Command & Control Agents within Battlefield Simulation for Training
|has title::Using Agent-based Simulation in a Decision Support System for Military Command & Control|
|Master:||project within::Computational Intelligence and Selforganisation|
|Student name:||student name::Torec Luik|
|Second reader:||has second reader::Nico de Reus|
Abstract KIM 1
Previously, TNO has developed a Multi-Agent System (MAS) for use in the military Command & Control (C2) domain. The C2 system currently in use by the Royal Netherlands Army (RNLA) is the Integrated Staff Information System (ISIS), from which commanders can issue orders to their units on the battlefield. Moreover, ISIS can also be connected to a Computer Generated Forces (CGF) simulator, like VR-Forces, which can model battlefield situations and simulate their outcomes. However, the orders sent from ISIS are adressed to high-level aggregate units (e.g. Battalion or Company), while the simulator requires input on a low-level (e.g. Platoon or single units). The MAS is currently used to transform the high-level orders output from ISIS to the low-level input orders for VR-Forces. To this end, the military hierarchy of Battalion, Company and Platoon commanders was modelled in the MAS using Belief-Desire-Intention (BDI) agents. The previous work on these agents has allowed them to succesfully perform the following (simplified) orders in the simulator: Attack, Defend and Seize, the latter being the combination of the first two. These orders are all manipulations of the moveTo-task and the settings of the units' Rules of Engagement (RoE), e.g. Fire at Will.
For further use of this system, TNO is now looking for an application of this MAS in the C2 domain, which can be either in training or in planning / decision support. We have decided that creating a (simple) decision support application would be more useful at this time than a (simple) training application. The main research question is therefore:
� "How can we (most optimally) create a (simple) decision support tool for military planning?".
Our research question entails multiple subquestions, such as: � "What part(s) of military planning can be supported using CGF simulation?",
� "What systems have previously been used to support (these parts of) military planning?",
� "What (extra) orders are needed to provide proper functionality in the decision support system?",
� "What refinements are needed in the current orders for them to provide proper functionality in the decision support system?",
� "What refinements are needed in the agent framework to provide proper func- tionality in the decision support system?",
� "How can these changes be implemented in the current system framework?",
� "How should we present the decision support system to the users?",
� "How should we present the results of the simulation in the decision support system?".
More specifically, for this Master's thesis we have identified a simple case where a military commander is supported by the MAS in the evaluation of his Course of Action (COA). In a discussion with four military simulation experts, requirements for this case have been defined. The current MAS will be expanded and an application will be built which allows the commander to compare routes for his units based on a chosen Measure of Effectiveness (MOE).
Abstract KIM 2
For this Master's thesis, we have created a prototype Decision Support System (DSS) in the military Command & Control (C2) domain for TNO. This DSS is meant to be used by military commanders, using simulation to support them in comparing multiple Courses of Action (COAs). For a DSS to be useful in C2, it needs to be easy and quick to use and it needs to provide realistic and reliable results. In order to achieve these results, requirements for the DSS have been acquired in both a review of related literature and in discussions with Subject-Matter Experts (SMEs) from the Royal Netherlands Army (RNLA) and TNO. The results provided in the DSS are obtained from simulation, so for realistic and reliable results we need a realistic simulation.
Therefore, we have extended Command Agents, a Multi-Agent System (MAS) created by TNO for interoperability between C2 and simulation, to provide more realistic tactical behaviour in the simulation. To this end, military tactical doctrine has been implemented in the Command Agents, based on Combat Instruction Sets (CISs) provided by the RNLA. These CISs describe in high-level steps the behaviour required for the correct implementation of certain orders and for correct reactive behaviour. We have chosen 3 main CISs to be implemented: Assault on an Enemy Position, Tactical Road March and Hasty Occupation of a Blocking Position. Each of these CISs showcase a different aspect of the agent's possibilities as a replacement of human input needed in a realistic simulation.