Every one . I am working on OVM7690 to AT91sam9260 interface, I am new in it. I have following queries and seek guidance from your side.
1. How can I check whether my cam is working or not? Is there any circuit through which i can connect it directly to pc and check its out put.
2. kindly give me its sample application.
I am a member of SENSORICA, an open enterprise, and we've been solicited to design a "Plant Counter for Maize Population and Yield Analysis", which relies on a laser-based range finder. Would you give us some guidance?
This project is part of an ongoing collaboration with Greener Acres value network, to design, produce, and distribute sensing technology for local food systems. We want this technology to be open source. All participants will be rewarded according to SENSORICA's value exchange principles.
SENSORICA is an open, decentralized, and self-organizing value network, focusing on providing sensing solutions, based on open innovation.
We are NOT another crowdsourcing scheme! We are about commons-based peer production.
@TiberiusB: Sounds like an interesting project. While I'm not personally available to help out, you're more than welcome to check out the Laser Range Finder Module (when it becomes available in a few weeks) and see if it will work in your application. It's fully open source, as well.
@Alexandre, @ghalib: I unfortunately don't have any technical support to give you. There is plenty of documentation available on using the CMUcam3/CMUcam2 - I'll be releasing the code I created for my proof-of-concept LRF device in a few weeks.
Just a quick update that Parallax officially kicked off production of the Laser Range Finder Module in July. Units (and full documentation) should be available by the middle of September!
I stopped by the manufacturing facility today to see how things were going. Fabrication and assembly are completed and they're now working through the test/calibration procedure. The procedure is labor intensive, as each device needs to be programmed, tested, laser diode aligned, laser diode glued, and then calibrated.
It's always exciting to see a design come to life. I've attached two pictures for your viewing enjoyment.
A tray of boards ready to be tested and calibrated.
The test/calibration fixture (hand crafted by Parallax!)
Yes! We should have the first production line modules in stock this week. I think I saw an e-mail that said they were shipping 125 units our way Monday or Tuesday. These were manufactured by another company in the San Francisco Bay Area because our production line was too busy.
The documentation drafts are in review internally and going through our editor. This process is organized and efficient, and wouldn't introduce a delay. I will see about getting the draft posted on this thread in the meantime.
Prices - we're looking at $129 as the target price. The product has a significant R&D and manufacturing cost. Volume discounts for commercial users will be under a hundred dollars.
I tried to see if I could sneek in an early order by phone today, but she said they wouldn't be available until monday.
And thanks for calling ratronic. You spoke with Chantal, one of our top brass salespeople who has worked for Parallax for almost 20 years.
I'm the one that told her "Monday" after talking to marketing, engineering, kitting, and editing. We have a few little tasks in the final stages related to web page posting, pricing, and documentation but it's all getting done today.
My LRF showed up. It's cool. I ordered a slipring assembly off eBay so I can spin the sensor like Neatos vacuum cleaner sensor.
But the update rate seems to be only about 1 Hz. Any way to speed that up to maybe 10 Hz?
Hi bkgable-
Thanks for buying an LRF. I have some ideas for code optimizations that may increase the rate by a few Hz, but probably not as high as 10. Optimizations are on my list of things to do, since I think that having a faster update rate will open up the doors to more uses of the LRF. I'll post on the forums when a new firmware update is ready.
Thanks for the reply. I assume the CMOS camera chip can be read at 30 fps. But the timing limitation is in finding the laser spot. Do you use all the cogs/cores to do that processing? Maybe assign sections of the image across multiple 80 Mhz cogs? I'm not too familiar with the Propeller.
Thanks for the reply. I assume the CMOS camera chip can be read at 30 fps. But the timing limitation is in finding the laser spot. Do you use all the cogs/cores to do that processing? Maybe assign sections of the image across multiple 80 Mhz cogs? I'm not too familiar with the Propeller.
There are a few limitations - one being the code portions written in Spin (all of the image processing, spot detection, and triangulation routines) and the timing of the frame grabber (written in PASM). The largest bottleneck is the timing/speed of the frame grabber. The camera itself can support 30fps, but the LRF is currently at 2.5fps (2MHz PCLK). The frame grabber currently runs in a single cog. Using multiple cogs to handle the frame grab may be possible and would probably have the most benefit, though it's a significant shift from my current design and would need some significant testing to ensure error-free frame grabs. I believe earlier in this thread, I discussed some of the timing issues I ran into when interfacing with the OVM7690 camera.
There's another [thread=134822]thread[/thread] that has some optimization suggestions, as well.
Just a thought, would this concept work with the single line image sensor and a red filter, plus gated control of the laser to "know" when a dot you are seeing is from the laser and not ambient?
