Shop OBEX P1 Docs P2 Docs Learn Events
PING))) for detecting people on a path? — Parallax Forums

PING))) for detecting people on a path?

OhmwareOhmware Posts: 11
edited 2011-12-07 15:43 in Accessories
I am working on a DIY home automation project and am looking for a way to detect that one or more people are on the path leading up to the entrance of my home. I would also like to detect if they are coming up the path towards the front door or merely cutting across the range of one of the sensors. Is an array of 3-5 PING))) sensors strategically positioned along the path a reasonable approach for accomplishing this?

I will need to discriminate between a person-sized or larger echo and something like a dog or squirrel. I also need a solution that is reasonably robust for outdoor installation. (It will be protected from rain, etc. but will be subject to a range of outdoor temperatures.)

All of these sensors will be on an IP network and feeding into a Mac to do the heavy computational lifting. (I'm going to try training an ANN with data from the sensor network to do the size discrimination and path-position sensing.)

I am also open to alternative suggestions, as I'm still in the specification stages of this project.

TIA!

Comments

  • Ed TEd T Posts: 50
    edited 2011-11-16 11:58
    Welcome Ohmware,
    You could get away with one PING))) if you can point it down the path and get measurements showing something at closer and closer range. Are there bushes or other things in the way that would give readings that you don't want? For detecting animals you could put the sensor at a height that would see a person but not a dog,cat,skunk, etc. Deer might still be a problem but you might be able to take into account the rate that the object is getting closer to distinguish someone walking down the path versus a deer standing in the path. You would not need measurements over too long a range to calculate the speed of the object down the path towards the sensor.

    --Ed
  • OhmwareOhmware Posts: 11
    edited 2011-11-16 12:22
    There are some obstacles that I believe would make a single sensor less reliable than I'd like, although to your point I should try it with just one to see. I've not worked with these before so this is new territory for me.

    I was just looking at the PADS page in your signature line. Is that your project? Very cool! Does it use a different sensor than the PING))) or just more processing power?
  • ercoerco Posts: 20,256
    edited 2011-11-16 13:21
    You'll want one or more PIR sensors. PINGs are good, but not weatherproof. The PIRs can be made weatherproof, as home IR floodlights clearly demonstrate.
  • Ed TEd T Posts: 50
    edited 2011-11-16 17:21
    Erco has a good point. At a minimum you would need a protective enclosure that kept it dry. The sound reflects well off of hard flat surfaces so you could for example point the Ping down and then reflect it out with a flat surface. Foam can also be used to protect it as the ultrasound can get through thin soft foam fairly well. The other concern is temp variation. The resonance frequencies of the transducers changes with temp. This is not compensated for and will result in reduced performance for big temp changes. Indoors it is not a big deal. In a moderate climate it might not be an issue either. I am not sure how well a PIR sensor will be able to distinguish between the various targets you mention, I have not done much with them, maybe Erco has more thoughts on that.

    Re the PADS sensor, yes that is my design. It uses the same type of piezo sensors as the Ping))). It does have more computing power as it uses the Propeller chip for all the signal processing. It also drives the output transducer with a lot more voltage to get more output.

    --Ed
  • OhmwareOhmware Posts: 11
    edited 2011-11-17 06:52
    Thanks for the informed advice! I could probably deal with keeping them dry, but it sounds like the temperature variations would be a real issue for the PING))). I'll look into using IR sensors instead.

    EDIT: Hmm, it doesn't look like the PIR sensor would allow me to do any discrimination between people and smaller targets like pets, nor would it provide any range information to work with. I am hoping to do a bit more than just light up the path any time something moves. Is there a better starting place than the PING))) that would still provide equivalent functionality?
  • TtailspinTtailspin Posts: 1,326
    edited 2011-11-17 08:31
    Sooo... what is the miniumum operating temputure of the ping sensor?

    I had no idea that my Ping would not work in cold weather, Darn it, I thought It was doing great for the past year too...
    Outside, on the porch, with frost on it...

    What should I do now? now that I found out it wasn't working for the last year,(I guess I only thought it was working.)
    I mean it would tell me when someone was aproaching the porch, but I guess it really wasn't working after all...

    Hrmm... what to do?, what to do?

