HN Reader

I have a Tesla with AEB that drives me up the wall. I'm often driving in heavy traffic, and occasionally need to quickly, but smoothly, stop within a few feet of the vehicle ahead. My car decides I'm cutting it too close, engages full automatic braking, jerking the car to sudden stop with a 10 foot gap. Sometimes the vehicle ahead will even start moving before my car is stopped, but it has decided to stop and stop it shall. Cannot override even by flooring the accelerator. I miss cars that did what they were told, when they were told to do it. For the record I've never rear-ended anyone.

If you are going to share your anecdotal experience about AEB, please also share your specific vehicle make/model/year.

The entire point of these tests by AAA was to measure the effectiveness of these systems, as compared to older cars/systems.

If you're saying "Well my AEB sucks" you might want to look at the tests, and consider that your system might be old, or just an individually sucky implementation.*

I've got a 2019 Subie Outback and the AEB is pretty good. It really only engages in very tight circumstances above a certain speed (and probably only lower than a certain speed) i.e. I'm pretty sure it will let me rear-end something if I'm crawling slowly enough. Prior to enaging AEB, the beeping is very loud with a very short interval.

(I traded in my 2017 Tesla Model S to get the Outback and I've never regretted it.)

With all this added tech and safety systems, why is per capita traffic deaths rising since 2018 and the highest since 2009?

Is all the added expense and impact on the environment worth it? Is the cost ever a concern in these mandates?

https://en.m.wikipedia.org/wiki/Motor_vehicle_fatality_rate_...

I very much doubt AEB saves anything like as many lives as "studies" claim, for the simple reason the "studies" will not be able to account for the number of times AEB hauls on the anchors, unexpectedly, for no reason at all. This has happened to me 3 times in the last few years, fortunately with nobody behind me, and likely causes more accidents than it avoids. It will be good when it's production ready.

I was in a Mazda dealer waiting for a checkup on my car and could overhear this young woman talking on her phone and to the service technicians. Her car had braked for no reason 3 times in the last few days and she was worried she would get rear ended if it kept happening. The mechanics did all sorts of test and drove her car around but couldn't find anything wrong. She had an 8 hour drive coming up and after exhausting other options decided to trade her car in on the spot for a new one! Extreme but at least her trade in wouldn't be considered defective.

>>The systems “are headed the right way,” says Greg Brannon, the director of automotive research at AAA.

They may have some promising data, but from what I've seen driving 2019 and 2021 model vehicles from Ford and Mazda, they are not even close to ready for prime-time.

I find both vehicles consistently falsely trigger the red-flashing-and-loud-beeping "COLLISION ALERT!!" dashboard warning to both small pavement cracks/potholes, and also to vehicles parked on the outside of a curve. This has happened at least a dozen times in the past ~30,000 miles driven. If those were instead automatic braking events, they would have caused an unnecessary rear-end accident least three times, as if I'd surprise "brake-checked" the driver behind me.

Yet, in an ACTUAL near-collision situation a few weeks ago, driving at night ~50mph on a rural road, a car ran a red light right in front of me, and I had to full-on threshold brake, years of road-race training reflexes kicking in before I was even aware of it. I barely avoided a collision, stopping with smoking brakes and a passenger with a pulled back muscle from having been not quite square with the seatbelt, my front bumper about a meter from their driver's door.

The car never made a peep — it completely missed the incident.

Of course, a working automatic braking system might have helped anyone in that situation, including me if I'd been a bit more sleepy or distracted. BUT IT DID NOT EVEN DETECT IT. The track record for the past 30,000+ miles is:

— 100% false positives — 100% false negatives

These 'collision detection systems' are 2-5 years newer than the systems AAA tested, yet the track record is awful. IDK what they are smoking but I do not want any.

I doubt the automakers are somehow holding back some magic solution, and would give them a LOT more time to get it right.

Just because a technology is promising does not mean it is ready to provide a benefit released in the wild on fast-moving multi-ton vehicles.

What does everyone think of the cost benefit?

>As a result, it should do more good; according to the NHTSA, if manufacturers deploy auto-braking systems that work at higher speeds, it would save at least 360 lives each year and prevent 24,000 injuries

>The government estimated that installing more advanced AEB systems on its cars would cost an additional $350 per vehicle. The auto lobbying group estimates prices could range up to $4,200 per car instead, and it has filed a petition to request changes to the final federal rules.

Over 13 million light duty vehicles were sold in the US in 2022, so +$4,550,000,000 in added costs per year.

I personally find ADAS technology to be deceptive, dangerous and terrifying from a vehicle control and privacy perspective.

Consider the fact that you essentially have a webcam both forward and cabin facing with advanced sensors scanning your face from the dash, eye position, monitoring your location at all times, recording audio (https://www.mobileye.com/privacy-roadclips/), capturing images of everyone around you all the time, taking and uploading a constant stream of data via an onboard SIM card.

