Tesla Autopilot and similar automated driving systems get ‘poor’ rating from prominent safety group

[fholbert]

The organization also rated hands-free highway driving systems from Ford and Nissan as “poor.” General Motors’ hands-free system, Super Cruise, was rated as “marginal.” Only Lexus’s Teammate with Advanced Drive system received a rating of “acceptable.” Even that rating, though, is still one step below the Insurance Institute’s highest possible rating of good.

“Some drivers may feel that partial automation makes long drives easier, but there is little evidence it makes driving safer,” Insurance Institute president David Harkey said in a statement. “As many high-profile crashes have illustrated, it can introduce new risks when systems lack appropriate safeguards.”

Story here: https://www.cnn.com/2024/03/11/cars/insurance-group-rates-tesla-autopilot-safety/index.html

 

[RTPEV]

I hesitate to comment here because it boils down to religious arguments, but here's where I stand on this:

These driving assistance aids can make driving safer if used properly. The problem is that most people use them improperly, which actually makes them more dangerous than not using them at all. And maybe "driving assistance" is the wrong term. As the quote says: "drivers may feel that partial automation makes long drives easier" and that may be true, but if they were thought of as "trouble avoidance aids" instead, maybe people would not over-rely on the automation.

 

[SidetrackedSue]

I'm old enough to remember stories like these about basic cruise control especially before TACC was a thing. No denying CC is a useful tool (especially for me because I have a heavy foot in great weather on an open road) but it needed to be used properly and with attention.

I also agree that there are very few if any situations where AP or FSD make driving safer than me driving myself especially since I strongly value defensive driving techniques of:

• travelling as close to the speed limit as I can (I won't set TACC to more than 10% higher than the limit), driving below the speed limit if conditions warrant it,
  
  
• leaving multiple car lengths in front of me at all times (2- 3 for stop and go traffic, 7+ on highways,
  
  
• starting to slow when I see a red light ahead when doing highway driving (and most city driving) and experiencing the joy of the light turning green for me before I get to the intersection so I don't have to start up from a full stop,
  
  
• choosing my lane carefully on multilane roads to eliminate last minute lane changes,
  
  
• signaling before I make a lane change,
  
  
• driving in the right lane for most 2 lane highway driving (obviously because I'm only going 10% over the limit) but moving into the next lane over if, and only if, there is a car merging from the on-ramp. If I'm in the middle lane because right lane is for semis, and there's a truck slightly in front of me or beside me, I'll pull into the left lane in order to give the truck room to move over if it is cut off by the merging car.
  

While I find AP's lane keep on highways less stressful at the end of hours of easy highway driving, my defensive driving techniques (especially getting in the right lane and letting others pass me) also make long drives less stressful than trying to squeeze seconds or perhaps minutes off my drive. It becomes a 'wash' for me: neck and shoulders are more relaxed at the end of drive because of less tension in steering, vs quieter brain because I didn't have to constantly on edge monitoring the car's behaviour, prepared to stop an improper dive into a right turn lane when I'm going straight, wondering if the car is actually going to stop for the stopped cars in front of me, and dealing with wipers that dry swipe when it isn't raining and won't turn on and/or adjust to the volume of water when it is raining.

In my ICE days my defensive driving would save me money by lowering my gas consumption and the same is true on EV road trips since I'd need to charge less. I also suspect it saves money on tires. But the big savings come from avoiding traffic tickets and collisions for 45 years (and counting). I've saved money on fines, insurance, and the cost of repairs/replacement. Avoiding all those things saves me time too. The time a traffic stop takes wipes out any time saved by driving way over the speed limit, not to mention how much time is saved not waiting for tow trucks and dealing with collision centres and arranging vehicle replacement while the car is being repaired.

 

[fholbert]

I hesitate to comment here because it boils down to religious arguments, but here's where I stand on this:

These driving assistance aids can make driving safer if used properly. The problem is that most people use them improperly, which actually makes them more dangerous than not using them at all. And maybe "driving assistance" is the wrong term. As the quote says: "drivers may feel that partial automation makes long drives easier" and that may be true, but if they were thought of as "trouble avoidance aids" instead, maybe people would not over-rely on the automation.

I could not agree more.

Some on this forum have complained about being unable to text or email without the auto pilot disengaging and giving them a strike. Safety of other drivers and pedestrians takes a back seat to their email needs. WTF is wrong with them?

