An Introduction to Game AI - Game Coding Complete - page 625

function IsHealthLowNode:Decide()
local hitPointPercentageLeft =
self._teapot.hitPoints / self._teapot.maxHitPoints;
if (hitPointPercentageLeft <= self._lowValuePercentage) then
return self._trueNode:Decide();
else
return self._falseNode:Decide();
end
end

The first node is the IsObjectCloseNode class, and it checks to see if the object ID
stored in _testObjId is

defined by the _closeDistance variable. This
replaces the hard-coded check to see if the player is close with a more generic ver-
sion. This is a great example of how you can parameterize nodes to make them more
reusable. For example, this node could be used to detect how close a health pack is
with no modifications.
The Decide() function is pretty straightforward and very similar to the hard-coded
block you saw earlier. If the object is valid and is found to be within the appropriate
distance, the true node

“ close, ”

s
Decide() function is called. This steps down a level in the tree and starts the pro-
cess again. IsHealthLowNode works in the same way. The only difference is the
actual logic.

' s Decide() function is called. Otherwise, the false node '

Use Percentages Instead of Absolutes

Notice the usage of the _lowValuePercentage variable. You might be
wondering why I m using a percentage here instead of an absolute value.

Using a percentage allows the max hit points for

'

the teapot to change

without having to update this logic at all. It will always consider anything

less than 34% of the max hit points to be “ low. ” If I used an absolute value,

I ' d have to update it whenever the max hit points of the teapot changed.

The only thing left is the brain itself:

DecisionTreeBrain = class(TeapotBrain,
{

_root = nil,
});

function DecisionTreeBrain:Init()
self:_BuildDecisionTree();
return true;
end