无人机避障——感知部分（Ubuntu 20.04 复现Vins Fusion跑数据集）胎教级教程

硬件环境：NVIDIA Jeston Orin nx

系统：Ubuntu 20.04

任务：跑通 EuRoC MAV Dataset 数据集

展示结果：

编译Vins Fusion

创建工作空间vins_ws

bash 复制代码

# 创建目录结构
mkdir -p ~/vins_ws/src

cd ~/vins_ws/src

# 初始化工作空间（生成 CMakeLists.txt）
catkin_init_workspace

# 克隆官网的代码
git clone https://github.com/HKUST-Aerial-Robotics/VINS-Fusion.git

cd ..

catkin_make

遇到了不少报错，参考以下内容可以基本解决前期的编译问题。

将c++11 修改为c++14

用VScode搜索set(CMAKE_CXX_FLAGS "-std=c++11")，全部替换为set(CMAKE_CXX_FLAGS "-std=c++14")，图中我已经全部替换过了。

指定find_package中的OpenCV的版本

用VScode搜索find_package(OpenCV REQUIRED)，全部替换为find_package(OpenCV 4 REQUIRED)，图中我已经全部替换过了。

添加头文件

camera_models>include>camodocal>chessboard>Chessboard.h中添加

cpp 复制代码

#include <opencv2/imgproc/types_c.h>
#include <opencv2/calib3d/calib3d_c.h>

在以下文件：

camera_models/include/camodocal/calib/CameraCalibration.h

loop_fusion/src/pose_graph.h

vins_estimator/src/featureTracker/feature_tracker.h

loop_fusion/src/ThirdParty/DVision/BRIEF.h

加入以下的头文件：

cpp 复制代码

#include <opencv2/imgproc/types_c.h>
#include <opencv2/imgproc/imgproc_c.h>

修改报错文件

在以下文件

VINS-Fusion/vins_estimator/src/KITTIGPSTest.cpp

VINS-Fusion/vins_estimator/src/KITTIOdomTest.cpp

其中，将 CV_LOAD_IMAGE_GRAYSCALE修改为cv::IMREAD_GRAYSCALE，同样的直接用VScode搜索 CV_LOAD_IMAGE_GRAYSCALE将其替换即可。

`然后编译`

bash 复制代码

usr/bin/ld: warning: libopencv_video.so.4.2, needed by /home/nvidia/vins_ws/devel/lib/libvins_lib.so, may conflict with libopencv_video.so.405
/usr/bin/ld: warning: libopencv_imgcodecs.so.405, needed by /home/nvidia/vins_ws/devel/lib/libvins_lib.so, may conflict with libopencv_imgcodecs.so.4.2
/usr/bin/ld: warning: libopencv_core.so.405, needed by /usr/local/lib/libopencv_video.so.4.5.5, may conflict with libopencv_core.so.4.2
/usr/bin/ld: warning: libopencv_video.so.4.2, needed by /home/nvidia/vins_ws/devel/lib/libvins_lib.so, may conflict with libopencv_video.so.405
/usr/bin/ld: warning: libopencv_imgcodecs.so.405, needed by /home/nvidia/vins_ws/devel/lib/libvins_lib.so, may conflict with libopencv_imgcodecs.so.4.2
/usr/bin/ld: warning: libopencv_core.so.405, needed by /usr/local/lib/libopencv_video.so.4.5.5, may conflict with libopencv_core.so.4.2
[ 98%] Built target kitti_odom_test
[ 98%] Built target kitti_gps_test
[100%] Linking CXX executable /home/nvidia/vins_ws/devel/lib/vins/vins_node
/usr/bin/ld: warning: libopencv_video.so.4.2, needed by /home/nvidia/vins_ws/devel/lib/libvins_lib.so, may conflict with libopencv_video.so.405
/usr/bin/ld: warning: libopencv_core.so.405, needed by /usr/local/lib/libopencv_video.so.4.5.5, may conflict with libopencv_core.so.4.2

编译虽然通过了，但是有警告，这里是因为在链接过程中存在 OpenCV 库版本冲突。系统检测到两个不同版本的 OpenCV 库（4.2 和 4.5.5）被同时链接。

段错误

如果不解决上述OpenCV 库版本冲突问题，后续运行vins_node将会出现段错误：

解决OpenCV 库版本冲突问题。

最好的解决方法就是全部统一成一个OpenCV 4.5.5。

bash 复制代码

# 卸载通过 apt 安装的 OpenCV
sudo apt purge -y '^libopencv.*' '^opencv.*'

# 清除残留配置文件
sudo rm -rf /usr/local/include/opencv* /usr/local/lib/libopencv* /usr/local/share/opencv* \
            /usr/include/opencv* /usr/lib/x86_64-linux-gnu/libopencv* \
            ~/.local/lib/python*/*/site-packages/cv2*

