1 of 7
1
Multitouch Zero Force Setup Based on IR Laser Light
Posted: 28 May 2008 02:27 AM   [ Ignore ]
Avatar
RankRankRank
Joined  2008-05-08
Total Posts:  342
Administrator

Multitouch Zero Force Setup Based on IR Laser Light

Here is the setup that I’ve been working on for a while now. It is based on front IR illumination by IR laser.

Camera

For this project I’ve been using my old Orange Micro iBot firewire camera with great success. This low-cost camera generates 640x480 at 30fps uncompressed images in Y800 format. I found that because of all of this and the IR sensitivity this camera is very suitable for IR multitouch application:

Camera1.jpg

After removing the original lens and mounting the new lens without IR blocking filter, I placed the double-sided tape on top of the lens and cut the opening in it:

Camera2.jpg

IR Bandpass Filter

In the past I used photo negative with some success, but found that using narrow bandpass filter is by far the best in removing any unwanted IR noise and visible light:

BandPassFilter.jpg

Here is the filter mounted on the camera:

CameraWithFilter.jpg

Table Setup

Currently I’m in the process of redesigning the table. Right now it is an open-frame setup:

TestSetup1.jpg

TestSetup2.jpg

IR Laser

And now the most important part. I use the 850nm IR Laser light to illuminate the front of the surface. When finger or any object interfere with the laser light, the camera below picks up the bright spot and runs it through the image processing algo to generate detected blob information.
Here is the view from the surface:

TestSetup3.jpg

The laser is mounted on a tripod on a tri-point adjustable platform so that it can be positioned is a way that it generates a light beam that is parallel to the glass surface:

IRLaserSetup.jpg

The IR laser is positioned just above the glass surface:

TestSetup4.jpg

Software

Even though I could use touchlib for blob processing/tracking, having a lot of experience in image processing, I found that it was easier for me to write a highly optimized version blob tracking software. Because of this, I achieve very low latency and low CPU usage running my blob tracking software (around 3-4% on my E6600 2.4GHz system).
Here are some screenshots of my blob tracking app: 

TrackingApp1.jpg

As you can see blobs are very bright and clean:

TrackingApp2.jpg

Here is my iPhone placed on the surface:

TrackingApp3.jpg

Video

Here you can see the system in action. While using FTIR the blobs become dimer during the fast movement, using the laser method the blobs are always bright

Short video of the system:

Image Attachments
BandPassFilter.jpgCamera1.jpgCamera2.jpgCameraWithFilter.jpgIRLaserSetup.jpgTestSetup2.jpgTestSetup1.jpgTestSetup3.jpgTestSetup4.jpgTrackingApp1.jpgTrackingApp2.jpgTrackingApp3.jpg
 Signature 

Computing is not about computers any more.  It is about living!

Code Laboratories
CL Studio Live - Video like never before
CL Eye Platform SDK

Profile
 
 
Posted: 28 May 2008 02:58 AM   [ Ignore ]   [ # 1 ]
Avatar
RankRankRankRank
Joined  2008-01-04
Total Posts:  676
Moderator

WOW! Alex, nice setup and video. I’m interested in the IR Laser setup so that means I’m going to ask alot of questions tongue wink . How does it perform with ambient light? I know that the laser is mounted parallel to the screen, but how does it create a uniform spread across the glass? I would have thought it would use something like a FTIR setup beamed into the side of some acrylic. Also, you mentioned you use glass, is it frosted or do you use a different diffuser?

 Signature 

Get Sparked at sparkon.net
ORION Multi-Touch Display (blog)
My YouTube

Profile
 
 
Posted: 28 May 2008 03:11 AM   [ Ignore ]   [ # 2 ]
Avatar
RankRank
Joined  2007-04-03
Total Posts:  241
Moderator

Looks good!

