Tutorials

Learn by building. These tutorials walk you through creating real-world behavior trees for games, robotics, and automation. Each tutorial includes a complete tree you can import into the editor.

🎮 Game AI Tutorials

Tutorial 1: Patrol Bot with Alert System

Goal: Create an NPC that patrols between waypoints, investigates disturbances, and returns to patrol after investigation.

Difficulty: Beginner | Time: 15 minutes

Tree Structure

Selector "PatrolOrInvestigate"
├─ Sequence "InvestigateDisturbance"
│  ├─ Condition "IsDisturbanceDetected"
│  ├─ Action "MoveToDisturbanceLocation"
│  ├─ Action "LookAround"
│  └─ Action "ReturnToPatrolRoute"
│
└─ Sequence "ContinuePatrol"
   ├─ Decorator "RepeatForever"
   │  └─ Sequence "PatrolLoop"
   │     ├─ Action "MoveToWaypoint_A"
   │     ├─ Action "Wait" (duration: 3s)
   │     ├─ Action "MoveToWaypoint_B"
   │     ├─ Action "Wait" (duration: 3s)
   │     └─ Action "MoveToWaypoint_C"

Step-by-Step Instructions

1. Open Behavior Tree Editor and clear the canvas
2. Drag a Selector node onto the canvas (this is your root)
3. Add two children to the Selector:
   • A Sequence named "InvestigateDisturbance"
   • A Sequence named "ContinuePatrol"
4. Under "InvestigateDisturbance", add:
   • Condition: IsDisturbanceDetected
   • Action: MoveToDisturbanceLocation
   • Action: LookAround
   • Action: ReturnToPatrolRoute
5. Under "ContinuePatrol", add a Decorator → RepeatForever, then add a Sequence with patrol actions inside
6. Export as JSON and test in your game engine

Tutorial 2: Enemy AI with Health-Based Behavior

Goal: Design an enemy that changes tactics based on health level: aggressive when healthy, defensive when injured, and fleeing when critical.

Difficulty: Intermediate | Time: 25 minutes

Tree Structure

Selector "EnemyBehavior"
├─ Sequence "FleeWhenCritical"
│  ├─ Condition "IsHealthBelow" (threshold: 20%)
│  ├─ Action "FindCover"
│  └─ Action "MoveToFleePoint"
│
├─ Sequence "DefendWhenInjured"
│  ├─ Condition "IsHealthBelow" (threshold: 50%)
│  ├─ Selector "DefensiveActions"
│  │  ├─ Action "ThrowSmokeGrenade"
│  │  ├─ Action "CallForBackup"
│  │  └─ Action "TakeCoverAndHeal"
│  └─ Action "EngageFromDistance"
│
└─ Sequence "AggressiveWhenHealthy"
   ├─ Condition "IsHealthAbove" (threshold: 50%)
   └─ Selector "AttackPatterns"
      ├─ Sequence "MeleeAttack"
      │  ├─ Action "ChargeAtPlayer"
      │  └─ Action "PerformMeleeCombo"
      └─ Sequence "RangedAttack"
         ├─ Action "TakeAimingPosition"
         └─ Action "FireWeapon"

Implementation Tips

• Use different threshold values for health conditions to create hysteresis and prevent rapid switching
• Add cooldown decorators to prevent spamming abilities (e.g., call for backup only once per minute)
• Test edge cases: what happens at exactly 50% health? Ensure conditions are mutually exclusive

🤖 Robotics Tutorials

Tutorial 3: Mobile Robot Navigation with Obstacle Avoidance

Goal: Build a navigation tree for a warehouse robot that safely navigates to delivery points while avoiding obstacles and humans.

Difficulty: Advanced | Time: 40 minutes

Tree Structure

Selector "SafeNavigation"
├─ Sequence "EmergencyStop"
│  ├─ Condition "IsEmergencyButtonPressed"
│  └─ Action "ImmediateStop"
│
├─ Sequence "HumanDetected"
│  ├─ Condition "IsHumanInPath"
│  ├─ Action "SlowDown"
│  ├─ Action "AnnouncePresence"
│  └─ Action "NavigateAroundHuman"
│
├─ Sequence "ObstacleDetected"
│  ├─ Condition "IsObstacleNearby"
│  ├─ Selector "AvoidanceStrategy"
│  │  ├─ Action "SteerLeft"
│  │  ├─ Action "SteerRight"
│  │  └─ Action "BackUpAndReplan"
│  └─ Action "ResumePath"
│
└─ Sequence "NormalNavigation"
   ├─ Action "ComputePathToGoal"
   ├─ Condition "PathComputed"
   ├─ Action "FollowPath"
   └─ Condition "ReachedGoal"

Sensor Integration

In a real robot, conditions would read from sensor data:

• IsHumanInPath: Computer vision + LiDAR fusion
• IsObstacleNearby: LiDAR point cloud analysis
• IsEmergencyButtonPressed: Digital input from hardware

ROS 2 Integration Example

# Launch file example
ros2 launch bteditor_runner bt_runner.launch.py \
  --tree_file navigation_tree.xml \
  --websocket_port 8765 \
  --tick_rate 10

Tutorial 4: Robotic Arm Pick-and-Place

Goal: Create a manipulation tree for a 6-DOF robotic arm to pick objects from a conveyor belt and place them in bins.

Difficulty: Advanced | Time: 50 minutes

Error Handling Strategy

Manipulation has many failure modes. Wrap each major phase in retry logic:

Decorator "RetryWithBackoff" (max_attempts=3, delay=1s)
└─ Sequence "PickObject"
   ├─ Action "CaptureImage"
   ├─ Action "DetectObjectPose"
   ├─ Condition "ObjectDetected"
   ├─ Action "PlanGraspTrajectory"
   ├─ Action "ExecuteGraspMotion"
   ├─ Action "CloseGripper"
   └─ Condition "ObjectSecured"

🏭 Automation Tutorials

Tutorial 5: Automated Testing Sequence

Goal: Design a test runner that validates a complex system through multiple test scenarios with automatic recovery.

Difficulty: Intermediate | Time: 30 minutes

Parallel Test Execution

Use Parallel nodes to run independent tests simultaneously:

Parallel "RunAllTests" (success_threshold=all)
├─ Sequence "TestDatabaseConnection"
│  ├─ Action "PingDatabase"
│  └─ Condition "ResponseReceived"
│
├─ Sequence "TestAPIEndpoints"
│  ├─ Action "SendHealthCheckRequest"
│  ├─ Condition "StatusCode200"
│  └─ Action "ValidateResponseBody"
│
└─ Sequence "TestMessageQueue"
   ├─ Action "PublishTestMessage"
   ├─ Action "ConsumeMessage"
   └─ Condition "MessageMatches"

📚 Learning Resources

Recommended Reading

• Getting Started Guide — Your first behavior tree in 5 minutes
• Node Types Reference — Complete catalog of available nodes
• Reactive Control Patterns — When to use behavior trees vs. state machines
• Robotics Patterns — Domain-specific best practices

Video Tutorials (Coming Soon)

We're producing video versions of these tutorials. Subscribe to our YouTube channel to get notified when they're released.

Community Examples

Check out these community-contributed behavior trees:

• Chess AI: Opening book → Middlegame evaluation → Endgame tablebase lookup
• Smart Home Automation: Morning routine → Energy saving mode → Security monitoring
• Drone Delivery: Package pickup → Route planning → Weather adaptation → Safe landing

← Docs index Getting started →

Last updated: 2026-07-16