SLAM 入门——gazebo仿真获取点云 + A-loam/Lego-loam建图

前言

本文仅为我在项目中遇见的一些常见的问题的记录，SLAM太难了QAQ。

准备工作

在进行SLAM项目实践前，需要进行简单的基础知识的了解

1. ROS基本用法 ，推荐古月居的21讲，我记得B站就有；
2. gazebo基本用法，推荐官方教程。本学习项目是通过仿真获取数据，故需要使用gazebo操作，如使用实体机器人或已有数据包，可忽略本部分内容 ；
3. Linux简单使用 ，会点命令行操作就行；

环境配置

基础：ros:noetic+catkin

仿真：gazebo 11.9.0

Lego-loam:gtsam

平台：husky(无实物仿真）

传感器：velodyne HDL32

项目来源

主要是跟着这个做的，但它传感器用的是realsense，需要自己替换一下

问题记录

I haven’t installed Chinese input method in my linux, so the part I wrote in in linux will be English.

Widely uesd :

• <opencv/cv.h>找不到：找到报错的地方，改成 #include <opencv2/imgproc.hpp>
• ‘CV_LOAD_IMAGE_COLOR’ was not declared in this scope: Replace CV_LOAD_IMAGE_COLOR, by cv::IMREAD_COLOR, and then CV_LOAD_IMAGE_GRAYSCALE by cv::IMREAD_GRAYSCALE in your source file.
• lego-loamCTRL+C停止后保存不了点云：[pcl::PCDWriter::writeASCII] Input point cloud has no data! rviz里选中 Map cloud再导入数据进行建图。
• catkin_make时，PCL报错：CmakeLists.txt里加上set(CMAKE_CXX_STANDARD 14)
• resource not found: xxx That means you didn’t get xxx package. Just use sudo apt-get install ros-noetic-xxx to install the package.
• ERROR: cannot launch node of type[AA/BB]:CC. Similar to the previous question, use sudo apt-get install ros-noetic-CC to install the package.
• When you load a new world in gazebo, it is always need a few minutes to finish the preparing process.

Special for this project :

• There are many topic mismatched, you may need to delete the ‘/’ signal in the source file.
• The husky source used in the project mentioned above is not simulating with radar sensor. It use realsense camera instead. However, A-loam can not utilize the pointcloud given by realsense camera. So, you have to change the sensor on husky.
• You may use the husky model provided by official website, and that model contains a right .xacro document about VLP-16.

Good Luck!