 Signature 

My multitouch blog: http://www.multigesture.net
Howto: Compile touchlib on windows XP/Vista
Howto: Compile touchlib on Ubuntu Linux
Downloads: Touchlib SVN builds

Profile
 
 
Posted: 28 May 2008 03:21 AM   [ Ignore ]   [ # 3 ]
Avatar
RankRankRankRank
Joined  2008-02-12
Total Posts:  1111
Member

A very interesting setup… love to see more details about it.

Taha

 Signature 

Company Website - Knode Research and Development
My Youtube

Profile
 
 
Posted: 28 May 2008 03:37 AM   [ Ignore ]   [ # 4 ]
RankRank
Joined  2008-02-04
Total Posts:  119
Member

Looks great..
I’also wondering how you get the ir light to spread out evenly, and doesn’t the ir light get blocked by the first obstacle/finger and the 2nd won’t show up if there behind each other?

Profile
 
 
Posted: 28 May 2008 04:06 AM   [ Ignore ]   [ # 5 ]
Avatar
RankRankRank
Joined  2008-05-08
Total Posts:  342
Administrator
Fairlane - 28 May 2008 02:58 AM

WOW! Alex, nice setup and video. I’m interested in the IR Laser setup so that means I’m going to ask alot of questions tongue wink . How does it perform with ambient light? I know that the laser is mounted parallel to the screen, but how does it create a uniform spread across the glass? I would have thought it would use something like a FTIR setup beamed into the side of some acrylic. Also, you mentioned you use glass, is it frosted or do you use a different diffuser?

Because of the high brightness and light intensity of the laser light, I found that this setup performs very well in both ambient and daylight. This is why my current test setup is in the form of the open box. The blob tracking pictures are taken in the room that was lit by a bright halogen light. As you can see there is some background IR light going through the bandpass filter. This is how you can see my hand on the other side of the glass.

Regarding the FTIR - I tried to beam the laser light into the acrylic, althought the light that was shining through was very bright, I didnt get any FTIR effect. I think this is due the high collimation of the laser light. Unless your acrylic is perfectly polished and its edges are perfectly parallel to each other, the light will escape the acrylic and you will not get enough light bounces within the acrylic panel.

In my setup I specifically use the fact that the laser light is very collimated to generate a thin plane of IR light right on top of the glass surface. I found that the closer you can get to the surface - the better. The IR light plane thickness is about 1mm throughout the glass area. When the tip of your finger touches the glass it is lit by the laser light. The blob brightness generated by this mehtod is many times higher in intensity than equivalent generated by FTIR, DI and other methods. This of course depends on the output power of laser used. In this particular setup I use a single IR laser that generates only about 3.5mW of light power. Because of its efficiency and simplicity, I prefer this mthod over the FTIR and DI.

Although I could have used frosted glass panel instead of the clear one, the difuser I’ve played with and got good results is a thin drafting/tracing paper (velum).

Because of its simplicity and small number of parts this method could be equally used in both horizontal and vertical setups. The one I’m particularly interested in is the storefront application of this method.

 Signature 

Computing is not about computers any more.  It is about living!

Code Laboratories
CL Studio Live - Video like never before
CL Eye Platform SDK

Profile
 
 
Posted: 28 May 2008 04:14 AM   [ Ignore ]   [ # 6 ]
Avatar
RankRankRankRank
Joined  2008-01-04
Total Posts:  676
Moderator

Great, thanks for clearing that up. smile Just on what Helza1 said about the shadow effect, does that come into play considering that there is only one laser beaming from a single point or is it reflected in some way? It certainly looks like there are no shadow effects from your pics/vid.

 Signature 

Get Sparked at sparkon.net
ORION Multi-Touch Display (blog)
My YouTube

Profile
 
 
Posted: 28 May 2008 04:55 AM   [ Ignore ]   [ # 7 ]
Rank
Joined  2007-04-16
Total Posts:  59
New Member

Looks very nice Alex.

