Publication
Simulation is frequently used in the study of multi-agent systems. Unfortunately, in many cases, it is not necessarily clear how faithfully the details of the simulated model represent the behavior of the physical system. Often, the effects of the environment in which the system is to be placed are even neglected entirely. Taking into account the entire system (including interactions with the target environment), establishing a clear hierarchy among multiple levels of modeling not only enhances the fidelity of the individual models, but also emphasizes the tradeoffs inherent in each. Understanding and leveraging the full spectrum of models allows the use of fast, high-level models for exploration in the parameter space, the results of which can be verified on more precise low-level models. Here, we demonstrate the generation of a family of models for a robotic wireless sensor network engaged in an acoustic detection task. Quantitative correspondence is shown between modeling levels and with the physical system.