We will extend our previous work to create artificial intelligence (AI) Reasoning Modules for planning, scheduling, characterization, machine learning, and fault detection/diagnosis/reconfiguration for spacecraft and their subsystems, each able to operate in standalone fashion or be easily integrated with one another to execute in a variety of computational environments, including in highly distributed situations. We will integrate our existing AI Modules within NASA’s core Flight System (cFS) so that they can be used (through cFS) on a wide variety of spacecraft, from large manned vehicles to small scientific instruments. We will also integrate the AI Modules on MSU’s RadPC (radiation tolerant processing CPUs) in an experiment onboard the ISS. In addition to an inflight demonstration of our AI modules, this will greatly accelerate the maturation of MSU’s RadPC, which replaces $200,000 RAD750 radiation hardened processing with equivalent processing power in $100 FPGA chips using soft-CPUs, quadruple redundancy, and FPGA reconfiguration for seamless recovery, achieving 3 orders of magnitude reduction in cost as well as significantly reduced CPU electrical power. The ISS experiment will fly for six months and feature two RadPC boards, one of which will be utilizing the full suite of AI Modules to monitor, detect, diagnose, and recover the other RadPC board as well as its own, providing an inflight demonstration for both RadPC and for the AI Modules.
The modules will utilize cFS’s Software messaging Bus (SB) and the networking version (SBN) to provide the integration mechanism for either local or distributed applications. A specific spacecraft mission could utilize the AI Scheduler merely by sending it tasks, resources, and constraints in the defined messaging format across the SB or SBN. A different application could use a different AI Module; Characterization, for example. A third might use all of the AI Reasoning applications.
Through cFS, a large number of future manned and unmanned spacecraft would benefit, including NASA GRC EPS Planning and Scheduling applications. With its ability to react to real-time events to autonomously create high-quality plans and schedules, the cFS AI Reasoning applications will illustrate their advantages over the status quo. There is a potential to automate the majority of subsystem management decision-making at NASA, The Phase II demonstration of the AI Reasoning modules in space onboard the ISS will greatly aid its adoption.
This technology can be sold to current Aurora customers and companies similar to them such as aerospace manufacturers, oil refineries, ship builders, mining operations, factories of all types, hospitals, auto makers, as well as commercial manned and unmanned spacecraft manufacturers and operators.