ORB_ Slam3 recurrence - Part I

Preface

Recently due to work needs , I was forced to transfer in SLAM field , It is helpless , To beg for food , Forced to give up the original dream , Cry, cry, cry , The recent discovery SLAM It's also fun , Ah , It's delicious ！ Don't talk much , Or that sentence , In order to urge myself to study hard , Record the first of the following recurrences SLAM Related projects .

1. ORB_SLAM3

ORB_SLAM It's based on ORB features 3D positioning and map building algorithm ,OBR_SLAM The family has been updated to the third generation , Today we will reproduce the following ORB_SLAM3, Although it has been more than two years , For the newcomer , Reproducing the latest version is the best choice .
Source connection ：https://github.com/UZ-SLAMLab/ORB_SLAM3
Recommend learning ：https://mp.weixin.qq.com/s/h1OIxdYQ5Eu-2OAHhsVcIQ
Now let's repeat it step by step , Students with poor conditions , Virtual machines are recommended Ubuntu18.04, The version is not recommended to be too high or too low , Go wrong .

2. Prepare the environment

2.1 C++11 Compiler

After installing the virtual machine , It is recommended to change to domestic sources , Open the terminal , Start installation gcc,g++,gcc yes C Language compiler ,g++ yes C++ compiler , Therefore, it is recommended to install the following . And then there's the installation Cmake Cross platform compilation tools .

1. gcc,g++

sudo apt-get install gcc
sudo apt-get install g++

2. cmake

sudo apt-get install build-essential
sudo apt-get install cmake

2.2 Pangolin

The installation process can be based on the process provided by the author , There may be many errors in the operation process , Everyone makes different mistakes , I suggest changing to domestic sources may cause less problems . official Install connections .

2.3 Opencv

It is worth noting that the official said opencv3.2 and 4.4 It's been tested , Can run through , But many students reacted , Only 4.4 It works , I suggest installing opencv4.4.

1. Download the installation package on the official website
   Official website Connect , The official website provides many ways to download , Pay attention to download Sources Ha
   
2. Unzip and switch to the path

unzip opencv-4.4.0.zip
cd opencv-4.4.0

3. Now start installing dependencies

sudo apt-get install git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev
sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev

4. Cmake once

mkdir build
cd build
cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D WITH_GTK=ON -D OPENCV_GENERATE_PKGCONFIG=YES ..

remarks ：

command	explain
CMAKE_BUILD_TYPE=RELEASE	The release
CMAKE_INSTALL_PREFIX	Installation path of dynamic library , You can customize
WITH_GTK=ON	libgtk2.0-dev Is the installation successful
OPENCV_GENERATE_PKGCONFIG=YES	Automatic generation OpenCV Of pkgconfig file

5. compile make
   Compile with multi system kernel , Can pass nproc Inquire about .

make -j2

6. make install

sudo make install

notes ： If something goes wrong in the middle , Need to compile again , You need to empty build file .

7. Add library path

sudo vim /etc/ld.so.conf

Add the following code to the file include /usr/loacal/lib, Remember to save and exit （ESC + : + wq!）. It can also be generated manually opencv.conf, open sudo vim /etc/ld.so.conf.d/opencv.conf, write in /usr/local/lib. Be careful opencv.conf, yes vim Manually generated empty file .

8. Make the path work

sudo ldconfig

9. Configure system environment

sudo vim /etc/bash.bashrc

Add at the end of the file ：

PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig
export PKG_CONFIG_PATH

Also use （ESC + : + wq!） Save and exit .

10. Make system variables effective

source /etc/bash.bashrc

11. Test whether the environment configuration is successful

pkg-config --modversion opencv4
pkg-config --cflags opencv4
pkg-config --libs opencv4

The above results appear , Congratulations on your success , If you still want to test , Whether the library can be called normally , Then you can test the code .
Create a new test file test.cpp. Input... At the terminal （ New file touch test.cpp ） （ Open file gedit test.cpp） Enter the following code to save .

#include <iostream>
#include <opencv2/opencv.hpp>
 
using namespace std;
using namespace cv;

int main()
{
    
    Mat srcImage = imread("image.jpg");
    imshow("Display Image window",srcImage);
    waitKey(0);
    return 0;
}

Pay attention to put a picture in the same level directory , And rename the picture image.jpg. Compile with the following commands .

g++ `pkg-config opencv4 --cflags` test.cpp  -o demo `pkg-config opencv4 --libs`
./demo

Come here and congratulate you , Hi Ti opencv4.4.

2.4 Eigen

install Eigen Relatively simple , There's nothing to say , Go straight to the code .

git clone https://gitlab.com/libeigen/eigen.git
cd eigen
mkedir build
cd build
cmake ..
make
sudo make install

3. Reappear ORB_SLAM3

3.1 Download code

git clone https://github.com/UZ-SLAMLab/ORB_SLAM3.git ORB_SLAM3

3.2 perform build.sh

cd ORB_SLAM3
chmod +x build.sh
./build.sh

3.3 test EuRoC Data sets

download EuRoC MAV Dataset Data sets , Take this data set as an example . Click on the link to download , stay ORB_SLAM3 Create a new file under the path Datasets（mkidr Datasets）, Rename the downloaded dataset to MH01 Put in Datasets In file . Pay attention to the path , Or you'll make a mistake .

Be sure to pay attention to the path correspondence

./Examples/Monocular/mono_euroc ./Vocabulary/ORBvoc.txt ./Examples/Monocular/EuRoC.yaml ./Datasets/MH01/ ./Examples/Monocular/EuRoC_TimeStamps/MH01.txt dataset-MH01_mono

There are the following problems , Is running the command again ../Examples/Monocular/mono_euroc ./Vocabulary/ORBvoc.txt ./Examples/Monocular/EuRoC.yaml ./Datasets/MH01/ ./Examples/Monocular/EuRoC_TimeStamps/MH01.txt dataset-MH01_mono

In addition, this problem may occur during operation , No error , There is no result , as follows , It took a long time to solve this problem .

The output in terminal is as follows.
Initialization of Atlas from scratch
Creation of new map with id: 0
Creation of new map with last KF id: 0
Seq. Name:
There are 1 cameras in the atlas
Camera 0 is pinhole
[ INFO:0] global /home/ber/Documents/software/opencv-4.5.4/modules/core/src/parallel/registry_parallel.impl.hpp (96) ParallelBackendRegistry core(parallel): Enabled backends(3, sorted by priority): ONETBB(1000); TBB(990); OPENMP(980)
First KF:0; Map init KF:0
New Map created with 271 points

terms of settlement , You can see Connect , Namely Examples/Monocular/mono_euroc.cc In this document false to true, Probably in this position .

Here are the results of normal operation .

notes ： What appears in the video image Small green box It is extracted ORB features , In the map video Green line path Represents the motion path of the camera , Small blue box Represents the motion process of the camera , That is, keyframes , Black spot Represents the road sign of the past , Red dot Represents the current road sign .

summary

Today's blog is written here , because , Too much blog content , Pages are always stuck , So the rest , I'll go on writing , I will test all the data later , And evaluate the results .