Generalized Grid Framework for multi sensor data fusion

This paper presents a methodology for multi sensor data fusion that uses the accumulation grid idea for the representation of data. A generalized grid framework is introduced to represent measurement data and fusion results in a common way. This allows the definition of standardized prediction and fusion operations and includes the variation between Cartesian and polar grids as well as the extension to 3D grids. It is shown that most of the typical object detection and tracking schemas can be represented by the framework. Two application examples including processing results from the field of automotive environment recognition are given to demonstrate special cases of the framework.