Hacker News Reader

Comment by u801e | Hacker News Reader

Since 1998, the number of children who died of heatstroke in vehicles has been around 30 to 50 a year [1]. Before that, the number was a bit lower (10 or less) [1]. Around that time, the NHTSA (National Highway Traffic Safety Administration) mandate on new vehicles having passenger side airbags as standard equipment came into effect.

To address the issue of children dying due to airbag deployments (which were designed to protect an unbelted 50th percentile size male in a frontal crash), the NHTSA recommended that children in car seats be placed in the rear seats in the vehicle.

Due to the fact that it's much easier to not notice a child in a rear facing car seat in the rear of the car, as opposed to one in the front passenger seat, the number of children who died of heat stroke each year increased from 10 per year to 40 per year.

[1] https://ichef.bbci.co.uk/news/624/media/images/76289000/gif/...

View on HN

We have keyless entry, lane departure warning, cruise control, and even autopilot. Can't we have a stupidly simple system that if a human is detected in the car and the temperature is over 37 degrees C:

- 0. Call the car owner's phone or alert them through any communication app they use

- 1. After 5 minutes, call the police

Ether or both would save lives. It doesn't seem like a very difficult technical task to me.

I am not sure detecting humans is the right solution... But -2. Open the windows -3. A/C...

I wouldn't trivialize the reliable detection of people, especially small children. I think we are just now getting at the point where that technology could be reliable, and from an automaker's perspective, it could be a similar level of investment as autopilot, and a much smaller benefit (more people would be saved by reliable autopilot, and it's also much more desirable to consumers). Even if they had such a technology, the automaker would certainly be afraid of the legal consequences of their system failing a real scenario, opening itself up for liability lawsuits and bureaucratic investigations.

wouldn't a motion detector be enough? It doesn't really matter if it's a kid or a dog.

No, it wouldn't. Not if the child is asleep. Some kids wiggle in their sleep in a carseat, but there are plenty who sleep like a sack of flour.

> I wouldn't trivialize the reliable detection of people, especially small children.

A fairly simple weight sensor as used for passenger seat belt alarms would seem likely to work (it's a little more complicated because you need to zero it for the weight of the child seat.) You'll get false positives if you carry goods on that seat, but if you have an installed child seat you are unlikely to do that.

The notification aspect is probably harder to get right than the sensor. Though an internally audible alarm and warning light for a short period when turning off the car with a detected child would probably itself be a big help to prevent the initial problem rather than reacting to it the way “call the owner and then the police” mechanisms do, plus be far simpler to implement.

Agreed, a reasonably designed 2W 2.4GHz radar in your car can detect both breathing and heartbeat. It couldn't tell the difference between a large dog and baby, but maybe you should be dealing with it either way...

Now I can see the car manufacturers don't want the liability for the a monitor failing to go off (false negative) due to any malfunction, when the number of false positives will be much higher. Thermometers and radars do fail, but humans will become dependent on them, and blame them and the manufacturers for anything that goes happens.

> Can't we have a stupidly simple system

No, because what you're describing isn't stupidly simple, but incredibly complex. Even just the 'call the police' bit - How? Which police? How do you communicate with them? What language do you use? How do you convey the problem? How do you detect and respond to emergency numbers that prefilter (where I am, call emergency and a human asks you for police/fire/ambulance, and then puts you through)? Do you use the standard voice channel, or an API? If API, whose API? What about areas that don't have an API (or much of a tech stack)? If voice, how do you make it not sound like a prank call? How do you provide extra data if the dispatcher needs more? Do you use a radio band instead? If so, which one? How do you ensure you're not spamming the channel? What if the police in the area don't work off GPS co-ordinates? I'm sure there are plenty more issues with just this part of the hypothetical system, but these are the ones off the top of my head.

Besides, while it sounds nasty, 30 deaths/year is but a drop in the bucket compared to other transport-related deaths. The amount of effort required to make a robust, reliable system to prevent those 30 deaths would be much more effective when applied to other transport problems resulting in death.

Tesla (of course!) has done this http://money.cnn.com/2016/09/21/luxury/elon-musk-tesla-hot-c...

A car that calls the police on its owner seems like a non starter.

I would be worried that it would erroneously call the cops on me and not purchase it.

If you mandated it I would do my best to see the people that voted for it out of office as soon as possible.

What you are discussing seems complicated, error prone, and expensive. My kid is an adult. I don't want your half assed system calling the cops to save my packages that I'm carrying home.

What if people are transporting their animal in the car and the the line is long when they go inside to pay. Shall we call the cops on them after 5 minutes?

