Do you need
What we have
  • Challenging training scenarios for military flight simulation?
  • Intelligent enemy fighter aircrafts (CGFs) for tactical air-air combat mission training?
  • An artificial intelligence module that can efficiently be coupled to your simulation platform?
  • Accessible tools for instructors or domain experts to design realistic, non-scripted behavior models for virtual opponents?

Need for smart virtual opponents

In tactical engagements it is indispensable for fighter pilots to be able to change their tactical plans and to adapt their behavior to the situation at hand. The training value of military mission simulations therefore increases significantly if pilots are able to practice such tactical engagements against enemy fighter aircraft. However, there is still much to gain in simulating the enemy who still exhibits rudimentary tactical behavior in nowadays training simulations.

Despite the far-reaching development of artificial intelligence, also for military applications, it remains a challenge to model human behavior in real-time mission simulators. The behavior of virtual opponents (‘bandits’) is a key component in flight simulation. At present, these simulated opponents usually act according to predefined scripts, meaning they have a limited ability to react dynamically and realistically to fighter pilots performing a training mission.

Smart Bandits AIR is an artificial intelligence module designed to allow the creation of intelligent behavior models for virtual opponents in tactical air-air combat simulation. By combining state-of-the-art artificial intelligence techniques with an easy-to-use interface, behavior models can be designed to specify virtual opponents attack plans and self-defense actions without the need for programmers. Behaviors models can be designed and executed fully independently of a specific simulation platform.

Effective behavior modeling

  • The software toolkit comes with an easy-to-use graphical editor through which behavior models for air platforms can be created efficiently.
  • Using the AI technique of hierarchical finite state machines, behavior doctrines present in the heads of domain experts can intuitively be translated to computational models without the need for programmers or technical expertise.
  • Predefined content is included in the form of behavioral building blocks for aircrafts to perform maneuvers, countermeasures or missile launches, or to reason based on visual, radar or radar warning receiver observations.

Flexible and easy to operate

Smart Bandits AIR can be used as an extension to any simulation platform and can be operated by instructors or domain experts to create and run training scenarios.

  • Scenario creation: operators can create scenarios by building behavior models for virtual opponents which can be tested directly in the target simulation platform.
  • Scenario execution: During a training session, operators can inspect and follow the mental states of Smart Bandits agents at run-time.

Easy to integrate

Smart Bandits AIR can be integrated within hours into COTS simulation platforms through a flexible integration process. Pre-built integrations have been created or are planned for STAGE, VBS2/3 and VR-Forces. Other platforms can easily be supported.

The development of Smart Bandits AIR was initiated by the Royal Netherlands Air Force (RNLAF). Current developments are focused on the use of new technologies such as machine learning and the application of Smart Bandits AIR for embedded training.

Video Clips

Related Issues

  • Artificial intelligence for airborne CGFs
  • Behavior modelling
  • Computer generated forces (CGF)
  • Embedded training
  • Instructor station
  • Intelligent agents for military mission simulation
  • Machine learning
  • Mission simulation
  • Scenario creation
  • Tools for behavior modelling
  • Virtual agents
  • Virtual combat training