Integration of advanced object properties

In several applications, it is necessary to learn more from the sensor data received than the time-varying geolocation of moving objects of interest. Rather, we wish to understand what the objects we observe are, i.e. we aim to learn as much as possible about their attributes in order to be able to classify or even identify them. Many relevant object attributes can be derived even from their purely kinematic properties such as speed, heading vector, and normal acceleration as well as from mutual interrelations inferable from multiple object tracks, as has been extensively discussed in the introductory chapter, Sect. 1.3.5.