Decision-making of UAV for tracking moving target via information geometry

In this paper, we consider the unmanned aerial vehicle (UAV) action decision-making problem for tracking a ground moving target. The studied problem was formulated as a Partially Observable Markov Decision Process (POMDP), and following a Kalman filter is derived to estimate the Markov chain system state. Exploiting an innovations approach, the total information gain in the UAV-to-target observation is used as an optimization criterion in a partially observable Markov decision process formulation to optimize the control strategy. In our method, we make UAV command decision to maximize the Fisher information in information geometry. The computer simulations validate that, comparing the simulation results with conventional methods, our proposed method has less time cost and smaller tracking error.