Hi folks,
I'd like to sense people moving around a huge space but I don't want to use computer vision also I though about acelerometer in each person but it could be expensive, so does anybody has different ideas?
thanks
Hi folks,
I'd like to sense people moving around a huge space but I don't want to use computer vision also I though about acelerometer in each person but it could be expensive, so does anybody has different ideas?
thanks
I am a newbie, so this might be stupid..
Depending on the size of the area, RF tags could be an option..
There is a company in South Africa TrolleyScan http://trolleyscan.com who uses RF tag to track trolleys in stores....
They also claim they can track animals ext.. maybe you can find some info in their concept..
That is if that fits the definition of huge...
Just my 10 cents ![]()
Radames:
I'd like to sense people moving around a huge space but I don't want to use computer vision also I though about acelerometer in each person but it could be expensive, so does anybody has different ideas?
Define huge space
!
Also important is the number of people expected (the ratio between space and space occupied by people) and the precision required. Do you need to understand the direction they face?
Other methods you might think about:
All these have advantages and disadvantages i.e.,
PIRs [http://www.seeedstudio.com/depot/pir-motion-sensor-module-p-74.html] are normally used by security systems to detect movement in a room. They cover a fairly wide field of view. But you could mount them on the ceiling and perhaps by playing with them a bit make their area of detection more focused, you could get a good sense of where people are in a room.
You could also use force-sensitive resistors on the floor [http://www.globalspec.com/FeaturedProducts/Detail/Tekscan/Custom_FlexiForce_Sensors/18353/0].
I would use a wide angle sonic ranger, constantly monitor its reading and maintains a running average and standard deviation. These calculations are fairly simple. A large (standard deviation) / (running average) or relative error indicates people moving around, a small value indicating no people or people not moving much. You then add some hysteresis. That was my ingredient for my music box project:
http://arduino.cc/forum/index.php/topic,49937.0.html
hey guys,
thanks the answers,
the place is about half a tennis court then people are going to dance around over this and we wish to make the illumination, lights over, change some way according to people movement!
I remember Seeedstudio sold a piezo electric wire sensor that you could measure when someone stepped on it, but I can't find anymore...
They also sell the sonic ranger I mentioned, without danger of wires wrapping around feet.
I worked on the Irisys People Counter:
which can track people reliably. They are quite expensive, though.
hey Leon,
the sensors of irisys looks very interesting however we're not going to spent to much money.
I guess that using a few of ultrasonic range finders like Maxsonar my problem can be solved as well as I don't need the precisely the position of the people but only the parameter of how people are moving around.
is there any open-source, open-hardware in you company? Irisys?
I would suggest the PIR's as well. Set up an array of them surrounding the space of people, and then you can manipulate various effects based on which sensor registers motion. Their max distance is 7m so you might need to get some up on the ceiling as well. But they are damn cheap compared to other proximity sensors so that's a plus.
Radames:
hey Leon,
the sensors of irisys looks very interesting however we're not going to spent to much money.
I guess that using a few of ultrasonic range finders like Maxsonar my problem can be solved as well as I don't need the precisely the position of the people but only the parameter of how people are moving around.is there any open-source, open-hardware in you company? Irisys?
No, it's all proprietary stuff with special sensors, ASICs, germanium lenses, and Kalman tracking software running on a DSP.