 

[Xenoilphobe]

The organization also rated hands-free highway driving systems from Ford and Nissan as “poor.” General Motors’ hands-free system, Super Cruise, was rated as “marginal.” Only Lexus’s Teammate with Advanced Drive system received a rating of “acceptable.” Even that rating, though, is still one step below the Insurance Institute’s highest possible rating of good.

“Some drivers may feel that partial automation makes long drives easier, but there is little evidence it makes driving safer,” Insurance Institute president David Harkey said in a statement. “As many high-profile crashes have illustrated, it can introduce new risks when systems lack appropriate safeguards.”

Story here: https://www.cnn.com/2024/03/11/cars/insurance-group-rates-tesla-autopilot-safety/index.html

Is there a link not on CNN, I don't want to give them any ad revenue. Nevermind I found it https://www.marketwatch.com/story/t...or-ratings-from-auto-insurance-group-05551b52

 

[3sr+buyer]

Why not go to the primary source (IIHS) if you want to know what they say?

First partial driving automation safeguard ratings show industry has work to do

Only one of 14 partial driving automation systems evaluated earns an acceptable rating for its safeguards designed to prevent misuse and lapses of attention.

www.iihs.org

Partial automation safeguard ratings

The Insurance Institute for Highway Safety (IIHS) is an independent, nonprofit scientific and educational organization dedicated to reducing deaths, injuries and property damage from motor vehicle crashes through research and evaluation and through education of consumers, policymakers and safety...

www.iihs.org

 

[DanCar]

What about Tesla's published safety reports that seem to indicate using autopilot is much safer than not?

 

[B_E_V]

The CNN and Market Watch articles are essentially rewriting a press release from the Insurance Institute for Highway Safety (IIHS).

Here are links to the press release First partial driving automation safeguard ratings show industry has work to do.

Their initial ratings Partial automation safeguard ratings.

Finally their Advanced driver assistance page.

Their FSD rating predates the recent driver monitoring recall, and was based on "Tesla Full Self-Driving (Beta), Version 2023.7.10 in a 2021-23 Tesla Model 3." A version after the recall might not have scored Poor for "Driver Monitoring" which according to IIHS should "Monitors both the driver’s gaze and hand position."

• Tesla FSD scored Acceptable for "Attention Reminders," and "Emergency Procedures" (unresponsive driver).
• Tesla FSD scored Poor for "Lane Change" because "Automated lane changes must be initiated or confirmed by the driver."
• Tesla FSD scored Poor for "ACC Resume" because "Adaptive cruise control does not automatically resume after a lengthy stop or if the driver is not looking at the road" is the requirement.
• Tesla FSD scored Poor for "Cooperative steering" because "Lane centering does not discourage steering by driver" is the requirement, and Tesla disengages if the driver steers.
• Tesla FSD scored Poor for "Safety Features" because the requirements are "Automation features cannot be used with seat belt unfastened" and "Automation features cannot be used with automatic emergency braking or lane departure prevention/warning disabled."

The Autopilot rating was based on "Tesla Autopilot, Version 2023.7.10 2021-23 Tesla Model 3" again before the recall.

• "Attention Reminders" was rated Poor instead of FSD's Acceptable rating.
• "Lane Change" was rated Good instead of Poor because Autopilot does not change lanes on its own, while FSD certainly does.

The "Lane Change," "ACC Resume," and "Cooperative Steering" categories are all justified by IIHS as maintaining "Driver Involvement." While in principle I agree that keeping the Driver Involved is a good thing, I disagree that restricting functionality is a good way to achieve that goal. Though in IIHS's defense, these ratings are intended for LEVEL TWO driver assistance systems.

 

[Xenoilphobe]

Once I get an insurance discount of 50%, I will buy FSD, until then.. no way...

 

[DDotJ]

• Tesla FSD scored Poor for "Cooperative steering" because "Lane centering does not discourage steering by driver" is the requirement, and Tesla disengages if the driver steers.

I've seen arguments for this type of system from a comfort/convenience standpoint. I see that IIHS argues that "The lane-centering feature should not switch off automatically when the driver makes manual steering adjustments within the lane, as that can discourage drivers from being physically involved in the driving, and physical involvement can help prevent mental disengagement" but I disagree.

I like the way Tesla approaches it as it is much clearer who is in control. It's binary, either the car has control or the driver with no ambiguity between the two and a loud audible alert when the responsibility is shifted in either direction. I don't agree that it is a safety issue, or that this makes it any less safe than the other systems.

Adaptive cruise control does not automatically resume after a lengthy stop or if the driver is not looking at the road

This one is fair for TACC, and seems relatively easy to implement.