激光SLAM：Faster-Lio算法编译与测试

来源：古月居

前言

Faster-LIO是基于FastLIO2开发的。FastLIO2是开源LIO中比较优秀的一个，前端用了增量的kdtree（ikd-tree），后端用了迭代ESKF（IEKF），流程短，计算快。Faster-LIO则把ikd-tree替换成了iVox，顺带优化了一些代码逻辑，实现了更快的LIO。在典型的32线激光雷达中可以取得100-200Hz左右的计算频率，在固态雷达中甚至可以达到1000-2000Hz，能够达到FastLIO2的1.5-2倍左右的速度。当然具体数值和计算平台相关。

FasterLIO使用了一种基于稀疏体素的近邻结构iVox（incremental voxels）。我们会发现这种结构用来做LIO更加合适，可以有效的降低点云配准时的耗时，也不会影响LIO的精度表现。

iVox也可以被集成到其他LO或LIO里，但是大部分方案里，最近邻并不是主要的计算瓶颈，gtsam/ceres什么的耗时相比最近邻那可太多了。、把iVox集成到Lego-LOAM里，、主要只是省了增量地图构建那部分时间，优化方面没什么变化（点少）。所以iVox与FastLIO倒是相性更好一些。

编译

部署系统：ubuntu20.04
ROS版本: noetic

github 地址：https://github.com/gaoxiang12/faster-lio

下载源码

git clone https://github.com/gaoxiang12/faster-lio

正克隆到 ‘faster-lio’…
remote: Enumerating objects: 224, done.
remote: Counting objects: 100% (108/108), done.
remote: Compressing objects: 100% (43/43), done.
remote: Total 224 (delta 76), reused 65 (delta 65), pack-reused 116
接收对象中: 100% (224/224), 38.13 MiB | 1.49 MiB/s, 完成.
处理 delta 中: 100% (97/97), 完成.

将原文件拷入ros工作空间

依赖

•ROS (melodic or noetic)

•glog: sudo apt-get install libgoogle-glog-dev

•eigen: sudo apt-get install libeigen3-dev

•pcl: sudo apt-get install libpcl-dev

•yaml-cpp: sudo apt-get install libyaml-cpp-dev

编译

catkin_make

报错1：

CMake Error at /home/jk-jone/jone_ws/build/livox_ros_driver/livox_ros_driver/cmake/livox_ros_driver-genmsg.cmake:14 (add_custom_target):
add_custom_target cannot create target “livox_ros_driver_generate_messages”
because another target with the same name already exists. The existing
target is a custom target created in source directory
“/home/jk-jone/jone_ws/src/faster-lio/thirdparty/livox_ros_driver”. See
documentation for policy CMP0002 for more details.
Call Stack (most recent call first):
/opt/ros/noetic/share/genmsg/cmake/genmsg-extras.cmake:307 (include)
livox_ros_driver/livox_ros_driver/CMakeLists.txt:46 (generate_messages)

如果工作空间中之前编译了 livox_ros_driver 的功能包，则需要删掉 faster-lio/thirdparty/livox_ros_driver 这个文件夹

再次编译

CMake Error at faster-lio/CMakeLists.txt:15 (add_subdirectory):
add_subdirectory given source “thirdparty/livox_ros_driver” which is not an
existing directory.

因为把那个文件删了，所以找不到路径

将 faster-lio/CMakeLists.txt 文件的第15行注释掉

add_subdirectory(thirdparty/livox_ros_driver)
改为
#add_subdirectory(thirdparty/livox_ros_driver)

再次编译

[100%] Linking CXX shared library /home/jk-jone/jone_ws/devel/lib/libfaster_lio.so
[100%] Built target faster_lio
Scanning dependencies of target run_mapping_offline
Scanning dependencies of target run_mapping_online
[100%] Building CXX object faster-lio/app/CMakeFiles/run_mapping_online.dir/run_mapping_online.cc.o
[100%] Building CXX object faster-lio/app/CMakeFiles/run_mapping_offline.dir/run_mapping_offline.cc.o
[100%] Linking CXX executable /home/jk-jone/jone_ws/devel/lib/faster_lio/run_mapping_online
[100%] Built target run_mapping_online
[100%] Linking CXX executable /home/jk-jone/jone_ws/devel/lib/faster_lio/run_mapping_offline
[100%] Built target run_mapping_offline

编译成功

测试

Faster-lio支持离线的测试与在线测试

离线测试

首先下载rosbag数据包到电脑

aviabags(https://drive.google.com/drive/folders/1YL5MQVYgAM8oAWUm7e3OGXZBPKkanmY1?usp=sharing)

ncltbags(https://drive.google.com/drive/folders/1VBK5idI1oyW0GC_I_Hxh63aqam3nocNK)

百度云盘下载地址：
BaiduYun:https://pan.baidu.com/s/1ELOcF1UTKdfiKBAaXnE8sQ?pwd=fekyaccess code: feky

