3D Mapping using SLAM

A repository of my learning of Visual & RGB-D SLAM used for 3D mapping of indoor environments with a
Raspberry Pi 5 & an External camera

Aim:

Mapping of indoor environments using RGB-D SLAM algorithm on a Raspberry Pi 5 with a Microsoft Kinect v1
as a camera module and using the RGB camera data along with its depth data to create and visualise a 3D structure of the environment.

References (Repos I have used / taken reference from):

1. OpenCV4:

https://github.com/opencv/opencv

2. OpenCV4 Contrib:

https://github.com/opencv/opencv_contrib

3. ROS2 Image Common:

https://github.com/ros-perception/image_common

4. ROS2 OpenCV:

https://github.com/ros-perception/vision_opencv

5. Pangolin:

https://github.com/stevenlovegrove/Pangolin

6. ORB-SLAM3:

https://github.com/egdw/ORB_SLAM3_Ubuntu20.04

7. ORBS_SLAM3 ROS2 Wrapper:

https://github.com/ozandmrz/orb_slam3_ros2_mono_publisher	[Forked]

8. Installation of ROS2:

https://github.com/ozandmrz/ros2_raspberry_pi_5

Software Required

1. ROS2 (Humble Hawksbill)
2. Ubuntu (20.04 - Focal)
3. Eigen3 (3.3.7)
4. G2O ()
5. Sophus ()
6. Pangolin (4.5.0)
7. OpenGL (3.1)
8. OpenCV (4.12.0-dev)
9. CMAKE (3.25.1)
10. ORB-SLAM3
11. Raspian OS (12 (bookworm))
12. Mesa (23.2.1-1~bpo12+rpt3)

Hardware Used

1. Raspberry Pi 5 ()
2. Raspberry Pi Fan & Heatsink
3. Raspberry Pi Power Adaptor (minimum 27 Watts)
4. 256 GB Mircro SD Card
5. Microsoft Kinect V1
6. External Web Camera

Build the Project

"The build commands in the file build.sh installs all the required software,
frameworks and any additional dependencies required by the user automatically"

• Clone the project:

    git clone https://github.com/eshan-sud/minor-project
  

• Run these conmmands on the shell

    cd ~/minor-project/
    chmod +x build.sh
    ./build.sh
  

• Additonal Instructions:

  • When prompted to choose mode when downgrading GCC version to 11 ; Enter the value '0'

Contact the author

• [email protected]