It would be too difficult to align a single dot with the linescan axis. A laser line, however, perpendicular to the linescan axis would be easier to detect without any fussy alignment. Here's a link that illustrates the principle:
would this concept work with the single line image sensor
Not to disagree with Phil's understanding of the technical problems, but it can be done relatively easily.
Use a laser module that throws a small, circular spot with a Gaussian intensity distribution. I suggest a module with a threaded housing and wire connections. Many of these lasers can be modulated using a digital signal to allow for exposure control (see later).
I would recommend using an Aluminum base plate with the laser firmly attached. A rigid structure is critical to holding the calibration and alignment. The detector board should have some adjustment with respect to a lens fastened to the same base plate. Ideally, the detector board should be tilted slightly to keep the spot in focus (fairly easy to calculate the correct angle using geometry). Alignment can be done by making the mounting holes in the board a bit bigger than the screws.
Line sensors are very fast and have a wide exposure range. Updates rates in excess of 25 times per second can easily be achieved. Now here's the real trick. You only need a sensor with 128 or 256 pixels. The reason is that with some basic maths you can get 10 times the resolution because the laser spot has a known intensity profile.
and a red filter
The red filter is a very good idea - it reduces the dynamic exposure range and increases contrast. Unfortunately, you can't use a narrow band filter because the incoming beam angle changes with distance and narrow band filters only work properly with a perpendicular beam. But a simple red plastic filter works fine.
plus gated control of the laser
Another way of increasing contrast is to use a higher power laser and switch it on only during the exposure time. That way you can have a very bright spot for a short time so that the sensor only needs a short exposure which consequently reduces the intensity of the background. An important point is that it does not affect eye safety because the laser is off for a significant time so the average power can be very low.
This kind of design still has the same range limit as the Parallax unit. The reason is purely geometrical as the resolution falls off rapidly with increasing distance. But an accuracy of better than 0.5mm out to 1m is no problem.
The gated control is not just for intensity and eye safety. It's so you can take differential exposures: one with the laser on; one with it off. Then subtract them to get just the laser response. I believe Joe does this in his firmware, too.
I suggest the linescan sensor because I assumed it would make the unit less costly to produce. One way or another, you *must* have precision mounted optics, so that isn't a consideration in my mind; I had a machine shop for several years and was accustomed to mundane things needing decent tolerances to work properly or repeatably.
Your right, gated control was intended to "find" the spot. I've thought about picking up one of those linescan sensors to play with.
First of all congratulations by your work!
I am trying to find documentation of the OVM7690 in the Internet without success, it has being very difficult to find quite simple documents like the device pinout and mechanical specifications, etc.. If possible, can you send me, or post, some material you have regarding this device?
Unfortunately, the OVM7690 data sheet requires a signed non-disclosure agreement with Omnivision, so I'm not able to release it. You can try to contact them directly or though your local manufacturer/sales representative to get more information.
First of all congratulations by your work!
I am trying to find documentation of the OVM7690 in the Internet without success, it has being very difficult to find quite simple documents like the device pinout and mechanical specifications, etc.. If possible, can you send me, or post, some material you have regarding this device?
Comments
Caught a Beta test review on Hack-A-Week and came searching. Guess I need to start watching this thread more closely...
Every one . I am working on OVM7690 to AT91sam9260 interface, I am new in it. I have following queries and seek guidance from your side.
1. How can I check whether my cam is working or not? Is there any circuit through which i can connect it directly to pc and check its out put.
2. kindly give me its sample application.
Waiting for reply.
Regards,
ghalib
I am a member of SENSORICA, an open enterprise, and we've been solicited to design a "Plant Counter for Maize Population and Yield Analysis", which relies on a laser-based range finder. Would you give us some guidance?
This project is part of an ongoing collaboration with Greener Acres value network, to design, produce, and distribute sensing technology for local food systems. We want this technology to be open source. All participants will be rewarded according to SENSORICA's value exchange principles.
SENSORICA is an open, decentralized, and self-organizing value network, focusing on providing sensing solutions, based on open innovation.
We are NOT another crowdsourcing scheme! We are about commons-based peer production.
Thank you for your attention!
Here's the link to the project's page
http://www.sensorica.co/home/projects/sensor-technology-to-support-local-food-systems/open-source-plant-counter
@Alexandre, @ghalib: I unfortunately don't have any technical support to give you. There is plenty of documentation available on using the CMUcam3/CMUcam2 - I'll be releasing the code I created for my proof-of-concept LRF device in a few weeks.
Joe
Just a quick update that Parallax officially kicked off production of the Laser Range Finder Module in July. Units (and full documentation) should be available by the middle of September!
I stopped by the manufacturing facility today to see how things were going. Fabrication and assembly are completed and they're now working through the test/calibration procedure. The procedure is labor intensive, as each device needs to be programmed, tested, laser diode aligned, laser diode glued, and then calibrated.
