An entropic framework for sensor modelling

The authors examine the general task of active sensing by defining an entropic measure of sensor efficiency for optimizing the rate of flow of information from a sensor. In particular the authors focus on data acquired using a 675 kHz pencil-beam sonar mounted on the Jason ROV while working in a challenging shallow water environment. In the authors´ case the entropic measure is defined based on a probabilistic three dimensional occupancy grid model of the underlying physics of the acoustic sensing modality. They demonstrate the utility of their entropic measure for examining the rate of information flowing from the sensor. This allows them to look at the rate of convergence and the stability of their mapping efforts in the face of typical uncertainties such as those associated with navigation and attitude. Their formulation is seen to be in very good agreement with independently obtained ground-truth data