Sir Airffic GoGet this map into your head and every other page becomes easier.
Five jobs of the flying brain
Machines now share the work a human pilot does in their head.
Sense: cameras, radar, LiDAR (a sensor that measures distance with lasers) and thermal cameras turn the surroundings into numbers.
Understand: Vision AI tells a power line from a person, and SLAM (building a map while working out your own position in it) tells the drone where it is.
Decide: choose which way to go, what to avoid and how to carry out the mission.
Act: adjust each motor precisely so the aircraft moves as decided.
Learn: retrain the AI with flight data so the next flight is better. Companies that turn this loop fastest grow strongest.
- Sense — Camera · radar · LiDAR
- Understand — Vision AI · localization
- Decide — Path · avoid · mission
- Act — Motors · flight control
- Learn — Flight data improves models
Five layers: from components to services
The market is a stack of five layers.
Sensors sit at the bottom, then compute chips, perception AI and autonomy software. At the top is the operations and airspace OS layer that flies many aircraft at once.
Lower layers are components, so they compete on scale and deep technology. Upper layers are services, so they compete on closeness to customers and volume of data.
Some companies own every layer (DJI, Skydio); others dig deep into one (NVIDIA, Echodyne). Which approach wins is the subject of the Strategy chapter.
- 5 Operations & airspace OS
Fly many aircraft safely at once (UTM, ops, data)
Airffic · Wing · DroneDeploy - 4 Autonomy & decisions
Plan paths, avoid obstacles, complete missions
Skydio · Shield AI · Auterion · PX4 - 3 Perception AI
Understand objects, people and position from images
Vision AI · SLAM · foundation models - 2 Compute chips
Run AI on board, fast and low-power
NVIDIA · Qualcomm · Ambarella - 1 Sensors
Measure the world with light, radio and lasers
Camera · radar · LiDAR · thermal
One drone's brain is not the sky's brain
One drone's brain keeps that drone from crashing. The sky's brain keeps many drones from crashing into each other — traffic control for the air.
The more aircraft there are, the less one drone's intelligence is enough. Self-driving cars still need traffic lights and road rules.
Airffic calls this sky-wide brain the airspace OS. It merges what each drone sees into one shared map and feeds mission control, video analytics and routing.
- Each drone senses and perceives
- The airspace OS merges one shared sky map
- Mission Control, Vision and Routing & Stock use it
Sir Airffic GoThe smarter each drone gets, the more valuable sky-wide coordination becomes. That is the most important cause-and-effect on this site.
Where the market begins and ends
We cover the technology that lets drones perceive the outside world (sensors, perception, autonomy) and the systems that run the sky safely with it.
Propellers, batteries and photo-editing apps are out of scope. Counter-UAS (detecting and handling unwanted drones) is in, because it uses the same sensors and AI.
| In scope | Out of scope |
|---|---|
| Cameras, radar, LiDAR, thermal and other sensors | Batteries, motors and propellers themselves |
| Computer vision, Vision AI, SLAM, foundation models | Photo-editing apps for aerial images |
| Autonomy and detect-and-avoid (DAA) software | Drone retail and insurance |
| Onboard AI chips and compute modules | Generic cloud servers on the ground |
| Operations, airspace OS (UTM), counter-UAS | Crewed air traffic control itself |
Frequently asked questions
What is the difference between automated and autonomous drone flight?
Automated flight follows a pre-planned route. Autonomous flight perceives its surroundings during flight and decides for itself — avoiding obstacles or changing route on the spot.
What is SLAM?
Simultaneous Localization and Mapping: building a map of the surroundings with cameras or LiDAR while working out where you are in that map. It matters most where GPS fails, such as indoors or under bridges.