Hacker News Reader

Comment by PeterisP | Hacker News Reader

It's not as simple. For example, look at the distance sensors used for parking and the data coming from them over the CAN bus.

That bus needs to be able to (at least indirectly) reach the brakes so that automatic emergency braking can take their measurements as inputs. And that bus needs to be able to reach the audio system to provide audible feedback to the driver when parking, and also mute the radio. Ergo, you have a bus that reaches both brakes and radio. Now, you might want to prevent the radio firmware from sending data to the brakes over that bus, but physically you have to have a connection between them, as we don't really want to make many separate buses for reasons of complexity, cost and maintenance; moving from a separate end-to-end wire for each specific purpose towards a shared bus was a great improvement.

View on HN

Parking assist is the very definition of bullshit convenience. In this case apparently it's dangerous as well?

>According to the National Highway Traffic Safety Administration (NHTSA), approximately 188,000 passenger vehicles were backing in crashes reported to the police in the United States during 2015…The combination of Rear Vision Camera and Rear Parking Assist reduced backing crash involvement rates by 42%. When Rear Automatic Braking was added to the Rear Vision Camera and Rear Parking Assist, vehicles with all three systems had backing crash involvement rates that were 78% lower than vehicles with none of the systems.[1]

The data would seem to contradict that statement.

[1] https://www.sciencedirect.com/science/article/abs/pii/S00224...

I think "Rear Parking Assist" would be better phrased as "Rear Proximity Sensors", as the "Automatic Parking Assist" (which is what I thought we were discussing) only brings a (likely statistically insignificant) marginal risk reduction of 2% and seems to fall under the "bullshit convenience" category. I do agree that the proximity sensors are themselves huge wins but (if I'm understanding correctly) this could be fairly trivially conveyed on the dashboard instead of the infotainment system.

As I see it, the issue here is that it's in automakers interest to tie together the safety improvements (which are great and benefit people) with the cost-reduction mechanisms consolidating controls and monitors into a single screen which seem to conflate operation with infotainment and, I would argue, reduce peoples' safety.

This is only tangentially related but different to the repair question, so I modified what I said a bit to reflect I was going off topic.

You misunderstood my intent, I'm saying that introducing these systems apparently produces a infosec threat significant enough to warrant federal legislation. In response to your post however I think we can dispense with the notion that folks backing into parked cars, light posts, etc, represents anything more than a nuisance to whoever owned the property that got backed into. Yeah it says "crash" in the writeup but we're not talking Michael Bay style 8 lanes of exploding cars here.

I don’t think it needs to be some Michael Bay crash to be significant. Let’s say that all these crashes are minor, as you state. If repairs ranged from $500 to $1,500 evenly, that is about $188 million in property damage a year. That’s not nothing. I agree that systems need to be better designed with infosec in mind.

It's not nothing, it's the consequences of people's general disinterest in operating a motor vehicle effectively. I think it -should- be expensive to be a bad driver. Good decision making is learned through poor decision making.

> I think it -should- be expensive to be a bad driver. Good decision making is learned through poor decision making.

You want less safety in order to teach some sort of moral lesson to people? Thank god we don't design planes that way, I guess.

Removing incentives to actually pay attention to where your two ton death machine is in space isn't how one introduces safety. What you're proposing is a crutch that rewards inattentive behavior. There's no moral to it.

A camera and sensors that give you more information about "where your two ton death machine is in space" and are literally proven to improve safety and reduce accidents means you pay attention less? So cars would be even better if we removed the side and rear view mirrors, since those are nothing but "a crutch that rewards inattentive behavior"? I mean, people have necks so shouldn't they just turn around and look instead of being lazy and relying on mirrors? I'm sorry, but I cannot take your position seriously.

Here's the thing: watching a monitor doesn't actually update one's spacial model of their surroundings and their vehicle's position therein. Under normal operating conditions an attentive driver has and is constantly updating a fairly accurate mental model of their vehicle and the objects near it that they are likely to interact with.

