CRII: CIF: A Sparse Framework Based Automotive Radar Sensing for Autonomous Vehicles

CRII：CIF：基于稀疏框架的自动驾驶汽车雷达传感

• 批准号: 2153386
• 负责人: Shunqiao Sun
• 金额: $ 17.5万
• 依托单位: University of Alabama Tuscaloosa
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-05-01 至 2025-04-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2153386&HistoricalAwards=false
• 关键词: CRII; CIF; Sparse; Framework; Based

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2). Millimeter-wave automotive radar has emerged as a key technology in autonomous driving in order to provide environmental perception under all weather conditions. However, successful deployment is facing several challenges. First, automotive radars are required to have high angular resolution in both azimuth and elevation directions in order to produce point clouds representing the shapes of objects and enable target identification. Enlarging antenna array apertures by simply increasing the number of antenna array elements involves both a huge cost and a large form factor, and is not feasible in automotive radar applications. Furthermore, as more vehicles are equipped with radar, the probability of mutual radar interference increases. This project aims to explore a novel joint sparse-frequency and sparse-array signal-processing framework that enables high-resolution environment perception for autonomous vehicles with low-cost, small form factor and low probability of mutual interference. The project will result in algorithms that are applicable to various radar-sensing applications, including the remote sensing of vital signs of patients in telemedicine, a crucial need during the COVID-19 pandemic. The proposed educational plan creates opportunities to guide senior Capstone designs, enriches curriculum in radar-signal-processing courses, and facilitates outreach for minority students through an existing multicultural engineering program. It is challenging to achieve high angular resolution by adopting sparse arrays synthesized via multiple-input and multiple-output radar techniques because the high side lobe associated with sparse arrays would result in angle ambiguity. In addition, conventional radar chirps occupying a large bandwidth with a constant pulse-repetition frequency greatly increase the chance of mutual interference. The technical aims of the project are organized into two tasks. The first task investigates a matrix completion-based array interpolation approach to fill the holes of both one- and two-dimensional sparse arrays. The relationship between the recoverability of low-rank radar data matrices and the irregular sparse-array geometry will be investigated. In order to effectively complete irregular sparse arrays, efficient iterative hard thresholding matrix-completion algorithms will exploit the structures and properties of the underlying low-rank Hankel and block Hankel matrices. The second task investigates a cognitive approach to sparsely allocate the radar chirps in both frequency and temporal domains in order to synthesize a high-resolution range profile while significantly reducing the probability of mutual interference. This task will design novel optimization methods to dynamically allocate the transmit chirps under both interference and range-Doppler peak side lobe constraints by relaxing integer variables for efficient computations.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该奖项是根据2021年《美国救援计划法》（公法117-2）全部或部分资助的。毫米波汽车雷达已成为自动驾驶的关键技术，以便在所有天气条件下提供环境感知。但是，成功的部署面临着几个挑战。首先，需要汽车雷达在方位角和高程方向上具有高角度分辨率，以产生代表对象形状的点云并实现目标识别。通过简单地增加天线阵列元件的数量来扩大天线阵列孔径既涉及巨大的成本又涉及大型外形，并且在汽车雷达应用中不可行。此外，随着越来越多的车辆配备雷达，相互雷达干扰的可能性增加。该项目旨在探索一种新型的关节稀疏频率和稀疏的信号处理框架，该框架能够对具有低成本，小型且相互干扰的概率的自动驾驶汽车进行高分辨率的环境感知。该项目将导致适用于各种雷达感应应用的算法，包括远程医疗患者生命体征的遥感，这是Covid-19-19大流行期间的至关重要的需求。拟议的教育计划为指导高级顶峰设计，丰富了雷达信号处理课程的课程提供了机会，并通过现有的多元文化工程计划为少数民族学生提供促进的范围。通过采用通过多输入和多输出雷达技术合成的稀疏阵列来实现高角度分辨率是一项挑战，因为与稀疏阵列相关的高侧叶会导致角度歧义。另外，占据恒定脉冲重复频率的大带宽占据较大的带宽大大增加了相互干扰的机会。该项目的技术目标被组织为两项任务。第一个任务研究了基于矩阵完成的阵列插值方法，以填充一维稀疏阵列的孔。将研究低排名雷达数据矩阵的可恢复性与不规则稀疏阵列几何形状之间的关系。为了有效完成不规则的稀疏阵列，有效的迭代硬阈值矩阵完成算法将利用基础低级别Hankel的结构和特性，并阻止Hankel矩阵。第二个任务研究了一种认知方法，以稀疏的频率和时间域分配雷达呼吸，以便综合高分辨率范围，同时显着降低了相互干扰的可能性。该任务将设计新颖的优化方法，以通过放松整数变量以进行有效计算，在干扰和范围多普勒峰侧叶限制下动态分配发出呼吸。该奖项反映了NSF的法定任务，并被认为是通过基金会的知识分子和更广泛的影响来评估的支持，并被认为是值得的。

项目成果

A Deep Reinforcement Learning Approach for Integrated Automotive Radar Sensing and Communication

• DOI: 10.1109/sam53842.2022.9827815
• 发表时间: 2022-06
• 期刊: 2022 IEEE 12th Sensor Array and Multichannel Signal Processing Workshop (SAM)
• 作者: Lifan Xu;Ru-dan Zheng;Shunqiao Sun

Fast Forward-Backward Hankel Matrix Completion for Automotive Radar DOA Estimation Using Sparse Linear Arrays

• DOI: 10.1109/radarconf2351548.2023.10149466
• 发表时间: 2023-05
• 期刊: 2023 IEEE Radar Conference (RadarConf23)
• 作者: Shunqiao Sun;Yining Wen;Ryan Wu;D. Ren;Jun Li

Spectranet: A High Resolution Imaging Radar Deep Neural Network for Autonomous Vehicles

Spectranet：用于自动驾驶车辆的高分辨率成像雷达深度神经网络

• DOI: 10.1109/sam53842.2022.9827798
• 发表时间: 2022
• 期刊: IEEE 12th Sensor Array and Multichannel Signal Processing Workshop (SAM
• 作者: Zheng, Ruxin;Sun, Shunqiao;Scharff, David;Wu, Teresa
• 通讯作者: Wu, Teresa

Coprime Visible Regions Assisted Angle Unfolding for Sparse ESPRIT

• DOI: 10.1109/radarconf2351548.2023.10149628
• 发表时间: 2023-05
• 期刊: 2023 IEEE Radar Conference (RadarConf23)
• 作者: Lifan Xu;Shunqiao Sun