Vehicle safety group tests 14 partial automated driving systems, none earn a "good" rating

[midian182]

In brief: Partial automated vehicle systems that can perform certain functions are supposed to make our driving experiences simpler and safer, but the results of a new study contradict automakers' claims. Out of the 14 systems that were tested, none was awarded a "good" rating.

The Insurance Institute for Highway Safety (IIHS), an independent non-profit founded in 1959 that is dedicated to reducing deaths, injuries, and property damage from motor vehicle crashes, carried out the study into partial driving automation features.

The features examined in the study aren't advanced driver assist systems like emergency braking. Instead, they cover the likes of adaptive cruise control, cooperative steering, lane-keep assistance, automated lane changing, and systems that sometimes allow a driver to take their hands off the wheel.

Also read: Carmakers are sharing driving habits with insurance companies, unbeknownst to owners

IIHS evaluated the vehicles' features based on driver monitoring, attention reminders, emergency procedures and other aspects of system design, awarding them a score of good, acceptable, marginal, or poor based on their safeguards.

Some examples of what would earn a good safeguard rating include monitoring a driver's gaze and hand position, using multiple types of escalating alerts to get a driver's attention, and automated lane changes being initiated or confirmed by the driver. Not a single one of the 14 systems was awarded a good rating: Only one was rated as acceptable, two were classed as marginal, and the rest were "poor."

The only systems not rated as poor

BMW, Ford, General Motors, Genesis, Lexus, Mercedes-Benz, Nissan, Tesla, and Volvo systems were all tested by IIHS. The Lexus Teammate with Advanced Drive system was the only one rated as acceptable, though another one of the company's systems, the Lexus Dynamic Radar Cruise Control with Lane Tracing Assist, received a poor rating. GM's Super Cruise and Nissan's ProPilot Assist were the two systems rated as marginal.

Tesla's Full Self-Driving (Beta) Version 2023.7.10 had the worst rating overall. Its system was rated as poor in six categories and acceptable in the other two. Tesla Autopilot was only marginally better, receiving a good score for lane changing instead of poor.

The "last-minute Halloween costume"

One of the tests involved covering the driver's face with a cheesecloth to see if a system could be activated when a person's face was obscured. IIHS also covered cameras with felt for another assessment, and attached weights to the steering wheel to simulate a driver's hands to try and trick the systems.

What kind of freakish hands would feel like ankle weights?

IIHS President David Harkey said that while the results are "worrying," all the tested systems did well in at least one category, which "means the fixes are readily available and, in some cases, may be accomplished with nothing more than a simple software update."

Permalink to story.

https://www.techspot.com/news/102242-vehicle-safety-group-tests-14-partial-automated-driving.html

 

[Endymio]

"One of the tests involved covering the driver's face with a cheesecloth to see if a system could be activated when a person's face was obscured...."

Yes, this is a far more important than features like emergency braking and lane-keeping.

/smh.

 

[Uncle Al]

No surprise about Tesla .... he's great at talking but always falls short in performance .....

 

[Theinsanegamer]

The big issue with these systems is requiring go drivers to focus on driving while not letting them drive. That is SO BORING. Have you ever tried to pay attention to something you have 0 input on? Your mind wanders constantly, because that is really hard to actually do.

With younger generations having vine brain, that will never work.

 

[Crunch]

I want the car with the most self driving and the least driver nagging. Looks like the test rewards driver nagging.

 

[maxxcool7421]

The issue is these are cars that people drive and then expect it to self drive well. you cannot have both.

Build a car from the ground up that only self drives and code / build around that and it will do much better.

Lastly .. until every car can connects to a mesh network to KNOW about other cars besides using totally fallible cameras this will never be eliminated as a problem.. but as someone who supports security products .. that is terrifying thought to connect a bunch of 6 thousand pound coffins wizzing around at 75mph just waiting for the 1st mesh exploit to kill a few hundred people.

 

[toooooot]

The issue is these are cars that people drive and then expect it to self drive well. you cannot have both.

Build a car from the ground up that only self drives and code / build around that and it will do much better.

Lastly .. until every car can connects to a mesh network to KNOW about other cars besides using totally fallible cameras this will never be eliminated as a problem.. but as someone who supports security products .. that is terrifying thought to connect a bunch of 6 thousand pound coffins wizzing around at 75mph just waiting for the 1st mesh exploit to kill a few hundred people.

I think code would do much more for safety than the tech itself. Good thought and planning first, and then everything built with that in mind next

 

[human7]

I want the car with the most self driving and the least driver nagging. Looks like the test rewards driver nagging.

Agreed. The tests have an agenda behind them. The aim is safety. But while the IIHS likes to claim that there is little to no data behind the auto industry's claims - that these automated systems improve safety - there is also little to no data that the tests the IIHS conducted will improve safety, either.

If a driver is actively tricking the car's systems to perform a task in an automated way that may or may not be unsafe, the problem is not the car. And the IIHS, because the only thing it can really do is judge cars (not people), will boo the cars for not catching on. And there is the agenda: the cars should tattle on and prevent bad drivers from driving, and these automated systems are a way to do that.

I read an article about this over on The Verge (https://www.theverge.com/2024/3/12/24098394/iihs-partial-automated-test-rank-ford-gm-tesla), which provided more detail into these tests, and they said this:

"The reasons were myriad, but overall the systems that were rated poor were found to be easily tricked and bad at monitoring driver attention. Some would work even when the driver wasn’t wearing a seatbelt."

I don't think the IIHS has gotten the memo: the whole point of automation is to remove the need for driver attention in the first place. So here we are, in a place where the technology isn't quite there yet to remove the need for driver attention completely, but it is advanced enough to remove some of that need. The IIHS would rather the tech be a nanny.

If a car company wants to take it on themselves and design their tech that way, fine. But these tests aren't truly measuring the capacity of cars to drive themselves. And that's what the general populace really gives a damn about, not if a cheesecloth or a covid facemask or a glance at a smartphone or whatever makes the assisted tech go bonkers and stops it from doing its job.

 

[whateversa]

Much of the safety features on my Ford Fusion is rubbish. It has tried to steer me out of my lane many times in the past on well painted roads - so I turned that feature off, it has not helped me once in years. It sometimes says my hands are not on the steering wheel while it is - its caused more danger to me as I look away from the road to see what the warning message is... its even happened on a highway, causing distracted driving. Then its warning alarm thing (where it thinks there is going to be an accident) has also never been helpful. I wonder how much extra I paid for that crap.