The subject disclosure is directed towards a sensor fusion interface that enables interaction between one or more entities of a physical environment and a computerized device component. A plurality of sensor modules generate multiple sensor input data associated with one or more entities in an environment and store such data in a shared library in accordance with a uniform and common schema. The multiple sensor input data is refined until a certain level of accuracy is achieved. Using the sensor fusion interface, entity state data is extracted from the shared library and exposed to the computerized device component.