Data association procedures for lateral velocity estimation

The observation of a carpsilas environment based on sensors is the premise for many modern automotive safety and comfort systems. Automotive radar sensors in the 24 GHz domain are used widely in this context as they can measure target range, azimuth angle and radial velocity simultaneously and with high accuracy even in multiple target situations. Lateral velocity, however, cannot be measured by these sensors in a single observation. This shortcoming is usually compensated by the application of tracking techniques to consecutive measurement data from which a velocity vector can be derived. For an automotive radar sensor almost all objects exhibit multiple reflection points due to its extensions in comparison to the sensorpsilas wavelength and range resolution. In this case the lateral velocity can be measured based on a single observation (Folster et al., 2007). The technical challenge of this approach is the correct mapping of all radar measurements to real objects, especially in multiple target situations.