How are you spreading the laser bean? as normally doesn’t it come out in a direct line.
From the Microsoft vid I think they were using 3 lasers. How wide is the IR coverage?

Just had a random thought: Have you tried fiducials? I know that the IR light is coming above the surface but maybe a clear plastic cube/puck would glow red and show the pattern.

Profile
 
 
Posted: 28 May 2008 05:44 AM   [ Ignore ]   [ # 8 ]
Avatar
Rank
Joined  2007-11-04
Total Posts:  85
New Member
Fairlane - 28 May 2008 02:58 AM

I’m interested in the IR Laser setup so that means I’m going to ask alot of questions tongue wink

Nice Fairlane so I will only have to read the anwsers!
I’m interested too!
Chris

 Signature 

My setup Description
My Website with video reviews of soft! UPDATED!
Sorry for my bad english, Im french!

Profile
 
 
Posted: 28 May 2008 07:22 AM   [ Ignore ]   [ # 9 ]
Rank
Joined  2007-10-06
Total Posts:  55
New Member

nice 1 !
How much mW is the laser?  5mW or more ?

Profile
 
 
Posted: 28 May 2008 08:11 AM   [ Ignore ]   [ # 10 ]
Rank
Joined  2007-04-16
Total Posts:  59
New Member
gally - 28 May 2008 07:22 AM

nice 1 !
How much mW is the laser?  5mW or more ?

Alex mentions that he is getting 3.5mW. But didn’t mention in terms of cost if thats what you mean.

Profile
 
 
Posted: 28 May 2008 08:24 AM   [ Ignore ]   [ # 11 ]
Avatar
RankRankRank
Joined  2007-09-22
Total Posts:  263
Sr. Member

Hi , this is really interesting and your setup looks great.How exactly are you dispersing the laser evenly on the surface?Where did you get the laser , the diode looks preaty small and seaing you said it’s 3.5mW , is it from a CD-ROM or did you buy it separately.

Could you please try out this setup without the bandpass filter but with touchlib’s background removal function , if it works whitout the bandpass filter i think we just got a better way of building tables.

Profile
 
 
Posted: 28 May 2008 09:34 AM   [ Ignore ]   [ # 12 ]
Rank
Joined  2008-03-08
Total Posts:  35
New Member

nice work!
it seems you found the way Microsoft did.
do you have any idea about the Qty of IR Laser Vs. Glass Area? what’s the size of your glass?
Hope to see more details soon! smile

Profile
 
 
Posted: 28 May 2008 09:42 AM   [ Ignore ]   [ # 13 ]
Avatar
RankRank
Joined  2008-04-27
Total Posts:  150
Member

is it possible for you to tell us the make and model of the IR laser? or did u modify it with a “filter” to generate the plane?

 Signature 

Pissed Right Off Genetically Engineered Nerd
ProgenLabs.com

Profile
 
 
Posted: 28 May 2008 10:53 AM   [ Ignore ]   [ # 14 ]
RankRankRank
Joined  2008-04-30
Total Posts:  321
Sr. Member

I decided to work with scale for a little while and I have come up with some rough dimensions of the size of his table or at least the area captured by the camera.

I got 33” wide and 24” high. or 40.8” inch diagonal ( THESE MEASUREMENTS ARE IN INCHES )

 Signature 

I’m bad about monitoring posts so if you have a question regarding something I’ve said, then just send me a PM.

Profile
 
 
Posted: 28 May 2008 12:30 PM   [ Ignore ]   [ # 15 ]
RankRank
Joined  2008-04-16
Total Posts:  183
Member

I too am interested in more info on the laser. I think AlexP said that he is using collimation to his advantage in this setup, so perhaps the beam is fanning out sufficiently by itself? I don’t know. But still, the results posted so far are phenomenal, compared to a lot of other MT setups I’ve seen, looking forward to seeing more!

 Signature 

Visit my research blog at:
http://cs.unk.edu/~webb/

Profile
 
 
   
1 of 7
1