TOF-VIO

April 28, 2020 · View on GitHub

ToF Camera Visual Initial Odometry

Video:

Changing Ambient Light Test	Exploration Test

Relevant publications:

Chen, S., Chang, C.-W., & Wen, C.-Y. (2020). Perception in the Dark—Development of a ToF Visual Inertial Odometry System. Sensors, 20(5), 1263. doi: 10.3390/s20051263

Introduction:

The work is the implement of the filter-based visual inertial odometry using a ToF camera input. The system has the capability to sense in the changing ambient light environment.

Prerequisites

Ubuntu + ROS
We have tested in the following environments:
Ubuntu 16.04 + ROS Kinetic
Ubuntu 18.04 + ROS melodic

Build and Run

Clone the repository to the catkin work space. E.g. /catkin_ws/src

cd ~/catkin_ws/src
git clone https://github.com/HKPolyU-UAV/TOF-VIO.git

Install ros-pcl

sudo apt-get install ros-kinetic-pcl-*
sudo apt-get install ros-melodic-pcl-*

Compile

cd ~/catkin_ws
catkin_make

Verify using Dataset

Using our recorded rosbag:

->Link1 Hand held test

->Link2 Lab test(Changing of environment lighting condition)

->Link3 UAV fly in corridor

The rosbag is compressed, depressed it before estimation.

Data Format of the rosbag

Topic Name	Content	Frequency
/image_depth	Depth image (u,v,z)	15
/image_nir	NIR image (u,v,i)	15
/points	Organized point cloud	15
/imu	IMU data	250
/gt	Ground truth captured by Vicon	50

Camera matrix and distortion coeffs of the Depth/NIR image

camera matrix		distortion coeffs
fx	211.95335	k1	0.57858
fy	211.95335	k2	-5.7317
cx	115.6517	p1	0
cy	87.125724	p2	0
		p3	10.0098

Place the .bag file into bag folder then modify the bag.launch file

<node pkg="rosbag" type="play" name="rosbag" args="$(find vio)/bag/nameofthebag.bag"/>

Run:

roslaunch tof_vio rviz.launch

roslaunch tof_vio bag.launch

Evaluation

TUM scripts can be used to evaluate the result, the following are the exported results:

Handheld test	UAV test

Experiment Platform Hardwar/Driver

ToF Camera:
PMD Flexx ToF Camera
drivers
ros wrapper
IMU:
Pixhawk
mavros

Maintainer:

Shengyang Chen(Dept.ME,PolyU): shengyang.chen@connect.polyu.hk