Neuromorphic Learning for Adaptive Control

The goal is energy-efficient in-situ learning and adaptation for space missions, starting with Lunar missions and extending to outer planet missions. The feasibility of this goal will be demonstrated in partnership with Intel using their next-generation Loihi processor and innovative machine learning algorithms developed at Ames. The demonstration will be done for a simulated lunar polar mission with the Tensegrity robot and benchmarked against previous offline adaptive approaches. This demonstration will initiate the path towards certification of Neuromorphic Spiking Neural Net in-situ learning processors with associated machine learning algorithms for space missions.

This innovation for in-situ energy-efficient learning with noise-resilient control is an enabling technology for missions like those to the Lunar poles. The properties of Lunar soil at the poles is poorly understood, and the alternative to an in-situ adaptive learning algorithm is to either have humans determine the appropriate control modality, or to collect sufficient data to train a controller on Earth before uploading it to the vehicle. The time to modify the onboard controller could negatively impact the performance of the mission. In Lunar operations, where surviving the Lunar night is a challenge, it may not be feasible to take this time to adapt an onboard controller. For missions beyond Cis-Lunar, such as outer planet moons, the need for insitu localized learning is even more compelling, as the energy required to communicate large datasets to Earth will be prohibitive – in addition to the time required for Earth-based adaptation.

This is a mature technology (TRL 7+) — the realistic path in is usually NASA's Technology Transfer Program: licensing an existing NASA patent, or a Space Act Agreement to use NASA facilities/expertise directly. NASA also runs a startup licensing program with no upfront fee for companies formed to commercialize a specific NASA technology.

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.