut in front, and drivers had to quickly apply the brakes themselves. One system reads speed limit signs and adjusts accordingly, a feature that takes getting used to, especially on highway exit ramps.

"There have been a few uh-oh moments," Reagan says. "We are careful to go over each feature with our drivers, but there's no substitute for on-road experience. One of the things we plan on doing is try to replicate the scenarios staff have flagged. Information like that will be useful in developing future ratings programs."

Did IIHS-HLDI staff trust the driver assistance systems they tried out?

Side-view assist ranked first in trust

What about adaptive cruise control?

Honda Civic system scored lowest in trust

Media contact

Email Joe Young
Director of Media Relations
office +1 434 985 9244
mobile +1 504 641 0491

For more information, visit our press room.