Studies find automatic braking can cut crashes over 40% • Hacker News Board

Re: Studies find automatic braking can cut crashes over 40%

#151

Post by KptMarchewa » Tue, Nov 15, 2022, 6:08 PM UTC

brookst wrote: post #84

hardolaf wrote: post #71

Earlier quoted context omitted.

I've never had a false alarm from the system on my car. But then, I actually follow the recommended minimum following distances and generally drive safely.

Well, it is either your superior driving behaviors, your car manufacturer's superior implementation, or just the way statistics work that not everyone sees every bug. My car auto-brakes at 5mph for parked cars that are backed into spaces, which I am sure reflects poorly on my driving in some way, if we're adopting the "bad things only happen to bad people" model.

Not having one false negative is infinitely more important than having false positives as you do, while you equalize them.

Re: Studies find automatic braking can cut crashes over 40%

#152

Post by mrguyorama » Tue, Nov 15, 2022, 6:15 PM UTC

dsfyu404ed wrote: post #30

melling wrote: post #9

Earlier quoted context omitted.

Yeah, everyone’s got that one story in a billion that tries to be the counter example. I remember back in the day when seatbelts were controversial because someone was thrown free from a car.

AEB is absolute magic for preventing rear endings by distracted drivers in stop and go traffic. People who's behavior doesn't tend to come with much risk of this will experience mostly false positives from AEB hence why "everyone" has a story.

Also, as shown in this very comment section, lots of people swear "the machine is just wrong" while being terrible drivers and tailgating

Re: Studies find automatic braking can cut crashes over 40%

#153

Post by lamontcg » Tue, Nov 15, 2022, 6:20 PM UTC

> The group found front-to-rear crashes were cut 49% when the striking vehicle had forward collision alert plus automatic braking, when compared with vehicles that didn’t have either system.

I wonder about front-to-rear crashes when the struck vehicle has automatic braking when the rear one doesn't.

I worry about people with automatic braking not paying attention and being on their phone and letting the automatic braking keep them from hitting people in front of them, while coming to abrupt stops and getting hit more from behind.

Re: Studies find automatic braking can cut crashes over 40%

#154

Post by giantg2 » Tue, Nov 15, 2022, 6:24 PM UTC

r00fus wrote: post #133

giantg2 wrote: post #61

Earlier quoted context omitted.

"I'm only one car-length away and still going 40 MPH. Another half-second and they'll be gone," These numbers must be wrong, and are terribly unsafe. Just off the top of my head, that's like travelling 60fps for .5 seconds is 30 feet, which exceeds 1 car length, even for my truck.

Anyone discussing braking margin/safety at speed in car-lengths ... is measuring the wrong thing. Your reactions don't correlate in distance (car-lengths), they're measured in time response (seconds). Always measure your safety buffer in the appropriate metric: time.

"Always measure your safety buffer in the appropriate metric: time."

Distance is the appropriate metric when you want to keep two objects a minimum distance apart.

Your response is only half the equation, then stopping performance comes into play. So while reactions are measured in seconds, you're converting those into distance traveled.

Measuring in following time is just a convenience for eliminating distance estimation biases. That time will also fluctuate based on the road conditions and vehicle performance.

Re: Studies find automatic braking can cut crashes over 40%

#155

Post by Workaccount2 » Tue, Nov 15, 2022, 6:34 PM UTC

alphabettsy wrote: post #83

Workaccount2 wrote: post #40

Earlier quoted context omitted.

Funny story, My sister got annoyed with the auto-braking on her new Subaru and turned it off. Three days later she rear ended someone, smashing up the whole front of her car. This was the third time she had rear ended someone. She keeps it on now.

> This was the third time she had rear ended someone. Pretty wild. Seems like license suspension or retraining should be required.

Knowing her, it's "Put your fucking phone down".

Re: Studies find automatic braking can cut crashes over 40%

#156

Post by meetingthrower » Tue, Nov 15, 2022, 6:44 PM UTC

CydeWeys wrote: post #27

branon wrote: post #10

Oof, my experience with with auto-brake in a 2021 Subaru Crosstrek (anecdotal, sample size of one, all that good stuff) is much the opposite. It's too sensitive and tends to brake shockingly hard during fairly mundane maneuvers. When backing up near a wall/obstacle, the auto-brake seems to kick in when anything is closer than three or four feet away from the bumper. I use my mirrors and have an accurate impression of…

Honestly, I feel like a lot of the people complaining about this feature are just bad drivers who regularly drive unsafely and are now getting called out on it. This in particular: > Sometimes, though, I will allow the distance between myself and the next car to close, for example, if they're turning and will be gone in a second anyway. The Subaru then ruins my well-calculated slowdown maneuver by slamming on the bra…

100%. We drive Subarus, highway and city. Have NEVER had an emergency brake triggered. OP is unsafe.

