Cuvillier Verlag

Publications, Dissertations, Habilitations & Brochures.
International Specialist Publishing House for Science and Economy

Cuvillier Verlag GmbH

De En Es
Postprocessing Architecture for an Automotive Radar Network

Hard Copy
EUR 28.00 EUR 26.60

E-book
EUR 0.00

Download
PDF (3 MB)

Postprocessing Architecture for an Automotive Radar Network (English shop)

Mark Schiementz (Author)

Preview

Table of Contents, Datei (34 KB)
Extract, Datei (200 KB)

A wide range of radar-based automotive applications is currently being explored in research or is already in the pre-production or series-production phase within the automotive industry. For several years, radar-based Adaptive Cruise Control (ACC) systems have been available on the market; initially reserved for the luxury segment, they have since become well-established there. Efforts are now underway to develop new radar-based applications designed to provide drivers with additional support, whether for comfort or safety. Examples include driver assistance systems such as parking maneuver assistants, lane-change assistants, and ACC/Stop & Go systems. Existing ACC systems on the market primarily utilize long-range sensors operating in the 77 GHz millimeter-wave band—a technology that, in its current form, does not meet the requirements of these new applications. Consequently, various 24 GHz short-range radar sensors have been developed in recent years. The intention is to integrate these with existing long-range sensors to form a radar network capable of satisfying the requirements of these new applications.
In road traffic, radar sensor systems must contend with extended targets in complex, multi-target environments containing a large number of both moving and stationary objects. Modern sensors are becoming increasingly adept at detecting these objects. To enable an appropriate response—either by the driver or by a driver assistance system—the measurement data from the various sensors must be processed using specialized signal-processing techniques. This ensures that the transmitted data contains a high proportion of relevant, error-free information. This stage of signal processing is known as “post-processing.” This work presents a post-processing architecture for a multi-sensor radar network. It extracts the information required by the application from the raw radar measurement data—filtering out false detections, stabilizing relevant information, and generating additional data by incorporating the host vehicle’s own state information. The study begins by outlining the operational constraints of radar sensing in road traffic environments and evaluating various 24 GHz radar network concepts. Subsequently, several estimation algorithms for determining the position and velocity of detected targets, as well as various tracking filters for monitoring them over time, are derived and assessed. Based on the insights gained from analyzing these estimation and tracking methods, a multi-target post-processing architecture is developed for a radar network consisting of two 24 GHz monopulse radar sensors. The data association process—essential for a multi-target system—is explained with reference to the detection characteristics of the sensors employed. In particular, the classification of a large number of detected targets into moving and stationary objects, utilizing the host vehicle’s own data, serves as a critical foundation for this association process. Furthermore, the architecture accounts for the fact that observed objects must be treated as extended, meaning they can generate multiple detections within a single measurement cycle. In addition to the analytical investigation of the various post-processing algorithms, an empirical evaluation of the proposed architecture—conducted using a test vehicle—forms a core component of this work.

ISBN-13 (Printausgabe) 3865375170
ISBN-13 (Hard Copy) 9783865375179
ISBN-13 (eBook) 9783736915176
Final Book Format A5
Language German
Page Number 192
Edition 0
Volume 0
Publication Place Göttingen
Place of Dissertation Göttingen
Publication Date 2005-07-13
General Categorization Dissertation
Departments Electrical engineering
Keywords Automotive Radar Networks, Radar Signal Postprocessing, 24 GHz Monopulse Radar Sensors, Multi-Sensor Data Fusion, Multi-Target Tracking, Kinematic State Estimation, Kalman Filtering, Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), Radar-Based Data Association, Radar Detection and Object Classification, Sensor-Level Clustering, Maneuver-Adaptive Target Tracking, Radar-Based Advanced Driver Assistance System