A framework to work with Industrial AMRs and perform SLAM using various sensors in challenging environments

In this repository, I have combined different frameworks to create a functional AMR that performs Robust SLAM in smart factories and similar environments:

For my AMR development, I’ve selected the following Open-Source Repositories:

1. Linorobot2 as the base repository for creating my AMR,
2. RTAB-Map as the main framework for sensor fusion and SLAM,
3. Kinematic-ICP for future work on enhancing ICP-based odometry,

Additionally, there are two more packages I created:

1. amr_main as the base package for running everything together,
2. laser_fusion which contains the necessary 2D-LiDAR fusion node (to fuse measurements from different 2D LiDARs).

• I’m still working on enhancing this project. It works well, but still needs some improvements to incorporate additional features (e.g., using Kinematic-ICP alongside RTAB-Map).

1. RTAB-Map parameter file: In this file, you can change the RTAB-Map parameters. You can also set which 2D LiDAR you want to use (front, back, all) and which Stereo Camera (front, back, both -> (Not available yet)).

   code ~/Multi_Sensor_SLAM/src/rtabmap_ros/rtabmap_launch/launch/rtabmap.launch.py
   

2. Change the Robot and Sensors' parameters or Add/Remove them:

   code ~/Multi_Sensor_SLAM/src/linorobot2/linorobot2_description/urdf/4wd_properties.urdf.xacro
   code ~/Multi_Sensor_SLAM/src/linorobot2/linorobot2_description/urdf/robots/4wd.urdf.xacro
   code ~/Multi_Sensor_SLAM/src/linorobot2/linorobot2_description/urdf/sensors/laser_new.urdf.xacro
   code ~/Multi_Sensor_SLAM/src/linorobot2/linorobot2_description/urdf/sensors/stereo_camera.urdf.xacro
   code ~/Multi_Sensor_SLAM/src/linorobot2/linorobot2_description/urdf/sensors/imu.urdf.xacro
   

3. 2D-LiDAR fusion algorithm:

   code ~/Multi_Sensor_SLAM/src/laser_fusion/laser_fusion/combine_laser_measurements.py
   

• In this demo, I have used only the front LiDAR (not both). I have also used the front stereo camera. Here, I'm performing VIO and ICP odometry, as well as visual and laser loop closures to create a 2D map. Additionally, RTAB-Map uses graph-based SLAM.

• This project requires ROS 2 (Recommended: ROS 2 Humble):

1. Clone the Repository:

   git https://github.com/ali-pahlevani/Multi_Sensor_SLAM.git
   cd Multi_Sensor_SLAM
   

2. Install the required Dependencies:

   rosdep update && rosdep install --from-path src --ignore-src -y
   

3. Build the Workspace:

   colcon build
   source install/setup.bash
   

4. Launch the Simulation:

   ros2 launch amr_main launch_all.launch.py