DEFDRONE-5: AI enabled Level 5 autonomous drone navigation and computing for security and defense

Alerion has developed its own technology for autonomous drone flight (patented), currently used for infrastructure inspections, e.g. wind turbines. Currently, drones developed by Alerion provide an autonomy level 4.

Alerion is looking forward to increasing resilience, robustness and reliability of their autonomous navigation system, by means of adding on the edge technologies. In this sense, the current needs and work proposed by Alerion is directed in three main lines :

  • Semantic scene understanding based on object detection and recognition approaches

  • Deep Learning based navigation under unknown conditions 

  • Autonomous navigation system developments to increase autonomy level to 5 

Surveillance applications can take high advantage of UAV technology combined with object detection and recognition techniques to be able to identify potential threats and risks, e.g. armed persons or unknown elements in populated areas or borders. With this technology, autonomous drones could be able to perform autonomous geofenced-based surveillance tasks. In this sense, object detection and recognition will allow us to adapt and deploy specific flight strategies. Potential risks or dangers that could affect the safety of the flight could be also detected and tracked in order to secure the return to a  safe point.  In this area, we envisage the usage of traditional techniques of pattern recognition in combination with the latest deep learning approaches for real time recognition, crucial for applications like autonomous industrial inspection, border surveillance, etc.

Deep learning techniques combined with artificial intelligence offer a big potential in path planning in unknown environments, or changing climatic conditions (e.g. flights during night, storm or windy conditions). Autonomous navigation can take advantage of deep learning capabilities to increase the security of the flight based on the intervention of the deep learning module to correct or provide alternative trajectories, e.g. collision avoidance.

The development of these approaches will lead towards the increment of the reliability of the whole autonomous navigation system, in the way to have a future solution in the next security certification level. So, in this sense those systems will provide an extra for our current 4 Level, increasing the robustness of the solution making it more independent from the pilot.

In order to achieve totally autonomous UAVs the safety-critical requirements increase exponentially, this means the necessity to explore drone safety certification within different safety standards as DO-178, DO-254 and so on. Nowadays, legislation for full autonomous UAV’s is missing and hence EASA certification over aircraft safety standards is mandatory. But as industry is going forward in this direction, Alerion's main interest in new developments is to fulfill industrial safety-critical requirements to increase reliability and resilience of the fully autonomous UAVs, giving us an advantageous position for this future standard.


Configuración de cookies

  • Necesarias Cookies impresincibles para el correcto funcionamiento de la web.
  • Publicitarias Cookies de terceros que guardan datos y preferencias del usuario.
Utilizamos cookies propias y de terceros para obtener datos estadísticos de la navegación de nuestros usuarios y mejorar nuestros servicios. Puedes aceptar la pólitica de cookies y aceptarlas para que se descarguen las cookies o puedes elegir las cookies que deseas descargar en las opciones.
Más información