    Oh wait, I know what to do. Keep using the Ping, Because I know the Ping will work in the cold weather. :thumb:

    -Tommy
  • Ed TEd T Posts: 50
    edited 2011-11-17 09:32
    Tommy,

    Glad to hear your Ping works successfully in your cold outdoor application. Good data point, but "working" is application dependent. I didn't say the Ping would not work in cold weather, I said the sensitivity of the Ping versus temperature might be a concern. In your application this is not an issue, in Ohmware's, I don't know. That is why I mentioned it as a concern.

    If you look in the data sheet you will see the Ping is spec'ed at 0 to 70 C. Sensitivity variation with temp is not given and I have not measured it.

    --Ed
  • OhmwareOhmware Posts: 11
    edited 2011-11-17 09:37
    It's clear I will have to do some experimenting anyway, and some device characterization will add to the fun! So I will have to test my setup over a range of outdoor temperatures… Good to know!

    Ed, I saw the thread where you were testing the range of your PADS design. Inspiring stuff! How did you generate those nice pretty graphs?
  • TtailspinTtailspin Posts: 1,326
    edited 2011-11-17 10:41
    Oh hey, sorry about my response, that was a bit harsh as i reread it...
    I sort of panicked when I thought I saw the Ping getting summarily dismissed because it was cold outside..

    The Ping can give false alarms in a very strong rain storm, and I understand that falling snow can be detected also.
    But I don't have any falling snow to test... I think we have laws about falling snow in Pismo Beach...

    Again, I apologize for my sarcastic response, sometimes my other hobby spills over into this hobby.

    -Tommy
  • ercoerco Posts: 20,256
    edited 2011-11-17 15:05
    Tommy: Maybe you forgot to tell your PING that it shouldn't work outside. Remember, bumblebees can only fly because we haven't told them how impossible it is. :)

    Good feedback on your PING installation, I would have thought frost & humidity would have ended its working career prematurely.
  • Ed TEd T Posts: 50
    edited 2011-11-17 16:41
    Tommy - no problem... I know you should not get between a man and his Ping)))
    Ohmware - I used a program called DPlot.I bought it a few years ago. I looked at the website (dplot.com) and it costs a little more than I remember paying.

    --Ed
  • OhmwareOhmware Posts: 11
    edited 2011-12-06 20:01
    I'm still using the PING))) for my prototype, but I did find a variety of ultrasonic distance sensors that are packaged for outdoor applications at <http://www.maxbotix.com/>. Pricey, though…
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2011-12-06 22:36
    Have you also considered the X-band motion detector? It can't distinguish coming, from going, from passing by, but for your purpose it might have an advantage in that it can see through walls. It would take a couple of them with overlapping fields to decode direction of motion, same as with the Ping)) or the PIR. For reliability, you might want multiple modalities.

    Here is a reference to doppler radar on the Memsic site that might interest you too. I'm not sure where that product stands at present or how well it works, but it sounds good. Can apparently be purchased here.
  • OhmwareOhmware Posts: 11
    edited 2011-12-07 10:38
    Tracy Allen wrote: »
    Here is a reference to doppler radar on the Memsic site that might interest you too. I'm not sure where that product stands at present or how well it works, but it sounds good. Can apparently be purchased here.

    Wow that looks awesome! I've put a call in to these guys and will likely be ordering one to play around with. Thanks for the link!

    My original concept for this project was inspired by a doppler radar system I worked with in the military. I did look at the X-band detector (and have one) but the lack of any range data makes it unsuitable for what I have in mind. This BumbleBee sensor is potentially ideal if it can be interfaced easily enough!
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2011-12-07 14:10
    Please do let me know how that turns out. I've been curious about those for some time but never followed up, and there doesn't seem to be any more about it on the Memsic site.
  • ercoerco Posts: 20,256
    edited 2011-12-07 15:43
    Tracy Allen wrote: »
    Here is a reference to doppler radar on the Memsic site that might interest you too.

    Funny, that reminded me of the book "MIG Pilot" when it detailed the crude but effective high-power radar in the Russian MIG fighter. It had a simple toggle switch. Nothing fancy, just on/off. After a flight, a maintenance worker was leaning into the cockpit to service something. When he felt a "heating" in his intestines, he realized that the pilot left the high-power radar on, and the focused beam was nuking the technician's innards.

    I guess one alternative solution to detecting people on a path might be to beam high energy microwaves onto the path (take the door off a radarange oven and bypass the door safety interlock), then listen for the screams of passers-by.

    No good?
Sign In or Register to comment.