基于环境感知的汽车毫米波雷达抗干扰研究

The interference reduction method of automotive millimeter-wave radar in complex environment is the key to ensure the all-weather accurate road target sensing of unmanned vehicles and active safety technology. With the increase of automotive radars in the road, the interference generated has random distribution and complex process, and overlaps with target features in time domain, frequency domain and spatial domain, thus leading to misjudgment of radars. Existing interference reduction models and methods have poor performance in reducing interference between radars and multi-path transmission of transmitted signals, and it is difficult to extract weak targets such as pedestrians from strong interference and reduce false targets. .In view of the existing shortcomings, this project intends to carry out the research on automotive radar interference sensing mechanism and interference reduction method from a new perspective of cognitive radar. Firstly, the interference sensing model is established to reveal the rules of mutual interference and multi-path transmission interference. Secondly, the inference reduction method based on adaptive waveform updating is studied to reduce the multi-path transmission interference of multi-radar transmitting signals with complex spatial and temporal distribution. The adaptive Doppler beamforming interference separation method is studied to identify weak targets that overlap with the interference in frequency domain and spatial domain. Finally, experimental studies were carried out to verify the effectiveness of the interference sensing model and the interference reduction method. .The study of this project is helpful to solve the problem of interference reduction of automotive radar, and has positive academic significance and social value to improve the future traffic safety of unmanned vehicle.

复杂环境下的汽车毫米波雷达抗干扰方法，是保障自动驾驶与主动安全技术全天候精确感知道路目标的关键。随着道路中汽车雷达增多，产生的干扰分布随机、过程复杂，且在时域、频域和空域中与目标特征重叠，进而导致雷达误判。现有的方法对雷达间强干扰、发射信号多径传输干扰抑制性能不足，难以从强干扰中提取行人等微弱目标并抑制虚假目标。.针对现有不足，本项目拟从认知雷达的全新视角开展汽车雷达干扰感知机理与抗干扰方法研究。首先建立干扰感知模型揭示雷达间相互干扰与发射信号多径传输干扰的规律；其次研究基于收发自适应的波形动态更新抗干扰方法，以抑制时空分布复杂的多雷达发射信号多径传输干扰；研究自适应多普勒频域波束形成干扰分离方法，以识别与干扰在频域、空域重叠的微弱目标；最后开展实验研究验证干扰感知模型与抗干扰方法的有效性。.本项目的研究有助于攻关汽车雷达抗干扰难题，对提高未来自动驾驶交通安全具有积极的学术意义和社会价值。

复杂环境下的汽车毫米波雷达抗干扰方法，是保障自动驾驶与主动安全技术全天候精确感知道路目标的关键。随着道路中汽车雷达增多，产生的干扰分布随机、过程复杂，且在时域、频域和空域中与目标特征重叠，进而导致雷达误判。现有的方法对雷达间强干扰、发射信号多径传输干扰抑制性能不足，难以从强干扰中提取行人等微弱目标并抑制虚假目标。因此本项目主要解决以下问题：1)研究干扰感知模型，从时域、频域与空域分析干扰信号的特征与规律，实现干扰信号采集与特征提取，提出基于拉丁超立方方法的Kriging替代模型解决干扰的表征问题，通过隐式关系分析干扰与目标特征的关系；2)研究感知抑制方法，提出基于自适应多普勒频域波束形成的干扰目标分离方法，基于收发自适应波束形成实现干扰抑制，基于扩展卡尔曼滤波实现干扰情况下目标追踪降噪，基于蒙特卡洛随机模型研究振动产生的干扰问题，基于归一化最小均方实现自适应滤波算法的人体特征检测；3)实验验证，通过基于TI公司AWR1642毫米波开发平台与Ettus公司的软件定义无线电USRP平台验证了提出的干扰感知模型与干扰抑制方法的有效性。在本项目的支持下，发表SCI论文6篇，中文期刊论文1篇，申请发明专利3项。