Don't forget this device is even on when your car is off as it is wired directly into the car powered by the battery.

I ripped mine out of my Honda and now there isn't a third party literally trying to control my steering wheel. The car drives noticeably better.

I don't think most people are aware this much data is being collected.

Look at Mobileye's new "SuperVision" platform that has literally ELEVEN Cameras... https://www.mobileye.com/solutions/super-vision/

https://ggim.un.org/unwgic/presentations/SS1_19Nov_Mobileye....

here's an interesting analysis for a lot more detail: https://lindseyresearch.com/wp-content/uploads/2021/05/NHTSA...

This is not feasible so likely a number of seperate broken systems will be implemented that will contribute to the injury and deaths of many till the NHTSA recognizes reality again. It usually takes them a few years, like when they tried to mandate vehicle-to-vehicle wireless broadcasting of identity and velocity vectors in cars by 2024 in the late 2010s and early 2020s.

It seems like there is a need to merge control loops. Humans have sensory input, a model of the situation, and react to all that. So also automated driving or driving assistance machines have input, a model, and reactions. Instead of having these systems compete, they could interact. Instead of simply slamming on the brakes or beeping loudly the automated system could display its current inputs and model of the world so that it would be easy to see if the issue were in front, to the side, or in the rear and what physical objects may be involved. Instead of competing for control of the vehicle cooperate to present the best sensory input, analysis, and modeling and have the driver and vehicle negotiate all that in order to control the vehicle in the best way.

Some of the complaints in this thread make me wonder if Mercedes has a really good automatic braking system, or if you are all much more erratic drivers than my wife and I to make them kick in all the time on your cars (or maybe a the systems are all temperamental, and we just happen to have driving styles that align with MB's AEB). Across a couple of their cars, I've only noticed it kicking in once, and in a situation where it definitely helped avoid an accident (car ahead braked rapidly and unexpectedly in fairly light traffic at around 40mph).

They have been recent cars though, so I'm inclined to agree with the comment theorizing that maybe those cars just aren't as new

On a related note, I've always wondered if straight ahead 60 to 0 braking tests were relevant to anything. Not many accidents happen where a driver is just unable to completely stop and end up hitting the other car at 10 mph.

I would think the insurance companies could come up with a better braking test, like 60 to 40 with a swerve, where the suspension response factors in as well as both tire lateral and longitudinal grip.

The AEB on my 2016 Honda Civic seems pretty good: I've only had it misfire one time, which admittedly is very jarring when it happens. It was down a winding road that had a very large brick wall winding around it: I guess the system didn't anticipate that I was about to turn. In fairness, it happened once.

Of course, I have most of the driving assistance features, like LKAS, disabled. They work alright, but when I got my windshield replaced, I couldn't easily get the part that had the all-important "H" logo on it, so they will refuse to re-calibrate the sensors. I think it's pretty funny that potentially life-saving driving assistance features could become useless because the glass on your car doesn't have the right logo on it. Maybe if there was something wrong with the glass, sure, but I mean, it is literally identical to the OEM glass as far as I understand it (same manufacturer, Pilkington...) and there's nothing preventing me from engaging the features now, they just are a bit misaligned. Given the direction that EVs are going, I'm guessing the enshittification of cars is just getting started, and safety will be a key scapegoat just like security often is for other electronics.

Especially during inclement weather (honestly even only mild weather), the system activates when it thinks i’m not gonna stop in time near a stop light. I don’t drive aggressive or reckless, but if the right mix of weather/sun/etc hits, the system is far too sensitive (despite adjusting settings)

Having this shit at 62 mph? Excuse my french but fuck no

Automatic breaking is perhaps the worst feature I've ever experienced on a car. It's jarring, takes control from the driver, and, in my case, was unnecessary to engage and dangerous.

62 MPH is insane. This is going to kill more people than it saves. I don't know what kind of idyllic driving scenarios the engineers and policymakers have in mind when they come up with this kind of stuff, but they certainly don't seem to have considered many I've experienced.

Perhaps I am in weird kind of commuter bubble by living down in the Houston area. None of this shit makes sense to me. If I were to take my car out onto I-45 right now (not even rush hour) and slam on my brakes in the main lanes, I can 100% guarantee there would be a serious multiple vehicle accident. It's actually against the law (white sign, black letters) to come to a complete stop in the main lanes on many freeways and tollways around here.

boomer automotive companies buying a perception system from the lowest bidder? that perception system also has full control authority over the brakes? what could go wrong

Toyota Safety Sense has kept me from a few fender benders. It’s active at up to 50mph and has been super reliable.

Antilock brakes are bad enough. I don't want to ever own a car with an automatic braking system.

Note to Wired: it's okay in 2024 to just use "100 km/hr" without pointless conversions.