Page 123 -

243_MasterPieces_02b.qxd 4/18/03 6:59 PM Page 95

The Learning Brick Sorter • Masterpiece 2 95

Having described the algorithm in plain words, it’s time to write some code again.
First, we have to solve a small problem: our knowledge base is organized as a bi-dimen-
sional array, but NQC only supports arrays with a single dimension.The solution, how-
ever, is not overly complex. Imagine assigning each cell a progressive number,
row-by-row, column-by-column, this way (notice that the bins and ranges in Table 2.5 is
start from 0 instead of from 1.This trick would work with 1-based arrays as well, but it
would require a bit more math.

Table 2.5 Converting the Knowledge Base into a Single Array Capable of Being
Translated by NQC

Bin 0 Bin 1 Bin 2
Range 0 Cell 0 Cell 1 Cell 2
Range 1 Cell 3 Cell 4 Cell 5
Range 2 Cell 6 Cell 7 Cell 8

We can now store the cells in a mono-dimensional array with nine cells, numbered
from 0 to 8, and address them using a simple formula:

cell_address = range x 3 + bin
We already have a constant called RANGES that defines the number of rows of Table
2.5. Similarly, we can create a constant for the number of bins and make our program
completely parametric. Now the declaration of the knowledge base becomes:

#define BINS 3
int kb[RANGES*BINS]; // the knowledge base

And we can address any cell with the formula:
cell_address = range x BINS + bin

Now we have all the elements to start coding the routine Analyze(), which reads the
color of the brick, assigns it to a range, and decides which bin it must be placed into.This
routine incorporates our previous work about defining ranges:

void Analyze()
{
int light_value;
// wait for coupled touch sensor released
while (SENSOR_1>LITE_THRSHLD);
light_value = SENSOR_1;
// search if the light value is contained in
// one of the already defined ranges
for(range=0;range<=top_range;range++) {
if (light_value>=min[range] && light_value<=max[range]) break;