Nuts & Volts - May 2004

Personal Robotics

Learn About Microcontrollers

Place Scotch tape on one end of the tube and thread your lens into the other, then aim it at your flashlight in a semi-darkened room.

At some point, looking at the tape with your eye loop, you will see your flashlight in focus, but upside down. In this way, you can see the distance at which a particular lens is forming an image. This will allow you to determine the dimensions of your lens board.

Back to the chicken/egg thing, it is somewhat difficult to get the lens centered on the sensor until you have working software and it is difficult to get the software going until you have working optics. Therefore, I highly recommend using the existing optics to flesh out the software.

Once you can take pictures of the surface of your desk, you can work on the optics. The surface roughness reading can be beneficial here, since a sufficiently rough surface will read as not rough if it is out of focus.

Again, the chicken/egg thing comes into play. A rough surface that is far enough away can still appear smooth if its features are too small to be recognized as features; think of a stucco wall at 20 feet away versus one that is two inches away. By working out the software on an optical system that is known to be good, you can test optical ideas once the software works.

A laser pointer with a Scotch tape diffuser can be helpful here to “break” the chicken/egg cycle, as well. It generates what amounts to infinite roughness. By aiming your laser at the sensor and looking at the motion registers or quadrature pins, you can see if you have broken anything. Aim the laser into the mouse and wiggle the pointer around. If you sense motion, then things are good.

Once you know where your new lens wants to focus, you can set up an LED and start taking pictures of it. To start, you will want to work in a darkened room. Set the mouse sensor and optic aiming upward. Then place

MAY 2004

the LED above it at some distance, centered on the mouse sensor.

It is easier if you have your ancillary lens focused at this distance. Next, take pictures and look at the data coming out of the lens, searching for a bright spot. Try to center this spot by moving the ancillary lens around.

Table 2 shows what the data coming out of my mouse looks like when it is centered on the lens. Notice the pattern of 63s. They are nearly centered in the data, indicating that the lens is nearly centered. Once you have it centered, you can affix your lens board with an adhesive, like hot glue. At this point, you can move the LED to a new distance, look at the data coming out, and focus it until you get the sharpest difference between low numbers and high numbers.

With these general optical tools at your disposal, you should have a reasonably easy experience getting the mouse to work.

The possibilities for this are really amazing. Imagine a line following robot that takes a picture of the line and uses an artificial neural network to determine how to steer in order to track the line. You could use a laser line generator; take pictures with and without the line generator on to detect terrain features.

You can even use it to do experiments in insect vision, called “optic flow,” where you read the quadrature outputs directly to determine the motions of surrounding objects relative to you.

In the software I have provided on the Nuts &Volts website ( www.nutsvolts.com), I have the IsoPod output generate HTML tables with colored backgrounds according to the pixel values of the image. You can save the output to a text file and as a UTF- 8 file with the extension “.HTML.” This will allow you to easily visualize the mouse’s output.