NVIDIA Jetson Xavier NX部署VINS-fusion-GPU

一、环境配置（Ubuntu 18.04）

1、Cuda 10.2的安装

sudo apt-get update
sudo apt-get install cuda-toolkit-10-2

安装好之后，在 .bashrc中配置环境变量。source之后，nvcc – version 即可查看cuda版本。

export PATH=/usr/local/cuda-10.2/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export CUDA_ROOT=/usr/local/cuda

2、Eigen（用ROS自带的也可以）

# Remove pre-built Eigensudo apt-get remove libeigen3-dev
cd ~/Downloads/
wget -O eigen.zip https://gitlab.com/libeigen/eigen/-/archive/3.3.7/eigen-3.3.7.zip #check version
unzip eigen.zip
mkdir eigen-build && cd eigen-build
cmake ../eigen-3.3.7/ && sudo make install
pkg-config --modversion eigen3 # Check Eigen Version

3、Ceres solver

cd ~/Downloads/
sudo apt-get install -y cmake libgoogle-glog-dev libatlas-base-dev libsuitesparse-dev
wget http://ceres-solver.org/ceres-solver-1.14.0.tar.gz
tar zxf ceres-solver-1.14.0.tar.gz
mkdir ceres-bin
mkdir solver && cd ceres-bin
cmake ../ceres-solver-1.14.0 -DEXPORT_BUILD_DIR=ON -DCMAKE_INSTALL_PREFIX="../solver" #good for build without being root privileged and at wanted directorymake -j3 # 6 : number of cores
make install
bin/simple_bundle_adjuster ../ceres-solver-1.14.0/data/problem-16-22106-pre.txt # to check version

4、Opencv 安装

关于Opencv 的安装需要特别谨慎，不建议频繁删除以前的库，因为多个项目使用的版本可能不同，所以可以将常用的版本安装在/usr/local下，将其余版本安装在自己Download或者home下。
如果需要删除原来的版本，执行以下命令：

# remove prebuilt opencv
sudo apt-get purge libopencv* python-opencv
sudo apt-get update
sudo apt-get install -y build-essential pkg-config## libeigen3-dev # recommend to build from sourcesudo apt-get install -y cmake libavcodec-dev libavformat-dev libavutil-dev \libglew-dev libgtk2.0-dev libgtk-3-dev libjpeg-dev libpng-dev libpostproc-dev \libswscale-dev libtbb-dev libtiff5-dev libv4l-dev libxvidcore-dev \libx264-dev qt5-default zlib1g-dev libgl1 libglvnd-dev pkg-config \libgstreamer1.0-dev libgstreamer-plugins-base1.0-dev mesa-utils     sudo apt-get install python2.7-dev python3-dev python-numpy python3-numpy

# To fix OpenGL related compilation problems cd /usr/lib/aarch64-linux-gnu/
sudo ln -sf libGL.so.1.0.0 libGL.so
sudo vim /usr/local/cuda/include/cuda_gl_interop.h# Comment (line #62~68) of cuda_gl_interop.h //#if defined(__arm__) || defined(__aarch64__)
//#ifndef GL_VERSION
//#error Please include the appropriate gl headers before including cuda_gl_interop.h
//#endif
//#else#include <GL/gl.h>
//#endif

安装新的版本（因为VINS-FUSION-GPU版本需要支持CUda的OpenCV，所以选择3.4.1版本）
Then once linking is done, go to Downloads to begin opencv installation

cd ~/Downloads/
wget -O opencv.zip https://github.com/opencv/opencv/archive/3.4.1.zip # check version
unzip opencv.zip
cd opencv-3.4.1/ && mkdir build && cd build
cmake -D CMAKE_BUILD_TYPE=RELEASE \-D CMAKE_INSTALL_PREFIX=/usr/local \     #这里的/usr/local可以换成自己的安装路径，到时候在Cmakelist里设置自己设置的路径即可。-D WITH_CUDA=ON \-D CUDA_ARCH_BIN=7.2 \     #对于NX和AGX,选7.2，Nano，TX2选6.2 -D CUDA_ARCH_PTX="" \-D ENABLE_FAST_MATH=ON \-D CUDA_FAST_MATH=ON \-D WITH_CUBLAS=ON \-D WITH_LIBV4L=ON \-D WITH_GSTREAMER=ON \-D WITH_GSTREAMER_0_10=OFF \-D WITH_QT=ON \-D WITH_OPENGL=ON \-D CUDA_NVCC_FLAGS="--expt-relaxed-constexpr" \-D WITH_TBB=ON \../
make  # running in single core is good to resolve the compilation issues
sudo make install
cd ../../ && sudo rm -rf opencv-3.4.1 # optional (can save 10GB Disk Space)
pkg-config --modversion opencv # Check opencv Version

OpenCV 在NX上的安装比较慢，差不多需要1个小时。OpenCV安装好以后，需要进一步编译CV-Bridge( cv_bridge是在ROS图像消息和OpenCV图像之间进行转换的一个功能包)。

5、CV-Bridge的编译

CV-Bridge的安装需要下载对应的版本，由于NX 是基于Ubuntu 18.04的ARM系统,因此下载melodic版本的CV-Bridge。首先，通过以下连接进入下载页面：

https://github.com/ros-perception/vision_opencv

如图所示:

