Signal processing of 4D high-resolution automotive radar for object detection

Cui, Hang

Permalink

https://hdl.handle.net/2142/117545 Copy

Description

Title

Signal processing of 4D high-resolution automotive radar for object detection

Author(s)

Cui, Hang

Issue Date

2022-11-04

Director of Research (if dissertation) or Advisor (if thesis)

Norris, William R

Doctoral Committee Chair(s)

Mehta, Prashant

Committee Member(s)

• Chowdhary, Girish
• Krishnan, Girish

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Signal Processing
• 4d High-resolution Automotive Radar
• Millimeter-wave Radar
• Object Detection

Language

eng

Abstract

Automotive radars were first deployed several decades ago and have been widely used in advanced driver assistance systems (ADAS). This was primarily due to their advantages from immunity to adverse weather conditions, such as fog, rain, snow, dust, and the capability to estimate a target’s velocity with Doppler measurements. In the past five years, the 4D high-resolution radar, or automotive imaging radar, has been under rapid development in R&D due to benefits from improvements in semiconductors. This thesis aims to explore the signal processing algorithms and pipelines on low-level radar ADC data, a.k.a. radar cube data, from both conventional and the latest 4D high-resolution automotive radar for object detection. The first contribution of this thesis is to explore the signal processing algorithms and pipelines of automotive radar for object detection. The single input multiple output (SIMO) antenna design of conventional automotive radar is introduced. Estimation algorithms on targets’ ranges, Doppler measurements, and angle of arrivals (AoA) are presented and applied on collected low-level radar ADC data together with practical techniques, such as constant false alarm rate (CFAR), static clutter removal and peak grouping. The signal processing of the micro-Doppler signature is also introduced not only for object detection, but potentially for object activity classification as well. The second contribution is to discuss and analyze the super-resolution algorithms, such as the Capon beamformer, a Multiple Signal Classifier (MUSIC) beamformer for AoA estimation of targets on collected low-level data of the latest 4D high-resolution automotive radar. The multiple input multiple output (MIMO) antenna design of 4D high-resolution automotive radar is introduced. Signal processing of radar point cloud, radar range-azimuth (RA) heatmap and radar range-Doppler (RD) heatmap are also explored on the 4D high-resolution radar. Experiments for conventional and 4D high-resolution automotive radar with different radar configuration parameters are specifically designed for both static and dynamic scenes.

Graduation Semester

2022-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/117545

Copyright and License Information

Copyright 2022 Hang Cui

Owning Collections