Video Lecture · WEPS Series · Lecture 2 of 3 · Wildfire Escalation Prevention
Containment Infrastructure
Original title: Infrastructure de Confinement. From detection to containment: assembling technologies that already exist — independently, in different countries — into a single architecture, and stating its regulatory limits honestly. Delivered in French, with English subtitles. Runtime: ~45–50 minutes.
Author Andy Kross
Language French (English subtitles)
Runtime ~45–50 min
Project WEPS
Full Lecture · FR · English Subtitles · ~45–50 min
Available on YouTube ↗
This lecture is available in
About the Lecture

Lecture Two opens where Lecture One left off — with a gap precisely formulated but not yet filled. Its argument is simple to state and easy to underestimate: every layer of the architecture this lecture proposes already exists, built independently, by different teams, in different countries. Nothing here is invented. What's missing is the assembly.

Four precedents ground this claim before the architecture is even described. Pyronear, a French open-source detection system, contained a fire outbreak at Fontainebleau to 6,500 square metres in April 2026. FireTracking, a commercial camera network, is targeting 95% coverage of at-risk massifs in Indre-et-Loire. Dryad Networks, a German company, has gone further still — tree-mounted sensors that detect combustion at the smouldering stage, before flame or visible smoke, now on trial with 120 units in the Gard. And Zhuque, a Chinese system already in serial production, pairs reconnaissance drones with a ten-drone containment swarm capable of a five-hundred-kilogram payload in a single deployment.

The lecture draws one distinction with particular care, because the entire WEPS thesis depends on it: a swarm is not the same proposition as a single drone. Dryad's architecture, however advanced, answers a different question than the one this cycle is built around — it optimizes response speed on one fire. WEPS optimizes presence across several fires at once, the exact structural gap established in Lecture One. Reconnaissance, confinement, and the full four-module platform — Detect, Contain, Network, Command — are walked through with this distinction as the throughline, and with an honest note on why Detect is viable today while Contain is not yet.

A minute-by-minute operational scenario follows, showing detection, confirmation, swarm engagement, and firefighter handoff inside the same eight-to-ten-minute window the French doctrine already targets — this architecture does not compete for that window, it fills it.

The lecture closes on its most consequential admission. Since January 2026, the legal framework governing emergency-service drones in France has been unsettled, and any swarm sharing airspace with a Canadair or Dash requires certified collision-avoidance before it can fly. That is not a caveat added for caution's sake — it is the only legally viable mode of operation today, and it is why this architecture is deliberately conservative rather than fully autonomous, in direct contrast to competitors who claim otherwise.

Lecture Outline
00:00–05:00
Opening. Four precedents, four countries, one missing architecture. Pyronear, FireTracking, Dryad Networks, and Zhuque — every brick already exists somewhere. "The problem isn't the technology — it's the architecture that connects it."
05:00–10:00
Confinement vs. Extinction. Why this architecture doesn't put out fires — it slows them down until aircraft arrive. A deliberate terminology choice, and the reasoning behind it.
10:00–20:00
The Architecture, Layer by Layer. The core of the lecture. Optical detection (Pyronear, FireTracking) versus olfactory detection (Dryad) at Layer 1–2. Reconnaissance (HEDRAF) at Layer 3. Then the sharpest distinction in the cycle: a single drone versus a swarm — Dryad's speed-on-one-fire architecture against Zhuque's precedent for presence-on-several. Layer 5 closes on human response.
20:00–28:00
Platform Modularity. The four-module structure — Detect, Contain, Network, Command — and why Detect is deployable today while Contain remains dependent on the regulatory timeline covered later in the lecture.
28:00–38:00
The Operational Scenario, Minute by Minute. Detection to firefighter handoff, walked through in real time, landing inside the eight-to-ten-minute window the French doctrine already targets.
38:00–48:00
Honest Limitations. The unsettled legal framework for emergency-service drones since January 2026. Why DGAC certification requirements make a conservative, human-confirmed mode the only legally viable option today — a direct contrast with Dryad's own claim of operating "without human intervention." Does not replace firefighters. Does not treat large fires.