ARCHARITHMS INC — Department of Defense SBIR Phase I: A20-078

Description

Advances in artificial intelligence (AI) and deep learning techniques, in conjunction with rapid growth in GPU hardware performance have opened up new possibilities for exploitation of AI to perform highly complex tasks with performance exceeding that of more traditional approaches. Potential applications within the Army Integrated Air and Missile Defense (IAMD) system include Identification and Classification of tracked objects, Defense Design, and Dynamic Planning and Tasking. To support AI / Deep Learning-based applications, the Army requires a robust, scalable architecture, framework and algorithm engine that can be utilized by multiple AI applications in an easily maintainable and extensible manner. The architecture and framework should include a combination of hardware and software that has a straightforward path of integration with current IAMD systems. The AI Engine should support integration and execution of multiple, simultaneous AI applications, as well as the ability to ingest, store, and process significant amounts of data. In addition to execution, the architecture, framework and engine should support training of the algorithms, which minimizes hardware and software costs as well as permits on-the-fly enhancements to the different applications.