Sensor management in multi-target tracking is commonly focused on actively scheduling and managing sensor resources to maximize the visibility of states of a set of maneuvering targets in a surveillance area. This project focuses on two types of sensor management techniques: - controlling a set of mobile sensors (sensor control), and - scheduling the resources of a sensor network (sensor selection). In both cases, agile sensors are employed to track an unknown number of targets. We advocate a Random Finite Set (RFS)-based approach for formulation of a sensor control/selection technique for multi-target tracking problem. Sensor control/scheduling offers a multi-target state estimate that is expected to be substantially more accurate than the classical tracking methods without sensor management. Searching for optimal sensor state or command in the relevant space is carried out by a decision-making mechanism based on maximizing the utility of receiving measurements. In current solutions of sensor management problem, the information of the clutter rate and uncertainty in sensor Field of View (FoV) are assumed to be known in priori. However, accurate measures of these parameters are usually not available in practical situations. This project presents a new sensor management solution that is designed to work within a RFS-based multi-target tracking framework. Our solution does not require any prior knowledge of the clutter distribution nor the probability of detection profile to achieve similar accuracy. Also, we present a new sensor management method for multi-object filtering via maximizing the state estimation confidence. Confidence of an estimation is quantified by measuring the dispersion of the multi-object posterior about its statistical mean using Optimal Sub-Pattern Assignment (OSPA). The proposed method is generic and the presented algorithm can be used with any statistical filter.
展开▼
