Postion detection of a moving object in a room.

I am currently working on a project. I am trying to detect the position of a moving body in a room and plot it on a graph. I am facing several difficulties.

What kind of sensor should I use ?
There are more than one objects in the room so I cannot use a sonar sensor since I want to plot the position of only the desired body. I am being told the the GPS module will not be accurate for detecting position in a room.

Someone suggested placing an IR LED(s) on the object. How will that work ? Should I keep an IR receiver at the four ends of the room ?

Also, is preprocessor the desired library for plotting graph ?

RFID tags, and measuring the RSSI level from multiple points.
Trilateration. It works quite well.

bing_007:
I am being told the the GPS module will not be accurate for detecting position in a room.

Its not that GPSs are inaccurate indoors, they usually just do not work at all.

RFID tags, and measuring the RSSI level from multiple points.
Trilateration. It works quite well.

Have you done this? I'd be highly dubious, because reflections from walls and other objects are likely to give very variable results.

Ultrasonic sensors with a moving transponder might work better.

Allan

An ultrasonic sensor can work for 1 axis detection of objects in a room. Scan the room, in 1 degree increments, store the scans, re-scan and compare depth measurements against original scan. I use a similar approach for my ultrasonic sensor. I, also, if the distance detected has changed fire off a 2nd ultrasonic detector at the same angle to confirm distance changed at angle.

LIDAR would be a better choice for what has been described.

Either a LIDAR or Ultrasonic sensor will return 'distance' and angle.

Two distance measurements units, laced at 90 degrees from each other would give a better room painting.

What exactly are you trying to track? an insect? a human? a bullet?
The way you are going to detect something will depend on what it is and also how fast it is going.

There are way too many approaches to list all of them, here's a few that might help:

  1. Visual detection

This can be achieved in lots of ways, and depends on the setup you want to have, object you want to track, how many axis, price, etc.
If you just want 1-2 axis then you could use an off the shelf camera and use some software or OpenCV to track it's position.
If you want 3 axis you would probably need a multi camera setup, you could use standard cameras but it would require a very good tracking software. Or you could use IR based tracking cameras, this is what Motion Capture uses but this is very expensive and would require physical trackers placed on the target.

  1. Auditory detection

You would need a multi microphone, possibly binaurial setup. And you would need a sound emitting device. This setup is prone to error in noisy environments and you would need to do a lot of calibration.

  1. Radio wave based detection

You would need a multi transmitter or receiver setup, 3 of these would give you 2D position while 4 gives you 3D (See trilateration). You need to keep the sources synchronized. This is prone to error in setups that are not good RF-wise (See reflection)

Overall visual detection can give you a cost effective solution and the accuracy can be "fixed" with software. The other solutions require more hardware but give you a solution with close to zero processing needed so they will be better in realtime applications.

Also I would like to give you this: IEEE 802.11mc - Wikipedia which would be a solution that uses existing hardware and software to achieve what you desire but the accuracy and reliability. is questionable

There are way too many approaches to list all of them,

No.

The only system that will work is this:-

It is not cheap but it can work. All other suggestions are theoretical solutions only.

The other system that works is active RFID as mentioned above.
Our test system worked well with three nodes in an office environment, while tracking 3/4 tags in an area of aroun 5-6m squared during demo to aspecific customer.

Accuracy was around 1-1.5m because of jitter & reflections and arbitrary attenuation withinin the scale of the test environment as suggested. (People sitting / walking around etc.

It performed better in a larger area because the signal was actually too strong, and more nodes would obviously enhance the accuracy.

In the test scenario, each node was connected back to the ‘spatial’ processor via serial, and tracked moving tags easily on a mock map.
(Thanks to Ray K for the software)

Look up the amg8833 sensor... Panasonic grid eye

LandonW:
Look up the amg8833 sensor... Panasonic grid eye

Come off it an 8 by 8 thermal camara is not going to cut it.

teunman:
What exactly are you trying to track? an insect? a human? a bullet?
The way you are going to detect something will depend on what it is and also how fast it is going.

