LETI: The 3D Modeling Method Developed at LETI Will Help Reconstruct Buildings Using Drones

The approach optimizes the simultaneous work of several quadcopters to generate 3D models of buildings under construction or renovation and can be used to create VR worlds and computer games.

Capturing images of objects with drones to create detailed 3D models today is actively used in various fields: design, diagnostics of infrastructure objects (buildings, pipes), reconstruction of buildings, neighborhoods, cities, creation of virtual reality environments, and some others. However, when the objects for modeling are large-scale, such as multi-story buildings, the use of a single drone requires a large amount of time, taking into account the need for recharging.

“For such situations, we have developed a method of generating 3D models of large buildings using a group of drones, which conduct a distributed mapping of the object for further digital reconstruction. Our algorithms allow the drones to work efficiently – not to do double work and not to miss any part of the object.”

Anton Filatov, Assistant of the Department of Software Engineering and Computer Applications at LETI
The scientists’ method allows them to calculate the space each drone should handle, taking into account the size of the quadcopter group, its battery charge, and the size of the building. The researchers also estimated how often photos should be taken to improve modeling accuracy and what trajectories drones should follow to work efficiently and avoid collisions. The results of the study were published in the Mathematics journal.

“At the same time, we can automate the process if we apply our method to drones, which can move independently along a pre-set and built-on-the-fly trajectory. First of all, the development will be useful when reconstructing large and tall buildings or constructing new ones, when it is possible to exactly determine what stage the work is at only from the air,” adds Anton Filatov.

Researchers are now working to make the developed method capable of implementing more complex tasks using a constellation of drones. For example, to automatically redistribute workload between quadcopters if one of them goes to change the battery.