# 更新系统库缓存
sudo ldconfig

# 更新下
sudo apt-get update

sudo apt-get upgrade

# 下载Opencv 4.5.5源码
wget https://github.com/opencv/opencv/archive/refs/tags/4.5.5.zip

# 解压
unzip opencv-4.5.5.zip	

# 安装依赖
sudo apt-get install cmake
sudo apt-get install build-essential libgtk2.0-dev libavcodec-dev libavformat-dev libjpeg-dev libswscale-dev libtiff5-dev
sudo apt-get install libgtk2.0-dev
sudo apt-get install pkg-config

# 编译安装
mkdir build

cd build

cmake -D WITH_TBB=ON -D WITH_EIGEN=ON -D OPENCV_GENERATE_PKGCONFIG=ON  -D BUILD_DOCS=ON -D BUILD_TESTS=OFF -D BUILD_PERF_TESTS=OFF -D BUILD_EXAMPLES=OFF  -D WITH_OPENCL=OFF -D WITH_CUDA=OFF -D BUILD_opencv_gpu=OFF -D BUILD_opencv_gpuarithm=OFF -D BUILD_opencv_gpubgsegm=O -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local ..

make -j8
sudo make install


# 环境配置
sudo gedit /etc/ld.so.conf.d/opencv.conf

# 添加以下路径
/usr/local/lib

# 保存后，终端执行
sudo ldconfig 

# 更新路径
sudo gedit /etc/bash.bashrc

# 最后添加
PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig
export PKG_CONFIG_PATH

# 保存后，终端执行
source /etc/bash.bashrc
sudo updatedb

然后编译Vins Fusion 发现没有 cv_bridge

bash 复制代码

ither need to install the package with the same name or change your environment so that it can be found.
CMake Error at /opt/ros/noetic/share/catkin/cmake/catkinConfig.cmake:83 (find_package):
  Could not find a package configuration file provided by "cv_bridge" with
  any of the following names:

    cv_bridgeConfig.cmake
    cv_bridge-config.cmake

  Add the installation prefix of "cv_bridge" to CMAKE_PREFIX_PATH or set
  "cv_bridge_DIR" to a directory containing one of the above files.  If
  "cv_bridge" provides a separate development package or SDK, be sure it has
  been installed.
Call Stack (most recent call first):
  VINS-Fusion/loop_fusion/CMakeLists.txt:9 (find_package)


-- Configuring incomplete, errors occurred!
See also "/home/nvidia/vins_ws/build/CMakeFiles/CMakeOutput.log".
See also "/home/nvidia/vins_ws/build/CMakeFiles/CMakeError.log".
Invoking "cmake" failed

如果用这条ROS相关的安装cv_bridge的指令还会报错：

bash 复制代码

# 安装 ROS Noetic 的 cv_bridge
sudo apt-get install ros-noetic-cv-bridge

原因是系统安装了OpenCV 4.5.5，但通过ROS官方源安装的cv_bridge是针对OpenCV 4.2编译的。而我们的代码需要OpenCV 4.5.5，因此可能需要重新编译cv_bridge以匹配我们安装的OpenCV版本。

重新编译cv_bridge以匹配OpenCV 4.5.5版本

bash 复制代码

# 1. 卸载旧版本(重要!)
sudo apt remove ros-noetic-cv-bridge

# 2. 创建工作空间
mkdir -p ~/cv_bridge_ws/src
cd ~/cv_bridge_ws/src

# 3. 获取vision_opencv源码(noetic分支)
git clone -b noetic https://github.com/ros-perception/vision_opencv.git

# 4. 关键：修改编译配置
sed -i 's/find_package(OpenCV 3 REQUIRED)/find_package(OpenCV 4.5 REQUIRED)/g' vision_opencv/cv_bridge/CMakeLists.txt

# 5. 编译安装
cd ~/cv_bridge_ws
catkin config -DCMAKE_BUILD_TYPE=Release -DOpenCV_DIR=/path/to/opencv4.5.5/build
catkin build cv_bridge

# 6. 添加环境变量(永久生效)
echo "source ~/cv_bridge_ws/devel/setup.bash" >> ~/.bashrc
source ~/.bashrc

如果安装完还是报错：

bash 复制代码

# 查找 cv_bridge 的实际安装位置
find / -name "*cv_bridge*Config.cmake" 2>/dev/null

得到下面这个路径

/home/nvidia/cv_bridge_ws/devel/share/cv_bridge/cmake/cv_bridgeConfig.cmake

在VINS-Fusion的CMakeLists.txt中（具体是loop_fusion/CMakeLists.txt）的find_package(cv_bridge)之前加入：

bash 复制代码

set(cv_bridge_DIR "/home/nvidia/cv_bridge_ws/devel/share/cv_bridge/cmake")

最后编译成功，没有warning也没有报错：

最后结果：

bash 复制代码

 
roslaunch vins vins_rviz.launch
 
rosrun vins vins_node src/VINS-Fusion/config/euroc/euroc_stereo_imu_config.yaml
 
rosrun loop_fusion loop_fusion_node src/VINS-Fusion/config/euroc/euroc_stereo_imu_config.yaml
 
# 数据集去Vins-Fusion github官网上进行下载
rosbag play MH_01_easy.bag