New FRC Smart Camera (Survey)

#1

Hi Everyone,

We are developing a powerful, easy-to-use smart camera for FRC. We are looking for your feedback, so we’ve written this survey to help us make the best decisions possible. It is a fairly short survey, and we would really appreciate your input.

Thanks!

#2

Color me interested. I dropped my email in the survey, looking forwards to seeing where this goes.

#3

replied. Sounds super neat!

#4

Very interested in this, it sounds super cool!. Excited to hear future updates for it!

#5

Interesting. This’ll be hard to pull off well and for cheap, but good luck.

#6

While I agree, I think if they can pull off the “well” part, teams will be less concerned about the “cheap” part. Considering teams might drop $300+ on a co-processor and $100+ on a decent camera and still not get vision processing to work, having an all-in-one solution that is easy to implement and works reliably would be VERY appealing to a lot of teams.

#7

having an all-in-one solution that is easy to implement and works reliably would be VERY appealing to a lot of teams.

Honestly, I don’t think “works reliably” is enough. Personally, I’d more or less flat-out refuse to use anything that we can’t directly modify and edit the code on unless it works flawlessly 100% of the time. Especially if we’re spending as much — or more — money on the smart camera as we did our x86_64 coprocessor that had a desktop GPU.

While some teams do drop that amount of money on vision gear, for basic contour based vision, you honestly only need a $35 Raspberry Pi and a $7 Playstation 3 Eye camera (they’re purpose built for CV: 640x480@60fps or 320x240@150fps, shockingly cheap). The sort of team that would be willing to spend $400 on vision is — I’d bet in general — more inclined to implement it themselves unless this solution truly is flawless.

I think making this open source would certainly spark demand.

#8

Have you looked at this? https://openmv.io/

Been playing with it, and it is a nice little gadget. Runs python natively. About the size of a large postage stamp.

#9

New to me. I’ll forward to our electronics mentor, I really like those prices and update rates.

The sort of team that would be willing to spend $400 on vision is — I’d bet in general — more inclined to implement it themselves unless this solution truly is flawless.
We’ve definitely spent that kind of money attempting to implement it ourselves and got to a 40% solution. I’d love to try an off-the-shelf 80% solution…

#10

Thanks for all of the great feedback! We will send an email in the coming days to talk about beta testing.

#11

Very interested to hear more on this.

We’d love something that is simple to use as the Pixycam… but with far better performance.

#12

Awesome, looking forward to it!