Sensor Management for Static Target Detection with Non-Binary Sensor Observations and Observation Uncertainty
Mark P. Kolba, Leslie M. Collins · 2007 IEEE/SP 14th Workshop on Statistical Signal Processing · 2007
Previously, a grid-based sensor management framework has been developed that is useful for directing the operation of a suite of sensors seeking to detect static targets. Earlier versions of the framework only allow the sensors to make binary observations-either "target present" or "no target present"-within the cells of the grid. This paper introduces the use of non-binary observations within the sensor management framework. Simulation results are presented which show that the presented sensor manager continues to outperform a direct search technique with the use of non-binary observations. The effects of uncertain sensor observations are also examined in this paper, and uncertainty modeling is introduced in order to allow the sensor manager to model uncertain non-binary sensor observations. Simulation results show that proper uncertainty modeling is crucial for maintaining robust sensor manager performance.