Game Development Reference
In-Depth Information
Figure 6.5
Longest possible path to freedom.
by the AI to implement heuristics. The coloring and the numbering will let us
see at a glance if the AI is working as we expect.
Evolution of the Evaluation Function
Our evaluation function will be part of the class we use to store the state data
instead of being part of the AI. Our evaluation function describes the fox's
situation. We use the concept of ''big numbers for big decisions'' [Dill08] in our
evaluation function. Any game board in Fox and Hounds can be categorized as
either good for the hounds or good for the fox. If the fox is hemmed in, it is good
for the hounds; if the fox has a path to freedom, it is good for the fox. There are
variations within each category, but there is no confusing middle ground. By
making the most important categories numerically far apart, we have sufficient
latitude for the variations. Our numbers will reflect this separation of the two
categories. A value of 0 means that the fox has won. A value of 127 means that the
fox is trapped and has lost. We need in-between numbers as well. Getting those
numbers is an evolutionary process.
Search Nedrilad ::

Custom Search