Redefining Millimeter-Wave Automotive Radar for Autonomous Driving: Advanced Signal Processing and Machine Learning Approaches

Millimeter wave automotive radars are highly reliable in all-weather environments and are indispensable for both advanced driver assistant systems (ADAS) and autonomous vehicles. The presentation will highlight recent advancements that continue to push the boundaries of what’s possible in automotive radar technology for fully autonomous driving. Key topics include the use of sparse arrays synthesized with multi-input multi-output (MIMO) radar technology to achieve enhanced angular resolution, innovative waveform designs for mitigating mutual interference between automotive radars, and high-resolution direction-of-arrival (DOA) estimation from a single snapshot. We will also address the challenges of applying MIMO radar theories in automotive contexts, including the use of model-based deep neural networks for high-resolution DOA estimation and the integration of domain-knowledge-guided deep learning in radar perception for autonomous vehicles. Finally, we will discuss future directions for automotive radar, with a focus on enhanced collaborative sensing through the use of multiple automotive radar systems. Speaker(s): Prof Shunqiao Sun Room: 466, Bldg: Packard Lab, 19 Memorial Drive West, Lehigh University, Bethlehem, Pennsylvania, United States, 18015