/Monocular-MiniSLAM

Real-Time Monocular Visual SLAM with Pose-graph optimization

Primary LanguagePython

Monocular-Visual-SLAM

Minimal implementation of monocular SLAM with pose graph optimisation (loop closing yet to be implemented)

To run the visual SLAM:

python run_slam.py --path ../KITTI/KITTI_gray/dataset/sequences/00 \
                   --optimize \
                   --local_window 10 \
                   --num_iter 100

where path is the path to the KITTI dataset (directory structure of code and data to be updated)

Task List

  • Build visual-odometry frontend with ORB descriptors and 2D-2D feature correspondences
  • Build G2O and Pangolin (check Installation.md) on few tips on installation and troubleshooting guide
  • Check working of front end of the slam system
  • Integrate pose-graph optimization backend using g2o (static)
  • Set up 3D plotter for visualisation of frames and point cloud
  • Integrate pose-graph optimization backend on-the-fly (dyanamic)
  • Intergrated bundled pose graph optimization (backend)