Neuromorphic Spacecraft Fault Monitor

The proposed work adapts the Neuromorphic Spacecraft Fault Monitor for a specific customer need where fast identifications of anomalous behavior are critical to their business model. NSFM leverages the novel computational efficiencies in neuromorphic hardware to provide next generation spacecraft fault monitoring. It augments existing approaches with contextually aware anomaly detection and diagnostics support. This contextual awareness is enabled by proprietary algorithms optimized for the non-Von Neumann architecture of neuromorphic processors. It saves users from having to investigate the same anomaly twice because it recognizes an anomaly as having happened before and informs the response. We have shown that our algorithms are comparable to state of the art in terms of accuracy while saving over 100x in energy delay product when compared to a Long Short Term Model (LSTM) Deep Neural Network on conventional processors, or equivalently, tackling 100x more complexity for the same power. This CCRPP extends this work to the Spinnaker chip (pictured right) and potentially one additional chip. This will enable the system to tackle larger problems, up to and including monitoring an entire space station the size of the International Space Station.  Technical Objectives Show state of the art accuracy in anomaly detection using Spinnaker while showing order of magnitude improved energy-delay product Show feasibility of implimentation on additional neuromorphic processor Space flight risk mitigation efforts (planning) Define/scope a space-qualified version We plan to meet these technical objectives by leveraging our experience developing for Intel's Loihi, Spinnaker 2, and our understanding of anomaly detection and anomaly characterization problems. We will develop a strong connection with our beachhead customer and demonstrate a product that improves a key part of their value stream to their customers. We will deliver reliable code and prototype hardware that is rigorously tested. We will develop a detailed plan for a future in-space demonstration based on lessons learned. Deliverables: Initial draft of customer requirements Refined draft of customer requirements including draft Software Definition Document (SDD) Software prototype using Spinnaker Prototype demonstration report describing power efficiency, compute time, and energy delay compared to existing approaches. Second software prototype using second hardware implementation Documentation of path to radiation tolerant hardware/software product Quarterly and Final Reports  

Potential NASA applications include onboard fault detection for satellites, lunar rovers, martian rovers and other power constrained systems. Particularly useful for systems with long light-time delay to Earth. Potential non-NASA applications include onboard satellites, particularly onboard large constellations of satellites with minimal onboard computation, limited power availability, and high reliability requirements.

This is early/mid-stage (TRL 5) — 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.