Shop OBEX P1 Docs P2 Docs Learn Events
Best sensor strategy for locating a person in a house? — Parallax Forums

Best sensor strategy for locating a person in a house?

scurrierscurrier Posts: 32
edited 2010-10-20 18:12 in Accessories
Hello,

I am new to parallax and the propeller chip, but I bought the education kit and the pro development board and I am really pumped to start making some useful gadgets! (First post on the forums!)

One idea that I have is a gadget for turning lights on and off in rooms as you move in and out of them. It might use X10 commands or some other method to turn the lights on and off. I am only just starting to think about this, but I am really at a loss as to what the best sensor strategy would be. I am thinking that each person would have to carry around some kind of dongle or something on their keys. Some issues:
1. How to keep the dongle relatively low power? Recharging nightly might be acceptable.
2. How to prevent interference between dongles and the detection method?
3. Would personalization be possible? i.e. being able to detect who is in what room. This could allow the system to personalize your light preferences (or temp or something in the future).

I'm sure I'll think of more problems to overcome. I was thinking that some kind of ultrasonic triagulation system might work. Just have the person carry around a small emitter at a certain frequency that chirps every eigth of a second or so. You could personalize based on the frequency. Drawbacks are you have to put up base stations all over. Maybe that's not so bad though.

What other sensor strategies are out there?

Thanks,
Shaun

Comments

  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2010-10-14 10:26
    If you're not trying to identify individuals, but just the presence of someone in the room, an IR motion detector would probably be adequate.

    -Phil
  • W9GFOW9GFO Posts: 4,010
    edited 2010-10-14 12:01
    This guy has done something similar. The details of the "pin" that the user wears are lacking though.

    Rich H
  • hover1hover1 Posts: 1,929
    edited 2010-10-14 13:10
    I didn't see the Altair 8800 so he can keep doing updates on 4K Basic :)

    Jim
    W9GFO wrote: »
    This guy has done something similar. The details of the "pin" that the user wears are lacking though.

    Rich H
  • hover1hover1 Posts: 1,929
    edited 2010-10-14 13:19
    As Phil mentioned, IR is used in quite a few energy efficient buildings right now. It's required in some states. (Come into a bathroom, light turns on, leave, the light turns off after 2 minutes). The Parallax PIR sensor is good for that scenario.

    http://www.parallax.com/StoreSearchResults/tabid/768/txtSearch/PIR/List/0/SortField/4/ProductID/83/Default.aspx

    For personalized entry, you would probably use a RFID Tag, Card, or Fob. The new Parallax Read/Write system would be good for that. The only problem is the reading distance. 3" max. The user would have to bring the card to within 3" of the reader. I'm sure Bill's cost a lot more.

    http://www.parallax.com/StoreSearchResults/tabid/768/txtSearch/RFID/List/0/SortField/4/ProductID/688/Default.aspx

    Jim


    scurrier wrote: »
    Hello,

    I am new to parallax and the propeller chip, but I bought the education kit and the pro development board and I am really pumped to start making some useful gadgets! (First post on the forums!)

    One idea that I have is a gadget for turning lights on and off in rooms as you move in and out of them. It might use X10 commands or some other method to turn the lights on and off. I am only just starting to think about this, but I am really at a loss as to what the best sensor strategy would be. I am thinking that each person would have to carry around some kind of dongle or something on their keys. Some issues:
    1. How to keep the dongle relatively low power? Recharging nightly might be acceptable.
    2. How to prevent interference between dongles and the detection method?
    3. Would personalization be possible? i.e. being able to detect who is in what room. This could allow the system to personalize your light preferences (or temp or something in the future).

    I'm sure I'll think of more problems to overcome. I was thinking that some kind of ultrasonic triagulation system might work. Just have the person carry around a small emitter at a certain frequency that chirps every eigth of a second or so. You could personalize based on the frequency. Drawbacks are you have to put up base stations all over. Maybe that's not so bad though.

    What other sensor strategies are out there?

    Thanks,
    Shaun
  • MikeKMikeK Posts: 118
    edited 2010-10-15 16:13
    I put an IR sensor in front of my TV, about 8' directly across from the couch. Sitting on the couch reading a book (TV off) there was not enough movement to keep the lights on. I think I had it on a 10 minute delay before it decided to turn off the lights. I removed the device before my wife and son decided to remove it for me.

    Alternately, if you position an IR sensor so that it sees when someone enters a room, you have to be sure it's not positioned so that it sees you when you walk past the doorway to that room.

    Mike
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2010-10-15 17:06
    'Probably need a "couch potato detector" like those seatbelt switches they put in car seats to sense someone sitting there. :)

    -Phil
  • zoopydogsitzoopydogsit Posts: 174
    edited 2010-10-17 03:59
    I recall reading in the past (probably 4+ years ago) about Roomba tracking using a bluetooth transmitter and two recievers providing the location via differential signal strength. They then plotted it to show a map of where in the room it cleaned.

    I searched and was not able to find and reference (the Internet is like that... presistence is not the same as human memory). However I found an interesting link of some folks tracking a roomba with WIFI signal strength.
    http://siibot.blogspot.com/2008/03/wheres-my-roomba-wifi-signal-measuring.html

    I haven't looked closely at the Zigbee/etc. but if it gives you a signal strength from the end node then you may be able to get the location.

    In practice I don't know if you will get it to work, much less work in a house or a larger area, nor how reliable it would be. Though if you have the time and put in the effort, then who knows.... If you did, then the really cool part is you could also use the bluetooth transmitter for other neat stuff like biometric data (hint - maybe use if for the location of someone who's fallen and what's their current health status). Or use it to send some other interesting data.

    Anyone else used the Zigbee and gotten signal strength data?
  • kwinnkwinn Posts: 8,697
    edited 2010-10-17 09:23
    IR works well in areas where people tend to be in motion. At home or in the office it tends not to work as well. Perhaps a combination of IR sensor and an IR dongle would work. At the end of it's timeout period the IR sensor module sends out a poll character, and the dongle responds.
    The ultimate of course would be a camera system that can see when the room is occupied.
  • lardomlardom Posts: 1,659
    edited 2010-10-17 09:31
    I agree with Phil. PIR should suffice. I would only think of a dongle if the system was required to 'unlock' something and that is an advanced project.
    I have thought about "ultrasonic triangulation" as a way to enable a robot to navigate around a room or a warehouse. I learned something about the way GPS works and I wondered if the basic principle could be applied without satellites.
  • scurrierscurrier Posts: 32
    edited 2010-10-17 23:31
    Some interesting ideas here guys. Sounds like ultrasonic might still be the best if the person might not be moving. Any of the radio ones I am worried might be too power consuming.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2010-10-17 23:36
    Back to IR for a sec. What these sensors detect is "body heat in motion." If the body heat isn't moving, perhaps jiggling the sensor a little (e.g. with a servo) could still detect it.

    -Phil
  • ElectricAyeElectricAye Posts: 4,561
    edited 2010-10-18 07:34
    Phil Pilgrim (PhiPi) wrote: »
    ...If the body heat isn't moving, perhaps jiggling the sensor a little (e.g. with a servo) could still detect it.

    I, too, have wondered if this would work - however, what I would like to test is something like a chopper wheel. Instead of moving the sensor, chop its field of view with a rotating mask. Y'think that might work? :) Please let me know if you try it.
  • lardomlardom Posts: 1,659
    edited 2010-10-18 07:46
    scurrier, let me guess. If you are sitting in a specific spot you want the sensors to detect it? You'd need a lot of resolution. I would think a camera would be involved as well as assembly code. It sounds advanced to me.
  • LukeHLukeH Posts: 22
    edited 2010-10-18 11:21
    ElectricAye wrote: »
    I, too, have wondered if this would work - however, what I would like to test is something like a chopper wheel. Instead of moving the sensor, chop its field of view with a rotating mask. Y'think that might work? :) Please let me know if you try it.

    That would work. It is how early generation heat seeking missiles were built. You'd need a lens to focus the energy too, but that is pretty trivial.

    Click here: http://www.ausairpower.net/TE-IR-Guidance.html and scroll a little more than halfway down, to the pictures of the different reticles, for a general idea of how it works.
  • scurrierscurrier Posts: 32
    edited 2010-10-19 20:51
    lardom wrote: »
    scurrier, let me guess. If you are sitting in a specific spot you want the sensors to detect it? You'd need a lot of resolution. I would think a camera would be involved as well as assembly code. It sounds advanced to me.

    Within 3-4 feet or so should be fine. I'd think you could do it with the speed of sound. Maybe you could tune the system by moving the sensor around to known locations and defining some kind of characterization map of your specific room geometry.
  • lardomlardom Posts: 1,659
    edited 2010-10-20 00:05
    I'm not quite a newbie anymore but I will be a student for a long time. I finished a project in Spin that was first coded in Pbasic. I thought I would have several sensors, an SD card reader and... I ended up with a user-programmable machine that didn't need any sensors or a card reader. I like the mechanical simplicity. There are fewer things to break down.
    From what you've described you could accomplish with IR, a camera or a 'couch potato sensor'. An ultrasonic system that requires you to move sensors around seems a bit much
  • John R.John R. Posts: 1,376
    edited 2010-10-20 18:12
    Phil Pilgrim (PhiPi) wrote: »
    Back to IR for a sec. What these sensors detect is "body heat in motion." If the body heat isn't moving, perhaps jiggling the sensor a little (e.g. with a servo) could still detect it.

    -Phil

    I worked with a company on a "zero force" sensing system to prevent people from being crushed. We used a two phase approach, one was to have a series of UV sensors at floor level, and monitor "feet" moving in and out. The other used PIR.

    Indeed, moving the sensor works to see a stationary body. We used a piece of "muscle wire" to rock the sensor back and forth.

    There is one type of scenario where the PIR doesn't work well at all, and that's when the ambient temperature reaches 90 Deg F or so. Depending on the sensor, you reach a point where the body isn't warm enough over the ambient to "see" the difference.

    The same thing can happen if the people are extremely well insulated, as in waring some really good parka gear.

    At the time, I was more involved with the software, and wasn't too involved on the hardware side, so I don't really have any more details than that.

    John R.
Sign In or Register to comment.