Normalization of Micro-Doppler Spectra for Cyclists Using High-Resolution Projection Technique

Radar-based functions for detection and classification of vulnerable road users (VRU) are primarily using Doppler and micro-Doppler information. The quality of this information depends significantly on the relative orientation of the target due to the radial measurement principle. Especially in case of intersecting VRU in the short range, this may cause a drastic signal deterioration that massively complicates reliable detection. This paper presents a signal processing method for normalizing the orientation-dependent dimensions of range and Doppler of a cyclist based on a high-resolution projection technique. The measured range ambiguities of micro-Doppler components are reduced significantly by creating an artificial range axis the spectral angle data points are mapped onto. Subsequent corrections of Doppler amplitudes result in spectra with normalized parameters. In addition, the proposed procedure enables improved detectability of features for potential behavioral indication of cyclists such as pedaling. The results show that standardization of angle dependency can be particularly crucial for the correct classification of crossing cyclists in the closer vehicle environment.