Efficient management of sensory data for an autonomous submersible utilizing a parallel processing architecture

The authors describe a systematic method of managing sensor information, utilizing a bit serial systolic array, so that irrelevant data storage and archival are rendered unnecessary. Improved sensor configurations and sensor postprocessing techniques designed to remove the operator from a control loop, creating a fully autonomous vehicle (untethered) are presented. an overview of submersibles and a discussion of underwater sensor system architecture are also presented. A systolic array for a fusion processor is discussed. It is found that systolic arrays provide a significant advantage to multisensor fusion research. Not only do the massive number of processors of an array accomplish repetitive and similar processing for numerous sensing elements simultaneously, but their mesh connectedness also promotes implicit data associativity without explicit software allocations