I am content and capable of reasoning about the real world and intelligently avoiding negative consequences by not doing stupid things.

I do not want to have to think about what conditions erroneous or real will cause my car to rat me out.

Not everything in life has a technical or legislative solution lots of things in life can only be avoided by rational people taking the time to do reasonable things every day like not leaving their babies in oven like cars and by and large do well at this. Out of 300 million Americans 30 manage to off their kids which is safer than bathtubs let alone the real risk of violent death in an accident.

If you imagine a unit cost of 500 usd and we have 263 million passenger vehicles outfitting them all would cost 130 billion usd to save 30 lives annually. If you imagine a time frame of 10 years you can imagine savings of 300 lives at a unit cost of 433 million per life saved.

I'd face a big fine any day over the death of another person that I caused inadvertently. The former sucks, but I'd get over it. The latter would wreck my life forever if it happened.

In any case, my suggestion was to call the owner first. They would have several minutes to respond. Or the product could only do that, and forget about the police part, which would also save lots of lives.

Or instead of calling the police, start the air conditioner and continue notifying the owner every few minutes if they do not respond to the first notification.

Its still an expensive feature that would get more expensive the smaller the number of units involved. Nobody would voluntarily pay for it especially those stupid enough to leave their kids in hot cars. Especially since in at least a portion of those cases the parents meant to leave the kid in the car they just didn't intend to do it for such a long time or didn't realize they would die from it.

I have a 3rd gen Toyota Prius which had the option of a 40W solar panel on the roof which runs a ventilator when it's parked. That and the white paint job really help it feel a lot cooler than previous cars I've owned, after leaving it parked for a few hours in the sun.

I'm not sure how much it would help here (no idea what the climate is like in this place), but I've sat inside when its 35c/95F outside, and it's just as comfortable as sitting under a tree - maybe more so as you can feel the wind.

It seems like such a simple feature (the solar system is completely isolated from the 12V and HV battery), I'm surprised more cars don't have it. Here is a picture - right is the (slightly opened) moon roof, and left is the solar panel: https://blogmedia.dealerfire.com/wp-content/uploads/sites/19...

There's an even simpler (if imperfect) solution:

Remind the driver to check the back seat if they opened one of the back doors before they got in the front seat.

Heck, most cars have seatbelt+weight sensors. Just give a warning if you detect anyone under 100 lbs in the back.

How many children were dying as a result of the airbag deployments before the recommendation I wonder?

Before the recommendation, the chart I linked in my previous post shows about 60 children died in 1995. More than 30 died each year in 1996 and 1997.

Edit: Updated my answer to answer the question asked instead of citing information about when the airbag mandate came into effect.

EDIT: Oops, I didn't initially notice that the chart in the grandparent post actually lists both airbag and heat fatalities...

---

Good question. As of 1996, it was only 21 total, according to the NY Times[1], but more than half of those were killed in the first 9 months of that year.

In 1999, Car and Driver magazine reported a total of 99 child fatalities. So, low "first-few-google-hits" data quality aside, that seems to show a rapidly rising trend[2].

Next, I got to this horrifying article about a University of Georgia study concluding that the government statistics on airbags are wildly wrong and airbags are actually killing us all[3]. (I gave up my research at this point and will sleep better if the next HN reader that comes along with a little time on their hands debunks that as publicity-seeking bad science that's now been discredited...)

[1] http://www.nytimes.com/1996/09/18/us/more-children-are-kille...

[2] http://www.caranddriver.com/columns/airbags-kill-more-kids-t...

[3] https://phys.org/news/2005-06-airbags-probability-death-acci...

An update from my coffee break:

Here's an IIHS/HLDI refutation of the study by Mary Meyer and Tremika Finney of the University of Georgia, that I mentioned above: http://www.iihs.org/iihs/sr/statusreport/article/41/2/2

> Next, I got to this horrifying article about a University of Georgia study concluding that the government statistics on airbags are wildly wrong and airbags are actually killing us all

The hyperbole isn't really necessary. Before "depowered" airbags were allowed by the NHTSA, the standard described in FMVSS (Federal Motor Vehicle Safety Standard) 208 stated that airbags must be able to protect an unbelted 50th percentile sized male in a frontal crash. This required that the airbag deploy at lower collision speeds and with more force than would be necessary with a belted occupant of smaller stature.

This lead to injuries and deaths due to airbag deployments in low speed collisions as mentioned in the study. My guess is that the study period was at a time when depowered airbags weren't as common as they are today.

Looks like we will need an invention to prevent heat stroke deaths from children getting stuck in cars.

Reply on news.ycombinator.com