Φ-sat-1, flown on FSSCat

Detected clouds in hyperspectral images on board, so that cloudy data need not be sent down.

L1 Launched: own statement. Rests on c9.
R2 Ran: own technical account. Rests on c4, c5, c6.

Operator	European Space Agency
Developer	University of Pisa, Ubotica, cosine and Sinergise, with ESA
Host	FSSCat
Where	Earth orbit
Class	perception
Hardware	Intel Movidius Myriad 2, on the Eyes of Things board, behind the HyperScout-2 imager
Model or software	CloudScout, a convolutional neural network for cloud detection
Launched	Sep 2020
Ran	during the mission; no date given in what was read

What was said, and by whom

“is the first experiment to demonstrate on-board feature detection using hyperspectral Earth observation Convolutional Neural Networks”

ESA Φ-lab: the operator. Φ-sats programme, ESA Φ-lab. Published undated (no date in the text; written before Φ-sat-2 was contracted). Read 4 Oct 2026, in part: four excerpts of 3,100 characters.
“has successfully delivered its first image and cloud mask”

ESA Φ-lab: the operator. Φ-sats programme, ESA Φ-lab. Published undated (no date in the text; written before Φ-sat-2 was contracted). Read 4 Oct 2026, in part: four excerpts of 3,100 characters.
“represents the first ever on-board AI deep CNN inference on a dedicated chip”

Giuffrida et al., the mission team: the mission team. The Φ-Sat-1 Mission: The First On-Board Deep Neural Network Demonstrator for Satellite Earth Observation, IEEE Transactions on Geoscience and Remote Sensing (Giuffrida and twelve co-authors, the mission team). Published 4 Nov 2021 (date of publication printed in the paper). Read 4 Oct 2026, in part: section V.D in full, the start of the conclusion, and about 20 excerpts of 70,500 characters.
“NN inference with an exemplar TinyYOLO [36] model was continuously executed, with all inference outputs exactly matching the reference values”

Giuffrida et al., the mission team: the mission team. The Φ-Sat-1 Mission: The First On-Board Deep Neural Network Demonstrator for Satellite Earth Observation, IEEE Transactions on Geoscience and Remote Sensing (Giuffrida and twelve co-authors, the mission team). Published 4 Nov 2021 (date of publication printed in the paper). Read 4 Oct 2026, in part: section V.D in full, the start of the conclusion, and about 20 excerpts of 70,500 characters.
“for live images acquired from the HyperScout-2 sensor”

Giuffrida et al., the mission team: the mission team. The Φ-Sat-1 Mission: The First On-Board Deep Neural Network Demonstrator for Satellite Earth Observation, IEEE Transactions on Geoscience and Remote Sensing (Giuffrida and twelve co-authors, the mission team). Published 4 Nov 2021 (date of publication printed in the paper). Read 4 Oct 2026, in part: section V.D in full, the start of the conclusion, and about 20 excerpts of 70,500 characters.
“The performance of an on-board inference engine based on a machine learning (ML) algorithm for cloud detection was validated in flight. This is the first time such an experiment has been conducted in space.”

Giuffrida et al., the mission team: the mission team. The Φ-Sat-1 Mission: The First On-Board Deep Neural Network Demonstrator for Satellite Earth Observation, IEEE Transactions on Geoscience and Remote Sensing (Giuffrida and twelve co-authors, the mission team). Published 4 Nov 2021 (date of publication printed in the paper). Read 4 Oct 2026, in part: section V.D in full, the start of the conclusion, and about 20 excerpts of 70,500 characters.
“the first artificial intelligence (AI) technology carried onboard a European Earth observation mission”

ESA, relayed by eoPortal: a directory run by the operator's agency. PhiSat-1 & -2 (Φ-Sat-1 & -2) Nanosatellite Mission, eoPortal (a mission directory run by ESA). Published 9 Aug 2026 (page metadata; the page is revised over time). Read 4 Oct 2026, in part: about 30 excerpts of 65,900 characters.
“The first satellite to run a deep neural network on board”

The Consciousness AI, a third-party tracker: another tracker. Orbital Compute Tracker Showing Every Satellite Running Computers in Space, The Consciousness AI. Published undated (no date on the page). Read 4 Oct 2026, in full.
“Launched in September 2020 and in August 2024 respectively”

eoPortal, an ESA-run directory: a directory run by the operator's agency. PhiSat-1 & -2 (Φ-Sat-1 & -2) Nanosatellite Mission, eoPortal (a mission directory run by ESA). Published 9 Aug 2026 (page metadata; the page is revised over time). Read 4 Oct 2026, in part: about 30 excerpts of 65,900 characters.
“The EoT board features the Intel Movidius Myriad 2 vision processing unit (VPU)”

Giuffrida et al., the mission team: the mission team. The Φ-Sat-1 Mission: The First On-Board Deep Neural Network Demonstrator for Satellite Earth Observation, IEEE Transactions on Geoscience and Remote Sensing (Giuffrida and twelve co-authors, the mission team). Published 4 Nov 2021 (date of publication printed in the paper). Read 4 Oct 2026, in part: section V.D in full, the start of the conclusion, and about 20 excerpts of 70,500 characters.

Where sources disagree

The accuracy figure often quoted for this mission, 88.4%, is for a synthetic test set on the ground. No in-orbit accuracy figure was found in the parts of the paper read. Noted.

Claims to a first

First AI on an Earth observation satellite: stands as worded. Each wording is scoped: to hyperspectral images and convolutional networks, to a dedicated chip, or to a European Earth observation mission. JPL describes onboard machine learning on EO-1 from 2003, which is earlier but is not a deep neural network on a dedicated chip, so the scoped wordings survive it.
First AI on an Earth observation satellite: unchecked. A third party's wording, wider than any of ESA's. It is recorded to show how a scoped claim grows when repeated. Nothing read establishes it.

Notes

The paper is peer reviewed, but its authors are the mission team, so it counts as their own technical account (evidence 2), not as independent confirmation. No catalogue object was found under the names tried.

Changed

Run raised from R1 to R2 after the mission paper was read.

This record as data