A Cognitive Architecture Using Reinforcement Learning to Enable Autonomous Spacecraft Operations

We propose an architecture to enable the modular development and deployment of autonomous intelligent agents in support of spacecraft operations. This architecture supports both training and application of artificial intelligence models. It particularly enables the use of deep reinforcement learning for each module independently and jointly. Deep reinforcement learning is a technique that enables the automated learning of plans of action and has recently successfully been used, for example, to learn strategies for games like Go. Our proposed architecture provides a "utility" layer for generalized learning and a provides for independent functional modules that can be added, modified, or removed easily. It also accounts for intensive multicore computational needs. Lastly, it allows for desired behavior to be learned independently or in the context of the broader system. In Phase I, we will deliver a preliminary cognitive architecture, a feasibility study, a prototype of an autonomous agent, and a detailed plan to develop a comprehensive cognitive architecture feasibility study.

The proposed technology can be deployed as a backup in current spacecraft and can play a foundational role in upcoming deep-space missions, which will require higher levels of autonomy than current missions. The system proposed can autonomously manage many spacecraft operations, including systems health, crew health, maintenance, consumable management, payload management, food production, and recycling.

The proposed technology can be applied in any setting that would benefit from robust, autonomous management, including airplane piloting, autonomous or semi-autonomous trucking, decision support, and medicine.

This is early/mid-stage (TRL 3) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.