[SystemSafety] State-of-the-art for Child Presence Detection ?

[Watts Malcolm (XC-CE/ENG4-AU)]

Hello all;

I'm trying to understand some existing (ISO 26262-based) hazard analyses relating to Child Presence Detection (CPD) according to Euro-NCAP standards, where this function is intended to prevent death and injury to children left in motor vehicles.  The severity of hazard seem indisputable : reports show between 20 and approximately 50 deaths per annum in the US alone, from hypo- or hyperthermia.

I have not been able to find publicly-available analyses that could be considered as state-of-the-art, and very little public information about the overlap of functional safety and CPD.

Q1 : Can any readers suggest references or published info to support hazard analysis of CPD ?

I am seeing different arguments related to Exposure to this hazard, and different Controllability arguments.

I have seen at least one (I believe fatuous) Controllability argument that goes like this : "Parents can simply not leave their children behind in cars. That is completely within the control of the driver: therefore that situation is highly controllable". Leaving aside cases where children access and become trapped a car without the knowledge of an adult, and the somewhat callous victim-blaming inherent in this argument, when does the "Controllability" related to this scenario begin ?  I might argue that it begins once the child is in the car, unsupervised and the CPD system should act. That is, considering an event where a child has been left behind in a car, and the CPD system fails in a way that prevents notification of this event, that's where the controllability begins.

Q2 : How do the experts see Controllability here ?

An Exposure argument that I have seen goes like this : 30 children die per year, 300 million vehicles in the USA, therefore rate = 1 in 10,000,000 (10^-7), therefore Exposure to this road situation is far below E1 (for example, if using frequency of exposure from (informative) Table B3 in current ISO 26262), therefore Exposure is estimated to be E0 (in ISO 26262 terms).  E0 is defined as "incredible" (in the sense of unbelievable), comparable to the chance of an aircraft landing on your local highway.
I have a big problem with this : 50 deaths per year in a single country alone is not "incredible".  But it is an argument based on measured incidents, and it follows the logic of estimating Exposure classes in ISO 26262 (an order of magnitude between classes). The figures used don't actually measure the real exposure to this situation; I think they are intended as an objective proxy for a case where we don't have a better measure of the real incidence of leaving children unattended in cars.

Q3 : Is "Incredible" really a valid and believable outcome here?   If not, what are the alternatives ?

The combination of arguments such as those above (particularly the argument of E0), leads to the quite odd conclusion that a safety system designed to prevent the deaths of children is not safety-related and ISO 26262 does not further apply.

Frankly, I don't believe it.  I'd appreciate the wisdom of colleagues here .

Thanks,

Mal.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.techfak.uni-bielefeld.de/pipermail/systemsafety/attachments/20250506/ca6a19b3/attachment.html>