Idahowalker:
An ultrasonic sensor can work for 1 axis detection of objects in a room. Scan the room, in 1 degree increments, store the scans, re-scan and compare depth measurements against original scan. I use a similar approach for my ultrasonic sensor. I, also, if the distance detected has changed fire off a 2nd ultrasonic detector at the same angle to confirm distance changed at angle.

LIDAR would be a better choice for what has been described.

Either a LIDAR or Ultrasonic sensor will return 'distance' and angle.

Two distance measurements units, laced at 90 degrees from each other would give a better room painting.

If I use ultrasonic sensor, how would I distinguish between other objects ? Like I said there are more than one moving objects in room but I only want the position of one..

Like I said there are more than one moving objects in room but I only want the position of one.

Still keeping what theses objects are a secrete then?

Only by knowing what you are trying to do exactly can we offer accurate help.

It is like saying you want to move some boxes to your brother's house and asking what is the best way to do it?

Then when we ask, how many boxes do you have, how heavy are they, how far away does your brother live, how quickly do they need to be delivered you won't say?

So we don't know whether to say, walk, drive or DHL.

Grumpy_Mike:
Still keeping what theses objects are a secrete then?

Only by knowing what you are trying to do exactly can we offer accurate help.

It is like saying you want to move some boxes to your brother's house and asking what is the best way to do it?

Then when we ask, how many boxes do you have, how heavy are they, how far away does your brother live, how quickly do they need to be delivered you won't say?

So we don't know whether to say, walk, drive or DHL.

These objects are humans..
I was looking into RFID tags, which ones should I get so that I have enough range.

Only by knowing what you are trying to do exactly can we offer accurate help.

Stone >>> Blood

Now when you say track do you mean position within the room? If so what degree of accuracy are you looking for?

There is my reply #7 you seem to ignore. There is the reply #8 that many people are dubious about. Having worked in a company that made a similar device, in this case something that registers if a person has walked outside of an area. I know it is not that simple to use a transponder ( the real name for an active RFID system ).

Just for the record, #8 was a proof of concept demo for a emergency equipment tracking system for assets and people on a fire scene.

It worked remarkably well in a room sized demo, but was intended for a ~200m x ~200m application to track assets and emergency responders in a 3D site (e.g. small building) with nodes located within three or more GPS equipped vehicles arnitrarily located around the site - providing inter-node co-location of the assets.

The actual area covered was approximately 20% larger than the area of the nodes themselves.
My involvement was over after the POC demo.

Grumpy_Mike:
Stone >>> Blood

Now when you say track do you mean position within the room? If so what degree of accuracy are you looking for?

There is my reply #7 you seem to ignore. There is the reply #8 that many people are dubious about. Having worked in a company that made a similar device, in this case something that registers if a person has walked outside of an area. I know it is not that simple to use a transponder ( the real name for an active RFID system ).

My exact motive is to display the real time position of the person (object) in a graph. By real time I mean to say, that when the person moves, I want the point on the graph to move along i;e showing the exact position.
I hope I made myself clear.

Grumpy_Mike:
Stone >>> Blood

Now when you say track do you mean position within the room? If so what degree of accuracy are you looking for?

There is my reply #7 you seem to ignore. There is the reply #8 that many people are dubious about. Having worked in a company that made a similar device, in this case something that registers if a person has walked outside of an area. I know it is not that simple to use a transponder ( the real name for an active RFID system ).

I want to show the exact position of the moving person (just one selected person, in a room which has more people) on a graph i;e when the person moves, I want the point on the graph to move as well showing real time co-ordinates.
Hope I made myself clear.

Hope I made myself clear.

Sigh,
No.

“The exact position “ is not something that can be acheaved because the word exact is not very well defined. You need to put a limit on that, like to the nearest millimetre or nearest meter, or nearest mile.

Why are you ignoring my suggestion? It is the only one that works and is remotely affordable.

Put a red hat on the person you want to track. Mount a camera in the center of the room ceiling pointing down. Use opencv to track the red object. You need a raspberry pi or PC to run opencv.

If you do not want to use opencv, there are robot computer vision cameras with dedicated ARM processors to do the pixel crunching.

https://pixycam.com/

The camera resolution is not very high so the location will not be precise.

See pixycam colors codes to avoid the problem of other red objects in the room.