EPFL wins the U.V. Helava Award in drone mapping

The small plane used by the reseachers in Vufflens (VD). © TOPO / EPFL

The small plane used by the reseachers in Vufflens (VD). © TOPO / EPFL

Drone software developed at EPFL lets surveyors generate high-precision maps of even the most inaccessible areas. This major breakthrough has just won the U.V. Helava Award for its novel approach.

Every field of research has its Holy Grail. In Geomatics Engineering, that would be an ultra-efficient method for generating a map of a given plot of land with centimeter-level resolution and accuracy. The advent of drones has taken the quest for such a method to a whole new level. Today’s drones can be equipped with aerial positioning sensors and fly in areas that GPS can’t reach, in order to map the hardest-to-get-to places such as dams tucked away in mountains and remote sections of highways and railways.

Three scientists at EPFL’s Geodetic Engineering Laboratory (TOPO) – Davide Cucci, scientist and lead author of the study, Jan Skaloud, senior scientist and corresponding author, and Martin Rehak, a PhD student – have recently made a major breakthrough in drone mapping technology. Their work, published in the August 2017 issue of the Journal of Photogrammetry and Remote Sensing, has earned them on 6 June the 2017 U.V. Helava Award, beating out hundreds of other articles appearing in the journal that same year. The Award’s selection panel highlighted the team’s “new approach and significant contribution” to their field. Their pioneering method for processing the data collected by drone sensors delivers more accurate results when mapping convoluted surfaces.

New processing method

The drone mapping process involves comparing a picture taken by a drone with the drone’s GPS positioning at the time the photo was taken and inertial observations made by the drone’s sensors. These are the same kind of sensors and filters used on smartphones to detect when the device has been turned from a horizontal to a vertical position. On drones, the sensors indicate the machine’s exact spatial orientation when a photo is taken. Here, accuracy is extremely important because an error of just a fraction of a degree in the drone’s areal positioning can translate into several meters of variance on the ground – sabotaging any attempt to generate accurate maps of critical areas, such as to calculate coastline erosion.

In conventional drone mapping systems, data points are processed one after the other. But the EPFL scientists have developed a method for processing the data all at once. Their technology – available as open-source software – views the data as a whole and results in higher-precision maps. The Award’s selection panel also pointed to the field tests carried out as part of the study; for the past several years, TOPO researchers have been flying drones over land dotted with sensors (near Vufflens, in Vaud Canton) to validate their approach.

Mapping what cannot be seen

The scientists’ search for their Holy Grail won’t stop there. Next, they plan to equip drones with lasers in order to more accurately pinpoint objects on the ground and map what cannot be seen: the thickness of snow, the length of a tree’s roots or the wood content of a tree trunk. They hope that one day their software, designed for maximum ease of use, will be available to all. It could be used to monitor dams, highways and railways, for instance, or to help firefighters better target rescue efforts during a severe blaze. At that point, their quest will be complete.

The U.V. Award ceremony took place in Riva del Garda, in Italy, with Martin Rehak, second from the left, and Davide Cucci, third from the left. © TOPO / EPFL 2018

Author: Sandrine Perroud