# 选择下载melodic版本到catkin_ws/src目录下，
gedit vision_opencv/cv_bridge/CMakeLists.txt
# Edit OpenCV PATHS in CMakeLists and include cmake file
# /usr/local/share/OpenCV 选择自己安装的OpenCV对应路径
find_package(OpenCV 3 REQUIRED PATHS /usr/local/share/OpenCV NO_DEFAULT_PATH COMPONENTSopencv_coreopencv_imgprocopencv_imgcodecsCONFIG
)
include(/usr/local/share/OpenCV/OpenCVConfig.cmake)  #under catkin_python_setup()
# Save and close CMakeLists
#  Build the package
cd .. && catkin_make

二、Vins-Fusion GPU编译安装

1、Vins-Fusion GPU

cd ~/catkin_ws/src && git clone https://github.com/pjrambo/VINS-Fusion-gpu #GPUsudo apt-get install ros-melodic-tf
sudo apt-get install ros-melodic-image-transport
sudo apt-get install ros-melodic-rviz# Edit CMakeLists.txt for loop_fusion and vins_estimator
cd ~/catkin_ws/src/VINS-Fusion-gpu/loop_fusion && gedit CMakeLists.txt##For loop_fusion : line 19
#find_package(OpenCV)
# /usr/local/share/OpenCV/ OpenCVConfig.cmake 选择自己安装的OpenCV对应路径
include(/usr/local/share/OpenCV/OpenCVConfig.cmake)cd ~/catkin_ws/src/VINS-Fusion-gpu/vins_estimator && gedit CMakeLists.txt##For vins_estimator : line 20
#find_package(OpenCV REQUIRED)
# /usr/local/share/OpenCV/ OpenCVConfig.cmake 选择自己安装的OpenCV对应路径
include(/usr/local/share/OpenCV/OpenCVConfig.cmake)cd ~/catkin_ws/
source devel/setup.bash
catkin_make

2、数据测试。

2.1公开数据集测试【以EuRoC（Stereo cameras + IMU）为例】

 roslaunch vins vins_rviz.launchrosrun vins vins_node ~/catkin_ws/src/VINS-Fusion/config/euroc/euroc_stereo_imu_config.yaml (optional) rosrun loop_fusion loop_fusion_node ~/catkin_ws/src/VINS-Fusion/config/euroc/euroc_stereo_imu_config.yaml
rosbag play YOUR_DATASET_FOLDER/MH_01_easy.bag

数据可以通过一下链接下载到：

http://projects.asl.ethz.ch/datasets/doku.php?id=kmavvisualinertialdatasets
如果下载慢的话，也可以通过以下连接获取KITTI
链接: https://pan.baidu.com/s/1UIoP1TgGLLWWpKYVrre9WA 密码: dbsb
EuRoC
链接: https://pan.baidu.com/s/1zvxyN1KlIk28_LvC36WU5Q 密码: f1rv
shapenet
链接: https://pan.baidu.com/s/1pdRk3FOxRe5YRBVGP0ACJw 密码: 4knl
TUM
链接: https://pan.baidu.com/s/1GFy8MZmH4TZGakzGWudyhQ 密码: ptga
数据来源原文链接：https://blog.csdn.net/weixin_43828675/article/details/117049952（感谢嚣张的叉烧包同志）

2.2 测试自己的设备（D435i+阿木P450）

在配置参数文件中加入以下参数，使得GPU起作用

In the config file, there are two parameters for gpu acceleration.
use_gpu: 0 for off, 1 for on
use_gpu_acc_flow: 0 for off, 1 for on
If your GPU resources is limitted or you want to use GPU for other computaion. You can set
use_gpu: 1
use_gpu_acc_flow: 0
If your other application do not require much GPU resources, I recommanded you to set
use_gpu: 1
use_gpu_acc_flow: 1

IMU和相机外参标定可以通过VINS实时在线标定或者使用Kalibr工具包进行标定，相关资料比多了，不再赘述。

下期给各位铁子们汇报在阿木实验室P450上部署VINS+Prometheus+Ggo-Planner的工作。

文献参考

1、无人机自主导航（ARM架构的vins-fusion-GPU部署）

https://blog.csdn.net/weixin_45600405/article/details/125249964?spm=1001.2101.3001.6650.3&utm_medium=distribute.pc_relevant.none-task-blog-2%7Edefault%7EESLANDING%7Edefault-3-125249964-blog-119830683.pc_relevant_landingrelevant&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2%7Edefault%7EESLANDING%7Edefault-3-125249964-blog-119830683.pc_relevant_landingrelevant&utm_relevant_index=4

2、安装环境OpenCV

https://github.com/arjunskumar/vins-fusion-gpu-tx2-nano

3、调内外参数

https://www.bilibili.com/video/BV1WZ4y167me?p=11&vd_source=1682913523e03074db0225c9475c4c71

4、技术分享 | 带你具体部署VINS_FUSION_GPU版本

https://mp.weixin.qq.com/s/FaE05W2vAvHCZc7R39M7Pw