-
实验在室外进行,测试中产生三次回波的三个目标分别为激光出射方向上距离20 m处1 m×1 m的20%漫反射靶板,以及距离10 m和15 m处放置的网孔为2.5 mm的金属网。经雷达探测处理后其成像效果图如图6所示。
由图可见3次激光回波均在图像中清晰、稳定显现,无丢失、错位情形。
-
图7为激光雷达对办公楼内约70 m长直走廊成像效果测试图,从图中可以看出长直走廊的弯曲变笔直了,以雷达近处最为明显,图像整体更接近走廊的实际情况。
-
图8(a) 为雷达对办公楼内走廊的靶板成像,在走廊中竖立的平板反射率分别为60%和5%,校正前后成像如图8(b)、图8(c)所示,由图可见,校正后60%的靶板回波强度(橙色)强于5%的靶板回波强度(红色),且同一靶板获取的强度值趋于一致,更客观的反映目标的反射率信息。
-
试验以竖立在距雷达10 m、50 m和120 m处的靶板为目标,靶板的反射率为60%,重复测量200次,距离R由全站仪标定,取200次测量结果Di,后求平均测距精度
$\sum\nolimits_{i = 1}^{200} {\left| {Di - R} \right|} /200 $ ,结果如表1所示。由表1可知,在120 m范围内测距精度优于1.5 cm。
表 1 精度实测试验结果
Table 1. Measurement results of range accuracy experiments
Range Range accuracy 10.005 m 0.76 cm 50.015 m 1.24 cm 120.878 m 1.12 cm -
雷达上电以单回波方式工作,使用NVIDIA Jetson AGX Xavier板卡模拟车载计算机与雷达连接,板卡运行Linux操作系统,使用Wireshark软件截取雷达上传的UDP包并统计,如图9所示,每秒的数据包数为7496.8包,考虑统计细微误差,数据包数应为7500包,如前所述,1024个激光脉冲回波距离和强度分3个包上传,数据点率应为7500*1024/3=2 560 K点/s 。
Design and application of high-speed and high-precision information processing of multi-line lidar
-
摘要: 为满足多线、高空间分辨率激光雷达大数据量、高实时性、高可靠性要求,研究设计了一种基于Zynq-7000器件的激光雷达信息处理系统。硬件设计充分利用Zynq内部资源配置了多个、多种接口,外接简易电路即实现多个外设控制和状态监控。不同于常用的SMP (Symmetrical Multiprocessing)方式,Zynq内双ARM核处理器的软件设计采用更灵活的AMP (Asymmetric Multiprocessing)方式,两核一主一从独立运行裸机程序,其中主核完成数据处理的时序调度,并与从核以乒乓方式交替计算,双核协同优化改进,在完成了AD全波形匹配法与TDC相结合的高精度测距基础上,进一步设计并实现了实时的距离和强度校正算法,算法不仅研究了校正量与距离和强度的关系,还引入了温度和探测单元差异影响因素。经测试,64线激光雷达可探测处理三次回波,在120 m范围内的测距精度优于1.5 cm,强度数据更客观反映探测目标的特性,雷达单回波数据点率达到2 560 K点/s。Abstract: In order to meet the demand for large-volume data, high real-time and high-reliability of multi-line, high spatial-resolution lidar, an information processing system based on Zynq-7000 was designed. In terms of hardware design, internal resources of Zynq were fully utilized to configure multiple interfaces connecting external simple circuit merely which implement controlling and status monitoring of many peripheral devices. In terms of software design, more flexible AMP mode was adopted, not popular SMP, that each processor, a master and a slave, ran bare-metal program independently, in which a master processor controls time sequencing of data conduct of system and computes alternately with slave processor in ping-pang scheme. Cooperation between processors was optimized and improved. Distance and intensity correction algorithm was proposed and realized further on the basis of high-accuracy combination method of full waveform matching of AD and TDC upon that system, which not only investigate the relationship between correction quantity and distance and intensity, but also study influence of temperature and difference among detectors. In experiments 64-lines lidar has ability of detecting and processing three echoes, distance accuracy of which was better than 1.5 cm within 120 m. Intensity data of lidar expressed target characteristics more objectively and data rate of single echo reached 2 560 K points per second.
-
Key words:
- high-speed and high-precision /
- multi-line /
- lidar /
- AMP /
- distance and intensity correction algorithm
-
表 1 精度实测试验结果
Table 1. Measurement results of range accuracy experiments
Range Range accuracy 10.005 m 0.76 cm 50.015 m 1.24 cm 120.878 m 1.12 cm -
[1] Liu Ruqing, Jiang Yan, Li Feng, et al. Design of multi-channel data acquisition system for real-time perception lidar [J]. Infrared and Laser Engineering, 2021, 50(5): 20200291. (in Chinese) doi: 10.3788/IRLA20200291 [2] Li Xiaolu, Zeng Jingjing, Wang Hao, et al. Design and real-time imaging technology of three-dimensional scanning LiDAR [J]. Infrared and Laser Engineering, 2019, 48(5): 0503004. (in Chinese) [3] Cheng Guimei, Liu Tao, Rong Peng, et al. Design of multi-detector data control and processing system. [J]. Infrared and Laser Engineering, 2016, 45(4): 0420002. (in Chinese) doi: 10.3788/irla201645.0420002 [4] Zhou Nan, Hu Juan, Hu Haiming. Development trend and Key techniques of multi-core processor [J]. Computer Engineering and Design, 2018, 39(2): 393-467. (in Chinese) [5] Tms320C6678 multicore fixed and floating-point digital signal processor SPRS691E[Z].Texas Instruments, 2014. [6] Zynq-7000 SoC data sheet overview[Z]. XILINX, 2018. [7] 何实. Xilinx all programmable Zynq-7000 SoC 设计指南[M]. 北京: 清华大学出版社, 2013. [8] Simple AMP: bare-metal system running on both Cortex-A9 processors [Z]. XILINX, 2014. [9] Xing Yanfang, Zhu Jinfu, Zhou Xiaomei. Design of muti-core processing based on ZYNQ [J]. Computer Technology and Development, 2018, 28(3): 60-66. (in Chinese) [10] Wang Xin, Qiu Xin, Mu Fuqi, et al. Study of MCAPI-compliant communication mechanism in embedded muti-core system[J]. Microelectronics & Computer, 2017, 34(11): 85-88. (in Chinese) [11] Tan Kai, Cheng Xiaojun. TLS laser intensity correction based on polynomial model[J]. Chinese Journal of Lasers, 2015, 42(3): 0314002. (in Chinese) [12] Cheng Xiaolong, Cheng Xiaojun, Li Quan, et al. Laser intensity correction of terrestrial 3D laser scanning based on sectional polynomial model [J]. Laser & Optoelectronics Progress, 2017, 54(11): 112802. (in Chinese) [13] Yang Shujuan, Zhang Keshu, Shao Yongshe. Ranging calibration model of terrestrial laser scanner based on gray level priority[J]. Infrared and Laser Engineering, 2019, 48(1): 0106007. (in Chinese) [14] Sun Xiaoli, James B Abshire, Adrian A Borsa, et al. ICESAT/GLAS altimetry measurements: received signal dynamic range and saturation correction[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(10): 5440-5454. [15] Bai Yang, Tang Wei, Xu Shiyue, et al. Real-time and high-precision ranging method for large dynamic range of imaging lidar [J]. Infrared and Laser Engineering, 2020, 49(S2): 20200179. (in Chinese)