In order to increase traffic safety, automotive radar based Vulnerable Road User (VRU) detection and classification approaches are drawing more and more attention in recent years. There has been a growing interest in the potential use of micro-Doppler features in discrimination of VRU targets, e.g. pedestrian s and cyclists.

In the paper, micro-Doppler of pedestrians and cyclists is examined by using actual data from Delphi 76GHz radar sensors as well as by computer simulations. Some practical issues in actual automotive radars, such as transmission gaps, thresholding, as well as the effect of target speed on micro-Doppler generation are discussed in the paper.

In addition, in order to apply micro-Doppler into VRU classification, the micro-Doppler pattern could be decomposed, and parameters could be estimated afterwards. In the paper, micro-Doppler decomposition and parameter estimation is discussed based on a SVD (Singular Value Decomposition) method. One of the SVD matrixes contains time dependent information, where the period/cycle of micro-Doppler pattern can be estimated. Another matrix contains information e.g. micro-Doppler spreads, symmetry, where relationship between different motion parts of the targets can be discovered.