Unmanned vehicles have become an increasingly important tool of the U.S. armed forces in recent military conflicts. It is critical that these vehicles be able to fuse data from a variety of heterogeneous, widely dispersed sensor platforms into meaningful information, and share that information in a timely manner among groups of friendly manned and unmanned units. Charles River Analytics proposes an effort to design and demonstrate the feasibility of a system supporting Data Integration over Distributed Autonomous Sensor Platforms (DIDASP). We will construct a system that uses Probabilistic Relational Models to fuse local sensor data on a simulated USV. This data will be shared across a network of simulated USVs, that will be able to incorporate the distributed data into their own fusion efforts in turn, thereby creating a clearer common operational picture for the entire network. Human operators will be able to query the network for specific information about any given sensor contact detected by the USVs. They will also be able to use a rules engine to establish a set of automatic update and alert criteria.
Keywords: Usvs, Sensor Fusion, Autonomous Vehicles, Distributed Fusion,
