I started wondering today about how I could increase the available video memory for my robot NINA.
The first possible solution I wondered about would be using some kind of external video memory card, if one existed. I figured I could buy a simple netbook for a small profile for NINA's on-board PC, then just plug in an external video card for the video processing (like face detection and stereo vision). I googled it and found some good results, but other results that were disappointing as well. Apparently most "external graphics cards" don't increase the video memory directly but just prepare video for better quality on an external monitor (but technically that is increasing performance, right? The main downside there is I would need an external monitor, and even if I bought just a tiny little LCD screen, which would look pretty cool, I don't know how that would connect to the actual vision processing on the PC.) Another disappointing finding is apparently external graphics cards were canceled after they were released because they were impractical (at least that was the case two years ago. I don't know about today). Anyway, that's the first solution: buy a netbook, get an external video card.
Second solution was just to find a robotic camera with video memory on-board to free up efficiency on the computer GPU. Unfortunately, I'm not sure any camera out there does it. I think the SRV Camera does something like this, but...
Anyway, I'm still thinking about how to perfect NINA (but perfection is a never-ending process--futile almost). Anyone know anything more about external graphics cards or other means to make the processing more efficient on an net book?
NINA uses OpenCV and two CyberSnipa Spotter webcams for vision and video processing.
I currently have a high-performance laptop doing her Central Processing, but it's too big to mount on NINA (no room on her chassis).