Risk-based sensor management for integrated detection and estimation

In this paper, we develop a risk-based sensor management scheme for unknown object detection and process estimation under limited sensory resources. Bayesian sequential detection and estimation methods are utilized for risk analysis. The objective is to find every object of interest in the mission domain and satisfactorily estimate the associated process dynamics with minimum risks. Two types of costs are taken into account for risk evaluation, i.e., the cost of making an erroneous decision regarding object existence or its estimates, and the cost of taking more observations for a possibly better decision. The Renyi information divergence is investigated to measure the information loss in making a suboptimal sensor allocation decision, which is used to formulate the observation cost. A set of simulation results are provided to confirm the effectiveness of the proposed sensor management scheme.