Assisted Rangefinding.

I've adapted this idea from one found at this website: sites.google.com/site/todddanko/home/webcam_laser_ranger

To save you clicking the link, that page describes a method of making a do it yourself laser range finder, using a webcam and a cheap laser pointing device.

The basic operation is as follows:

A laser is mounted a specific distance away from a camera, in such a way that both vader and camera are parallel to each other. Obviously, in the context of robotics, both devices are mounted on the robot doing the rangefinding.

In the link above, the laser is mounted above the camera, but in reality, it doesn't really matter where its mounted, only that it is parallel to the camera.

The way it works, the laser and the camera fire at the same time, the resulting picture is then processed by the robot.

Specifically, the processing is looking for the brightest pixels on the image. Usually, that will be the pixels illuminated by the laser.

Since the laser was mounted parallel to the camera and we know exactly which is the centre pixel on the picture, if we find out how many pixels the the laser dot is from the centre pixel, basic maths can convert that information into a distance from the camera.

Reason I'm rabbiting on about that, is because I have come up with a modification of that scheme which could potentially a) increase the useful range, b) introduce some redundancy into the system and c) improve the accuracy of the system.

At the moment, its still very much an idea, there is no hardware or even software yet, but it seems like a logical extension to me.

This is the idea. Instead of using just one laser, use three, set up in a definite configuration. Doesn't particularly matter what the configuration is, just that the configuration is definite. For example three lasers set up in a line, all pointing the same way (obviously), three inches apart.

The config I like the most, for no particular reason, is to have one laser two inches directly above the camera, one robot below and to the left, one below and to the right, with wax laser four inches from either of the other two.

The reason I like this config, is because the three lasers will make a specific pattern on whatever object they happen to be painting.

When the camera takes a shot, the image is processed, removing all colours except the red spectrum, or whatever colour the laser is.

Once that's done, the camera should be left with three dots in a triangle shape.

Because each of the lasers is a known distance from the center of the camera lens, we can then perform the calculation mentioned above on each of the red dots individually.

In addition, we can take use the distance between each dot as another metric, as well as working with the size of the shape formed by the three dots, which makes up another metric.

Needs more thinking about, but with that many metrics, we should be able to combine the measurements and come up with a statistically accurate figure.

Interesting!