Command Agents using BML for Tactical Decision Support

From Master Projects
Jump to: navigation, search

has title::Command Agents using BML for Tactical Decision Support
status: finished
Master: project within::AI and Communication
Student name: student name::Robbert Bronkers
Start start date:=2010/10/01
End end date:=2011/02/01
Supervisor: Tibor Bosse
Second reader: has second reader::Philip Kerbusch
Company: has company::TNO
Thesis: has thesis::Media:Thesis.pdf
Poster: has poster::Media:Media:Posternaam.pdf

Signature supervisor



General officers in the Dutch army face tactical decisions very often. Training is needed before actions are performed in real combat, due to the fact that in real combat many casualties are possible. However the problem often is that training a general officer, can become very expensive. Usually they have control over a high level in the hierarchy of the army. This creates the problem of high costs because of the large amount of soldiers and vehicles needed, combined with repeating simulations. As a solution to this problem, it is possible to let the general officers train with simulations of all sorts of scenarios. This method of training allows the officers to run a single scenario for many times, to find out how different choices would have influenced the outcome of the simulation. It benefits on the fact that many different scenarios, with combinations of all sections available in the army, on all sorts of terrains, can be practiced during these simulations. To run a scenario in a simulation, the general officer will give several orders which need to be performed in the simulation. These orders are given in a certain form; it is done with use of Battle Management Language (BML). This language is unambiguous and is used to control military operations.

A problem with running simulations is that the total hierarchy of the army must be programmed within the simulation. A tank battalion exists out of many tanks which individually need tasks. When an order is given it has to be passed to the subordinates. This creates many tasks to be programmed. This can be performed by a Multi-Agent System, to organize faster simulations. Such a system could be used to perform tactical reasoning on several levels below the commanding unit.

The main goal of the master project is to create a prototype of a Multi-Agent System which will make tactical decisions based on a command of a commanding unit. This is done to perform research on the possibilities of tactical decision support with use of a Multi-Agent System. Using such a system creates faster simulations; however there are many sorts of divisions within the army. Each division has a hierarchy which is comparable to others, combining different types of vehicles with different sort of soldiers. By creating a prototype of the system an idea can be given on the feasibility of the system to be implemented on all divisions, levels, and soldiers.