PDA

View Full Version : giving up on the web



jmm
01-08-2010, 02:24 PM
i have no idea what im getting into and i have looked on every web site i can find. i know this mught be a simple project for some of you out there but like i said i have no idea...

im a furniture builder and i have built some flatscreen stands with accuators befor but i have had a request for one that will house a smaller television, lift out of its housing with an accuator and will follow the owner maybe with a motion sensor around the room so that it can be seen. it will only be 180 degrees of movement. i have never heard of anything like this befor but i figured that this website would be a good place to start. any suggestions? if im in the wrong place im sorry but i had to try...

Quantum
01-08-2010, 03:08 PM
Lots of people want this type of sensor that will follow a person. But it dosent exist and there isnt a real simple solution. Detecting motion is one thing and not that hard to accomplish but tracking is a whole different thing.

You would need multiple motion sensors and arrange them so each detects a portion of the 180 degree field of sight. This still wouldnt make tracking perfect but is probably the simplest way to set this up. Im sure there are camera and computer setups with some advanced software that would make this work but it would probably cost 2 -3x the cost of the television. If not more.

darkback2
01-08-2010, 03:35 PM
They were talking about a TV that looks back at you, and routes the sound to where you are sitting. If you think about it, a person watching TV should have their face turned towards the TV. Why not use a webcam with face recognition software. The camera could track the persons face, and turn the TV so that their face was centered.

It would definately be expensive...

DB

zoomkat
01-08-2010, 04:34 PM
If the person would be ok with using a tv remote control as a becon to identify his location, then the solution may be more affordable and less complex.

jmm
01-10-2010, 06:34 PM
i like the beacon idea. how can this be done... what will i need?

MikeG
01-10-2010, 06:59 PM
If the person would be ok with using a tv remote control as a becon to identify his location, then the solution may be more affordable and less complex.

Zoomcat, how does that work? How do you use a TV remote to identify location?

darkback2
01-10-2010, 07:50 PM
What if you mounted 2 leds on the TV. Then you could use a wiimote to send an angle to the tv turn motor, which could turn the TV until the LEDs were the right space apart...Then again I like the webcam doing face tracking better.

DB

MikeG
01-10-2010, 08:48 PM
darkback2 and zoomkat can you please explain to the simple folks how the heck to do this?

darkback2
01-10-2010, 10:10 PM
Ok...so we are trying to figure out the position of the viewer, and the position of the screen.

So to start with the screen needs to be mounted on a turn table. I would probably go with a really strong linear servo to turn the TV on its turn table.

Now we need to determine the position of the viewer. If you are going with the webcam idea, then you mount a camera on top of the TV facing out. THe camera would feed video to a computer running face recognition software. The software would hopefully be able to pick out a person's face, and determine where in the camera's field of view the face was.

I would then divide the video feed into zones...Lets say 3, from right to left from the computer's point of view. If the detected face is in zone 1, then the servo would turn the TV and camera so that the face was in zone 2. So it would turn the screen and camera to the right. If the face was in zone 3, then the servo would turn the TV and camera so that the face was in zone 2. If the face is in zone 2 then the servo doesn't move. You would probably need a relatively wide angle camera, and you would have to set things up so that the computer didn't end up searching for a face if it momentarily looses detection. Another thing to consider is what if there is more than one face. In that case I would set it up to split the difference.

Another idea would be to put RFID sensors throughout the room in places where the viewer usually goes. So the table, the bar, the fridge, the couch...you get the point.

The computer would then locate the viewer based on the movement of the RFID tags from one place to another, and could move the TV to match up with the location of the detection.

I hope this is helpful.

DB

zoomkat
01-11-2010, 12:07 AM
The idea of an IR beacon is fairly simple. IR remote controls put signals modulated at ~38khz. Below is a radio shack modulated IR detector. A detector such as this could be mounted such that it is very directional, and it panned left and right until it gets a peak in detected IR (the remote control). Another approach could be a passive overlapping array of the could be used with the individual outputs monitored, with the result of the array Attached to the tv) being moved until the center detector had the highest output. Also below are some electronics using a remote control that could be used to manually raise/lower and pan the tv. Bottom is a PIR module that could be used similar to the modulated IR detector, except it would be looking for a person's body heat. The most important question is how much is the person willing to spend on his tv setup? Lots of ways to do things, from mild to wild.


http://www.radioshack.com/product/index.jsp?productId=2049727
http://store.qkits.com/moreinfo.cfm?Product_ID=387
http://store.qkits.com/moreinfo.cfm/QK142
http://store.qkits.com/moreinfo.cfm/Fk509

MikeG
01-11-2010, 08:39 AM
Thanks zoomkat, I understand using an array of sensors to triangulate location. For some reason when I read your original post and it sounded like this is a easy task. When in fact there are several technical hurtles to consider.

darkback2, do you have face recognition software?

