I've been fascinated by optical flow ever since I learned that you can do some pretty good navigation work with the chip taken from an old optical mouse. I think this is probably going to be one of the methods we use to take Blimpduino navigation to the next level (it's ideal for any unknown environment, as well as GPS-denied ones). As the picture above suggests, optical flow is the method that many animals, especially insects, use to navigate in complex space.

Here's a writeup from Hizook about the latest research in the field. The authors describe the advantages of this technique:

"The main benefit of this method is that the information about the optics and scene depth is wrapped into a model that is easily learned from video from the robot. No calibration grids or complicated lens distortion models are required! The primary limitation from a practical standpoint is the assumption of constant scene depth, which would break down if, for example, the robot turned to directly face a very close wall, departed from the ground plane, or moved from a narrow corridor into a large open area. "

Here's a video that shows how it works:

CVPR 2009 from Richard Roberts on Vimeo.

[Fly image at the top from the Max Planck Society]

Views: 947

Comment by Mathew krawczun on March 13, 2010 at 1:04pm
thank you for posting this, I have also found optic flow a very interesting idea and love to read more on it.
Comment by lionel on March 13, 2010 at 1:14pm
Yes, it is very important not only on Earth but for planetary UAVs too.
Comment by Jack Crossfire on March 13, 2010 at 2:35pm
As Reed said, sometimes it works & sometimes it doesn't but maybe someday it'll be practical.
Comment by Rich Weaver on March 13, 2010 at 2:54pm
I agree, it takes a fair amount of silicon to implement it. Some of the vision guys I work with have been using the Xilinx FPGA version of it, same as in the following article:

Comment by Earl on March 13, 2010 at 5:35pm
I understand all of it with the exception of the start to finish.
Comment by Earl on March 13, 2010 at 5:44pm
Wait a minuet...I think I got it !.....w.r.t. means ...".with respect to ".....
Yea...NOW I understand !
Comment by Rana on March 13, 2010 at 11:56pm
Chris, this could really be a great breakthrough in the DIY autopilots, thanks for sharing !
Comment by abey on March 14, 2010 at 2:01am
Shameless plug: web page on our optic flow navigation for aerial vehicle research: http://lis.epfl.ch/microflyers

And a video of our latest results in this domain (merging GPS nav with OF obstacle avoidance): http://www.youtube.com/watch?v=OmkTZxOe-bE
Comment by David Ankers on March 14, 2010 at 2:08am
This is why I love this site.

Abey, that was very impressive, love to hear you on a future podcast! Amazing.
Comment by Rich Weaver on March 14, 2010 at 10:48am
Looks like I have some studying to do. I'd love to replace our Xilinx implementation. Thanks for sharing.


You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service