A research team from the Public University of Navarre’s Smart Cities Institute is working on the communications system of an autonomous drone, operated by an expert system, for use in emergency missions. The work is part of the EMERAL project, in which the UPNA is participating alongside the companies Tesicnor, Sistemas Navarra, i3Code Solutions, Naitec and Fuvex. The objective is to arrive at drones capable of taking off autonomously in weather conditions with the potential to cause natural disasters, fly according to a pre-set flight plan, collect information of interest in the area and return to the control centre to transmit the information to the emergency services.
The project ‘EMERAL: EMERgency uAvs for Long range’, funded by the Government of Navarre following its latest call for strategic projects, aims to use autonomous drones during floods and to replace helicopters —the use of which involves greater cost and risk for emergency personnel— to gather information during both preventive stages, to draw up contingency plans, and emergency situations, to launch rescue missions or other actions on the ground, and even following disasters, to analyse the effects in the area.
From left to right, Ion Goizueta Crespo, Jesús Villadangos Alonso, José Ramón González de Mendivil Moreno, Alberto Córdoba Izaguirre, Federico Fariña Figueredo, José Javier Astrai Escola and Francisco Javier Falcone Lanas
The work in progress is based on a previously developed web app which is already operational, the flood emergency plan management app Noe, designed by Tesicnor, a system which serves as an early warning in the event of emergency according to a set of predefined parameters. The research team wishes to include autonomous drone flight operated by an expert system in the EMERAL project, which is part of this application. ‘We do not know of any other initiative in the world which covers these possibilities,’ explains Jesús Villadangos, an ISC researcher working in the project, which also involves the UPNA researchers Alberto Córdoba, José Javier Astrain, Francisco Falcone, José Ramón González de Mendivil, Ion Goizueta, Federico Fariña, Almudena Ochoa Lainez, Pablo Medrano Fernández and Sandu Ilco.
‘The idea is for an expert system to inform when it’s recommendable for a drone to make a reconnaissance flight, a flight plan to be generated automatically, the plan to be loaded on the vehicle and the flight to start. We’re working towards a scenario in which everything can be done autonomously but in practice final authorisation from a person in charge of the system will always be needed according to current regulations.’ This is actually positive because it will mean that the system continues to learn over time, Villadangos adds.
The main challenge the researchers face is the development of the expert system to analyse the prevailing weather variables and recommend the action to take. ‘For this, we’ve used a rule system based on fuzzy logic, similar to a neural network but capable of modelling human knowledge. In the end, what we do is introduce the knowledge of the emergency technicians in the system and model their assessments on the basis of existing parameters, which results in specific decisions,’ the researcher summarises.
Overcoming geographical obstacles
Another challenge the researchers face is analysing the terrain and generating a flight plan which is able to overcome any possible obstacles in, for example, mountainous areas. The system relies on LiDAR (light detection and ranging) data bases, an optical remote sensing technique which gathers information on terrain using laser technology and the data from which most European states manage in publicly accessible databases. It also incorporates other data sources such as satellite imagery from the Sentinel-2 mission or public data on land-use layers. Using these data, the system will have to automatically generate flight plans, which involves numerous difficulties.
‘Planning in a desert is easy, but in mountainous areas it’s very complicated, among other things because in most countries the regulations require constant communication between the drone and the operator so that manual operation can be activated if needed even when the drone is flying remotely,’ explains the researcher.
To get over this problem, the members of the project have designed a flight procedure for emergency missions using a second drone which serves as a communications liaison with the reconnaissance drone. Finally, to ensure maximum safety in missions, the UPNA research team is working to improve the security software on the drone flight controller.