Re: Studies find automatic braking can cut crashes over 40%

#157

Post by nradov » Tue, Nov 15, 2022, 6:46 PM UTC

dsfyu404ed wrote: post #41

nradov wrote: post #29

Earlier quoted context omitted.

True, but the latest generation of electronic stability control systems with brake-based torque vectoring are pretty good at handling those situations. Many cars already include that feature and it will probably be universal in a few years. https://www.motorbiscuit.com/do-you-know-what-torque-vectori...

Nothing will save you from physics. The tires can only apply so much force to the vehicle in order to change direction or speed. A stability control system fundamentally trades distance traveled forward in exchange for not getting sideways. ESC is good at preventing people who don't know how hard their vehicle can actually turn from going "all in" with the brakes and winding up off the road because they robbed the ca…

I think you missed the point. Adding brake-based torque vectoring to ESC does more for crash avoidance than just ESC alone. It allows the vehicle to rotate faster by maximizing use of the available traction at all four wheels. Even drivers who know how to get the most out of their vehicles aren't able to selectively brake the inside wheels while turning. This isn't magic and won't prevent all crashes, but will prevent some.

Re: Studies find automatic braking can cut crashes over 40%

#158

Post by brookst » Tue, Nov 15, 2022, 7:06 PM UTC

notacoward wrote: post #138

brookst wrote: post #64

Earlier quoted context omitted.

Reducing the odds of getting rear-ended by absorbing the slack of a conservative following distance rather than braking to maintain the same following distance from 35 to 20 mph.

You can't control what the person behind you does, but you can control what you do and rushing people through turns is unsafe. If the person in front of you signaled (or did anything else to indicate) their turn in sufficient time, then you should be able to do likewise for the person behind you with a light tap on the brakes. The odds of being rear-ended probably won't change by any significant amount, and IMX peopl…

These systems are imperfect. People are imperfect, too.

But please consider that not everything is evidence that everyone else is reckless driver. One day your system will fail in a surprising/dangerous way, and you'll tell someone "wow, it mistook a reflection for tail lights", and they'll say "well I am a perfectly safe driver so that could never happen with my careful driving, you must have been really reckless", and you'll have an epiphany.

Re: Studies find automatic braking can cut crashes over 40%

#159

Post by comfypotato » Tue, Nov 15, 2022, 7:07 PM UTC

proggy wrote: post #87

Implementation matters. The automatic braking on my compact SUV has falsely triggered hundreds of times over the course of 3 years, usually when transitioning from a moderately-pitched-down driveway into a roadway. The result of this scenario is the car coming to a dead stop with its front end halfway into a traffic lane, putting me and my passengers at greater risk of a collision from oncoming traffic approaching fr…

This is my gripe with the safety of self-driving cars generally. The net crashes are much lower (very good thing) but the crashes that do happen take place under different circumstances. A good driver paying close attention will probably be better than the computer for a long time. The crashes being avoided are dumb read end collisions where the crash happened because the driver didn’t realize someone was stopped at the red light in front of them (just one example of an unfortunately common crash).

Still looking forward to it when I’m older and a bad driver anyway. Can’t wait to have a personal vehicle that takes me everywhere.

Re: Studies find automatic braking can cut crashes over 40%

#160

Post by toss1 » Tue, Nov 15, 2022, 7:43 PM UTC

proggy wrote: post #87

Implementation matters. The automatic braking on my compact SUV has falsely triggered hundreds of times over the course of 3 years, usually when transitioning from a moderately-pitched-down driveway into a roadway. The result of this scenario is the car coming to a dead stop with its front end halfway into a traffic lane, putting me and my passengers at greater risk of a collision from oncoming traffic approaching fr…

Yup. Our cars don't have auto-braking, but the "Pre-Collision Warning" or "Collision Alert" in the Ford Ranger and Mazda MX5 are both very annoying. They create enough of a distraction that I'm concerned that it could lead me into an accident. Yet it triggers too late to do anything about it, unless I simply train myself to reflexively hard brake whenever it goes off, which would almost surely get me rear-ended, especially since it's always a false alarm.

They both mis-trigger fairly reliably if I'm in a bend in the road and there is a car or object on the outside of the bend, so it seems that they don't take into consideration that I've turned the wheel, or the steering angle. I've also had total mystery alerts, seemingly from spots or a shovelfull of dirt on the road surface.

OTOH, the distance-keeping cruise control on the Mazda is very good, and I've used it for dozens of miles at a time on country roads, with and without traffic.

These emergency auto-braking systems are clearly not ready for prime-time, except maybe for highway use, which is where they can do the most good anyway. I sure as heck don't want my car to auto-brake in any of the situations I've seen the "Collision alert" triggered. Some would have definitely caused accidents.