navtech-radar-slam

Radar SLAM: yeti radar odometry + ScanContext-based Loop Closing

What is Navtech-Radar-SLAM?

• In this repository, a (minimal) SLAM problem is defeind as SLAM = Odometry + Loop closing, and the optimized states are only robot poses along a trajectory.
• Based on the above view, this repository aims to integrate current available radar odometry, radar place recognition, and pose-graph optimization.
  1. Radar odometry: Yeti open source that implemented cen2018 and cen2019 methods with considering motion distortation for RANSAC.
     • The odometry modules consumes file-based input (not ROS subscription) in this example. See odometry/yeti_radar_odometry/src/odometry.cpp for the details.
     • However, to seamlessly connect the motion estimation result with the later place recognition module, we added ROS publishing lines to the original odometry.cpp code. Also, see odometry/yeti_radar_odometry/src/odometry.cpp for the details.
  2. Radar place recognition: Scan Context open source
     • In MulRan dataset paper, the radar scan context is also proposed, but in this repository we use a Cartesian 2D feature point cloud (extracted via cen2019 method) as an input for the original Scan Context (IROS2018) method and it works.
     • The Scan Context-based loop detection is included in the file pgo/SC-A-LOAM/laserPosegraphOptimization.cpp.
  3. Pose-graph optimization
     • iSAM2 in GTSAM is used. See pgo/SC-A-LOAM/laserPosegraphOptimization.cpp for the details (ps. the implementation is eqaul to SC-A-LOAM and it means laserPosegraphOptimization.cpp node is generic!)

How to use?

Dependencies

• Yeti: OpenCV and SC-PGO: GTSAM

Steps

First, clone and build.

$ mkdir -p ~/catkin_radarslam/src && cd ~/catkin_radarslam/src
$ git clone https://github.com/gisbi-kim/navtech-radar-slam.git && cd ..
$ catkin_make 

Second,

• Download a sequence from the MulRan dataset (you need to download polar_oxford_form.zip)
• Change this line in the yeti launch to your downloaded and unzipped radar data directory path.

Then, enjoy!

$ source devel/setup.bash
$ roslaunch src/navtech-radar-slam/launch/navtech_radar_slam_mulran.launch

Examples

• The examples are from MulRan dataset, which is suitable to evaluate the radar odometry or SLAM algorithm in complex urban sites.
  • The MulRan dataset provides the oxford-radar-robotcar-radar data format (i.e., meta data such as ray-wise timestamps are imbedded in an radar image, see details here)

1. KAIST 03 of MulRan dataset

• Video (youtube link)
• Capture:

  

2. Riverside 03 of MulRan dataset

• Video (youtube link)
• Capture:

  

Related papers

If you cite this repository, please consider below papers.

• Yeti open source for radar odometry:

  @ARTICLE{burnett_ral21,
      author = {Keenan Burnett, Angela P. Schoellig, Timothy D. Barfoot},
      journal={IEEE Robotics and Automation Letters},
      title={Do We Need to Compensate for Motion Distortion and Doppler Effects in Spinning Radar Navigation?},
      year={2021},
      volume={6},
      number={2},
      pages={771-778},
      doi={10.1109/LRA.2021.3052439}}
  }
  

• Scan Context open source for place recognition:

  @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 }
  }
  

• MulRan dataset:

  @INPROCEEDINGS{ gskim-2020-mulran, 
      TITLE={MulRan: Multimodal Range Dataset for Urban Place Recognition}, 
      AUTHOR={Giseop Kim and Yeong Sang Park and Younghun Cho and Jinyong Jeong and Ayoung Kim}, 
      BOOKTITLE = { Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) },
      YEAR = { 2020 },
      MONTH = { May },
      ADDRESS = { Paris }
  }
  

TODO

• About utilities
  • support ROS-based input (topic subscription)
  • support a resulting map save functions.
• About performances
  • support reverse loop closing.
  • enhance RS (radius-search) loop closings.