Micro-Doppler Extraction of Bicycle Pedaling Movements Using Automotive Radar

Abstract

The specific detection and classification of vulnerable road users in the closer vehicle environment is the foundation for safe autonomous driving of the future. In addition, the capability of predictive intention recognition is of fundamental interest in achieving the goal of ``Vision Zero''. High-resolution radars in the short-range enable the detection of cyclist-identifying micro-Doppler distributions. This paper presents a method for the extraction of micro-range-Doppler pedaling motions of cyclists in highly relevant movement scenarios. An adaptive ellipse fitting algorithm separates the high-resolution micro-Doppler components of the wheels from the pedaling motions and enables their robust extraction for the potential use of anticipatory safety functions. The proposed procedure is applied to real radar measurements of a cyclist. The results show that the time-dependent motion behavior of the pedals and contributing parts of the legs can be obtained from the radar data.