Graph based SLAM for multiple cameras using SuperPoint feature detector. Uses (zed_wrapper, rtabmap_ros, LibTorch and ROS noetic).

1. ROS (Offical Guide)
2. zed_wrapper (Offical Guide)
3. zed_ros_example (Offical Repository)
4. LibTorch (Offical Installation Guide)
5. rtabmap + rtabmap_ros (Offical Installation Guide)
6. Download SuperPoint Weights .pt file (Offical Reposoitory)
7. catkin build workspace and source to .bashrc.

1. modify launch file to configure camera frames and link for SLAM
2. Change translation and orientation of cameras w.r.t base_link (Both tf packages and zed_multi_camera nodes)

roslaunch multi_camera_superpoint_semantic_slam_zed_odom.launch

Point Cloud reconstuction and real-time pose graph optimization occurs at 1 Hz.

View all topics on RVIZ under rtabmap namespace.

1. Integrate SuperGlue Feature Matching.
2. Integrate Semantic Segmentation for Multi-Camera SLAM System.
3. Integrate Bi-Directional loop closures and develop more roboust pose-graph optimizations. (Intial files in mergeScans.h and mergeScans.cpp)