I'm here again with the same question as always. Is there a way to correctly get the number of people (and identification of them) in a room? I don't need a 10cm accuracy but it needs to detect when somebody enters a room or exits.
Technology of course did advance a bit in the last few years and I've read a lot about Bluetooth LE. Here is actually a design for a small bluetooth LE beacon: Dr. Michael Kroll
The real question is ... can this (combined with a Bluetooth reciever) actually be used to detect when someone is in the room or not?
I know that RSSI can't really be used for indoor positioning-distance calculation if you have a single reciever. But could this module work? You get an RSSI value from each reciever and in theory you should be able to get the distance. Of course I have no idea if this could work in a real world. These things have a range of 8m so in theory you wouldn't even need RSSI because a person would probably drop out of range when leaving the room (maybe it's worth mentioning that I live in an older house with brick walls and I would be using this to control my home automation system).
The below seems to indicate that signal strength may not a reliable indicator of distance. If signal strength were a good indicator of distance, you would see it being a lot.
PeterH:
If it's just for home automation, do you really need to know anything other than whether the room is empty?
You don't really. But even figuring that out is hard. PIR sensors are useless from my experience because they assume that somebody is moving and you can't have an immediate feedback when somebody leaves the room. You also have a lot of blind spots with just one sensor in one room. I've tried counting the number of people that enter and leave the room but even that is unreliable.
And I would love to actually identify who is in the room.
zoomkat:
The below seems to indicate that signal strength may not a reliable indicator of distance. If signal strength were a good indicator of distance, you would see it being a lot.
I don't really get how this is connected to my problem. This is collecting signal data from wifi devices (or only access points - I am not sure) around the city. And that really doesn't have much to do with localization of one or multiple sources of RF transmission.
Is it one room in isolation, or (seems more likely) detecting this for multiple adjacent rooms? Any RF based solution is likely to be problematic if you're trying to cover multiple rooms, since signals won't go through walls when you want them to but definitely will when you don't.
PeterH:
Is it one room in isolation, or (seems more likely) detecting this for multiple adjacent rooms? Any RF based solution is likely to be problematic if you're trying to cover multiple rooms, since signals won't go through walls when you want them to but definitely will when you don't.
It's for multiple adjacent rooms. Well for the moment I just want one room. I need to know when a person enters/leaves this room. So I don't need range to cover the whole house, but I need to limit it to one room only. Or at least accurately distinguish between a person that's in the room and a person that isn't.
Are the people allowed to be forced to carry a device to identify them?
The reason I ask is while it may be unreliable to count the entrance and exits, maybe if you do something like an RF scanner at the doorways with unique rfid tags on each person. That way when a person passes through the gate, then you know, theoretically, where they are. Though you have to make the assumption the room/rooms were empty.
Note, my knowledge of RFID is mostly academic and possibly been corrupted by science fiction, so I am not 100% confident in this solution and have no clue where to begin.
You could put weight sensors under the carpet and under the furniture to try to detect whether anyone is standing/sitting in the room. That won't tell you who it is.
You can use a PIR sensor to detect whether anyone is moving in the room. That would give you some clues about occupancy but it'd give you a lot of false positive and false negative results, and wouldn't tell you who it was.
You could position an IR camera somewhere that covered the places people tended to occupy while excluding major heat sources, and use video object detection to detect warm bodies. That wouldn't tell you who was there.
You could use visual face recognition to recognise when there was a face in the room, and (if the face could be recognised) who it was. I don't think it is currently feasible to do this well enough to be reliable, but it would certainly be possible to get something crude working, and it's an area which is improving rapidly.
You could give all of your friends hats that had a modulated IR emitter and use an IR sensitive camera to detect and identify them. That would tell you which hats were in the room, but not which people.
You could install a web cam and pay somebody to watch the video and tell you who was visible in it whenever you asked.
You could put security turnstiles on all the entrances and exits and require people to log in and out of the room.
Otherwise, I think the feasible options have already been covered in this thread and the other threads you have created previously for the same question.
mirith:
Are the people allowed to be forced to carry a device to identify them?
The reason I ask is while it may be unreliable to count the entrance and exits, maybe if you do something like an RF scanner at the doorways with unique rfid tags on each person. That way when a person passes through the gate, then you know, theoretically, where they are. Though you have to make the assumption the room/rooms were empty.
Note, my knowledge of RFID is mostly academic and possibly been corrupted by science fiction, so I am not 100% confident in this solution and have no clue where to begin.
Yes, you can force people to wear a tag.
Your solution is doable but passive RFID has a bit too short of a range for it to work like this - or it would require for a person to actually "scan" the tag.
PeterH:
Otherwise, I think the feasible options have already been covered in this thread and the other threads you have created previously for the same question.
The reason why I made this topic again is because there were a few advances in RF communication during that time. Both Bluetooth LE and NFC have become available for a reasonable price. And both of them could be used for this purpose (Bluetooth LE probably more than NFC). I am just interested if anyone actually did any experiments with these technologies for this purpose.
I don't know your purpose for detecting if people are in the room, but I do have a brief story to relate about occupancy sensors on light switches.
My doctor had motion sensing lighting in their offices. I was in a room waiting for doctor a long time, when the lights went out because the sensor was not able to detect where I was sitting. Waving my arm didn't help.
My first thought was to just wait in the dark, when I decided I should investigate why it was taking so long. I was on crutches at the time, so was hesitant to try to get up in the dark to try to re-activate the light. I finally figured I could just waive my crutch in the direction of the light switch, which worked. It turned out the staff had forgotten to tell the doctor I was waiting, and I would have been waiting indefinitely if I had not acted.
It worked out OK for me. But, I can certainly see situations where it might not go so well. An elderly person with unsteady gait might may not be able to figure a way to get the lights back on. They also may not be as steady on their feet in pitch darkness and again when they are suddenly blinded by light when it does come back on if they do stand.
cadcoke5:
I don't know your purpose for detecting if people are in the room, but I do have a brief story to relate about occupancy sensors on light switches.
My doctor had motion sensing lighting in their offices. I was in a room waiting for doctor a long time, when the lights went out because the sensor was not able to detect where I was sitting.
-Joe
Bad installation which contravenes guidlines.
Normall used in large offices to turn off the majority of lights. Some have to remain on though for safety as you discovered.
Is there a way to correctly get the number of people (and identification of them) in a room?
Then the so called occupancy detectors would not cut it.
If they are the ones used in buildings they are just PIR ( passive infra red) detectors and they require movement to trigger them.
We had one in a toilet at my university and when you were in a stall and they cut out there was little you could do to activate them. At night there was no other light in the room and no one about to trigger it.
