Real-time futures graph tracking visualization and analysis tool

[摘要] A hybrid envisionment is a novel representation of a simulated state graph, specifying all possible states and transitions of the system, characterized by both qualitative and quantitative state variables. The Deep Green project creates a hybrid envisionment, called a futures graph, to depict all possible occurrences and outcomes of a combat engagement between friendly and enemy units on a battlefield. During combat, Al state estimation techniques are utilized to efficiently track the state of the battle in a futures graph, giving the commander an up-to-date analysis of what is taking place on the battlefield and how it the battle could turn out. Because state estimation of highly complex hybrid envisionments is a relatively unexplored and novel process, it is important to ensure that it is handled efficiently and accurately enough for usage on the field. This paper explores an approach for discerning the behavior in state estimation through the use of an analysis suite. By accompanying Deep Green state estimation with the analysis suite developed, estimation techniques could be benchmarked and analyzed over various implementations through both numerical and graphical metrics. The metrics generated greatly helped to improve the estimation algorithm over the course of its development.

[发布机构] Massachusetts Institute of Technology