A-LOAM

Advanced implementation of LOAM

A-LOAM is an Advanced implementation of LOAM (J. Zhang and S. Singh. LOAM: Lidar Odometry and Mapping in Real-time), which uses Eigen and Ceres Solver to simplify code structure. This code is modified from LOAM and LOAM_NOTED. This code is clean and simple without complicated mathematical derivation and redundant operations. It is a good learning material for SLAM beginners.

Modifier: Tong Qin, Shaozu Cao

1. Prerequisites

1.1 Ubuntu and ROS

Ubuntu 64-bit 16.04 or 18.04. ROS Kinetic or Melodic. ROS Installation

1.2. Ceres Solver

Follow Ceres Installation.

1.3. PCL

Follow PCL Installation.

2. Build A-LOAM

Clone the repository and catkin_make:

    cd ~/catkin_ws/src
    git clone https://github.com/HKUST-Aerial-Robotics/A-LOAM.git
    cd ../
    catkin_make
    source ~/catkin_ws/devel/setup.bash

3. Velodyne VLP-16 Example

Download NSH indoor outdoor to YOUR_DATASET_FOLDER.

    roslaunch aloam_velodyne aloam_velodyne_VLP_16.launch
    rosbag play YOUR_DATASET_FOLDER/nsh_indoor_outdoor.bag

4. KITTI Example (Velodyne HDL-64)

Download KITTI Odometry dataset to YOUR_DATASET_FOLDER and set the dataset_folder and sequence_number parameters in kitti_helper.launch file. Note you also convert KITTI dataset to bag file for easy use by setting proper parameters in kitti_helper.launch.

    roslaunch aloam_velodyne aloam_velodyne_HDL_64.launch
    roslaunch aloam_velodyne kitti_helper.launch

5. Docker Support

To further facilitate the building process, we add docker in our code. Docker environment is like a sandbox, thus makes our code environment-independent. To run with docker, first make sure ros and docker are installed on your machine. Then add your account to docker group by sudo usermod -aG docker $YOUR_USER_NAME. Relaunch the terminal or logout and re-login if you get Permission denied error, type:

cd ~/catkin_ws/src/A-LOAM/docker
make build

The build process may take a while depends on your machine. After that, run ./run.sh 16 or ./run.sh 64 to launch A-LOAM, then you should be able to see the result.

6.Acknowledgements

Thanks for LOAM(J. Zhang and S. Singh. LOAM: Lidar Odometry and Mapping in Real-time) and LOAM_NOTED.

Basalt 第二节课的作业

本次作业的主要目的是，验证Basalt的视觉重投影的求导方法和VINS-MONO的求导方式虽然不同，但是最终的效果是相同的

补全jacobian.cpp 中的Basalt和VINS-MONO的求导方式，需要注意的是我们对VINS-MONO的求导方式稍微有些不同，我们对旋转施加的是左扰动，而VINS-MONO 实施的是右扰动

*　实验的输出结果

Pt_homogeneous: 
-0.449721 -0.354484  0.819812    0.1095
d_uv_dp: 

 1.21979        0 0.669137
       0  1.21979 0.527433
d_uv_d_rel: 

 0.133567         0 0.0732705 -0.237198   1.30093  0.432396
        0  0.133567  0.057754  -1.18697  0.237198 -0.548566
Basalt
jaco_h: 
  0.108058  0.0785089  0.0732705  -0.889287   0.976457    0.26524
-0.0785089   0.108058   0.057754   -1.05323  -0.479054  -0.535411
jaco_t: 
 -0.108058 -0.0785089 -0.0732705   0.956562  -0.913049  -0.432396
 0.0785089  -0.108058  -0.057754     1.0997   0.505784   0.548566
VINS-MONO
jaco_h: 
  0.108058  0.0785089  0.0732705  -0.889287   0.976457    0.26524
-0.0785089   0.108058   0.057754   -1.05323  -0.479054  -0.535411
jaco_t: 
 -0.108058 -0.0785089 -0.0732705   0.956562  -0.913049  -0.432396
 0.0785089  -0.108058  -0.057754     1.0997   0.505784   0.548566

将VINS-MONO projection_factor.cpp 3D点对绝对位姿的导数jaco_i，jaco_j修改为Basalt的求导方式，也就是3D点先对相对对位姿求导，然后相对位姿对绝对位姿求导

*　实验要求,修改后的VINS-MONO 可以正常运行

其中"utility/Twist.h"提供了SE(3)的伴随矩阵的接口(SE3Adj())