Self-Driving Cars Can Avert Hot-Car Deaths

Dr. Lance Eliot, AI Insider

Image for post
Image for post

Sadly, yet another hot-car death occurred recently, involving a man that left his 1-year-old twins in the backseat of his car while he went to work for the day, occurring in the sweltering heat of upstate New York.


Besides human deaths due to hot-car conditions, there are also pets that have either died or been injured due to being left alone in a parked car.

There are some people that knowingly leave a living entity in their car such as their pets and apparently don’t realize the dangers involved, and there are some people that mistakenly do so.

In the case of the man that left his twins inside the car, he has indicated that he forgot they were there, having assumed that he had dropped them off elsewhere already, and so mistakenly left them in the car.

How Hot-Cars Deaths Occur

There is actually a greenhouse effect that happens inside a locked car that has the windows rolled-up.

According to various charts, if the outside temp is 70 degrees Fahrenheit, a car can readily increase the internal temperature by about 20 degrees in 10 minutes. Thus, the internal temp is now 90 degrees.

By 20 minutes, the temp will have risen another 30 degrees, and so it would likely be around 120 degrees inside that car. I realize that you can argue with these numbers by saying that the circumstances vary as to how high the temp would go and how quickly — all I’m trying to say is that the temperatures inside tend to get hotter than we think, and you can’t judge this by the outdoors temperature. It could still seem to be a reasonable 70 degrees outdoors, and meanwhile inside that car it’s a boiling 100 degrees or more.

Get ready for this scary statistic: The leading cause of non-crash car deaths for children under the age of 15 is heatstroke.

Usually, a parent or a caregiver has left the child inside a locked car, which is parked outdoors, and the child is left in the car such that the greenhouse effect eventually kills the child. Horrible to even contemplate.

For those that think that leaving a small crack in the car windows will prevent heatstroke, they are quite mistaken in this belief. The slightly open window is not effective to dissipate the heat.

Simple Ways To Prevent Hot-Car Deaths

One technique that some say should be used involves always putting something “essential” into the backseat such as your wallet or purse.

The odds of leaving behind your wallet or purse is relatively unlikely. You’ll be forced somewhat to turn to the backseat and in so doing be reminded that your child is there.

For the leaving of pets in a car, it is rare that someone forgets they left a pet in the car, and more the time is just sheer ignorance or outright foolishness to have left the pet in there.

Certainly, there are some cases where a parent or caregiver leaves a child in the car to then be able to dash into a grocery store, and those too fall into the shocking aspects of ignorance or foolishness categories.

The act of doing these kinds of things are often subject to criminal laws.

It is generally against the law to leave a living being in a locked and insufficient air circulating and temperature controlled environment. Society has wised up to this kind of behavior, and the “see something, say something” has helped to deal with the numbers of instances where this continues to recur.

Some have proposed that a device should be included into our cars to help prevent this from happening.

For example, a camera pointed inward that can “see” that there’s a living being in the car, and then perhaps bleat a loud alarm or honk the horn, and possibly even be able to automatically unlock the doors. Maybe even have it be able to roll down the windows. Or, restart the car and turn-on the air conditioning. Instead of a camera, it could be a heat temperature gauge that is combined with a motion detector, and if there is motion inside the car and the temperature has gotten too high, and the car is parked, once again some kind of safety action is undertaken by the car.

These are all potential solutions to the problem.

Unfortunately, almost no one is going to be willing to pay for these elements.

It would raise the cost of a conventional car.

To retrofit a conventional car with these elements will be somewhat costly too.

Even if these all worked as prescribed, how many people would be willing to admit that they might someday leave a living being inside their locked car? I’d dare say that no one walks around thinking this is going to happen.

Autonomous Cars And Averting Hot-Car Deaths

What does all this have to do with AI self-driving driverless autonomous cars?

At the Cybernetic AI Self-Driving Car Institute, we are developing AI software for self-driving driverless autonomous cars and this includes devising AI that can aid in averting hot-car deaths.

Most AI developers would say that this is an “edge” problem.

An edge problem is considered at the edges of the core aspects of something that needs to be done, and as such those edges can be worried about at some later date.

Worrying about a living being left inside a parked car is just not very high on the priority list.

Let’s also consider the future and see what might happen as AI self-driving cars begin to become available and pervasive.

We are going to have parents that will put their child into an AI self-driving car and tell the self-driving car to take the child to day-care. The parent won’t be riding with the child. Instead, the AI system is taking the child, alone, over to day-care. I know this seems incredible to contemplate, and I admit as a father it is unimaginable that I would do this, but I am sure this is the direction of society.

It’s convenient for the parent. No need to drive out of your way. No need to leave work. Just have the AI self-driving car take care of your driving needs. This is predicated on the notion that the AI self-driving car is at a Level 5. A Level 5 self-driving car is considered an AI system that can drive the car as a human can drive a car and that there is no human intervention needed for the driving of the car.

