Game Development Reference
In-Depth Information
Figure 6.1
First two moves of Tic-Tac-Toe.
but it's too late for him to do anything about it.'' In terms of gameplay, the AI
programmer can be tempted to use a variable depth limit to change the skill level
of the AI in what seems to be a realistic manner. Be warned that small changes in
look-ahead depth can cause major changes in the effectiveness of the AI. In the
case of Fox and Hounds , we will see that five moves of look-ahead are all the fox
ever needs; more depth does not help. With four or fewer moves, the fox may
wander too far from the vicinity of effective moves to ever see them. Tuning the
AI via look-ahead depth is effective only in games where incrementally more
look-ahead produces incrementally better AI.
Heuristics give guidance in ambiguous situations. Think of heuristics as general
rules, often based on experience. Heuristics are very helpful in game AI, and
evaluation functions need all the help that they can get. At some point, the AI will
hit the look-ahead depth limit, and the evaluation function will have to pass
Search Nedrilad ::

Custom Search