Particle filter based information-theoretic active sensing

This work addresses the task of active sensing, or information-seeking control of mobile sensor platforms. Formulation of a control objective in terms of information gain allows mobile sensors to be both autonomous and easily reconfigurable to include a variety of sensor and target models. Tracking a moving target using a camera mounted on a fixed-wing unmanned aircraft is considered, but the control formulation is not specific to this choice of sensor or estimation task. A control formulation is developed which minimizes the entropy of an estimate distribution over a receding horizon subject to stochastic non-linear models for both the target motion and sensors. Previous similar work has been restricted to either a stationary target, a horizon of length one, or Gaussian estimates.The prediction of conditional entropy is shown to be inherently complex, and a computationally efficient sequential Monte Carlo method is developed. The entropy prediction depends on this Monte Carlo method as well as a novel approach for entropy calculation in the context of particle filtering. These methods are verified through simulation and post-processing of experimental flight data.