Why? 20 ms is a very long time in the world of digital electronics. 1 MHz = 1 microsecond ticks, and most embedded MCUs run 10x to 100x faster than that. Sensors and ADCs commonly have microsecond-range response times, too.

The mechanical part (rapid inflation) is more interesting, but for a long time, they did it the "obvious" way: a precisely-calibrated explosive charge, just enough to fill the bag without tearing it apart. Now, I think they're moving toward compressed air, which probably requires some clever engineering to build a sufficiently responsive electronically-controlled valve.

It's like you answered your own question.

The parent described it as "mind-blowing"; I was curious about that.

Maybe they meant it literally, if your head's in the wrong place.

20ms isn't all that long a window over which to filter all that sensor data and determine that you are actually in a crash (and didn't just take a speed bump a little too fast).

False-positive airbag deployments are not viewed very positively by the affected user (even if its strictly better than a single false negative)

A false positive is essentially a false negative (the airbag cannot inflate twice), plus the consequences of knocking the drivers hands off the steering wheel, obscuring their sight and causing a shock that will in most cases cause an accident.

So a false positive seems strictly worse than a false negative to me.

Yes, I guess at speed that is likely the case. It's more the 5-mile-an-hour fender-bender that tends annoy the user when it harmlessly/needlessly deploys the airbags

> around 20 ms after impact.

that's the standard for all airbags

https://www.iihs.org/topics/airbags