Nishi
01-11-2010, 12:17 PM
What about something like 'The NorthStar Navigation System'? It's an infrared localisation system which uses infrared points of light as orientation points. An infrared sensor determines its position and orientation from the position of two points of light.

EDIT: http://www.evolution.com/products/northstar/

zoomkat
01-11-2010, 06:33 PM
Thanks zoomkat, I understand using an array of sensors to triangulate location. For some reason when I read your original post and it sounded like this is a easy task. When in fact there are several technical hurtles to consider.

Actually, there is no triangulation involved. Two IR detector/transistor setups each controlling a DPDT type relay, with relay each driving a pan motor could be used to align the tv screen towards an IR remote control. The below link shows the detector/transistor setup. After a little more looking, the bottom link is probably the best solution. Expanding on this setup the person could raise/lower and pan the tv using a universal IR remote control (assuming his equipment isn't Sony). Very inexpensivly too.

http://www.rentron.com/remote_control/Fire-Stick-I.htm
http://www.rentron.com/remote_control/IR-D15A.htm

Quantum
01-11-2010, 07:05 PM
Actually you do need to make an array. The IR detectors dont have any signal detection. This wont work with IR signals. There either is a signal or there isnt.

You need to make an array with several sensors. The array needs a housing that lets say each sensor has a 10 degree range of view. With 15 sensors you now have a field of 150 degrees. Your program then compares which sensor/sensors are triggered by the remote and you have your tracking array.

The program that monitors the sensors then pans center to where the signal is coming from. As the amount of sensors increase and the range of view they observe is decreased the accuracy of the array increases.

zoomkat
01-11-2010, 07:26 PM
Actually you do need to make an array. The IR detectors dont have any signal detection. This wont work with IR signals. There either is a signal or there isnt.

You need to make an array with several sensors. The array needs a housing that lets say each sensor has a 10 degree range of view. With 15 sensors you now have a field of 150 degrees. Your program then compares which sensor/sensors are triggered by the remote and you have your tracking array.

The program that monitors the sensors then pans center to where the signal is coming from. As the amount of sensors increase and the range of view they observe is decreased the accuracy of the array increases.
Well, an array of two detectors side by side, seperated by a center partition that allows the left detector to see from front around to the left, and the right from center around to the right. The center partition only allows both to see the IR transmitter when the tv is directly facing the IR transmitter. The idea is to eliminate any "program" and the associated hardware to run it.

MikeG
01-11-2010, 10:01 PM
May years ago I built a simple sun tracker. Sort of the same concept you propose, zoomkat. The sun tracker employed two photo-transistors separated by a partition. Measure the error and update accordingly - all components - no software.

So your idea is more of an active receiver. The device is always looking for an IR signal present on both receivers. When both sensors have a signal the motor is pointing to the source. So I guess the length and shape of the divider determines accuracy.

Noog
01-11-2010, 10:16 PM
The TPA81 IR thermopile is a single sensor containing 8 pixels in an array and I understand that they would have no problem detecting a human heat signiture out of the ambient room temp. Each pixel covers a 5.12 degree field of view (41 degree field of view in total) and 4 of these units together would provide you with a little over 160 degrees of coverage. Although, at over $100 (CAD) each this could get pricey. Depending on the layout of the room you may not need this kind of coverage.

www.robot-electronics.co.uk/htm/tpa81tech.htm

An Arduino board could easily handle the basic software requirements to read from the sensors (I2C), figure out the location of any viewers in range and then send out commands to a servo system to reposition a TV. I think it would be neat to do some advanced coding with this setup to solve solutions *like what to do when there are multiple viewers.
*
Noog

zoomkat
01-11-2010, 11:15 PM
May years ago I built a simple sun tracker. Sort of the same concept you propose, zoomkat. The sun tracker employed two photo-transistors separated by a partition. Measure the error and update accordingly - all components - no software.

So your idea is more of an active receiver. The device is always looking for an IR signal present on both receivers. When both sensors have a signal the motor is pointing to the source. So I guess the length and shape of the divider determines accuracy.
Yes, this is a simple tracking setup. It is simple, but does not address how to realign the tv when it is time for it to go back down into its slot. That is why I think the IR-D15A is the best control solution while still being affordable. The person with the remote could pan and position the tv as desired as well as having up/down lift control.