3

I was curious and wanted to see, what it might take to build your own desk availability solution. It seems the way to go is to use some kind of lowband pir sensors, but my Google-fu is rather poor. I can find many sensors and solutions, but all the ones I have found so far is packaged/bundled solutions, that consist of sensors, and then a subscription to a service.

Isn’t it possible just to buy a bunch a sensors, and a hub, and then setup your own independent solution.

I guess I could build it from scratch using something like arduino + sensors, and the have it manufactured yourself, but that is a too big step - no need to reinvent the wheel.

4
  • Perhaps you should look for automotive seat sensor or mower seat sensor parts. Let us know if you found any. Perhaps if a direct search for that does not find anything, you'll have to find the parts manuals for some automobiles or riding mowers and find the part numbers that way. Perhaps for tractors too. Let us know what you found if any. – kalyanswaroop Apr 5 at 20:22
  • 1
    don't forget that a desk, with nobody seated, is not necessarily un-occupied – jsotola Apr 9 at 4:07
  • Hah ! You're right. There's a lot of people standing at their desks now a days ! Maybe it will need a smart camera that detects a human figure and breathing or other movement ! – kalyanswaroop Apr 13 at 15:43
  • I believe what @jsotola means is that someone can take a desk, put their stuff there, and then temporarily leave (to go to a meeting, to talk to a colleague at his desk, to go to the bathroom, to go eat...), which would be detected as an unoccupied desk when it’s in fact not available. – jcaron May 2 at 7:35
1

Desk occupancy detection is one of those things that may seem to be deceptively easy.

The standard sensor for detecting presence is indeed the PIR sensor, because it's cheap, and draws very little power, so it is well suited to running on battery for extended periods of time.

However, be aware that:

  • PIR sensors do not detect actual presence, but movement. And only movement in a plane orthogonal to the axis of the sensor (i.e. if your sensor is facing down, only horizontal movement), and often even in a single direction (e.g. left-right and not front-back). So detecting that someone arrives is easy (that's usually well enough movement to be detected), but detecting that someone is still there is less easy (you usually have to keep a longish timeout).

  • PIR sensors have a fixed field of view. The little translucent dome is really a Fresnel-like lens which brings the IR from direction of each of the facets to a single sensor. If the field of view is too narrow, you may miss movement (quite rare with most sensors). If the field of view is too wide, you may detect movement that is outside the area you're trying to cover, like people walking past a desk, or people at a neighbouring desk. Height will have a large influence on the floor area covered, and desk size will influence whether you may catch neighbours.

  • Sensor placement and orientation are thus very important, but there are cases where it just won't work.

There are tons of PIR sensors out there. Zigbee, Z-Wave, BLE, LoRaWAN, Sigfox, probably Wi-Fi... I would personally recommend Zigbee over the other solutions, but YMMV.

Note that many such sensors are really designed as "intrusion alarms" rather than presence detection, so their behaviour may be subtly different, enough to be annoying.

There are plenty of other technologies that can be used to detect presence, including:

  • Seat movement ("vibration").
  • Seat pressure (with pressure-sensitive resistors or similar things in the seat, like in cars, but that's a bit more complex to set up)
  • Thermal array imaging
  • Shape or face detection using video
  • Various proximity/ToF sensors but they're probably difficult to place and calibrate to get accurate results
  • And many more

Some will detect actual presence rather than just movement. Some will be able to run on battery while others require external power. Some are very cheap while others can be very expensive. Some raise privacy issues. Some can be used to monitor a larger area (a group of several desks) with a single sensor and can be carefully adjusted and calibrated, others much less so.

The choice of the solution depends a lot on what you want to use the sensors for (e.g. if you need realtime updates, or you're just interested in reporting), the environment (desk height, density...) and more.

Hope this gives you a few pointers.

Not the answer you're looking for? Browse other questions tagged or ask your own question.