let’s meet & talk today

UFPUS Project

Unified Fire Protection Units and System-UFPUS

ARTIFICIAL INTELLIGENCE SYSTEM AND METHOD FOR FIRE IDENTIFICATION AND EXTINGUISHING IN REAL-TIME

Project Owner: EXINN Technology Center & Think Bank Albania Foundation

Knowing that forest fires represent a constant threat to ecological systems, infrastructure, economy and human lives, it is necessary to undertake joint actions towards increasing the level of protection against forest fires.

The main scope of the invention is to provide an Artificial Intelligence system and method for fire identification and extinguishing in real time, by combining and embodying existing tools, technology systems and products puzzle referred to as an innovation algorithm. This technology in the future should serve to protect the planet from natural fire emergencies and is ready to be test first in Albania by developing a pilot testing area/zone in one of our national forest parks in the country.

Technical Readiness Level –TRL 5

Technology validated in relevant laboratory environment.

After we concluded on the AI theoretical model, we achieved its code implementation (a prototype), and because we are lacking a NRT fire map feed, we tried to simulate it by using conventional methods of online maps. Furthermore, on the hardware side, we have achieved to operate the drone programmatically (with computer code), thus the AI system is able to automatically give coordinates to the drone (or also known as Planning a flight route).

The following is a list of our main achievements until now, regarding the software-hardware communication side: 1. AI theoretic model 2. AI implementation of the model (software prototype) 3. AI simulations, feedback and calibration of the model 4. AI notification system for personnel 5. Software communication with drones 6. Pro-grammatical plan of drone flight path from AI.

With further effort, we plan for the AI system to also be able to command the drones to drop the fire-extinguishing materials on the fire, once the drone has arrived at the coordinates, as one of our main goals is to make the whole system automatic. Note that the drones, once arrive to the fire location, they don’t need any more guidance from the satellite, as they are able to “see” it by themselves. Again, this is how we design to complete our AI model and system to automatically coordinate the drone to drop the material exactly on the fire source.

How does it work?

We intend to make use of an existing satellite technology, which provides NRT (Near-Real-Time) map feed with information about fires on the terrain. These satellites, technologically speaking, are able to detect and reflect fires on the map with red-dots. Now this is where comes the AI model that we designed, which, provided a real-time map feed, is able to detect little red-dots on the map as fire by itself, with no need to be supervised by human factor. After it has identified the fire/s, it notifies for its finding/s and provides the drone with coordinates for it to follow. The AI system, provided a near-real-time fire feed map, is able to scan, identify and automatically coordinate drones with the location of the fires.

Added value

The drone/s has an attached smartphone on its body, for being able to communicate with the central AI system, even when out of range to its respective drone-house, for getting real-time commands and streaming live image from the environment. The smartphone also gives the drone semi-autonomous behavior and decision-making capabilities, to self-optimize its flight path and further increase the detail of path of dropping the fire-extinguishing balls into the source of fire, by taking advantage of the drone’s camera, the smartphone’s computational power and by using image-recognition technology, specifically trained to detect fires and identify their source. In case that the AI system identifies a fast-spreading fire, it should dispatch a fleet of nearby drones to try and contain the situation.