Hi all, I have some thoughts for a project and my lack of arduino experience (and, sensors in general) is leading me to dead ends. I'm new to the arduino so my reading has been focused with Google searches, wikipedia, and these forums.
What I want to do is be able to uniquely identify someone walking (or jogging) by, let's call it a "base station", at a range of 1 meter with a budget of around $200 for a single person solution.
RFID comes to mind as a possible solution, but my reading indicates that the 1m range becomes expensive. I was thinking that having an active RFID tag would work, but those readers seem to be expensive as well. Does anyone know of any RFID solution that might work?
Another idea that came to mind is using a magnetometer to detect when a person is in range. I've seen this done before. Then some item on the person would send out an ID signal to the "base station" (I haven't seen this done). Perhaps XBee modules to communicate?
Using a camera may work, but I have to somehow deal with removing false positives. I wouldn't want to know when an animal gets close to the object, for example. Do you think that having a QR code on the side of a leg and a camera would work? Could I use an IR camera and have an object on the person that outputs a unique IR signature and detect that?
Take a step back and try to be very clear and precise about what you want to do.
When I read the sentence above I think you want to tell if Jane ran past as opposed to John.
Is that really true or do you just want to know that somebody ran past?
RFID is only going to work if people carry tags.
You them seem to branch out into image recognition and an arduino is not going to do that.
An arduino will tell you if something has come close to your sensor but you are right it might be a deer not a person.
To tell if someone is there or not, you can use a PIR detector, passive infrared, but
if you want to specifically identify them, you either need to hang a unique tag of some
sort on them [RFID might not work at 1m], or a large barcode like a license plate, or
use computer vision facial recognition, which is still in the research stage, I think,
as far as reliability goes [unless maybe you're the FBI]. If you look at OpenCV, I
believe they have some facial recognition routines, but it runs on a PC.
Unless you are hell-bent on a diy solution, I'd recommend going for a turnkey solution from whomever supplies this kind of thing for marathons and cycle-races.
RFID would still be the most elegant solution. The readers are expensive only because they are popular right now. It is not especially difficult to just build a reader from scratch and that is what I suggest you research. Hobbyists were doing this long before commercial readers of any type were within a reasonable price.
Here's a place to start. But obviously, you are going to need to ampllify the signal much more than this solution. But the rest remains the same.
Does the tag has to be invisible? If so, you can use IR reflective/emitting ink to paint a pattern on shirt and use IR camera to read that pattern. I have done some code reading fairly complicated patterns (up to 1024 unique patterns). Unfortunately, that project is not open source so I could not release the code here.
Of course IR inked QR pattern and QR algorithm can work, too.
One thing about IR ink is that it is easier to separate background noise and pattern.
codeman9:
What I want to do is be able to uniquely identify someone walking (or jogging) by, let's call it a "base station", at a range of 1 meter with a budget of around $200 for a single person solution.
RFID comes to mind as a possible solution, but my reading indicates that the 1m range becomes expensive. I was thinking that having an active RFID tag would work, but those readers seem to be expensive as well. Does anyone know of any RFID solution that might work?
Why does your range have to be 1m? An RFID sticker tag inside a shoe and a reader on the ground should be able to reduce that range considerably. Even better would be two sticker tags, one in each shoe. Doesn't matter if they are not both the same ID, as you just want to be able to identify the person. Place the reader where the path is narrow.
Henry_Best:
Why does your range have to be 1m? An RFID sticker tag inside a shoe and a reader on the ground should be able to reduce that range considerably. Even better would be two sticker tags, one in each shoe. Doesn't matter if they are not both the same ID, as you just want to be able to identify the person. Place the reader where the path is narrow.
This is an interesting idea...How fast is the acquisition time of one of these readers if the shoe passed over the reader quickly? Could I make the reader "long" somehow? That is to say, more like a pill shape vs. spherical? Or would I need a couple readers next to each other to cover a 3 foot wide path?
codeman9:
This is an interesting idea...How fast is the acquisition time of one of these readers if the shoe passed over the reader quickly?
You'd have to ask the manufacturer or supplier. I've no idea.
Could I make the reader "long" somehow? That is to say, more like a pill shape vs. spherical? Or would I need a couple readers next to each other to cover a 3 foot wide path?
Place your feet together and measure the width of both feet. Mine are about 9 inches. If the reader is in the middle of the path then the nearest foot would be about 9 inches away. Most people don't walk or run with their feet close together or wide apart so I guess your 'acquisition' area would be in the range 0 to 10 inches.
To reduce this, you could narrow the path at this point with (temporary?) obstacles to form a chicane, or build a kissing gate http://en.wikipedia.org/wiki/Kissing_gate with the reader in the 'D'.
As you haven't said where this will be deployed (public or private land), it's hard to give you a definitive answer.
You'll need to protect the reader from damage from being stood on, so I suggest covering it with a hard plastic sheet. You'll also need to ensure that it isn't a trip hazard.
I know little about RFID, so all I'm doing is suggesting alternative methods of placing the devices. You'll have to check that they are practicable.