scancontext-pybind

News

• Currently (Apr. 2022), not a set of full features are made yet. For checking the features currently supported, see wrapper.cpp.

Purpose

• Goal of this repo is
  • For fast python support for lidar-based place recognition
    • ps. the native python version (see https://github.com/gisbi-kim/PyICP-SLAM) was already supported, but it was slow.
  • Original C++ code: see https://github.com/irapkaist/scancontext

Build

• Please ensure eigen3 is installed (e.g., via sudo apt install libeigen3-dev)
• Other dependencies (nanoflann and pybind11) are downloaded automatically (for details, see CMakeLists.txt)
• Use the following commands to build:

   $ mkdir build
   $ cd build
   $ cmake ..
   $ make
  

• Install the Python package via pip with

  $ cd build
  $ make pip-install
  # or install manually with
  $ cd build/python
  $ pip install .
  

Use examples

• For the hands-on exploration of the supported features, for example (you need numpy and open3d),

   $ python3 examples/test_make_desciptor.py
   $ python3 examples/test_compare_scd.py
   $ # to be added ...
  

Docker support

• see docker/docker_run.sh. Modify the REPOSITORY_PATH with your own path.
  • For the docker-based test tutorial, see this video (TBA)

Application: PyICP-SLAM2

• TBA ...

Cite

  @ARTICLE { gskim-2021-tro,
      AUTHOR = { Giseop Kim and Sunwook Choi and Ayoung Kim },
      TITLE = { Scan Context++: Structural Place Recognition Robust to Rotation and Lateral Variations in Urban Environments },
      JOURNAL = { IEEE Transactions on Robotics },
      YEAR = { 2021 }
  }

  @INPROCEEDINGS { gkim-2018-iros,
    author = {Kim, Giseop and Kim, Ayoung},
    title = { Scan Context: Egocentric Spatial Descriptor for Place Recognition within {3D} Point Cloud Map },
    booktitle = { Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems },
    year = { 2018 },
    month = { Oct. },
    address = { Madrid }
  }

Ack

• Thanks to the author of KCP. The pybind wrapper src and the directory structures were borrowed from KCP.