Mars Rover AI Hits 93% Laser Targets as Satellites Cull Cloudy Images Onboard
Updated
Updated · spacedaily.com · Aug 10
Mars Rover AI Hits 93% Laser Targets as Satellites Cull Cloudy Images Onboard
2 articles · Updated · spacedaily.com · Aug 10
Summary
NASA’s AEGIS software let Curiosity choose ChemCam laser targets from its own camera images, hitting scientist-desired material more than 93% of the time versus about 24% when teams guessed targets in advance.
A 4-to-24-minute one-way radio delay to Mars makes real-time control impossible, so the software solves a timing problem rather than making broad scientific judgments.
ESA’s 2020 Φ-Sat-1 used an onboard neural network to detect cloud-covered Earth images and discard them before downlink, preserving scarce power, storage and transmission time.
OPS-SAT’s SmartCam classifier reached about 95% accuracy filtering bad frames, and one engineer proposed, built, tested and flew it in under two weeks before routine use.
Across rovers, Earth-observation satellites and large constellations, AI is being deployed as narrow automation for delay, bandwidth and scale constraints—not general spacecraft intelligence.