IIHS-HLDI research finds little evidence that partial automation prevents crashes

Crash records and insurance data offer little evidence that partial automation systems are preventing collisions, research from the Insurance Institute for Highway Safety and the Highway Loss Data Institute shows.

“Everything we’re seeing tells us that partial automation is a convenience feature like power windows or heated seats rather than a safety technology,” IIHS President David Harkey said.

The clearest evidence so far comes from studies of BMW and Nissan vehicles that have been on the road for a number of years. HLDI studied these vehicles in 2021. Now a new study of the same vehicles from IIHS confirms that partial automation in these vehicles doesn’t confer additional safety benefits beyond those of crash avoidance features like front automatic emergency braking (AEB).

More than half the new models for sale in 2023 were available with partial automation systems as an optional or standard feature, despite mounting concern over a series of high-profile crashes that occurred while drivers were using the technology. Using cameras and other sensors, these systems can keep your car moving down the road in the center of the lane, navigating curves, slowing down to avoid other vehicles and then accelerating again when the way is clear.

But vehicles equipped with these systems are far from self-driving. They can’t manage many routine roadway features and traffic situations, so drivers have to pay close attention to what’s happening on the road and be ready to take over at any time. That’s a big challenge because the technology can encourage a false sense of security and induce boredom, causing drivers to tune out.

There’s a key difference between partial automation systems and the crash avoidance features that are usually included with them but also sold separately.

Crash avoidance features like AEB, blind spot warning and lane departure prevention only come into play when a potential danger arises — slamming on the brakes to avoid rear-ending another vehicle, for example. Because they’re unobtrusive under normal circumstances, most drivers who use them leave these features switched on all the time.

In contrast, a partial automation system works constantly to keep the vehicle in the desired position on the road. Intended for use on highways and other limited-access roads, such systems must be switched on whenever the driver wants to use them. Most drivers do so only occasionally.

Using insurance claims data, HLDI has conducted multiple investigations into the potential safety benefits of crash avoidance features. Broadly speaking, the studies have all shown that features that warn or intervene in an emergency reduce the frequency of insurance claims, and the reductions increase incrementally as one feature is stacked on another.

Partial automation could also theoretically help prevent crashes. One of its component features, adaptive cruise control (ACC), keeps the vehicle traveling at a driver-selected speed when the road is clear and slows and accelerates to maintain a set distance from vehicles ahead. It is associated with longer following distances, less tailgating and fewer lane changes — positive driving behaviors that could reduce risk. The other main component of partial automation, lane centering, could potentially do a better job in preventing side-swipe and run-off-road crashes than lane departure prevention, since lane centering theoretically would preempt such departures rather than intervening as they occur.

So far, there’s little evidence that’s happening, the studies of BMW and Nissan vehicles show.

HLDI found that property damage liability claims — which are for damage to other vehicles hit by the insured driver — were 8% lower for 2017-19 Nissan Rogues equipped with forward collision warning and AEB. However, there was no additional benefit associated with ACC or Nissan’s ProPILOT Assist partial automation system, which adds lane centering on top of ACC. Changes in claim rates under collision coverage — which is for damage to the insured driver’s own vehicle — were small for all the technologies.

Similarly, forward collision warning and AEB were associated with a 7% reduction in collision claim rates and a 13% reduction in property damage liability claim rates for 2013-17 BMW and Mini vehicles. BMWs and Minis that were also equipped with ACC showed a larger, 25% reduction in property damage claims and no greater change in collision claims. As with the Nissan vehicles, there were no additional statistically significant reductions associated with BMW’s Driving Assistant Plus partial automation system.

HLDI’s claims data, collected from insurers representing 85% of the private passenger vehicles in the U.S., don’t show whether the partial automation system was switched on during a crash, nor do they include the type of road where the insurance claims occurred. That means any potential benefits from partial automation, which is generally designed to be used on high-speed roads, would be diluted by the large volume of insurance claims for low-speed fender benders.

Jessica Cicchino, senior vice president for research at IIHS, tried to determine if such safety benefits might be hiding in the HLDI data. She compared police-reported crash rates for the same BMW and Nissan vehicles that HLDI studied in 17 U.S. states during 2013-22. Although she also had no way of knowing whether the features were switched on at the time of the crash, she was able to restrict her study to the front-to-rear and lane departure crashes that partial automation could potentially prevent. She looked at crashes on limited-access interstates, freeways and expressways and then looked separately at crashes on other roads.

Like HLDI, she found substantial reductions in crash rates associated with crash avoidance features.

Front-to-rear crash rates were 49% lower for Rogues with forward collision warning and AEB and 54% lower for Rogues with forward collision warning, AEB and ACC than for vehicles with no crash avoidance features. There was no significant effect on lane departure crash rates from lane departure prevention.

Unlike HLDI, Cicchino found larger reductions associated with partial automation. Front-to-rear crash rates were 62% lower for Rogues with ProPILOT Assist than for vehicles without any crash avoidance systems. Lane departure crash rates were 44% lower for Rogues with ProPILOT Assist and lane departure prevention than for unequipped vehicles.

When she looked into those numbers more deeply, however, she found that the apparent benefits from ProPILOT Assist were the same on high-speed roads where IIHS research shows partial automation is most likely to be switched on and low-speed roads where the added convenience it provides is minimal at best. Below 37 mph, in fact, ProPILOT Assist’s lane-centering feature only works if you are following another vehicle. That suggests that other characteristics of the equipped vehicles or their drivers were responsible for the reduction.

One such characteristic may be the vehicles’ headlights. Noting that Rogues of that vintage were equipped with poor-rated headlights unless the buyer opted for a premium package, Cicchino turned to the crash records again and discovered that the rate differences were greatest in the dark.

From Nissan marketing materials, she ascertained that the 2018 and 2019 Rogues with ProPILOT Assist were more likely than unequipped vehicles to have acceptable-rated headlights. Acceptable-rated headlights themselves reduce single-vehicle nighttime crashes by about 15%, compared with poor-rated ones, other IIHS research has shown.

For the BMW vehicles, Cicchino examined only lane departure crashes, because the vehicles with partial automation came with a more advanced front crash prevention system than those without partial automation, making it impossible to isolate the effect of the partial automation system on front-to-rear crashes. She found that neither lane departure prevention alone nor the same feature combined with partial automation had a significant effect on crash rates, either on limited-access highways or on roads with lower speed limits.

The vehicles in these studies range from five to 11 years old, and it’s possible that newer partial automation systems are more effective from a safety perspective. On the other hand, the many years of data that have accumulated for these vehicles make the findings more compelling.

“With no clear evidence that partial automation is preventing crashes, users and regulators alike should not confuse it for a safety feature,” Cicchino said. “At a minimum, safeguards like those IIHS promotes through its rating program are essential to reduce the risks that drivers will zone out or engage in other distracting activities while partial automation is switched on.”