Poor Occupancy Detection

Overnight occupancy detection of my daughter’s room compared to the master bedroom. The results are the same with default room size and custom room size. Yes my daughter’s bed is closer to the sensor, but our bed is in the middle of the room. What is the solution here?

A scale drawing or picture would really help better understand what is going on here. The field of view of a single switch can be somewhat limited. So it is hard to say if this is expected or weird without a proper drawing of space.

In any space where you want detection to be rock solid almost all the time it will always be a good idea to have multiple means of detecting occupancy with different perspectives on the room.

The other thing that can happen with beds is thick covers and mattresses can make it hard for mmWave to see you.

In the case of the smaller room, my daughter’s head is nearest the switch. In the larger room, are heads are furthest from the switch. I acknowledge that only our heads are likely visible to the switch due to covers obscuring us.

If you flip through the manual there is a field of view image and if I understand your diagrams your daughter is probably at the edge of the fov, and smaller so harder to see.

2 Likes

The smaller room is the one that detects occupancy all night long while my daughter is in bed. It is the larger room that fails to detect us in bed.

How far away are you from the switch? I put one in my living room that is having issues with presence and after some reading it looks like the chip is:

and it says 1.5m for heart beat. I did not see anything for presence but it seems like sitting is maybe 15ft but laying down is likely under 10ft. Even on their specs it says that if you monitor someone in bed it needs to point down at them. I’m guessing they are relying on the chest moving with breathing and heart beat and need a more direct on angle to get a stronger signal change.

My understanding of how these work is that they send out a signal, then listen for the reflection. Time it takes gives distance, and the difference between antennas gives angle. Then the change of distance gives motion. When you are sitting up and take a breath you are a large fairly flat surface that gets closer so that would be a large reflection that moves changing distance. When you are laying down and you breath your chest would be moving up and be a smaller change at any given distance and since the amount reflected at any given time would be so much smaller it would be harder to filter out from noise.

All that to say that the sensor in your daughters room is probably getting a better angle and is close enough that the signal has not faded by the time it gets back to the sensor.

I don’t think these will work well for presence while sleeping, but you could use a door sensor and set a rule to log presence for your automatons based on the sensor, but not clear it if the door is closed.

Depending on what you’re trying to do with bed presence detection, I’ve had very good luck with these sensors (mounted on the mattress itself).

Similar to most other users with the Inovelli mmWave, I have excellent tracking when sitting on the bed, but not very good tracking when under the covers and sleeping. You can crank up the stay life to see if it helps, but it’s likely that your placement is just too far away for it to notice those micro-movements while you are sleeping.

1 Like

My wife is the closest to the sensor. Her head would be 9ft away, while her feet are about 6.5 ft away. I think you are correct in that the movements would need to be frequent or large enough under the covers to detect presence.

Primarily I am trying to make the light remain in auto off and on presence detection mode without thinking we are leaving and entering the room all night (i.e. not turn on the lights while we are asleep.

This may be a separate issue in that even though I turn the light off at the switch, when it senses presence it turns on the lights. I would think that if you turn the light off by the switch, it will no longer turn itself on when presence is detected unless you turn the light switch on.

Truly I just want the switch to know that we are in the room so that the lights do not turn on when we are supposed to be asleep (think that we are entering the room every few minutes). As per my response to cfoos1, this may be a separate issue. Shouldn’t the light remain off even if presence is intermittent if you turn the light off at the switch?

I think presence is going to change the switch state when presence state changes even if you turn it off with the switch and has to be that way. Otherwise manually changing the light would lock out presence changing, or you need to add more and more logic on the switch.

I think you are better off disabling the local presence toggling on that switch and handling it on the hub. So long as your mesh is healthy and your hub not overloaded any rules should process fast enough that you should not see a difference. Then you can also add in the bed sensor rohan linked and a door sensor to improve the logic and do things like if you are in bed and the switch detects motion do nothing, but if you get out of bed and it does, preset the light to 5% then turn on so you have dim lights for rr visits. Add a PIR under your bed and you can further prevent false lightups by having it only turn on if it also detects your feet.

I think the onboard presence is best for small spaces you stay for a while, but are not constantly entering and leaving. I set one up in my bathroom and it has been flawless. I even added a rule that if the light is on for more than 2 min to turn on the fan. I have a fan auto off rule where if the humidity is less than living room humidity+8 it executes with a 1hr delay then off command to the fan.

You can switch the mode from occupancy to vacancy to achieve this. In vacancy mode, the switch will never turn on the light after it’s turned off. The whole point of occupancy mode is that the switch comes on and off automatically with presence.

I ended up creating an automation in home assistant that changes the presence detection behavior during night hours. I have it on 0 during sleeping hours, and then it changes to 1 during the day.

i also decided to utilize them in smaller rooms than I originally intended. They work beautifully in my walk in closet and laundry room.