Presence detection



  • Has anyone looked into or played with thermal image sensors like Omron d6t? It can detect stationery human presence which would be a huge advantage over PIR or Ultrasound. The module is reasonably priced and is Ic2 output, so it should work with Arduino. Any thoughts?


  • Contest Winner

    @pbcstudios

    The physics of a PIR are essentially accomplishing the same thing, at a fraction of the cost.

    I guess I'd want to understand what benefit you seek in presence detection that has an advantage over the simple, low cost, reliable PIR sensors employed in millions of homes and businesses today?



  • I think the one problem I see the most with PIR is the lack of detection with stationary presence, whether it be watching tv, reading a book, etc. I have read posts where people have to "wave their hands in the air" to turn lights back on after timing out. I have a similar situation in my kitchen, where the lights turn off often while I am preparing a meal. I can reach everything I need while standing in one spot, only my hands are moving (sadly, this motion scenario can be applied to many rooms in my house.). So, I worked around this by setting up an ultrasound sensor to act as a motion sensor where I'm standing.

    I am now working on a sensor to detect presence in my sun room based on sitting at the computer, opening doors, walking through the room, etc. using multiple sensors. I'm curious.... is this the solution? Is it the best or only solution? A custom sensor in each room to detect presence. What happens if I move to a new place? Or rearrange the room?

    Is there a way to make a general sensor that works in every room that detects presence, moving or stationary? After all, turning lights on/off plays a large role in why most people automate their home.

    So, how have others worked around this?


  • Contest Winner

    @pbcstudios said:

    I have read posts where people have to "wave their hands in the air" to turn lights back on after timing out.

    that's an interesting approach to that problem... In our offices at work, we have PIR sensors. When the light goes out while you are 'working' in your office, it is pretty embarrassing. We can't adjust the sensitivity of those sensors!

    You should try it!



  • https://xethru.com/ When this comes down in price it seems like it would be a good solution. Right now it is still a bit pricey.


  • Hero Member

    @pbcstudios said:

    m

    I did something for this for my wife. I have a motion sensor in the study so the lights go on when someone comes in. But the timeout is never right, sometimes the room will be occupied for minutes, other times for hours. My wife got particularly irritated at the lights going out while she was on the PC. She 's very small and the motion sensor couldn't pick her up properly. So she had to get up and wave her arms to get the lights back on. Interesting to watch, as long as you don't laugh out loud :), but irritating for her.

    So I rigged a Door/Window sensor to seat bottom of the desk chair. Now when anyone is sitting in the chair the sensor is tripped suppressing the timer and the lights stay on until they leave.

    But it would be nice to have some sort of passive occupancy detector, not just for this situation but for any room in the house. Right now in the Family Room i have an override schedule that prevents the lights from going off between 7PM and midnight but if someone is there watching TV or reading in the afternoon and doesn't move enough to trigger the PIR then the lights go out. Again, depending on the size of the person and their location they may nor may not be able to get the PIR to trigger with just a wave of the arm.

    I don't know what kind of technology would solve the passive room occupancy problem, and if it exists is it affordable.



  • Another way is to track the amount of people that enters the room. This can be done with a double IR beam horizontally placed next to each-other (IR-A and IR-B). Now when a person walks into the room, you know that IR-A triggers before IR-B, and you increment the room-occupants-counter with 1. If the person leaves the room, IR-B will trigger before IR-A and you decrease the counter.

    The IR barrier can be installed inside the door casing.

    Using this technique you don't need a PIR sensor, unless the room has a window and you want to detect movement without someone passed the IR beams (a burglar maybe?)

    Combine the IR beams with a lightsensor and you can also define that the light should go on only below a certain light level.

    My 2 cents...



  • This site has rather a good review of techniques http://dreamgreenhouse.com/projects/2013/presence/index.php

    From a processing point of view, my favourite is the "Wasp in a Box" algorithm.


  • Mod

    @FreakOfNature said:

    https://xethru.com/ When this comes down in price it seems like it would be a good solution. Right now it is still a bit pricey.

