Lidar with Velocity

A robust camera and Lidar fusion based velocity estimator to undistort the pointcloud.

​ This repository is a barebones implementation for our paper Lidar with Velocity : Motion Distortion Correction of Point Clouds fromOscillating Scanning Lidars . It's a fusion based method to handle the oscillating scan Lidar points distortion problem, and can also provide a accurate velocity of the objects.

​ Here is a Wiki to give a brief intro about the distortion from TOF Lidar and our proposed method. For more infomation, u can also check out the paper arXiv.

1. Prerequisites

Ubuntu and ROS. Tested on Ubuntu 18.04. ROS Melodic

Eigen 3.3.4

Ceres Solver 1.14.0

Opencv 3.2.0

2. Build on ROS

Clone the repository and catkin_make:

cd ~/catkin_ws/src
git clone https://github.com/ISEE-Technology/lidar-with-velocity
cd ../
catkin_make
source ~/catkin_ws/devel/setup.bash

3. Directly run

First download the dataset and extract it.

replace the "DATASET_PATH" in config/config.yaml with your extracted dataset path (notice the "/")

Then follow the commands blow :

roslaunch object_undistort start.launch

there will be a Rviz window and a PCL Viewer window to show the results, press key "space" in the PCL Viewer window to process the next frame.