(duration: 11/2011 - 03/2013 /// funding: Agentschap NL, Innovatie voor Maatschappelijke Veiligheid, thema Fysieke Bescherming)
In the spring of 2011, several fires scourged the dune area in the province Noord-Holland. Such large fires in natural reserves are difficult to locate and monitor. Even when witnesses report smoke, the fire department has a difficult time accessing the area and finding the actual site. Personnel has to move into the area, which can be dangerous because of toxic smoke and fast-moving fire.
It is important to speed up and improve detection and localisation of fires. If not spotted and combatted in time, large dune fires can be quite a disaster, both socially, economically and environmentally. The FireSwarm project brings together the Bergen Fire Department, the Safety Region Noord-Holland-Noord, and several academic and technology partners to solve this problem.
A Swarm of Unmanned Aerial Vehicles
FireSwam proposes to use robots to take over the dangerous task of detecting and monitoring large fires. FireSwarm will introduce swarms of small, autonomous flying robots: Unmanned Aerial Vehicles, or UAV's. UAV's are already being used in several comparable application areas, but these projects use large, expensive drones. FireSwarm will use smaller, cheaper drones, which has several advantages.
A large swarm of small drones can cover much more ground than a single UAV, and individual drones are expendable, both technically and financially. A swarm is also highly scalable: if the area to be covered is bigger, you can simply launch more drones. Finally, small drones are safer; they cause less harm if they crash.
The FireSwarm partners will develop a UAV that meets all the project's criteria: it must be light-weight, energy-efficient, cheap, and it has to carry several types of sensors. Further challenges lie in the software domain:
Autonomy: it is impossible to control a swarm of robots from the ground, so the drones must be able to fly by themselves.
Coordination: drones must be able to fly in a swarm, and respond to other drones and objects in the air, and conditions such as side-winds. Further, they must be able to quickly cover the entire designated area.
Robust communication: the drones themselves will function as a network, allowing communication amongst themselves and with human users. The network must continue to function, even when single drones malfunction.
Pattern recognition: using sensor fusion, the drones must learn to recognise fire or smoke from the sky, and use sensor imput for coordinated flying, and determining location.
GUI: Fire fighters must be able to quickly deploy the UAV's to a specific area. Collected information must be quickly and accessibly shown to the fire fighters.