Receding horizon stochastic control algorithms for sensor management

The increasing use of smart sensors that can dynamically adapt their observations has created a need for algorithms to control the information acquisition process. While such problems can usually be formulated as stochastic control problems, the resulting optimization problems are complex and difficult to solve in real-time applications. In this paper, we consider sensor management problems for sensors that are trying to find and classify objects. We propose alternative approaches for sensor management based on receding horizon control using a stochastic control approximation to the sensor management problem. This approximation can be solved using combinations of linear programming and stochastic control techniques for partially observed Markov decision problems in a hierarchical manner. We explore the performance of our proposed receding horizon algorithms in simulations using heterogeneous sensors, and show that their performance is close to that of a theoretical lower bound. Our results also suggest that a modest horizon is sufficient to achieve near-optimal performance.