0. 环境说明(非常重要)
本教程适用于:
-
Ubuntu 20.04
-
ROS Noetic
-
VMware 虚拟机(无独显)
-
不使用 RViz(避免崩溃)
-
使用官方数据集(rosbag / LVBA dataset)
1. 系统基础环境安装
sudo apt update
sudo apt upgrade -y
sudo apt install -y \
build-essential cmake git wget curl \
python3-pip \
libeigen3-dev libpcl-dev libopencv-dev libsuitesparse-dev
2. 安装 ROS Noetic
sudo apt install ros-noetic-desktop-full -y
配置环境变量:
echo "source /opt/ros/noetic/setup.bash" >> ~/.bashrc
source ~/.bashrc
初始化 rosdep:
sudo apt install python3-rosdep -y
sudo rosdep init
rosdep update
3. 安装 Sophus(必须)
cd ~
git clone https://github.com/strasdat/Sophus.git
cd Sophus
git checkout a621ff
mkdir build && cd build
cmake ..
make -j$(nproc)
sudo make install
4. 创建 catkin 工作空间
mkdir -p ~/catkin_ws/src
cd ~/catkin_ws
catkin_make
5. 安装 Vikit
cd ~/catkin_ws/src
git clone https://github.com/xuankuzcr/rpg_vikit.git
cd ..
catkin_make
6. 安装 FAST-LIVO2
cd ~/catkin_ws/src
git clone https://github.com/hku-mars/FAST-LIVO2.git
编译:
cd ~/catkin_ws
catkin_make -j$(nproc)
source devel/setup.bash
7. 关键配置(必须关闭 RViz)
编辑 launch 文件:
roslaunch fast_livo mapping_avia.launch rviz:=false
8. 下载论文数据集(必须)
推荐:
FAST-LIVO2 Dataset
LVBA Dataset
(官方 Google Drive)
9. 运行论文数据(核心步骤)
开启 FAST-LIVO2:
source ~/catkin_ws/devel/setup.bash
roslaunch fast_livo mapping_avia.launch rviz:=false
新终端播放数据:
rosbag play your_dataset.bag
10. 运行结果说明
运行后会自动生成:
~/catkin_ws/src/FAST-LIVO2/Log/pcd/
里面文件:
1661398631.xxx.pcd
1661398632.xxx.pcd
...
为什么有很多 pcd?
每个 pcd = 一帧局部地图
用于滑动窗口建图
最终拼成完整地图
11. 查看 pcd(推荐方法)
方法1:CloudCompare(推荐)
安装:
sudo apt install cloudcompare
打开:
CloudCompare
GUI操作:
File → Open → 选择 Log/pcd/ → Ctrl+A → Open
方法2:命令行合并(论文最终结果)
cd ~/catkin_ws/src/FAST-LIVO2/Log/pcd
CloudCompare -SILENT -O *.pcd -MERGE_CLOUDS -SAVE_CLOUDS
输出:
cloudCompare_merged.bin / .ply
方法3:单帧查看(轻量)
sudo apt install pcl-tools
pcl_viewer 1661398631.922543.pcd
12. 常见问题解决
RViz崩溃
解决:
rviz:=false
虚拟机必须关闭 RViz
CloudCompare打不开 pcd
解决:
sudo apt install cloudcompare
或用:
pcl_viewer
ros master 断开
重新:
source ~/catkin_ws/devel/setup.bash
roscore
13. 最终论文级流程(最重要)
# 1. 启动算法
roslaunch fast_livo mapping_avia.launch rviz:=false
# 2. 播放数据
rosbag play dataset.bag
# 3. 找到输出
~/catkin_ws/src/FAST-LIVO2/Log/pcd/
# 4. 合并点云(最终地图)
CloudCompare -O *.pcd -MERGE_CLOUDS
# 5. 导出论文图
File → Save As → map.ply
14. 最终结果你会得到
完整3D地图(论文级)
trajectory map
LiDAR-Visual fusion result
可导出 PLY / LAS
一句话总结
FAST-LIVO2 在你当前环境中正确流程就是:
rosbag → FAST-LIVO2 → pcd序列 → CloudCompare合并 → 论文地图