There's probably no reason why these kinetic switches can't also be used for detecting other events like doors opening/closing etc. I feel like a radio signal is a bit more reliable and easier to detect than high frequency sound.
I also think calling these a "sensor" is a bit of a stretch. They detect events but have no knowledge of the current state of the thing they're sensing. E.g. the can detect a door opening/closing, but have no idea if the door is open or closed at a given time
1) 93.75% success rate in controlled conditions, 92.1% in a somewhat-realistic deployment scenario - too low for reliability. I wouldn't use something like that to trigger smart home automations.
2) Range hardcapped at ~1m due to how ultrasound works, you can't centralize detection. Their answer is to give everyone in the household a wearable receiver, which is eeeeeeeh idk, doesn't look consumer-friendly to me.
3) Paper suggests a mix of durable and consumable parts for the transmitter. Their numbers show that the 3d-printed PLA cantilever needs to be replaced every 900 cycles or so. Should work fine, but...
4) ...every transmitter pair needs to be tuned per-setup, every time. Not a plug&play in the consumer sense.
Also I guess this might be annoying for pets that can hear well beyond 20 kHz.
If so, will it penetrate through walls?