I am sure it will be years of proof that AI self-driving cars are at a Level 5 and of sufficient proficiency before people will trust putting their baby into the backseat and letting them ride alone.

You might be willing to put a teenager into the car and figure that if somehow the self-driving car had an issue, at least the teenager could potentially take action. A baby or small child has not much chance of being able to take their own action. I’m not suggesting that these occupants could do any driving, only that once they are old enough, they could potentially get out of the car if they needed to do so.

I have often pointed out though that this ability to get out of the car is a dual-edged sword.

You put your ten-year-old into an AI self-driving car. You tell the child to not get out of the car. Keep in mind that no matter how good the AI might be; the self-driving car is still a car. This means that the self-driving car is subject to mechanical breakdowns. Suppose the AI self-driving car breaks down on the freeway, halfway to the desired destination with the child inside the car.

Do you want your ten-year-old to sit inside that car, which presumably is now a sitting duck in the emergency lane of the freeway? Maybe yes, maybe no. Suppose that the child decides they are going to get out of the car on their own, under the thinking that it is safer, and steps out into the oncoming traffic?

That’s not good either. It’s a conundrum.

Ethical Aspects Of Autonomous Car Capabilities

As a society, we need to consider the societal and ethical implications of AI self-driving cars.

There are some aspects that the technology of a self-driving car can be leveraged that might help in some means.

First, it is likely we’re going to have cameras pointing inward in true AI self-driving cars.

This will be pretty much built-in. If you are planning on using your self-driving car for ridesharing purposes, I’m sure you’ll be happy to know that there’s a camera that can be used to catch someone messing with the inside of your self-driving car. This also raises apparent privacy issues.

The AI self-driving car is certainly going to be equipped to start the car on its own, drive the car on its own, and otherwise control the car. This is a given.

As such, suppose that the camera detects that a child is in the backseat, and otherwise unattended, and if needed the AI could automatically start the car and start the air conditioning. This could either happen because the AI figured out to do so, or that the owner of the self-driving car was alerted and the owner perhaps then instructed the self-driving car to take such action.

We’re also going to have some means of communicating with the AI of the self-driving car.

More than likely, it will be a verbal interface akin to what we do today with Alexa and Siri.

The notion is that suppose an occupant wants to find out what the AI is doing, or wants to change the plans of the AI system as to driving the car, the human would talk with the AI. I might be inside my AI self-driving car, I tell it to drive me to work, and then along the way I see a Starbucks and so ask the AI to momentarily stop there for me to get some coffee.

If a child is in the AI self-driving car, the child can potentially talk with the AI.

The child might say they need help and ask the AI to call their parents. Or, maybe the child says they are bleeding and to go to a nearby hospital. Now, we’ll obviously need to have some savvy AI because the child might say that they don’t want to go to school today and to take them to the beach. It can’t just be that whatever an occupant utters that the AI will blindly obey per se.

The AI of the self-driving car can also draw attention to the car, if needed, for the benefit of the occupants.

Someone is in the self-driving car and they somehow get hurt, and so maybe the AI starts honking the horn or turning the headlights on and off to attract attention and aid.

Also, if the self-driving car has V2V (vehicle to vehicle communications), it can alert other nearby self-driving cars to come to the rescue by heading to wherever the self-driving car is parked (assuming you want this to happen).

The AI could also call the police or the fire department, and even drive to a designated location to meet a police officer. Etc.

The microphone inside the self-driving car can be used to detect not only words, but other sounds such as the sound of someone getting hurt or in pain.

Some self-driving cars will have an internal temperature gauge that comes with the normal operation of most cars anyway.

By using a combination of an inward pointing camera, temperature gauge, potential motion detector, audio microphones, and other such devices, the AI can possibly be a lifesaver for the occupants.

I say can be because if the AI is not developed to do this, it won’t be happening by magic.

Thus, as stated earlier, it’s considered right now an edge problem and only efforts like ours and others that see this as important are working toward solving it.


See something, say something,

it’s a motto for humans and for the AI of self-driving cars.

Once we begin having self-driving cars toting around our children, elderly, and other such occupants, I am betting we’ll want the AI to be a helpful caregiver to watch over and protect those occupants. This though has to be done in a manner that doesn’t overstep what we as a society would want the AI to do.

Having an AI system that imprisons the occupants is bad, but at the same time “good” if it is done in a manner to protect the occupants properly and appropriately. Anyway you cut it, we want to avert the chances of anyone dying due to being left inside a parked self-driving car.

Let’s see if we can get the AI to help us on that sensible goal.

For free podcast of this story, visit:

The podcasts are also available on Spotify, iTunes, iHeartRadio, etc.

More info about AI self-driving cars, see:

To follow Lance Eliot on Twitter:

For his blog, see:

For his Medium blog, see:

For Dr. Eliot’s books, see:

Copyright © 2019 Dr. Lance B. Eliot

Written by

Dr. Lance B. Eliot is a renowned global expert on AI, Stanford Fellow at Stanford University, was a professor at USC, headed an AI Lab, top exec at a major VC.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store