Introduction to AI Programming for Games

The post introduces game AI through a Monster Wars example. It starts with simple if/else decisions for attack, defense, and unit building, then fixes movement with steering behaviors, and finally refactors the enemy strategy into a finite state machine.

Reading notes

#

• AI in games can be as simple as normal coding if the game only needs basic decisions.
• The first enemy logic compares army value to choose attack or defense, then picks units by countering the player’s unit mix.
• The tutorial uses Monster Wars, a Cocos2D iOS starter project, to keep the focus on AI.
• The original movement code moves monsters directly toward targets and causes jagged behavior, rounding issues, and sticking.
• Steering is presented as the AI side of movement, alongside pathfinding, with seek, flee, arrive, wander, separate, and path following as examples.
• Seek sets acceleration toward a target at max strength, but can overshoot.
• Arrive adds target radius, slow radius, and deceleration so units stop at the destination instead of flying past it.
• Separate adds acceleration away from nearby allies so units do not stack on top of each other.
• The state machine section defines four states: Massing, Countering, Defending, and Rushing.
• Defending reacts to enemies near the base and builds counters while under attack.
• Countering waits for an attack, then spends saved money to respond with overwhelming force.
• Rushing attacks when the army is strong enough and keeps building Quirks while ahead.
• The refactor moves state behavior into separate AIState subclasses with enter, execute, exit, and name methods.
• The tutorial ends by noting that game AI should serve fun and pointing to books and an example project for further study.