Artificial Intelligence - Infantry Squad Behaviors

The Norwegian Defence Research Establishment (FFI) is using Bohemia Interactive Simulations’ AI tool VBS Control Editor to develop a library of behavior models for semi-automated forces in VBS3. As the main institution responsible for defense-related research in Norway, FFI conducts simulations of land force operations for experimentation and analysis purposes. They presented the results of their work at I/ITSEC 2018 and have published a paper online titled “Modelling Human Behaviour using Behaviour Trees” here with more details.

Behavior trees are a relatively new way of creating AI behaviors for non-player characters in simulations and computer games.

“Behavior trees are well suited for developing human behavior models of moderate complexity for semi-automated forces in constructive simulations,” said Per-Idar Evensen, Senior Scientist, FFI. “Because they are modular and composable, they can represent more complex actions and sub-trees can be reused.”

VBS Control is a new software technology designed by Bohemia Interactive Simulations (BISim^TM) to provide next-generation AI for VBS3 and other VBS-based products. The core of VBS Control is the VBS Control runtime library (also known as the “AI Library”), which has been included for free in VBS3 since version 17.2. The library ensures that any VBS3 station can evaluate VBS Control AI behaviors without the need for an additional license. VBS Control technology is currently used in two distinct products: The VBS Control Editor, which is sold as part of the VBS Simulation SDK and allows users to create custom AI behaviors for use in VBS3, and VBS Control Behavior Pack 1, a set of pre-programmed artificial intelligence behaviors for the most common training use cases that can be added to VBS3 through a license sold separately.

End users of VBS Control Editor like FFI can create highly complex AI behaviors using the industry-leading behavior tree technology and Lua scripting. The behavior creation process is based on combining primitive behaviors provided by the VBS Control runtime library (such as navigation, tactical analysis, cover usage or weapons handling) into a more complex decision-making process, such as a full military doctrine.

FFI is using VBS Control to create a library of behavior models for the most important battle drills for mechanized infantry platoons, including behaviors for dismounted soldiers and combat vehicles like infantry fighting vehicles and main battle tanks. In the video above, FFI presents an example of an infantry squad behaviors created using VBS Control AI to simulated a reaction to contact battle drill. In this example, once the squad is engaged in combat by the opposing force, they form a line and then split into two fire teams. The teams then conduct backwards leapfrogging until they find a position to break off contact and find cover from the enemy.

“The composability and modularity of BT-based behavior models open up opportunities for collaboration on development and sharing of behavior models of battle drills between NATO and partner nations that have similar doctrines,” Evensen said.

For more information, download the FFI report, “Modelling Human Behaviour using Behaviour Trees.”