Recently, there’s been a lot of activity related to eliminating or at least minimizing the tragic deaths of children accidentally left in the back seat of a car, which can reach interior temperatures over 120ºF/~50ºC on a hot day. Although it’s easy to say, “How can you forget a child in the back seat?” It does happen: there were 53 confirmed deaths in the US in 2018 (plus many more cases of extreme but not fatal heat stroke), and 26 as of August 2019 (note that about 25% are due to the children crawling in unnoticed and then getting trapped rather than being forgotten).
Auto makers are under pressure to “do something” although the specific shape of that “something” is hard to define. You can get full details with dates, places, and names on these deaths at NoHeatStroke.org, which makes for depressing reading. There’s a parallel discussion of whether it really is the responsibility and legitimate function of the auto company to somehow remind drivers about children left in cars, but that’s a discussion for another time.
This is actually not an easy problem to solve. In principle, it’s enough to have the electronics check the car seat in some way (such as by weight), and then tell the driver to check out that back seat. The reality is that each proposed solution has subtle tradeoffs, as they usually do.
For example, if you simply measure the weight of the seat, what if the additional weight is a package rather than a child? Do you look for some sort of occasional motion—but what if the child is sleeping soundly? How do you “tare” the seat so the weight of the empty child car seat is taken out of the equation, and only the child or package is weighed? Is it enough to flash the car lights and sound the horn to alert the driver, or does the car decide to roll down the windows if the rear door is not opened to check within a certain time-out period? Or should the car send a text alert to the driver’s phone, and what if the phone is off or they don’t have their phone with them, or they are not the usual driver but driving the car for someone else and doing pickup?
Other proposed simpler solutions skip the use of analog sensors and instead rely “door-sequencing logic” to track when the rear door has been opened at the beginning of the trip and whether it was opened again at its conclusion; it sounds like a miniature version of a programmable logic controller (PLC). But the logic system would reset if the car is stopped and shut off, if the user stops to get gas, for example. Some auto manufacturers are already offering (or will do so soon) some type of rear-seat occupancy alert, but these initial efforts are not standardized across all vehicles. That adds to the potential confusion if drivers use multiple vehicles, of course.
Some “experts” have said the only viable solution to have a camera with image recognition aimed at the back seat. Besides the obvious expense, there are issues of reliability, placement, blocked imaging, and performance to consider for such an advanced approach. A standardized approach also means that the designated solution has to be very effective if every car vendor is expected to adopt it.
The irony is that occupancy detection can be an easy or hard problem, depending on the circumstances of sensed location, installation constraints, moving or static targets, and other factors. Complicating the problem is that the car seats themselves are not permanently installed; they are sometimes removed and certainly upgraded as the child gets bigger. Further, any system that generates false positives will soon be ignored (or even shut off, if that is possible).
The problem and its solutions are not new. In fact, you can buy aftermarket add-ons which purport to serve the same function. Among them the RVS Brilliant Backseat Reminder System, which targets both packages and children (Figure 1). In this unit, “a moderate audio alarm is activated when you leave your car without your cargo. If this alarm is not turned off in 40 seconds, the system will activate the vehicle’s horn which will get your attention.” The website is vague about how it detects their presence and that they have been left behind.
Then there’s the Cybex Sirona M Sensorsafe 2.0 Convertible Car Seat, which uses a smart chest clip that is synced with the installed vehicle receiver and the caregiver’s smartphone (Figure 2). It senses and provides an “alert through both the vehicle receiver and the caregiver’s cellphone when the driver accidentally leaves the child behind in the car but also alerts the driver if a child unbuckles themselves while the vehicle is in motion; if the back seat has become too warm or too cold; and if a child has been seated for too long —all through a user-friendly app.” Note that the “receiver” is plugged into the cars OBD (onboard diagnostics) port – another complication.
In early September of this year, the Alliance of Automobile Manufacturers and the Association of Global Automakers (trade groups that represent the bulk of domestic- and international-made cars and trucks in the U.S. market) agreed to implement basic rear-seat reminder systems on new cars by model year 2025.
At a minimum, these alerts will include auditory and visual cues to remind drivers to check the back seat before leaving the car, and will also include the door-sequencing algorithm – although that doesn’t help in the 25% of the cases where the child climbs into the unattended car and gets locked in. Lawmakers are pressing for a more comprehensive solution, but the problem is that there is no clear way to solve this without adding a lot of cost, complexity, and potential for confusion.
I have no idea what form the final rear-seat alert system will take, or if it will be uniform across the industry. Once it goes beyond simple door sequencing, it becomes an area for analog sensors and interfaces, with all the benefits and downsides they inherently bring, including the need for sensor self-test, which is often difficult to do.
What’s your view on solving the problem of developing a rear-seat alert system? What approach would you use? Will it add to the car’s “vampire drain”? What percentage or rate of false positives and missed positives would you accept? Or is this a case of lawmakers demanding that engineers, once again, somehow come up with a “near-perfect” solution to a problem, despite the practical issues and harsh realities?