When your attention is drawn to a rear camera system you've now ditched that spacial model entirely and have changed to modeling the video game on the screen in front of you. This context switch is -very- expensive to unwind and at least momentarily costs you your awareness of where everything else except the back bumper is in space, and what's in front or to either side of the vehicle.

As far as mirrors go, they aren't to be relied on solely as they do not provide a full picture of the vehicle's surrounds. Checking one's blind spots by turning your head and actually -looking- in your blind spot before merging is a thing. same goes with the rearview.

Hope that clears up your confusion.

Some portion of those incidents were backing over pedestrians and children, not just backing into stationary objects.

We know someone who backed over their child and the child died.

I assure you, that was not a nuisance.

I know three guys that got hit by lightning. I still don't consider wild-ass outliers when making decisions.

Luckily for you, you don’t need to. Consider outliers.

The reason they the US mandated backup cameras was due to accidents like this.

They were that common.

from your original point.

> In response to your post however I think we can dispense with the notion that folks backing into parked cars, light posts, etc, represents anything more than a nuisance to whoever owned the property that got backed into. Yeah it says "crash" in the writeup but we're not talking Michael Bay style 8 lanes of exploding cars here.

Given the complete lack of any empathy at all you've shown, I don't expect you to change your mind on this, but please try to consider if we have a way to prevent deaths, then maybe we should try and not just consider a parent backing over their child to "represents anything more than a nuisance to whoever owned the property that got backed into."

We're getting pretty far afield from the original point of decoupling control systems from network connected infotainment systems, which doesn't necessarily eliminate backup sensors or cameras. The current everything-in-one-bucket implementation is a matter of convenience and cost savings for the manufacturers, not a hard technical requirement.

I could go on at length about the consequences of trying to play whack-a-mole with human stupidity and how that just breeds stronger idiots. There have been no significant changes to vehicle-related death statistics since the 90s when airbags were mandated, despite significant improvements to crash protection and countless here-let-me-help-you-with-that safety features that have been introduced in the last 30 years. So what do you call it when improving safety doesn't actually improve safety?

> So what do you call it when improving safety doesn't actually improve safety?

USofA exceptionalism?

Road Deaths per 100,000

    Year  USofA    Australia
    1990  17.88    13.66 
    2021  12.89     4.4

https://en.wikipedia.org/wiki/Motor_vehicle_fatality_rate_in...

https://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_i...

> There have been no significant changes to vehicle-related death statistics since the 90

I mean, if you look at the data, you'll very quickly see you're wrong

https://tc.canada.ca/en/road-transportation/statistics-data/...

Fatalities have gone only down over teh past 20 years.

Your assertion that car deaths haven't moved is false

I looked at the data for the continental us and deaths per vehicle-hour is flat as a fucking strap when graphed over the last 30 years.

Current cars usually have gateways that route messages around different busses. This would protect you from having unwanted messages arriving on secured buses, but pass through the wanted messages.

Unfortunately, those gateways are pretty poor when they do exist. There are some thieves in the UK popping off headlights to access the CAN bus, where they then inject a message that tells the system that a valid key has been used so go start the engine.

> And that bus needs to be able to reach the audio system to provide audible feedback to the driver when parking, and also mute the radio. Ergo, you have a bus that reaches both brakes and radio

A $2 speaker would work just fine. You would still hear it over the radio. If your radio is so loud you couldn't hear it while parking you have bigger problems. A shared bus might be more efficient, but it is not more redundant. I would rather spend $2 on redundancy here.

Seems like that would be trivially managed with a firewall (canbus security gateway). Is it purely cost that they don't lock down or segment off the canbus? Being able to steal a car by talking to the immobilizer through the headlights comes to mind.

> Ergo, you have a bus that reaches both brakes and radio.

But do you? Why not have a plain old audio connection that goes to the audio system? It doesn't have to be on the CAN.

Reply on news.ycombinator.com