The increasing importance of urban combat and operations in modern warfare drives a need for effective simulation of the urban environment for training, testing, and planning purposes. One of the defining elements of the urban environment is a dense population of people and vehicles. This presence adds many considerations to urban operations including concealment of hostile forces among noncombatants, sensor confusion, the risk of civilian casualties, and navigational difficulties. Attempting to simulate this civilian presence at an entity level, rather than an aggregate level, poses many difficulties. The volume and density of the entities to be simulated necessitates the minimization of computational expense and network traffic generated per vehicle. The high amount of resources required to simulate any significant fraction of a city's population suggests a distributed approach. Distributing this task, however, is difficult as fluctuating traffic patterns unbalance loads and the large coverage areas of modern sensors prevent many obvious simplifications and optimizations. In support of the Urban Resolve Experiment for the Joint Experimentation Directorate of the US Joint Forces Command (JFCOM J9), a simulation application named CultureSim was developed to handle the simulation of the background population of a city. CultureSim has been under continual development since 1999. Today, working the computational and network limitations of its simulation environment, it has successfully generated over half a million simultaneous entities to populate an urban environment. Each entity moves independently, using road networks to follow a behavioral profile that generates realistic traffic patterns. As the simulations it participates in grow in sophistication, especially with the inclusion of modern 3D views of the battle space, there has been an increased pressure to improve the realism of the movement of these urban vehicles. Latency from the distributed environment, the risk of desynchronization between simulators, uneven vehicle distributions, changing traffic patterns, the need to react to other participants in the simulation, and the need to produce a reasonable traffic flow from abitrary vehicle layouts all make this task difficult. This paper details many of the problems and obstacles faced by CultureSim in producing realistic urban traffic and the solutions implemented to address these problems.
展开▼