[JFR 2022] EIL-SLAM:Depth Enhanced Edge-based Infrared-LiDAR SLAM

This repository contains code for a Depth enhanced edge-based Infrared-LiDAR SLAM system for ROS. It integrates the open source works of rebvo and Lego-LOAM to generate this work.The system takes in point cloud from a Velodyne VLP-16 Lidar and thermal infrared images as inputs. It output 6D pose estimation in all-day.

System requirements

-- C++11

-- Linux, X11, v4l2

-- OpenGL development libraries (GL,GLU,glut)

-- PCL1.7

-- TooN 2.2 mathematical library (http://www.edwardrosten.com/cvd/toon.html - ZIP provided in the repo)

-- Lapack (for advanced TooN functions)

-- LibAV (Video Codecs)

-- Libgd (sudo apt-get install libgd-dev)

Compiling

Compiling TooN2.2

./configure && make && sudo make install

Compiling Rebvo

cd rebvo/rebvolib/

/usr/lib/x86_64-linux-gnu/qt5/bin/qmake -spec linux-g++-64 -o Makefile rebvolib.pro

cd rebvo/

/usr/lib/x86_64-linux-gnu/qt5/bin/qmake -spec linux-g++-64 -o Makefile rebvo.pro

make REBVOFLAGS=-m64

Compiling Calc2.0

The system use the Calc2.0 to infrared scene recognition. To get start, refer the Calc2.0 to configure the environment and train the model.

Compiling Ros

cd rebvo/ros/

catkin_make

Run the package

1. configure the related parameters in rebvo/ros/src/thermal_loam/cfg/rebvo_thermal.yaml
2. Run the launch file:

roslaunch rebvo run.launch

3. Play existing bag files:

rosbag play *.bag --clock --topic /velodyne_points /camera/image_raw 

if completing the configuration of Calc's environments, run the python file:bash

cd calc2.0/
python loopDetector.py

4. Datasets

https://pan.baidu.com/s/1Gq-6ztc59uVvobezGktLVQ 提取码: dgtk

Citation

If you think this work is meaningful, please cite:

@article{chen2022eil,
  title={EIL-SLAM: depth-enhanced edge-based infrared-lidar SLAM},
  author={Chen, Wenqiang and Wang, Yu and Chen, Haoyao and Liu, Yunhui},
  journal={Journal of Field Robotics},
  volume={39},
  number={2},
  pages={117--130},
  year={2022},
  publisher={Wiley Online Library}
}