When harm comes from safety tech, what to do?

Among safety features being rolled out in new vehicles, automated emergency braking may hold the most potential to reduce injuries and save lives.

These systems use cameras and radar to detect obstacles in the path ahead — and deploy the brakes to avert collisions. But last month, federal safety officials opened a probe into a system installed on more than 553,000 Nissan Rogues from the 2017 and 2018 model years that, according to complaints filed by consumers, are slamming on the brakes even when no obstacles exist.

More than 800 motorists have lodged complaints about false positive incidents with Nissan Motor Co. and NHTSA.

Fourteen crashes have been attributed to the braking malfunctions, resulting in five injuries. Those numbers perhaps pale compared with the benefits of automated emergency braking systems: In February, the Insurance Institute for Highway Safety reported that automatic emergency braking "reduces the frequency of property damage liability claims by 13 percent, rates of rear-end crashes by 50 percent and rear-end crashes involving injuries by 56 percent."

"You have technology as a way to prevent these crashes, but then it fails in new ways that we need to understand," said Sean Kane, founder of Safety Research & Strategies. "It's complex, and there are no real answers."

Safety advocates are concerned about the prospect of false positive incidents mushrooming. In 2016, Nissan and 19 other automakers reached a voluntary agreement with NHTSA to make automated emergency braking a standard feature on all new vehicles by September 2022, raising the probability that millions more of these systems will soon be on the road, at a time when the problem of false positives is just beginning to be understood.

"There's a real concern that this issue is emblematic of a bigger picture for driver-assist systems, whether it's AEB or some other features that have such tremendous potential," said Jason Levine, executive director of the Center for Auto Safety, which petitioned NHTSA to open a preliminary evaluation of the Rogues in March. "But because of the haphazard way in which they're being introduced and the total laissez faire nature of NHTSA's oversight, there's a real risk of undermining public confidence in these safety features and technology."

Concerns are already proliferating among consumers.

In the complaints to federal safety officials, patterns emerge. Many drivers reported multiple instances of phantom braking and expressed fears of being crushed after abrupt stops on highways as well as a desire to turn off or remove safety systems for which they often paid a premium.

"At one time on the highway, it almost caused an accident due to sudden and abrupt, unneeded braking, slowing the car unexpectedly, nearly causing cars behind me to run into me. This vehicle behavior has occurred multiple times," one motorist from Grosse Pointe Park, Mich., wrote in a complaint filed with NHTSA on March 5.

Another driver, from Zebulon, N.C., wrote on Feb. 26 that "I was told the issue was fixed. Then the other day, it happened again. I was driving at 60 mph and it just breaked (sic). Nothing in front of me or beside me. I have asked Nissan to return the car because I do not feel safe driving it."

While some consumers reported the incidents occurred either on highways or at highway speeds, others noted that specific types of infrastructure, such as bridges, railroad tracks and parking garages, appeared to be triggers.

Axel Nix, former senior director of Harman International's advanced driver-assist system division, says these are widely known throughout the industry as potential triggers for false positives.

"Parking garages are somewhat notorious because if you are on a ramp, there's a lot of metal in the ceiling, and a radar sensor might think you're about to drive into a steel wall," he said. "For us, we say, 'Of course, I'm on a ramp driving up.' "

Cameras can counter the inability of radar sensors to distinguish between metal that poses a threat and metal in surrounding infrastructure, but they also have their drawbacks. For example, AEB systems might initiate braking if a camera sees steam rising from a manhole cover, while a radar sensor would perceive no threat.

Sensors can offer redundancy and counterbalance each other — initiating an automatic brake, for instance, only if both sensors are in agreement that an obstacle exists. Fusing the data in nuanced software can be an expensive proposition for automakers that seek to add emergency braking features.

Nix said there's essentially a spectrum on which these systems can be calibrated. Engineers can determine their sensitivity. That is, they can choose to reduce instances of false negatives — situations in which real obstacles might potentially be missed — by making the systems more responsive to perceived threats, which in turn makes them susceptible to false positives. Or vice versa.

Finding a balance depends on the philosophy of individual automakers and suppliers.

"Our theory was that we looked very closely to see how dangerous it would be to have a false positive," he said. "Hey, if we go into full emergency braking but only at less than 50 kilometers an hour, maybe that's an OK trade-off to brake for a ghost that isn't there. On the other hand, if you have a braking at 150 km, we might limit that braking because that risk equation is very different."

The Rogue evaluation isn't the industry's first brush with problems associated with false positives in automated emergency braking systems. It isn't even Nissan's first encounter with them.

Based on two complaints associated with inappropriate braking, NHTSA opened a preliminary evaluation in July 2012 into the Intelligent Brake Assist feature on the 2013 Infiniti JX35. In one complaint, a pregnant motorist driving with her 7-year-old son said her leased vehicle came to a "dead stop" on a bridge in Ocean Township, N.J.

After talking to a mechanic at her local dealership, she wrote to NHTSA, "I am told that this particular bridge is the only culprit in the entire country because it has too much metal."

In the ensuing federal evaluation, Nissan told federal safety officials that "unique road environment anomalies" may cause brake activations even when no threat is imminent. Investigators replicated the problem during their four-month probe but nonetheless concluded the braking interventions were "extremely brief and may interrupt travelling speeds, but do not stop the vehicle."

NHTSA said the inappropriate braking posed a "low risk to motor-vehicle safety." Investigators were satisfied a customer service campaign conducted by Nissan, which included a software reflash of the electronic control units that govern radar and Intelligent Brake Assist functions, would fix the problem.

It is unclear how the automated emergency brake system design evolved or changed from the 2013 Infiniti JX35 to the 2017 and 2018 Nissan Rogues. Documents from both the 2012 and ongoing 2019 federal evaluations show a matching 284E7 prefix on parts numbers associated with the driver-assist feature. It is not yet known how — or whether — the parts were modified in the interim. Parts numbers from the radar units change from 28437 to 28438 between the two investigations.

“Nissan is committed to the safety and security of our customers and their passengers," the automaker said in a statement. "Nissan has investigated this issue extensively and, in consultation with NHTSA, launched field actions notifying affected customers of a software update that improves MY17-18 Rogue AEB/FEB system performance.

"Customers are invited to bring their vehicle to an authorized Nissan dealership where the update will be applied at no cost to the customer.  As always, Nissan will continue to work collaboratively with NHTSA on all matters of product safety.”

With regard to the reported problems with the Rogue, Nissan has sent seven communications, including a Technical Service Bulletin, to dealers with instructions on how to reflash the ECUs associated with the radar and driver-assist functions.

That guidance started coming out Sept. 6, 2016. The most recent came Aug. 17, 2018, informing dealers of a customer service campaign to update the software. But some wonder why the automaker hasn't issued a recall of the affected Rogues to ensure they receive the fix.

"At least here, we have an example of a fix that can be applied, hopefully," Levine said. "I think that NHTSA, if they are not writing a standard, they could at least set the bar that an AEB defect is a safety concern that requires a recall. At least you'd have a floor to work from that will hopefully enhance safety going forward."

NHTSA has championed its voluntary agreement with automakers as a means to usher in an era of automated braking faster than if the agency had proposed regulations and gone through a yearslong rule-making process. On the flip side, there are no standards that determine how these automated emergency braking systems should perform.

"Nissan's not alone," Kane said. "We see a fair number of complaints across a variety of manufacturers. I see these problems creeping up. They'll get better over time, and the technology will improve. But for now, it's a question of, 'How do you put these technologies in low-cost cars?' and it's a race to the bottom."

Issues with automated emergency braking systems may provide a glimpse of future complications, as fully automated vehicles rely on some of the same sensors and fusion software.

"The AEB issue is one microcosm of a larger set of issues that's sitting out there for the industry to drive with no real oversight, and at the same time, it has this significant effect," Kane said. "The bottom line is, these new technologies that come out, when you don't set some baseline regulations that ensure the safety of the underlying technology, you end up with problems."