Phenomenological Study of Radar Scattering at High Millimeter-Wave Frequencies in Support of Autonomous Vehicles

There has been significant investment in realizing autonomous vehicles technology. Despite tremendous efforts and investments, the initial optimism in achieving total autonomy has not yet been materialized. This is not totally surprising as the required false alarm rate and probability of detection are very stringiest for a complex navigation problem which includes fast-changing scenarios of many dynamic obstacles of different kinds. This situation is often exacerbated by inclement weather and road conditions. In an effort to improve sensors’ capabilities, this thesis focuses on examining the applications and advantages of polarimetric J-band (220 – 325 GHz) radars for driverless cars.

Automotive radars operating at lower J-band frequencies are envisioned to replace the current 77-GHz systems in vehicles for their superior resolution, compact size, and wider bandwidth. This dissertation aims at studying the polarimetric radar response of typical traffic scenes at this frequency. For highly automated vehicles, identifying each target on the road and its vicinity is crucial in planning the navigation and ensuring safety for all. Characterization of the radar backscatter from various objects enables their identification in traffic scenes and is necessary for optimizing the design of automotive sensors. The polarization signature of different targets can help with sematic mapping of targets on the road, especially for those targets that may be in the same range and Doppler bin.