Wednesday, April 20, 2011

USB 3.0 for machine vision

A little over a year ago Point Grey started talking about their USB 3.0 cameras. Around the same time the AIA started to explore what might be the most appropriate protocols for machine vision USB 3.0. So what’s happened since?

On paper, and as discussed in this excellent article from Advanced Imaging magazine from March 2010, USB 3.0 has a lot going for it as a machine vision interface. Obviously, it’s going to provide fast data transfer, which will help as resolutions and frame rates continue to rise, but it also offers more power, which makes life easier for camera designers and those users who want only a single cable. But we’re approaching the middle of 2011 and I’m not sensing a groundswell of support for 3.0. Frankly, both consumers and the machine vision industry seem pretty uninterested.

My feeling, and I’d like to hear your views on this, is that with GigE, CameraLink, CameraLink HS and CoaXPress we have more than enough options. (And that’s not forgetting that FireWire still has a lot to recommend it for lower-end applications.) In other words, for machine vision USB 3.0 is something of a dead end.

Unless something was to come along to change that, which I think might happen.

I have to wonder if there’s a new and improved version of the Kinect in the works. This would provide higher resolution and higher speed, which would in turn require bigger and faster camera sensors. Follow that line of thought and you’ll see that Kinect II might have to be a USB 3.0 device. If that happens, expect to see vision integrators using it for robot guidance applications, with the costs of such systems tumbling.

USB 3.0 for regular machine vision? No, but for 3D work it may be the way forward.

No comments: