Modeling and Simulation of Radar Sensor Artifacts for Virtual Testing of Autonomous Driving

Abstract

Automotive radar measurements are prone to noisy sensor readings due to artifacts, such as false detections, clutter, and aliasing. Simulation models of sensors that are deployed in virtual testing pipelines must demonstrate a high degree of fidelity. Capturing artifacts correctly can be seen as a criteria for the trustworthiness of the model. In this work, we outline a comprehensive overview of different types of artifacts in radar along with their characteristics. Next, we allocate their formation process in the radar signal processing chain from which we draw implications about their correct modeling. We also present mathematical models that describe the artifacts along with their implementation in a radar simulation model.