Updated
Updated · spacedaily.com · Aug 10
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.

Insights

If AI filters out half of satellite imagery as unusable, what groundbreaking anomalies might we be blindly deleting in orbit?
With AI executing hundreds of thousands of orbital maneuvers, what happens when two autonomous satellites choose conflicting escape paths?
Could shifting data processing from Earth into orbital networks create an autonomous space infrastructure that operates entirely beyond human control?