AVRcam as a noncontact optical displacement sensor (mouse)

This forum is used for discussing hardware, software, and technical details of the AVRcam embedded system.

AVRcam as a noncontact optical displacement sensor (mouse)

Postby Kinema » Wed Dec 01, 2004 5:50 pm

Would it be possible to use the AVRcam hardware to implement a noncontact displacement sensor similar to an optical mouse? Does the processor have enough power?

A few thoughts:
* Maybe this would be a task more suited to a DSP
* Agilent owns a few well guarded patents on the subject
* Maybe it would be easier/cheaper to adapt an optical mouse sensor

--adam
Kinema
 

Agilent link

Postby Kinema » Wed Dec 01, 2004 5:53 pm

More info on Agilent's optical navigation products can be found at http://we.home.agilent.com/USeng/nav/-536893499.0/pc.html.

--adam
Kinema
 

Postby Kinema » Wed Dec 01, 2004 6:46 pm

I applogize for the noise. I should have done more reading before posting (RTFM and all).

Upon closer inspection of the data sheets at Agilent I would have to guess that this application is beyond the capabilites of the AVRcam. Agilent's standard sensor is capable of a frame rate of 1500 fps and a maxim linear velocity of about 12 in/sec (about 0.3 m/s). Their high perfomace sensor is capable of 6000 fps (damn!) and 40 in/sec (about 1 m/s). I would have to say that this type of perfomance is beyond virtually all non application specific implementations.

Thanks for humoring me though.

It seems to me that the best solution to this problem is a custom lens and light source for the Agilent sensor. Unfortunatly I know nothing of optics so I do not know if it would be feasible to design and manufacture a lens that would allow the sensor to be more then a few inches from the target surface.

--adam
Kinema
 

optical tracking mouse

Postby techcare » Thu Dec 02, 2004 10:03 am

Several years ago I designed an optical X-Y encoder board to track motion around an Agilent HDNS-2000 optical mouse chip using simple optics to enhance the focal point of the sensor to about 1.5 inches from the floor surface. The unit outputs quadrature signals for both X and Y motion, and enables my robot to track it's position over most surfaces without concern about wheel slippage.

You can use the guts of an optical mouse to start experimenting with

All you need to do is mount a small adjustable lens like those used for single board cameras, in the optical port and use a high brightness red led pointed down at the surface to be measured, Adjust the lens to obtain best tracking results at the distance your sensor is off the floor.
techcare
 

Postby Kinema » Thu Dec 02, 2004 3:55 pm

So it is possible. Thats funny. After I posted this I headed over to sci.optics and asked if it could be done and linked the data sheets. I recived the following response:

It would be easy to change the focal length to one which
would allow a larger object distance, but I don't believe you can get the
resolution you need over the range 2 to 10 cm. Defocus will blur the image
too much over that large a range. You need a fast lens (NA at least 0.1 or
~F/5) to get the resolution you need, but fast and large depth of field are
incompatible.


Thanks for the info. I'll have to see if I can dig up a suitable lens. The guy that responded did say that if it was possible that an aspheric lens such as those found at http://www.thorlabs.com/SelectGuide2.cfm?Guide=47&Section=2&Ref=2 would work best. I'll have to read up a bit on optics terms and try it out..

--adam
Kinema
 

optical x-y endcoder

Postby techcare » Fri Dec 03, 2004 12:36 pm

I believe the lens I used was a laser collimator taken off an old laser printer scanner head. It had a theraded insert to allow the focal distance to be changed.

The focal point of the lens only gave less than 1/5" of depth of field that was usable for tracking, so the sensor has to stay within that tolerance of the focal distance to the surface you are tracking, or you would have to provide some method of adjusting the focus on the fly.

The stock lens of an optical mouse only gives about 1/16" of an inch usable distance.
techcare
 

pinhole lens, high contrast surface and a bright light works

Postby hamjudo » Mon Dec 13, 2004 5:04 pm

but fast and large depth of field are incompatible.


Compensate for a slow lens by either using a brighter light, a higher contrast scene, or both.

I took a small piece of aluminum foil and poked a hole in it. I bent it over the lens of an optical mouse. The aluminum foil totally covered the LED, and covered the lens except for the tiny pinhole. I aimed a 100 watt light bulb, with reflector, at a checkerboard pattern about a foot away. The modified mouse can track that checkerboard from about 18 inches away.

Note that the resolution is inversely proportional to the distance from the scene. Normally the lens is about a quarter inch from the surface. When it is a factor of 75 times farther away, the resolution is also worse by a factor of 75.

I like the mouse as a mouse, so I won't try to improve the performance of the pinhole. I think the pinhole lens would work much better as a replacement for the original lens, instead of being crumpled on top of the lens. This is an experiment that I'll leave for someone with an extra mouse.
--
Paul Haas, paulh at hamjudo dot com
hamjudo
 
Posts: 6
Joined: Mon Dec 06, 2004 11:49 am
Location: Ypsilanti MI USA


Return to AVRcam Embedded System

Who is online

Users browsing this forum: No registered users and 9 guests

cron