An FP400 owner reported that two people can merge into one on the map when they pass each other; another user said four people tracked briefly, but did not have longer-term data. That makes a one-person walkthrough a weak test for a shared room.
Before using a person count or zone to turn lights off, I would log what happens when two people cross paths, one leaves, and the other sits still. A momentary count drop should not be treated as proof the room is empty. I would test a sustained whole-room absence condition and make sure a manual light choice can hold through a tracking error.
For people using multi-person presence tracking, which transition causes the most false offs, and what rule has actually kept the lights stable?
Real-time occupancy counting is only available in the Aqara app and not in other systems like Home Assistant.
This means that in other systems, you can only rely on the presence sensor, which reports “occupied” or “unoccupied”. If the sensor merges two people into one and this one person leaves the area, the second person is no longer detected by the sensor and presence is no longer reported. Logs, timers, and monitoring multiple zones don’t help here, as it just appears as though the last person left the zone or zones normally.
Even if you tracked entries and exits across multiple zones to count how many people came in and went out, issues would arise if, for example, several people entered the area individually but left together.
With real-time occupancy counting and multi-zone monitoring, a complex solution could be developed, but it would actually make more sense if the sensor simply didn’t ignore the remaining person in the room.
That’s a useful correction to my proposed test: if the sensor drops the person still in the room, logs can reveal the failure but cannot keep the light on at that moment. I wouldn’t let that single occupied/unoccupied signal turn lights off in a shared room without another independent check or a manual hold. Do Aqara Home automations expose the live person count as a condition, or have you only been able to see it in the app?
I haven’t noticed any errors in the log. It only shows entries like “Presence”, “Someone is approaching”, “Someone left”, “Someone is gonig away” or “Someone entert”.
If unoccupied after a set amount of time, the light turns off; if I didn’t set it up that way, I’d have to turn the light off manually with a switch, and that really defeats the whole purpose. At the moment, the light just turns off and can then be turned back on manually via a switch or voice assistant. That doesn’t happen very often, so I can live with it. That’s better than having to turn off the lights manually every time.
What exactly do you suggest? Always using two sensors?
It cannot be used as a condition in the automations. The current number per zone is displayed in the app, and there is a kind of log showing when and how many people were there, but that is only for viewing in the app.
It is turned off, but it still happens occasionally.
I wouldn’t add a second sensor for an occasional miss you can live with. Your timeout is a reasonable tradeoff; I should have been clearer that I meant avoiding an immediate off from one brief false vacancy, not removing the off rule entirely. If the misses become disruptive, I’d first check whether they happen in the same position or when people cross paths; a second sensor is only worth considering if it covers a repeatable blind spot. Thanks for confirming that the app’s person count isn’t available as an automation condition.