8. Artificial Intelligence

8. Artificial Intelligence

Describe the behavioral logic of enemies, allies, and NPCs, as well as their relationship with difficulty, performance, and anti-cheat measures.

8.1 Opponent / Enemy AI

8.1.1 AI Taxonomy

• Types: Patrol, Assault, Defensive, Ranged, Summoner, etc.
• Boss & Elite AI: Unique mechanisms and phase transitions.

8.1.2 Perception & State

• Perception System: Sight cones, hearing radius, and threat generation.
• State Machine:
  • Required: State Transition Diagram (Idle → Alert → Chase → Combat → Flee).

8.1.3 Behavior Logic

• Architecture: Selection of Behavior Trees, GOAP, or Scripted Logic (Behavior Trees recommended for complex AI).
• Decision Frequency: Update rates and performance budgeting.
• Difficulty Scaling: Impact on behavior (smarter tactics vs. just higher stats).

8.1.4 Group AI

• Coordination: Flanking, covering fire, and role distribution.
• Aggro Management: Threat distribution and token-based attack slots.

8.2 Non-combat Characters

• Ambient NPC Schedules: Daily routines, patrol paths, and interaction nodes.
• Service NPCs: Interaction flows for Merchants, Quest Givers, and Function NPCs.
• Atmospheric NPCs: Crowd density and background behaviors to enliven the world.

8.3 Friendly Characters & Companions

• Roles: DPS, Healer, Tank, Buffer.
• Follow Logic: Distance maintenance and pathing relative to the player.
• Synergy: Combo triggers and elemental reactions with player actions.

8.4 Support AI

8.4.1 Player and Collision Detection

• Colliders: Hitboxes, hurtboxes, and physics layers.
• Tolerance: "Coyote time," edge correction, and forgiving hit detection.

8.4.2 Pathfinding

• Algorithm: NavMesh, Grid-based, or Waypoints.
• Complex Geometry: Handling destructible environments, verticality, and dynamic obstacles.
• Authority: Division of responsibility between Client (prediction) and Server (validation).

8.5 Difficulty & Adaptivity

• Modes: Design philosophy for Normal, Hard, Nightmare, etc.
• Dynamic Difficulty Adjustment (DDA): Scaling enemy strength/count based on player performance.
• Balancing: Safety nets for new players vs. skill ceilings for veterans.

8.6 Performance & Anti-cheat Considerations

• Budget: Maximum active AI agents per frame and tick rate limits.
• Server Authority: Critical states and outcomes validated server-side.
• Trust Model: Defining what data the client can see vs. what it can influence.