Making the game act.
WHAT THE GOAL IS
Opponents that are interesting to play against, not opponents that are optimal.
WHY THAT DISTINCTION MATTERS
A perfect opponent is frequently unenjoyable.
WHAT APPROACHES EXIST
State machines, for simple distinct behaviours Behaviour trees, for more complex decision-making Utility systems, scoring options and choosing the best Planning systems, sequencing actions toward goals
WHAT BEHAVIOUR TREES PROVIDE
Readable, editable decision structures that designers can modify.
WHY THEY ARE COMMON
They scale better than state machines without requiring programming for every change.
WHAT PATHFINDING SOLVES
Moving through a world with obstacles.
WHAT THE STANDARD APPROACH IS
Searching a graph representing navigable space.
WHAT A NAVIGATION MESH IS
A representation of walkable surfaces, generated from the level.
WHAT TO BE CAREFUL WITH
Pathfinding cost with many agents Agents crowding and blocking each other Paths that are technically correct and look unnatural
WHAT MAKES BEHAVIOUR READABLE
Telegraphing: showing what is about to happen.
WHY THAT MATTERS
Players must be able to react, which requires warning.
WHAT TO ADD DELIBERATELY
Imperfection, so opponents feel fallible.
