AI-ML and space exploration at NASA
summary
A machine‑learning pipeline aims to give the MOMA mass spectrometer on ExoMars on‑board science autonomy by turning raw LDMS traces into 1,000–to‑2000‑bin 1D arrays plus seven method parameters, filtering garbage via UMAP + clustering, then matching spectra to labeled samples with baselines (random forest, k‑NN) and a tuned custom neural network. The lab ETU yields ~500,000 raw spectra reduced to ~30,000 scientific spectra after cleaning. A working Earth prototype clusters and classifies data and is being integrated into mission software; on‑board decisioning is still future work.
AI-written summary · about
chapters (7)
- 0:00Introduction and background
- 4:31Challenges in mission operations
- 9:01Science autonomy concept
- 11:07ExoMars MOMA instrument case
- 16:44Machine learning tools and results
- 22:49Other projects and challenges
- 24:15Q&A session
related talks
40mExploring deep space and deep ocean with AI guidance
20mMining Large Data Sets of Biomedical Omics Da...
41mExploring Dark Energy & Star Formation w/ ML
31mFrom Radar Echoes To Real-Time Recognition: Deep Learning on Another Planet -An Open Source Workflow
45mFrom Radar Echoes To Real-Time Recognition: Deep Learning on Another Planet an Open Source Workflow
41mAccelerated Discovery to Knowledge using Machine Learning
30mCultivating a company-wide machine learning culture
51mWhen the Universe Calls