Wednesday, June 3, 2009

Standardized Solutions to Arbitrary Problems

On a previous episode of I Heart Robotics, we attempted to fix the targeting system optics of our automatic Nerf™ turret.

The computer vision system previously has had several issues with locking targets, the biggest problem has been that the target had to stand about one meter in front of the turret for it to get a lock on the target's face. This distance was considered by researchers to be too close for live fire practice on researchers. Test subjects were then obtained by the researchers. Unfortunately after disclosing the experimental procedures to our test subjects, as required by law, they became uncooperative and unwilling to continue the experiment.

To continue the experiments, the targeting range of the Nerf™ N-Strike Vulcan EBF-25 computer targeting system needed to be increased to allow the projectiles to decelerate to a reasonable velocity.

The next experiment involved duct taping a pair of cheap binoculars to the web camera as shown in the image below. Unfortunately the results were not of an acceptable quality, as they did not allow targets to be locked at a distance.

Next we began looking at other DIY research and solutions for adapting telephoto lenses to web cameras. This information may prove useful for achieving our goal of connecting a webcam to our microscope.

At this point it was noted that there appeared to be a standard lens mounting system, called C-Mount, for microscopes and security cameras. Since these lenses are roughly 1 inch in diameter, they are perfect for computer vision applications.

Interestingly, after further research, it was found that many web cameras use their own standard lens mounting system with a thread of M12x0.5 being used for many Logitech and Creative devices. It appears that many Firewire cameras also support the same lens mounting system but with Firewire performance.

Unfortunately Carl Zeiss does not appear to sell webcam lenses online, however they will custom build lenses for industrial applications.

I will provide further updates once our new lenses are received and our improved targeting system is tested. Also, tomorrow I will dissect the QuickCam Pro 9000 to determine if it can be adapted for this project.

No comments: