<< return to Pixycam.com

CMU Cam 6, Pixy 2 Roadmap

Pixy Hardware
#1

Hi guys,
do you have a time plan/road map for future hardware releases of the pixy board? Or do you plan to keep it the way it is as long as possible (such as the raspberry pi guys)? I am not very experienced in the topic, but I had the impression (from reading forum posts) that for some interesting tasks (high rate usb communication, SLAM, stereo vision etc) memory is a little bit on the short side. Would be interesting to hear a little bit about your plans.

Thanks.

#2

Hi Basil,

CMUcam 6 is a long ways off, we just released CMUcam 5 =)

There is 264K bytes of RAM and 1M bytes of Flash, which is pretty good, just not for storing raw images. There is still a lot more we can do with image compression to be able to store images on camera. This was never really the point though, since Pixy is meant to be more of a sensor than storage.

Scott

#3

Hi Scott,
thanks for your reply. So I guess then that the actual image processing has to happen on another computer anyway. But that seems to make sense.

Basil

#4

It depends on what kind of image processing you have in mind. For example, we’re able to perform face detection on Pixy, but there are are other algorithms that are too intensive for this hardware (we’re still exploring the limit). For now, Pixy is great for detecting color signatures, but eventually it will work with much more than just that. Just don’t plan on storing lots of video or images on Pixy :slight_smile:

Scott

#5

Although I know the goal of pixy is to make visual processing as easy as possible (hence the task specific demo jobs) sometimes I think it would be even better if there was a way to turn the visual processing “system” into a “stream” that you could plug various pre-made processing “modules” together so that the “stream” can be optimized for the end-user’s desired tasks. Similar to some of the computer vision tools and APIs out there.

#6

Also agree with those wishes, want to get closer to something with more practical applications:

  • Stereo vision, more RAM on the HW side (work with Intel Real Sense technology as an example)

  • Point cloud mapping and SLAM on the FW side

#7

Hello Nathan,
We keep track of these requests, so thanks for the feedback!

Edward

#8

Any update on next version, tentative timeline?

Swamy