Enabling dynamics during environmental generation to support correlated simulations

ABSTRACT
The production of compiled environments used by various visual, constructive, or serious game modeling and simulation runtimes generally leave the production of dynamic events, weather, time of day, and AI entity movement to the runtime environment. When a OneSAF constructive simulation or VBS2 serious game runtime is created from source data, the defacto state of the environment is fixed. For military simulation, from the initial state, various actions during the course of the simulation will have an effect on the training environment. In particular, weapons effects on cultural features (buildings, walls, roads) and infrastructure (light and telephone poles, urban models) are calculated in the simulation, causing appropriate changes to the 3D environment. This can include surface cratering, partial or complete destruction of buildings, downing of utility poles, and so on. Ideally, the visual effect that shows the amount of damage should be easily computed and portrayed to the simulation participant. Likewise, the semantics of the weapon effect (e.g., loss of mobility or connectivity) needs to be updated in the simulation to affect future planning and events.

Traditionally this has been accomplished in a variety of ways, usually tied to unique capabilities of each runtime and the general nature of the event, communicated over the shared simulation network using a minimal (DIS / HLA) description. Further, the correlation of the effect of common events is, of course, limited to the ability of the runtime to express the same geometric and visual changes. Even simple terrain modifications, such as weapons cratering and bulldozing to create revetments, can both be time consuming and make it difficult to achieve spatial correlation across runtimes. These are fundamental runtime issues that we do not address in this paper.

However, in this paper we do explore the static generation of environments that can be used to support correlated events across multiple simulations. We limit our discussion to the generation of 3D structures with multiple precomputed destruction states that can be exported across multiple constructive and serious game environments. We use parametric generation from geospatial source data to simultaneously generate the individual structures as well as a family of destruct states that can be selected by the runtime. We also describe the process to generate simple 3D clutter models that can be parametrically generated. This process is particularly attractive in simulations where complete physics-based approaches are computationally expensive, but some level of analysis is available to select among a collection of dynamic events.

We will illustrate this work with examples using OneSAF and VBS2, achieving correlation in dynamic events between these two disparate simulation systems.

VITA
Mr. Daniel Hershey received a B.S. in Computer Science and minor in Physics with University Honors from Carnegie Mellon University in May 2006. Since joining TerraSim, he has contributed to a number of products involving urban environment generation. These products include automated building interior generation, parametric roof generation, and urban details processing. He developed the TerraTools OneSAF UHRB building generation process and the automated generation of geometry for other SAF systems. He has also contributed to TerraSim’s serious game export capability.