How Can Flights Out Of Sight And Indoors Become Even Safer Without A GNSS Signal?

An important factor for safer autonomous travel is minimising the risks that can lead to accidents. One major risk is the loss of satellite-based positioning signals (GNSS signal). Especially indoors and when flying out of sight and in rough terrain, a loss of connection can be fatal. In the best case, a homecoming function is activated, but an uncontrolled crash is also a likely scenario. So the goal is to ensure that the flight can continue smoothly even in this case. Something we work on every day at Spleenlab. An interplay of machine learning, AI and sensor data is essential here. 

The goal is to calculate one’s own position in real time and to build a 3D model of the environment.  This can happen from the interaction of various sensor data or, thanks to so-called visual odometry, be generated directly from camera images. This approach is revolutionary as far as the use of sensor technology is concerned, since all automated air and ground vehicles have corresponding cameras. The use of a lidar laser can be dispensed with. 

Crucially, Visual Odometry in real time and embedded in the respective system (e.g. via an Nvidia Jetson or Xavier AGX board or integrated in the Auterion Skynode and AI Node) enables these functions on the vehicle without the need for a controller or ground station to intervene.  In its compactness and complexity, this concept is easily adaptable to a wide range of domains (vehicles of all kinds, UAV, or for fast and high-flying manned aviation, e.g. helicopter or air taxi) and operational scenarios. Regardless of whether indoor or outdoor.