UAS behavior modeling based on high level abstraction

This paper presents a telemetry compression algorithm for Unmanned Aerial Vehicles (UAV) based on high level abstraction. A typical communication channel usage is not optimized when it is used for telemetry because it requires persistent communications to the ground station. This method spends large bandwidth in the communication channel having an inefficient cost in terms of power consumption. We have noticed that unmanned aerial vehicles behave following patterns easy to describe by a characterizer abstracting the trajectory of the maneuvers. This behavior is also valid for aircrafts commanded by the autopilot or when the aircraft is flying a 4D navigation. The aim of this algorithm is not only to have an efficient channel usage but to reduce the delay due to the distance and to increase the accuracy without increasing the sampling rate. It is possible to rebuild the trajectory from a small amount of information if we have a good model of our aircraft and this might be very helpful for collision avoidance. Let us recall that inferring intent is a big issue in UAS detect sense and avoid. The solution presented in this paper uses a so called PID characterizer. The PID characterizer is a method which solves the six degrees of freedom motion equations for a small time interval while flying. These equations are modeled in a transfer function designed by a set of parameters that we called ephemeris. The channel usage will be low because only the ephemeris will be sent to the ground station. This algorithm is useful for monitoring the vehicle position and attitude but nevertheless, it is not so useful for remote control due to the stochastic nature of the control loop. Sudden trajectory changes need to be recalculated in flight when a significant disagreement with the predicted trajectory is detected. Transition motion and stochastic sequences must be covered by traditional telemetry.