Randomized Path Optimization for the Mitigated Counter-Detection of UAVs

UAVs provide exceptional capabilities and a myriad of potential mission sets, but the ability to disguise where the aircraft takes off and lands would expansively advance the abilities of UAVs. This paper describes the development of a nonlinear estimation algorithm to predict the terminal location of an aircraft and a trajectory optimization strategy to mitigate the algorithm's success. A recursive Bayesian filtering scheme is used to assimilate noisy measurements of the UAVs position to predict its terminal location. We use a blackbody radiation-based likelihood function tuned to the UAVs known endurance limitations to assimilate the position measurements. A quadratic trajectory generation method with waypoint and time variation is used to produce a parameterized family of potential aircraft trajectories. The estimation algorithm is then used to assess parameterized UAV trajectories that minimize certainty of the true terminal location. The KL divergence is used to compare the probability density of aircraft termination to a normal distribution around the true terminal location. Results show that the greatest obfuscation of path directly correlates to variations in time of flight with respect to the vehicle's maximum possible flight time.