Satellites
Pass overhead a few times a day and see a wide area at once. A small fire between passes is invisible, and cloud or thick smoke can hide one that is directly beneath.
Wildfire Ahead · Drone Operations
A wildfire is easiest to stop in its first hour.
The hard part is knowing which hour, and where to look. This is the part of Wildfire Ahead that turns a drone flight into evidence a fire agency can act on — and keeps the decision to fly firmly in human hands.
Licensed stock photography. Not Wildfire Ahead equipment or operations.
The problem
Everything that watches for fire has a blind spot.
No single way of looking is enough on its own. Understanding why is the whole reason this subsystem exists.
Fixed cameras
Watch continuously, but only the arc they point at. Beyond the ridge in front of them, they see nothing at all.
People on the ground
See the most and are the most trusted — and there are never enough of them, least of all on the days when everything is burning at once.
An aircraft can answer one question, at one place, at one time. The intelligence is knowing which question is worth the flight.
Licensed stock footage of a real wildfire. It was not detected, predicted or responded to by this platform.
What actually happens
Five steps, and a person in the middle of them.
This is the entire system in one picture. Read it once and you know what we do — including the two connections we have designed but have not finished building.
- Where to look. The platform ranks places by how much an observation would actually settle — how likely an ignition is, how uncertain the current picture is, what is downwind, and how long it has been since anything looked.
- A person decides. The ranking is a suggestion put in front of a named human. It is never an instruction to an aircraft, and there is no configuration of this system in which it becomes one.
- The agency's aircraft flies. Under the agency's own pilots, procedures and aviation approvals, using whatever drone they already operate.
- What came back. Imagery arrives with the record of how it was captured — where, when, by what sensor, in what coordinate system, calibrated or not. Anything that cannot say is held back rather than counted.
- What we now know. The new evidence is weighed against satellite, lightning and ground observations, and the ranking in step one changes. The loop tightens.
Where to look
Knowing where you cannot see.
"We saw nothing" and "we could not see" are different statements. A system that confuses them will eventually reassure someone about a place it never observed — so the map of what has been observed is kept separately from the map of what was found.
Your fleet, not ours
We do not sell you a drone.
Agencies already own aircraft, and rarely all of one make. A provincial fire service, a utility and a county emergency office will each have bought differently, for good reasons, years apart. Most software makes you pick a side. We built the layer underneath instead, so the fleet stays your decision — including the decision to change your mind later.
These aircraft already fly over the same country every week. A utility inspecting a transmission line in August is flying through the same landscape a fire agency is worried about — and the picture it brings back is the same kind of evidence. One interface is what makes that possible rather than merely imaginable.
A drone adapter
Discover a fleet, read what each aircraft can do, submit a route, follow a mission. Every method either reads state or makes a proposal — there is no method that commands an aircraft, and that absence is the design, not an oversight.
A sensor adapter
Turns whatever came back into one record: where and when it was taken, by what sensor, at what accuracy, in which coordinate system, calibrated or not. A thermal image that cannot state its calibration is held back rather than scored.
Capability profiles
A versioned statement of what an airframe and payload combination can genuinely do, so nothing plans a mission an aircraft cannot fly. Two profiles exist today, for the hardware we have modelled.
What is actually built
Judge this the way an engineer would.
Wildfire software is easy to demonstrate and hard to trust. So here is the inventory, with the stage of each item stated plainly. Nothing below is a plan; every line is code that runs today.
Built— Working code you can run today. Flight-plan validation
Uploaded route files are checked against archive, size, structure, namespace and coordinate rules before anything reads them. A malformed or hostile file is rejected with the reason.
Simulated— Runs against synthetic data, never a real aircraft. A protocol twin
A complete stand-in for the drone dock and aircraft, exact enough to develop and test against without an aircraft in the air. Every run is deterministic and replayable, so a defect found once can be reproduced exactly.
Built— Working code you can run today. Fail-closed preflight
Route, schedule, altitude, weather, restriction, battery reserve and operator approval are each checked separately. The most permissive answer the system can give is ready for a human to release — never "cleared to fly".
Built— Working code you can run today. Replayable receipts
Every assessment can be replayed later to show whether an input or a policy changed since. An answer you cannot re-derive is not evidence, and after an incident that difference is the only thing that matters.
Designed— The contract is specified and reviewed. It is not built yet. The link to the wider platform
Ranked places to look flow out; captured observations flow back. Evidence in both directions, authority in neither. Specified, reviewed, and not yet built.
Gated— Blocked on an approval only a regulator or agency can give. Flying near a real fire
Requires authorizations from Transport Canada or the FAA and from the agency running the incident. No software here substitutes for one, and no future release will.
Take part
Who this is for.
- Fire and emergency agencies
- Bring the aircraft and the approvals you already have. See what the evidence looks like before committing to anything.
- Drone manufacturers and integrators
- The second adapter is the one that proves the interface is real. If that is your fleet, we would like it to be yours.
- Researchers
- The hard open problem is deciding where a flight is worth it. The data to answer that well does not exist yet, and saying so is how we intend to get it.
Access grants a view of a synthetic development environment. It is not an aviation approval and confers no authority to operate an aircraft.