Artificial Intelligence in Unreal Engine 5

# CHAPTER 12

Artificial Intelligence in Unreal Engine 5 #

1. Introduction #

2. Learning Objectives #

• Generate a NavMesh Bounds Volume to define walkable areas.

• Create an AI Controller to "possess" a Character.

• Understand the relationship between a Behavior Tree and a Blackboard.

• Implement AI Perception (Sight and Hearing).

• Script basic AI logic: Patrol, Chase, and Attack.

3. The NavMesh (The Map) #

• The Problem: If you tell an AI to walk to a tree, it will walk in a straight line, smashing endlessly into a brick wall that happens to be in the way.

• The NavMesh (Navigation Mesh): You place a "NavMesh Bounds Volume" over your entire level. The engine calculates all the floors, stairs, and obstacles, and generates a green invisible carpet. The AI uses this carpet to calculate complex pathfinding around walls (A* Algorithm). *Press 'P' in the editor to see the green NavMesh!*

4. The AI Controller (The Brain) #

5. Behavior Trees and Blackboards (The Logic) #

• The Blackboard: The AI's memory. It stores variables (e.g., TargetActor, IsPatrolling, LastSeenLocation).

• The Behavior Tree: A top-down flowchart that reads the Blackboard and makes decisions.

• Selector Nodes: Move from left to right. They ask, "Can I Attack? No. Can I Chase? No. I will Patrol."

• Sequence Nodes: Execute actions in order. "Walk to Target, THEN Play Attack Animation, THEN Wait 2 seconds."

6. AI Perception (Sight and Hearing) #

• You add an AIPerception Component to the AI Controller.

• You configure "AI Sight" (e.g., 90-degree field of view, 1500 units distance).

• When the player walks into that vision cone, the component fires an event: On Target Perception Updated. The AI Controller takes the Player Actor and writes it into the Blackboard variable TargetActor. The Behavior Tree immediately switches from "Patrol" to "Chase."

7. Visual Learning: AI Behavior Tree Flow #

[ ROOT ]
   |
[ Selector: What should I do? ] (Checks left to right)
   |
   +-- [ Sequence 1: Attack ] (Condition: If 'TargetActor' is within 2 meters)
   |       |-- Face Player
   |       |-- Play Attack Animation
   |
   +-- [ Sequence 2: Chase ] (Condition: If 'TargetActor' is Set)
   |       |-- MoveTo: 'TargetActor'
   |
   +-- [ Sequence 3: Patrol ] (Fallback: If 'TargetActor' is NOT Set)
           |-- Find Random Point on NavMesh
           |-- MoveTo: Point
           |-- Wait: 3 seconds

8. Best Practices #

• Use Tasks, Services, and Decorators: Inside a Behavior Tree, logic is modular. A Task is a specific action (e.g., "Attack"). A Decorator is a condition (e.g., "Only run this branch IF Health > 50"). A Service runs in the background while a branch is active (e.g., "Continuously update the player's location every 0.5 seconds while chasing").

9. Common Mistakes #

• Forgetting the NavMesh: A developer spends hours coding a brilliant AI brain. They hit play, and the enemy stands completely frozen. The most common cause? The developer forgot to put a NavMesh Bounds Volume in the level, so the AI literally doesn't know how to take a single step.

10. Mini Project: The Wandering Ghost #

1. 1. In your level, add a NavMeshBoundsVolume from the Place Actors panel. Scale it to cover the floor. Press 'P' to verify it turns green.

1. 2. Create a Blueprint Character named BP_Ghost. Place it in the level.

1. 3. Open BP_Ghost's Event Graph. Off Event BeginPlay, drag to a Delay node (set to 2 seconds).

1. 4. Drag from Delay to a Get Random Reachable Point In Radius node (Radius: 1000).

1. 5. Drag the output to an AI MoveTo node. The Pawn is Self, the Destination is the Random Point.

1. 6. Connect the On Success output pin back to the Delay node (creating a loop).

1. 7. *Result:* The ghost waits 2 seconds, picks a random spot on the green NavMesh, walks to it, and repeats endlessly.

11. Practice Exercises #

1. 1. What keyboard shortcut reveals the invisible Navigation Mesh (the green carpet) in the UE5 Viewport?

1. 2. Explain the relationship between the Blackboard and the Behavior Tree.

12. MCQs with Answers #

13. Interview Questions #

• Q: Explain the A* (A-Star) pathfinding algorithm conceptually. How does the NavMesh in Unreal Engine utilize this to navigate complex terrain?

• Q: Compare a "Selector" node to a "Sequence" node inside an Unreal Engine Behavior Tree. How do they handle Failures differently?

• Q: Walk me through the exact workflow of setting up AI Vision. How do you transition an AI from a "Patrol" state to a "Chase" state when they see the player?

14. FAQs #

15. Summary #

16. Next Chapter Recommendation #