A Spatio-Temporal Memory System for Multi-Modal Sensor Fusion
Authors:
- Dr. James Crowder, CAES APD
-
Dr. Michael Hirsch, ISEA TEK LLC
Abstract:
Testing and assessment of complex systems requires the ingest,
correlation, and analysis of a variety of heterogeneous data. This is
particularly true in the testing, analysis, and reporting of multi-sensor data.
Complex sensor systems available on Unmanned Air System (UAS)
platforms require advanced techniques to enhance their resiliency and
survivability. This includes autonomous, artificial intelligence (AI)
architectures that can process, analyze, and provide actionable intelligence in
terms of understanding and reporting on actions and events focused on
system failures, either through degradation over time or operational mistakes.
Unmanned vehicles must operate in unstructured environments that are
inherently unpredictable and dynamical. An autonomous UAS must have
some degree of cognitive intelligence [1] to undertake tasks without direct
and continuous human involvement, especially in unknown environments.
This includes storing, correlating, and retrieving spatial and temporal tags
and information associated with sensor readings. Here we present the initial
design and testing of a Spatio-Temporal database system capable of storing
and correlating complex spatio-temporal information that allows inferences
to be made across both time and space (geography) to provide situational
awareness to sensor processing analysts. An example is provided
demonstrating the power and effectiveness of the Spatio-Temporal Database
Memory (STDM) and associated correlation algorithms.
Full access to this whitepaper requires completing and submitting the following form: