RESEARCH BASELINE · OCTOBER 31, 2026

A large sky survey can leave unresolved detections without discovering extraordinary aircraft. The Galileo Project’s infrared commissioning study offers a concrete example: roughly 500,000 reconstructed trajectories, with 144 still ambiguous after review. The authors describe those remaining cases as likely mundane. This is a retrospective examination of work published in Sensors in 2025, not a new October 2026 data release.

What the commissioning exercise measured

Laura Dominé and colleagues tested an array of eight long-wave infrared cameras over five months. Their calibration used aircraft positions from ADS-B broadcasts, alongside simulated tracks and hand-labelled observations. They assessed both whether observable aircraft were recorded and whether recorded aircraft were successfully detected.

A preliminary filter selected trajectories with unusually winding two-dimensional paths. About 16 percent were flagged for review. The 144 ambiguous survivors were not identified as exotic vehicles: distance, motion estimates or additional sensing channels were needed to clarify them.

Why the denominator matters

The primary paper gives the measurement pipeline a published baseline. Independent Universe Today reporting carried by Phys.org also highlighted the unresolved tracks and the team’s aim of obtaining distance through separated detectors. That report corroborates what the research team said; it is not an independent repetition of the experiment.

Our assessment is that the most useful number here is not an unexplained-object tally in isolation. Readers also need to know which targets entered the field of view, which were detected, and how often ordinary objects triggered the selection rule. Without those distinctions, a detector’s blind spots can be mistaken for a property of the sky.

Ambiguity is a research task

A path drawn across an image is a projection. Establishing its physical motion requires more information than the shape of that line. Additional observations may identify an ordinary target or leave a more compelling anomaly; the original flag alone does not decide between those outcomes.

The project’s Harvard-hosted publication entry identifies the study and summarizes its methods. That entry and arXiv describe the same research and should not be counted as independent replications. The value of the study is a measurable starting point against which later instrument performance can be compared.

Source boundary: we reviewed the published study record and reporting, not the raw trajectory archive, and did not reproduce the classification pipeline. The original AI-generated hero is a generic sensor concept, not an engineering drawing of Galileo hardware.