کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

Automotive radar target classification using micro-Doppler features

نویسندگان :

Amin Aghatabarroodbary¹ Mohammad Hassan Bastani² Fereidoon ‌‌Behnia³

1- دانشگاه صنعتی شریف 2- دانشگاه صنعتی شریف 3- دانشگاه صنعتی شریف

کلمات کلیدی :

Radar target classification،automotive radars،feature extraction،micro-Doppler signature،advanced driver assistance systems

چکیده :

The rapid advancements in automotive radar tech- nology have enabled robust object detection and classification in challenging environments, contributing to the development of autonomous vehicles and advanced driver-assistance systems (ADAS). Beyond basic detection, classifying radar targets, such as distinguishing between pedestrians, cyclists and cars, is critical for improving situational awareness and safety. Micro-Doppler signatures, which capture subtle motion characteristics of targets caused by small-scale movements like limb motion or rotating wheels, provide a rich source of information for classification tasks. This paper explores using micro-Doppler features for radar target classification, leveraging signal processing techniques and machine learning models. In this paper, 5 features are extracted from micro-Doppler signatures to classify a 4-class classification problem. Moreover, an ensemble learner with 100 weak learners is used to evaluate classification accuracy. Simulation results with different signal to noise ratios (SNRs) demonstrate that the proposed features extracted from micro-Doppler signatures of four classes of targets significantly enhance the accuracy and reliability of automotive radar target classification and can discriminate the targets with about 98 percent of accuracy for SNR of 30dB.