Detection and tracking based on a dynamical hierarchical occupancy map in agent-based simulations

摘要

Agent-Based Models in military simulation need a model for detection and tracking other agents. It has been suggested that statistical models, such as occupancy maps or particle filters, can be used for that purpose. An occupancy map is one possibility for this task. The more volume of space, however, in a simulation, the more the computational demand of using occupancy maps grow and the more benefit could be obtained by the ability to switch to a coarser granularity in at least some parts of the volume. Using both possible benefits of an occupancy map, fine granularity in tracking and detection where needed and less computational demand by switching to low granularity where possible, parts of the volume will be transferred to a new occupancy map on a higher hierarchal level with coarser granularity. Only the most interesting areas in the simulation have fine granularity. The main contribution of this research will be to provide an improved algorithm and a prototype for using a hierarchy occupancy maps in agent-based simulations involving large volumes of simulated space.