Syllabus
1. Introduction to human-like AI for training • Introduction to Virtual Battlespace 4 and VBS Control AI framework (military simulation standard used by military organizations in more than 60 NATO and NATO-friendly countries) • Determinism and explainability of behaviors • Military VS Other domains • VBS Control Editor setup
2. Behavior Tree Primitives I • Basic control nodes • Basic leaf nodes • Leaf nodes scripting customization • Messaging • Definition of modules
3. Individual behaviors I • Non-autonomous: Move, Fire, Mount, Dismount
4. Behavior trees primitives II • Advanced control nodes • Subtrees • Sensors • Orders
5. Group behaviors & emulation of the command structure • Non-autonomou fire+team behavior
6. Integration practice • API design, architecture design • Reusable trees VS single purpose trees
7. Behavior tree primitives III • Asynchronous queries • Path-planning • Cover system • Defilade system
8. Individual behaviors II • Autonomous: Occupy, Fast Move, Safe Move • Reaction trees
9. Group behaviors II • Non-autonomous squad behavior: Move in column • Autonomous fire-team behavior: Fast Occupy, Safe Occupy • Cover system on group level
10. Information aggregation & high-level tactical analysis • Platoon non-autonomous behaviors • Platoon sensors
11. Integration practice II • API design, architecture design • Best practices in maintainable behavior tree design • Reusable trees VS single purpose trees
Annotation
This subject follows up on Human-Like Artificial Agents (NAIL068) with the goal to teach good principles in designing complex, group-based, controllable, believable, and structured agent behaviors in artificial environments. It introduces advanced behavior tree functionalities and present best practices of their design.
The course is highly practical, modeling most of the challenges in a military simulation environment, which has high standards in regards to stability and functionality of agent behaviors.