It's always exciting to see a design come to life. I've attached two pictures for your viewing enjoyment.
A tray of boards ready to be tested and calibrated.
The test/calibration fixture (hand crafted by Parallax!)
Until next time,
Joe
Yes! We should have the first production line modules in stock this week. I think I saw an e-mail that said they were shipping 125 units our way Monday or Tuesday. These were manufactured by another company in the San Francisco Bay Area because our production line was too busy.
The documentation drafts are in review internally and going through our editor. This process is organized and efficient, and wouldn't introduce a delay. I will see about getting the draft posted on this thread in the meantime.
Prices - we're looking at $129 as the target price. The product has a significant R&D and manufacturing cost. Volume discounts for commercial users will be under a hundred dollars.
Ken Gracey
Shouldn't take long to enable on-line sales.
Ken
And thanks for calling ratronic. You spoke with Chantal, one of our top brass salespeople who has worked for Parallax for almost 20 years.
I'm the one that told her "Monday" after talking to marketing, engineering, kitting, and editing. We have a few little tasks in the final stages related to web page posting, pricing, and documentation but it's all getting done today.
Ken Gracey
Visit the product page or search on the Parallax site for #28044. All the technical documentation is also available from that page.
I want to publicly thank everyone at Parallax who has been working so hard behind the scenes (and putting up with me) to make this project a reality.
We look forward to seeing what sorts of cool projects you all use this for!
Enjoy!
Joe
But the update rate seems to be only about 1 Hz. Any way to speed that up to maybe 10 Hz?
Hi bkgable-
Thanks for buying an LRF. I have some ideas for code optimizations that may increase the rate by a few Hz, but probably not as high as 10. Optimizations are on my list of things to do, since I think that having a faster update rate will open up the doors to more uses of the LRF. I'll post on the forums when a new firmware update is ready.
Take care,
Joe
There are a few limitations - one being the code portions written in Spin (all of the image processing, spot detection, and triangulation routines) and the timing of the frame grabber (written in PASM). The largest bottleneck is the timing/speed of the frame grabber. The camera itself can support 30fps, but the LRF is currently at 2.5fps (2MHz PCLK). The frame grabber currently runs in a single cog. Using multiple cogs to handle the frame grab may be possible and would probably have the most benefit, though it's a significant shift from my current design and would need some significant testing to ensure error-free frame grabs. I believe earlier in this thread, I discussed some of the timing issues I ran into when interfacing with the OVM7690 camera.
There's another [thread=134822]thread[/thread] that has some optimization suggestions, as well.
Take care,
Joe
It works fine in the near-field, but by spreading the beam thus, you lose quite a bit of sensitivity for long-distance measurements.
-Phil
Use a laser module that throws a small, circular spot with a Gaussian intensity distribution. I suggest a module with a threaded housing and wire connections. Many of these lasers can be modulated using a digital signal to allow for exposure control (see later).
I would recommend using an Aluminum base plate with the laser firmly attached. A rigid structure is critical to holding the calibration and alignment. The detector board should have some adjustment with respect to a lens fastened to the same base plate. Ideally, the detector board should be tilted slightly to keep the spot in focus (fairly easy to calculate the correct angle using geometry). Alignment can be done by making the mounting holes in the board a bit bigger than the screws.
Line sensors are very fast and have a wide exposure range. Updates rates in excess of 25 times per second can easily be achieved. Now here's the real trick. You only need a sensor with 128 or 256 pixels. The reason is that with some basic maths you can get 10 times the resolution because the laser spot has a known intensity profile.
The red filter is a very good idea - it reduces the dynamic exposure range and increases contrast. Unfortunately, you can't use a narrow band filter because the incoming beam angle changes with distance and narrow band filters only work properly with a perpendicular beam. But a simple red plastic filter works fine.
Another way of increasing contrast is to use a higher power laser and switch it on only during the exposure time. That way you can have a very bright spot for a short time so that the sensor only needs a short exposure which consequently reduces the intensity of the background. An important point is that it does not affect eye safety because the laser is off for a significant time so the average power can be very low.
This kind of design still has the same range limit as the Parallax unit. The reason is purely geometrical as the resolution falls off rapidly with increasing distance. But an accuracy of better than 0.5mm out to 1m is no problem.
-Phil
Your right, gated control was intended to "find" the spot. I've thought about picking up one of those linescan sensors to play with.
First of all congratulations by your work!
I am trying to find documentation of the OVM7690 in the Internet without success, it has being very difficult to find quite simple documents like the device pinout and mechanical specifications, etc.. If possible, can you send me, or post, some material you have regarding this device?
Thanks by now!
Humberto
Unfortunately, the OVM7690 data sheet requires a signed non-disclosure agreement with Omnivision, so I'm not able to release it. You can try to contact them directly or though your local manufacturer/sales representative to get more information.
Take care,
Joe