This project builds a unified neural network model that simultaneously performs positioning and Doppler estimation using imaging radar data. Through a self-supervised machine learning methodology, the project aims to significantly enhance perception performance compared to traditional methods.