B-VLOS operations require certified “detect and avoid” technology, so that the drone independently can avoid obstructions and other air traffic.
In pure Sense and avoid approaches, the dominant sensor is vision. However good sense and avoid behavior is difficult to realize with vision, since most targets of interest are approaching radially and are difficult to detect in wide field-of-view imaging sensor output.
The most promising sensory systems for long-range sense and avoid in uncontrolled environments would appear to be radar. In our view, it is an open question whether the best approach is a local one in which the drone makes no assumptions about potential obstacles and merely avoids them in a reactive manner based on sensory input, or a model-based one in which the drone keeps track of cooperative agents in its airspace.
WP5 will develop sensor HW and SW components for sensor data processing and avoidance behavior. Together they constitute a sense and avoid framework that can be applied to safety-cases based on small multi-rotor and fixed-wing drone platforms. The development process contains some non-trivial tasks such as finding the optimal sensor types and determining the optimal behavioral approach.