GOAP AI Engine

Documentation | Support

This package requires AI Navigation package (com.unity.ai.navigation) from Package Manager

GOAP Engine lets your AI agents automatically plan and execute sequences

of actions to reach their goals — without hand-crafting state machines or

behaviour trees. The planner uses A* search to find the cheapest valid

action chain from the current world state to the desired goal state.

HOW IT WORKS

Configure goals, actions, requirements, and effects entirely from the

Inspector. The planner handles everything — finding the optimal action

sequence at runtime, re-planning reactively when the world changes, and

locking resources to prevent conflicts between multiple agents.

PLANNING SYSTEM

• A* search with admissible heuristic (guaranteed optimal plans)
• Async (threaded) planning — A* runs on a background thread
• GC-optimized — flat float[] state arrays, node pool, state pool
• Plan cache — repeated state+goal combos skip the search entirely
• Relevance-based re-planning — only re-plans when relevant state changes
• Reactive re-planning triggered by state change events (no polling)

INSPECTOR-BASED SETUP

• Goals, actions, requirements, and effects — all configured in Inspector
• No code required for basic AI setups
• Action Creator window generates boilerplate-free action subclasses
• BehaviorPreset ScriptableObject for shareable behavior configurations

SENSOR SYSTEM

• RangeSensor — sphere-based detection with Physics.OverlapSphere
• LineOfSightSensor — FOV cone + raycast obstruction check
• Configurable poll interval, output modes (Boolean/Count/NearestDistance)
• Writes directly into behavior state — triggers reactive re-planning

MULTI-TARGET NAVIGATION

• TargetBinder component — attach to any scene object
• Automatic resource locking — prevents two agents claiming the same target
• Shared targets — allow multiple agents simultaneously (waypoints, etc.)
• Dynamic costs — distance-based or context-based cost per planning cycle
• Target wait timeout — hold plan and retry before switching goals

FEATURES

• DSL Expressions — compact string syntax for requirements and effects
• Conditional Effects — And/Or condition guards on individual effects
• Parametric Actions — one class, multiple Inspector-configured variants
• Plan Ordering — enforce action order with name DSL ([After]/[Before])
• Goal Evaluator — custom A* scoring function for urgency/distance bias
• IncludeOnPlanning flag — disable actions without removing from Inspector

SHOOTER ENEMY ACTIONS

Eight ready-to-use action classes: PatrolAction · ChaseAction ·

FireWeaponAction · TakeCoverAction · ReloadAction · AlertAction ·

InvestigateAction · PickUpAction

CHARACTER SYSTEM (Optional)

Includes Character component and plug-in abilities:

Movement · Run · Jump · Rotate · Input · Weapon · PickUp · Health ·

Animation · IK · HUD · NavMesh · Camera

GOAP GRAPH EDITOR

Visual node graph of the full action/goal/state dependency network.

• Design-time: full dependency view, Inspector panel, Edit Script button
• Play-mode: live state values, running action highlight, runtime overlay
• Undo/redo node moves, subclass field support