    Given that they sell a simple FTDI "USB connection board" for $350 (roughly 100x overprized) the module itself could sell for $4, if it's also overprized 100x ;-)



  • Thanks for the replies:

    So basically, there really is no such thing as a smart home. Pleased to find out I'm not alone. Custom sensor and algorithms is the only way to go. I asked the question because I'm dating again and after the past year of living alone, I might have company sometime. To be honest, as a single occupant, I already have this figured out. Like the "Wasp in the box", it's easy to track my movements in the house.
    Where am I?
    Most likely in the room where the last sensor went off. Which means shut down all the lights in the house, except for that room. I guess if that special someone comes over, I'll have to override the system and leave a certain amount of scenes disabled. Thanks for the input.



  • @pbcstudios good luck with the dating :wink:


  • Hero Member

    @clippermiami said:

    So I rigged a Door/Window sensor to seat bottom of the desk chair.

    That's gold John!!!!

    Another way for this solution, would be don't turn off the lights if the screensaver is not active ...assuming the major use of the room is to use a PC...



  • Hi,

    Just a small input (I didn't read the article yet) but would not RFID or NFC be an option? At least an alternative for the IR in the door's frame?
    In some Hospitals it exists a RFID system to detect where a newborn or a child is, with a RFID bracelet and sensors in the doors.

    At the end I think that all the solutions are essentially a question of having sensors in entry/exit points and using algorithms.

    Cheers



  • There are two ways to approach this problem. Either with BT decibel measuring, provided everyoen har something with BT.

    The system thinks you are not in the room before the signal strength of the BT device reaches a certain level.

    And also as pbcstudios suggested with thermal detection
    Can this be used?:
    http://www.dx.com/p/mlx90615-ir-human-body-infrared-non-contact-thermometer-module-405701#.VflnUN_tlBc



  • Actually, I think there are thousands of approaches to this problem... it's certainly been well addressed in other HA fora.

    I was rather intrigued by this scientific paper...

    Through-the-Wall Sensing of Personnel Using Passive Bistatic WiFi Radar at Standoff Distances
    http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6020778

    ...which claimed to detect presence using Doppler analysis of ambient WiFi signals at 2.4 or 5 GHz.

    Not that there's a MySensors node for that yet, but I'd quite like to prototype one... or anyone else? (I was wondering if the NRF radio could be called into service as the RF front-end?)



  • I'm glad to hear people are still talking about this post. There are a lot of great suggestions. I'm really surprised that the home automation industry has not really figured out a way to do this. My original thought using the thermal sensors, specifically the Omron d6t, does not look like a promising option. Although the company that produces this item claims that static presence detection is a perfect use for it, I cannot seem to find anyone who actually uses the sensor in that manner. Further searching shows that the module is really only able to detect a present from a few feet away so putting it in a large room would be useless. This is further complicated by the fact that the module is sensitive to light coming in the windows. I will say the non-contact a thermal module posted yesterday does look promising for a few uses.

    For me, I'm probably going to stay away from Bluetooth or RFID sensors because it assumes that everyone has a transmitter on their person. In the end it looks like custom sensors and customer logarithms for each room is really the only way to go. All I can say is I couldn't possibly afford to do that if it wasn't for the work of Henrick!



  • I think what ericvdb said with the IR sensors is probably close to the best solution. If you figure out a way to count people at each doorway you could do a sudo presence detection.

    So, if you count people as they enter, and at each threshold remove a person from the room they are leaving and add to the one they are going to you should have something that is rather accurate.

    This could have issues if you carry someone from one room to another, but you could add pir to rooms to correct exceptions. Something along the lines of, you carry someone into the living room, it counts one person there, you go to kitchen, it reduces to 0 and turns off lights, but you have an exception that if there is motion and zero people it changes to one person in the room. Would just have to be sure your hvac does not set of the pir.


Log in to reply
 

Looks like your connection to MySensors Forum was lost, please wait while we try to reconnect.