Master Thesis

Neural Aerial Social Force Model: Teaching a Drone to Accompany a Person from Demonstrations

Work default illustration


Recent advances in computing power, sensor technology and battery duration, have motivated the development of commercial Unmanned Aerial Vehicles (UAV), most commonly found in the form of quadrotors. Quadrotors have never been so cheap and easier to manufacture, which makes available to the wide public. Although commercial drones are in general not very sophisticated, they are incredibly versatile, and there is a huge potential of applications in which they could be deployed in the future.

However, there are some technical aspects that prevent this from happening right now. While the hardware technologies found in UAVs are advanced, software technologies have still quite a long distance to reach the hardware's level. One of the main issues, which is still not successfully solved, is the task of navigation. Robot navigation is one of the greatest challenges in Robotics, and particular case of aerial navigation is even harder. On one side, Aerial robots must deal with an extra dimension of movement, and on the other side, there is little research about this topic because aerial robots haven't been considered traditionally

The work is under the scope of the following projects:

  • ColRobTransp: Colaboración robots-humanos para el transporte de productos en zonas urbanas (web)