Geon has successfully developed a prototype of our Multi-Input Data Fused Image Processing System (MIDFIPS) under the Navy Technology Acceleration Program. MIDFIPS is an Artificial Intelligence/Machine Learning (AI/ML) based image-processing payload designed to provide a complete operating picture for surveillance and reconnaissance tasks. MIDFIPS analyzes multi-modal imagery to perform object detection and fuses results to enhance decision making processes. It incorporates a model-reasoning mechanism within the “black-box” components which is leveraged to mitigate adversarial attacks.
Envisioned for low SWaP hardware, MIDFIPS’ image processing capability leverages existing COTS solutions shown to perform advanced computational imaging. The system ingests and analyzes multi-modal imagery to perform object detection and utilizes a data fusion component to derive a new image classification decision. The underlying framework anticipates adversarial attacks that can negatively impact one or more of the sensors that MIDFIPS is processing. As a result, the data fusion component performs “sensor selection” based on score metrics of the analyzed data of individual pipelines to inherently carry failover and mitigate loss of surveillance capability.
In a dynamic battlespace where adversaries contest access and project power in disruptive ways, the ability to obtain reliable intelligence and paint a unified operating picture is crucial. MIDFIPS aims to bridge sensor technology gaps in unmanned platforms, embracing the concept of persistent awareness with advanced analytic technology facilitated by ML. Geon’s solution addresses the limitations of individual sensor information by performing data fusion and maintaining situational awareness through the application of the multi-modal sensor paradigm at the tactical edge.
To view the results from our testing or to get a demonstration of the MIDFIPS system, please contact us today!.