OneDrive下载地址：
OneDrive：https://1drv.ms/u/s!AgNFVSzSYXMahcEZejoUwCaHRcactQ?e=YsOYy2

Call run_mapping_offline with parameters to specify the bag file and the config file like:
通过下面的指令 运行 run_mapping_offline 文件 并且加载对应的rosbag文件 和对应的配置文件

./build/devel/lib/faster_lio/run_mapping_offline --bag_file your_avia_bag_file --config_file ./config/avia.yaml

其中 your_avia_bag_file 路径需要更换为下载的数据包路径

同样对于nclt数据包可以运行下面的指令。数据是机械式激光雷达velodyne的数据

./build/devel/lib/faster_lio/run_mapping_offline --bag_file your_nclt_bag_file --config_file ./config/velodyne.yaml

your_nclt_bag_file 路径需要更换为下载的数据包路径

运行FasterLIO，然后退出的时候 会在终端打印FPS和time

像下面这样：

I0216 17:16:05.286536 26492 run_mapping_offline.cc:89] Faster LIO average FPS: 1884.6
I0216 17:16:05.286549 26492 run_mapping_offline.cc:91] save trajectory to: ./src/fast_lio2/Log/faster_lio/20120615.tum
I0216 17:16:05.286706 26492 utils.h:52] >>> ===== Printing run time =====
I0216 17:16:05.286711 26492 utils.h:54] > [   IVox Add Points ] average time usage: 0.0147311 ms , called times: 6373
I0216 17:16:05.286721 26492 utils.h:54] > [   Incremental Mapping ] average time usage: 0.0271787 ms , called times: 6373
I0216 17:16:05.286731 26492 utils.h:54] > [   ObsModel (IEKF Build Jacobian) ] average time usage: 0.00745852 ms , called times: 25040
I0216 17:16:05.286752 26492 utils.h:54] > [   ObsModel (Lidar Match) ] average time usage: 0.0298004 ms , called times: 25040
I0216 17:16:05.286775 26492 utils.h:54] > [ Downsample PointCloud ] average time usage: 0.0224052 ms , called times: 6373
I0216 17:16:05.286784 26492 utils.h:54] > [ IEKF Solve and Update ] average time usage: 0.342008 ms , called times: 6373
I0216 17:16:05.286792 26492 utils.h:54] > [ Laser Mapping Single Run ] average time usage: 0.530618 ms , called times: 6387
I0216 17:16:05.286800 26492 utils.h:54] > [ Preprocess (Livox) ] average time usage: 0.0267813 ms , called times: 6387
I0216 17:16:05.286808 26492 utils.h:54] > [ Undistort Pcl ] average time usage: 0.0810455 ms , called times: 6375
I0216 17:16:05.286816 26492 utils.h:59] >>> ===== Printing run time end =====

默认点云会以pcd文件的格式保存下来

在线测试

用之前建立的仿真环境下的 mid360雷达的数据进行一个初步在线测试

打开仿真环境

faster-lio 里面没有 mid360 雷达的 配置文件和启动文件 ，有avia的，都是livox的固态雷达，基本雷达，仿照avia的写一个就行

mid360.yaml 如下

common:
  lid_topic: "/livox/lidar"
  imu_topic: "/livox/imu"
  time_sync_en: false     # ONLY turn on when external time synchronization is really not possible
  time_offset_lidar_to_imu: 0.0 # Time offset between lidar and IMU calibrated by other algorithms, e.g. LI-Init (can be found in README).
                 # This param will take effect no matter what time_sync_en is. So if the time offset is not known exactly, please set as 0.0


preprocess:
  lidar_type: 1        # 1 for Livox serials LiDAR, 2 for Velodyne LiDAR, 3 for ouster LiDAR, 
  scan_line: 4
  blind: 0.5


mapping:
  acc_cov: 0.1
  gyr_cov: 0.1
  b_acc_cov: 0.0001
  b_gyr_cov: 0.0001
  fov_degree:  360
  det_range:   100.0
  extrinsic_est_en: false   # true: enable the online estimation of IMU-LiDAR extrinsic
  extrinsic_T: [ -0.011, -0.02329, 0.04412 ]
  extrinsic_R: [ 1, 0, 0,
          0, 1, 0,
          0, 0, 1]


publish:
  path_en: false
  scan_publish_en: true    # false: close all the point cloud output
  dense_publish_en: true    # false: low down the points number in a global-frame point clouds scan.
  scan_bodyframe_pub_en: true # true: output the point cloud scans in IMU-body-frame


pcd_save:
  pcd_save_en: true
  interval: -1         # how many LiDAR frames saved in each pcd file; 
                 # -1 : all frames will be saved in ONE pcd file, may lead to memory crash when having too much frames.

launch文件如下

运行该launch文件

roslaunch faster_lio mapping_mid360.launch

初始位置的情景和点云模型

飞一圈后整个地图模型

审核编辑：汤梓红