MIKROS SYSTEMS CORPORATION — Department of Defense SBIR Phase I: N193-A02

Description

As Unmanned Surface Vehicles (USV) and Unmanned Undersea Vehicles (UUV) shift control to on-board autonomy in operator-on-the-loop or fully autonomous modes, the quality of the sensor data fed to machine learning or artificial intelligence (AI) systems has significant impact on Guidance, Navigation and Control (GNC) decision-making and mission success. When continuous communications with external control authorities (C2) is not possible, the vessel must handle in-stride detection and corrective action for degraded or failed sensors. Mikros Systems Corporation proposes a Sensor In-stride Diagnostic, Correction, and Confidence Component (SID3C) to mitigate degraded sensors in the context of the Unmanned Maritime Autonomy Architecture (UMAA). The UMAA Architecture Design Description (ADD) prescribes a modular, extensible platform with onboard systems existing as a software component with access to existing services and data streams. While the modular UMAA architecture fosters flexibility and independent components, it also requires a interoperable data structure where SID3C can adapt to changing vessel and payload configurations. A new approach merging subject matter expert (SME) domain knowledge with artificial neural network (ANN) machine learning is proposed to adapt to agile platform configurations without complete system redesign.