Much of the promise behind automated and semi-automated vehicles is the promise of safety. In theory, these systems are better drivers than humans, who are prone to all sorts of inattentiveness, rule-breaking and poor judgment. But is that theory a reality?
Not yet, according to the traffic safety group AAA. For the second time since 2018, the group has been testing driver-assistance systems in new vehicles to see if they are actually safer than human drivers. It tested all the high-end models and found they are far more prone to error than people might suppose.
This year, the researchers tested five auto makers’ systems over a distance of 4,000 miles. On average, the cars ran into trouble once every eight miles.
Most problems involved the lane-keeping systems in these vehicles, which are all semi-autonomous. There were also issues with the driver-assist systems cutting out with little warning, leaving inattentive drivers potentially in an emergency.
Worst of all, however, two-thirds of the time, the vehicles failed to notice disabled vehicles in their path and struck them at an average of 25 mph.
This was little better than the results in 2018, when AAA recommended that automakers stop including the technology on additional models.
“AAA has repeatedly found that active driving assistance systems do not perform consistently, especially in real-world scenarios,” said a AAA spokesperson. “Automakers need to work toward more dependable technology including improving lane keeping assistance.”
The models AAA evaluated
This year, AAA tested five models with semi-autonomous features, often called driver assistance systems:
- 2019 BMW X7 SUV with the Active Driving Assistant Professional system
- 2019 Cadillac CT6 sedan with the Super Cruise system
- 2019 Ford Edge SUV with the Co-Pilot 360 system
- 2020 Kia Telluride SUV with the Highway Driving Assist system
- 020 Subaru Outback SUV with the EyeSight system
In 2018, AAA also tested a 2017 Tesla Model S with the Autopilot system, but no Tesla was in this year’s lineup.
Each system had trouble with lane-keeping, often coming too close to other vehicles or to guardrails. The Kia, the BMW and the Subaru all had trouble spotting a simulated disabled vehicle most of the time. However, the Ford and Cadillac systems couldn’t be tested on the track because they only work on divided highways. Additionally, the systems often shut down with barely a notification to the driver.
These systems are becoming more mainstream and that could be a problem
Despite AAA having sounded the warning in 2018, automakers have continued to expand the models that include these driver assistance systems. Unfortunately, that could be more problematic than is immediately obvious.
This is because those who already own vehicles with semi-autonomous properties are often early adapters of technology, who tend to have a fuller understanding of what the systems are capable of. Unlike those without as much technologicial know-how, they may know not to fully rely on the driver assistance.
As the systems become ubiquitous, however, people may buy vehicles without realizing the dangers of relying on the driver assistance systems. They may just assume that the systems work like an autopilot, which they do not.
In response to AAA’s findings, each of the five manufacturers defended that drivers must always be fully in control of their vehicles. Driver assistance systems are not meant to take over for drivers; merely to assist them. AAA notes that most of the owners’ manuals actually note that the driver assistance system may not notice stationary objects like stalled cars.
BMW, for example, stated that the X7’s Active Driving Assistant Professional system performed as designed.
If the systems, performing as designed, cannot keep within a lane and cannot avoid hitting parked vehicles, are they safe enough to be used at all?