I was watching a great YouTube channel named “DIY Perks”. One of the things I saw him doing was salvaging old laptops’ webcams and soldering a male USB cable to them. This way we has a very thin and compact webcam to use. I was thinking of how this would help in FRC.
Here is my list why I think they can be beneficial:
- Easily securing it behind a transparent plastic (PC for example) to see the inside of (for example) the storage/hopper to count the power cells.
- Easily securing them to box extrusions due to their rectangular & thin shape.
- Saving weight as they will probably be much lighter than any other USB camera.
- They are super cheap and can be easily replaced. (around 3 dollars)
I do not know if these USB cameras’ settings are configurable to use with vision. However, even if not I think they would be a viable option for driver/operator cameras.
I am looking forward to your ideas and how this would be a good/bad idea in FRC.
Likely the biggest problem with this from an FRC perspective would be the shear quantity of different chip-sets and sensors that are likely used across all these laptop webcam boards, and trying to support advanced functionality across any decent range. In my experience, FRC tends to find one decent, consistent, product and rolls with supporting that as much as possible, like the PS3 Eye Cameras for Chameleon Vision.
I’d also suspect there may even be inconsistency in one board to the next from the same laptop model since the quality of laptop webcams aren’t usually high concern for a manufacture as long as it fits the basic spec, unlike standalone webcam products, and so they may use whatever chipset and sensor is cheap, available, and compatible without worry about what exact spec or functionality is present.
Keep in mind, the PS Eye Camera is one of the better USB cameras you can get for vision processing, and costs about $14 currently.
It’s cool that you can extract these parts from dead laptops, and could be a fun educational project. Especially if you like tinkering and taking things apart (as I definitely do).
Looking purely from a cost perspective… it may be more worthwhile to get a $10/hour job and use the wages to purchase functional cameras, rather than spend your time extracting and configuring and “robustifying” a variety of cameras. Then again, that assumes one could find such a job…
An additional cost thought to keep in mind: Assuming a $5000 registration cost and 12 matches, each match that your robot is in costs $416. That’s a lot of money to pay if the robot sits dead on the field doing nothing. Generally, you want to get your $400’s worth of robot action. As I see it, spending a bit more on critical components (ex: data source for a vision processing algorithm) is insurance for ensuring that registration fee is utilized to its best.
Each team will have a different pain tolerance for this, and “tipping point” for their cost-related decisions. But, the point being, the cheapest option in the short run doesn’t always equal the best option in the long run.
Not to say it shouldn’t ever be done. To me, it’s the educational value that would make it worth it to do this. But, if you take away the educational value, I tend to say either spend a bit extra to buy a proven off-the-shelf component, or scrap that portion of the robot entirely and invest the resources elsewhere.
For people considering the PS Eye, realize that it requires a coprocessor, since there isn’t a driver compiled for the roboRIO.
I honestly don’t like the PS Eye. I bought one because it is cheap and other people liked it. However, I think it is bulky and has kind of poor image quality.
My team’s goto camera is the Logitech C930e. Right now, it is expensive (but all webcams are), but it has good image quality and a very wide field of view (90 deg diagonal).
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.