Laser slam:logo-loam --- code compilation, installation and gazebo test

LeGO-LOAM brief introduction

LeGO-LOAM Our English full name is lightweight and ground optimized lidar odometry and mapping. Lightweight lidar odometer with ground optimization and mapping

The framework is as follows , In general, and LOAM It's consistent

LeGO-LOAM Is based on LOAM Improved version , Its main purpose is to realize the positioning and mapping of the car under changeable terrain , A series of improvements have been made for both the front end and the back end .

In the original LOAM For example, the application scenario is ：AGV Driving in the grass , Points in the grassland will probably be extracted as corners , But this kind of point is unstable , Compared with buildings 、 Stable corners such as trunk , I hope to remove the corners on the ground .

LeGO-LOAM More lightweight , Reduce the calculation load without affecting the accuracy , It ensures the real-time performance of the embedded platform

comparison LOAM Improvements

front end ：

• 1 Classify and extract ground points , Avoid the extraction of some edge points
• 2 A simple point cloud clustering algorithm is applied , Eliminate some possible outlier
• 3 Solve the front-end odometer in two steps , Reduce the calculation load without affecting the accuracy , It ensures the real-time performance of the embedded platform

Back end ：

• 1 Use slam The back-end part is reconstructed with the concept of key frame in
• 2 The concepts of loop detection and pose map optimization are introduced , Make the global consistency of the map better

LeGO-LOAM Code compilation installation

github The address is ：https://github.com/RobustFieldAutonomyLab/LeGO-LOAM

Has been tested ROS Version has ：

• indigo
• kinetic
• melodic

Its dependent libraries are gtsam. So you need to install it first gtsam

wget -O ~/Downloads/gtsam.zip https://github.com/borglab/gtsam/archive/4.0.0-alpha2.zip
cd ~/Downloads/ && unzip gtsam.zip -d ~/Downloads/
cd ~/Downloads/gtsam-4.0.0-alpha2/
mkdir build && cd build
cmake ..
sudo make install

cmake It can be changed into the following instructions

cmake -DGTSAM_BUILD_EITH_MARCH_NATIVE=OFF -DGTSAM_USE_SYSTEM_EIGEN=ON ..

The purpose is to set some macros
DGTSAM_BUILD_EITH_MARCH_NATIVE Is open css Accelerated macros , To shut down . Avoid other libraries （ such as PCL Or your own code ） If it's not turned on , It will cause abnormal conditions during operation
DGTSAM_USE_SYSTEM_EIGEN This is installed with the system eigen Compiled macro , To turn on .OFF Yes, I can gtsam Self contained eigen, There are two versions of code running eigen It's easy to go wrong

Then download LeGO-LOAM Of ROS Function pack

git clone https://github.com/RobustFieldAutonomyLab/LeGO-LOAM.git

And then on ROS Working space for

Then compile , Be careful The first compilation should use

catkin_make -j1

You don't need to compile later -j1 了

The reason for this is LeGO-LOAM There are some customized msg.
-j1 Will be compiled by single thread . Put custom msg Compile to .

If you use it directly catkin_make According to how many cores the computer has , Compile several threads .

If the compilation is successful, it will generate LeGO-LOAM Of Executable file

LeGO-LOAM Gazebo test

Here's a direct build gazebo Test scenarios Yes LeGO-LOAM Conduct a preliminary test

Before running, it needs to be based on The type of lidar used , stay utility.h Set some variables in the file
The location of the file is include Under the folder ：

Point cloud of lidar Topic name

extern const string pointCloudTopic = "/velodyne_points";

The parameters of lidar include

• Wire harness
• Horizontal points
• Horizontal corner resolution
• Vertical angle resolution
• Angle with horizontal direction
• The number of laser lines scanned to the ground

stay gazebo The simulation scenario built in is used Velodyne-16 Laser radar of , Its configuration is as follows ：

// VLP-16
extern const int N_SCAN = 16;
extern const int Horizon_SCAN = 1800;
extern const float ang_res_x = 0.2;
extern const float ang_res_y = 2.0;
extern const float ang_bottom = 15.0+0.1;
extern const int groundScanInd = 7;

And then run in launch Under folder run.launch file
Then start the simulation scenario , Do a preliminary test

The result is as follows ：


stay rviz The medium green is the corner , The pink one is pastry
There is no abnormality in feature extraction

Construction of drawings