BRENNEN

Industries / Transport and vehicles

Aviation

Aviation is the business of flying passengers and cargo, together with the airports, air traffic control and maintenance shops that every flight depends on.

1914

On 1 January 1914 the St. Petersburg–Tampa Airboat Line flew its first scheduled trip across Tampa Bay, Florida, with Tony Jannus as pilot and the former St. Petersburg mayor Abram Pheil as passenger; IATA's history page calls it "the world's first regularly scheduled heavier-than-air airline".

iata.org

40

Milestones achieved across 7 domains.

14

To watch for: announced, not yet confirmed outside.

14

Open deliverables. Nobody has shipped them.

In one minute

Where it started

On 1 January 1914 a Benoist flying boat carried one paying passenger across Tampa Bay on the first trip of a scheduled airline, which flew for about four months. From 1918 the US Post Office flew scheduled airmail, and its contracts later kept the first private airlines going. Britain opened the first control tower at Croydon in 1920, and computers began booking seats and building crew schedules in the 1960s.

Where it stands

IATA, the airlines' trade association, expects airlines to carry 5.1 billion passengers and 71.7 million tonnes of cargo in 2026, on revenue of $1.165 trillion, and to keep 2.0% of it as profit, half the 2025 margin, because of fuel costs and disruption in the Middle East. ICAO, the United Nations aviation agency, counted 85 accidents on 38.08 million scheduled flights in 2025, or 2.23 per million departures, down from 2.56 in 2024, though deaths rose from 296 to 387. Much of the daily work, such as stacking bags in holds and rebuilding crew rosters after a storm, is still done by hand or with closed software.

What is open

Nobody may experiment on a live aircraft or in live airspace, so the open jobs here are in public data, simulation and bench tests. This research found no public benchmark for crew recovery, transcribing controllers' radio, solving conflicts on real traffic, matching contrails to flights, reading maintenance logbook pages or converting cargo messages to IATA's new data standard. Low-cost robot arms, robot models that learn from a few demonstrations, and AI models that read documents and web pages put each of these within reach of a small team.

Domains

AchievedWatch forOpen

Narrower industries

Helicopter servicesAirlinesAirportsAircraft maintenancePrivate aviationFlight schoolsAir taxis

Figureheads

1 person

Advances to build with

2 research notes

Open in aviation

3 of 14 open deliverables

Open · Flight operations

Contrails matched to flights

A public test set of real contrails seen in satellite images, each matched by people to the flight that made it, using public ADS-B tracks (ADS-B is the position report each aircraft broadcasts). Anyone could then score the automatic contrail-to-flight matching that avoidance trials rely on.

Built with what exists

GOES-16 satellite images (the images behind OpenContrails); OpenSky Network ADS-B history; pycontrails, an open-source (Apache 2.0) Python library for modeling aviation's climate effects, to predict where each flight could have made a contrail; an open-source labeling tool such as CVAT.

Credit · unclaimed

Open · Air traffic control

Public ATC transcription scoreboard

A fixed test and public scoreboard for turning recorded ATC radio into text, using only audio anyone can download, with one scoring script for every system and new speech-to-text models added as they appear.

Built with what exists

the free one-hour ATCO2 test set; the ATCOSIM corpus (about 10 hours of simulated controller speech from Graz University of Technology and the EUROCONTROL Experimental Centre, posted on Hugging Face); OpenAI's open Whisper models; ElevenLabs Scribe v3 (see the Note "Fewer word errors"); a Hugging Face Space to host the table.

Credit · unclaimed

Open · Airports and ground handling

Robot bag stacking, measured

A repeatable bench test of a robot stacking loose bags in the lower hold of a narrowbody jet. Build a short plywood section of a narrowbody lower hold at reduced scale, using the hold height and width in the manufacturer's public airport planning document. Fill it with scaled bags (weighted foam and cardboard boxes, drawn from the reference set below), teach a low-cost robot arm to stack them, and compare it with a person stacking the same bags in the same mock hold. No aircraft and no airport are involved.

Built with what exists

The SO-101 low-cost arm with Hugging Face's LeRobot software (see the Note "No download"); a robot model that learns a task from a few demonstrations, NVIDIA GR00T 2 (see the Note "Five demonstrations per task"); a depth camera; plywood; the Airbus A320 or Boeing 737 airport planning document for hold dimensions.

Credit · unclaimed