Framework to enable accelerated development and deployment of AI/ML Solutions

The goal of this proposed Phase II effort is to implement a specific use case for PEO STRI that uses the HyperBlox frameworkĀin the R&D cycle to help accelerate the development of moreĀspecific AI/ML solutions. Using this novel framework, HyperBlox can facilitate the rapid creation of software for Army applications, regardless of use-case. With current methods, manually developing and deploying an AI/ML software for a specific Army use-case may take 1-3 years HyperBloxÆs framework will be capable of expediting this process using AI/ML toĀaccelerate software development andĀenable deployment within months rather than years. As part of the Phase IĀstudy, a feasibility analysis of the HyperBlox framework was completed.ĀA live demonstration was given to the PEO STRI stakeholdersĀusing two sample applications that were developed and deployed using the HyperBlox framework. Two new use cases have been identified by PEO STRI within the Synthetic Training Environment (STE) for the potential Phase II effort: Resource optimization:ĀThe RVCT consists of high-tech, interactive equipment - including a heads-up display, high resolution monitor and representational controllers - that connects to the STE to activate collective, mixed-reality training scenarios. Availability of hardware resources and network bandwidth limits the ability for real-time or near realtime distributed simulation and also limits the ability to deliver an effective synthetic training experience. The use case involves developing a suite of AI/ML algorithms to learn and predict resource utilization (network bandwidth, power consumption, cpu/memory performance) for the various scenarios. These predictive resource analytics are then used to drive resource reservation andĀscenario scheduling across the RVCTs within the distributed environment toĀplan, prepare, execute & assess multi-echelon collective training and improve user experience. 1. Cognitive automation:ĀThe Stinger Training System supports live, collective force-on-force training that measures the gunner's ability to properly execute all tasks for the Stinger operation. If the training program is aidedĀwith AI/ML improvements, they can help improve weapon handling and firing procedures to develop and maintain gunner proficiency at a high level and enable a more dynamic training experience for gunners.ĀThe use case involves developing an AI/ML application that analyzes the video feed andĀmeasured data received from the gunner, and based 2. ĀĀĀ ĀĀĀĀĀĀĀĀĀĀ on an AI/ML model,Āpredicts the steps the gunner should have taken to achieve the intended result.ĀThe predicted steps can also be compared against the actual steps taken by the gunner to provide feedback on the improvements that could be targeted by the gunner.ĀAnother use of AI/ML can be to analyze the video feed and measured data and provide real-time live feedback to the gunner to reduce the cognitive burden. This canĀimprove gunner proficiency, shorten training time and effect savings in resources.