Detection and tracking of extended targets for a 24 GHz automotive radar network

Sie sind hier: Homepage > Suche

Detection and tracking of extended targets for a 24 GHz automotive radar network

Fölster, Florian / Oprisan, Dan / Görner, Stefan / Rohling, Hermann

Future applications for comfort and safety in the automotive area will be based on radar sensors. The currently most established application for automotive radars is Adaptive Cruise Control (ACC), which considers target ranges up to 200m in a small azimuth angle area with a range resolution of 1 m. Additional applications like Stop and Go, Pre-Crash or Parking Aid will require a large azimuth angle area, ranges up to 30 m and range resolution of 20 cm. Therefore radar sensor networks consisting of four radar sensors behind the front bumper have been developed to fulfil these general system requirements. The target extension of a normal car is much larger than the range resolution of the radar sensors. Therefore targets can not be handled as point-like targets anymore. In this paper three different methods are presented to adapt the signal processing process to extended target situations. The first approach uses adapted target models to filter ghost targets and to describe the detected target for a single measurement cycle. The target position is defined by a specific center point for each model. The second approach benefits from a longer observation period and describes targets by their intersection probability density. The target position is derived by correlating the measured intersection density with reference target patterns. The third method uses a quality function, such as target probability, calculated over the observation area to estimate positions of extended targets and to segment them.