Optimal information collection for nonlinear systems- An application to multiple target tracking and localization

Active sensors need to be optimally deployed or configured so as to make the best measurements at any given time. Then of immediate concern is the problem of characterising or measuring sensor performance before the measurement has been taken. In this paper, the well known concept of Mutual Information is chosen as the utility function for optimizing the sensor parameters. The standard problem of multiple target tracking using a configurable sensor mounted onto a UAV, or a ground based robot vehicle, is formulated in an optimal control framework such that the trajectory traced by the UAV maximizes the Information gain. An example of airplane tracking using an UAV is illustrated.