Onboard Quadrotor Flight using Our ESVIO as State Estimator in HDR Scenarios

Peiyu Chen, Weipeng Guan

ESVIO is published in IEEE RA-L with IROS2023 presentation option. (The IEEE RA-L pdf is available here and the arxiv pdf is available here).

@article{ESVIO,
  title={ESVIO: Event-based Stereo Visual Inertial Odometry},
  author={Chen, Peiyu and Guan, Weipeng and Lu, Peng},
  journal={IEEE Robotics and Automation Letters},
  year={2023},
  volume={8},
  number={6},
  pages={3661-3668},
  publisher={IEEE}
}

@inproceedings{EIO,
  title={Monocular Event Visual Inertial Odometry based on Event-corner using Sliding Windows Graph-based Optimization},
  author={Guan, Weipeng and Lu, Peng},
  booktitle={2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
  pages={2438-2445},
  year={2022},
  organization={IEEE}
}

If you feel like ESVIO has indeed helped in your current research or work, a simple star or citation of our works should be the best affirmation for us. 😊

Our video demos are available on Bilibili (click below images to open) and YouTube¹.

The performance evaluation videos in various datasets (including the comparisons with ORB-SLAM3, VINS-Fusion, and Ultimate-SLAM) as well as testing of outdoor large-scale and onboard drone flights can also be found on Bilibili ^{1 2 3} .

Compare ESIO with ORB-SLAM3, and Ultimate SLAM

1.1 Ubuntu 20.04 with ROS Noetic.

1.2 Ceres Solver Follow Ceres Installation, remember to make install and use the version 1.14.0 (Download Link).

1.3 we use catkin build, and all the dependency files are stored within the folder dependences.

mkdir -p catkin_ws_dvs/src
cd catkin_ws_dvs
catkin config --init --mkdirs --extend /opt/ros/noetic --merge-devel --cmake-args -DCMAKE_BUILD_TYPE=Release
cd ~/catkin_ws_dvs/src
git clone git@github.com:arclab-hku/ESVIO.git --recursive

You should modifie your .bashrc file through gedit ~/.bashrc, add the following codes in it:

source ~/catkin_ws_dvs/devel/setup.bash
alias esviobuild='cd ~/catkin_ws_dvs/src && catkin build esvio_estimator feature_tracker pose_graph -DCMAKE_BUILD_TYPE=Release -j8'

After that, run the source ~/.bashrc and esviobuild command in your terminal.

Our datasets for evaluation can be download from our One-drive or Baidu-Disk. We have released totally 9 rosbag files for evaluating ESVIO, with the introduction of these datasets can be found on this page.
For the convenience of the community, we also release the raw results of our methods in the form of rosbag (link).

After you have downloaded our bag files, you can now run our example:

• For the ESIO (event+imu) version:

roslaunch esvio_estimator esio.launch 
rosbag play YOUR_DOWNLOADED.bag

• For the ESVIO (event+image+imu) version:

roslaunch esvio_estimator esvio.launch 
rosbag play YOUR_DOWNLOADED.bag

We thanks the rpg_dvs_ros and DV ROS for their intructions of event camera driver. We add some modification for the code, and the driver code of the event camera is available in link. User can choose either one.

• Step 1: Install libcaer (add required repositories as per iniVation documentation first):

sudo apt-get install libcaer-dev

*Step 2: Create a catkin workspace and copy the driver code:

mkdir -p ~/catkin_ws_dvs/src
cd ~/catkin_ws_dvs
catkin config --init --mkdirs --extend /opt/ros/noetic --merge-devel --cmake-args -DCMAKE_BUILD_TYPE=Release`
cd ~/catkin_ws_dvs/src

And then copy the code from our link, or directly use the driver code in the dependences folder

• Step 3: Build the packages:

catkin build davis_ros_driver  (if you are using the DAVIS)
catkin build dvxplorer_ros_driver  (if you are using the DVXplorer)

source ~/catkin_ws_dvs/devel/setup.bash

• Step 4: After source your environment, you can open your event camera:

roslaunch dvs_renderer davis_mono.launch` (if you are using the DAVIS)
roslaunch dvs_renderer dvxplorer_mono.launch` (if you are using the DVXplorer)

• Step 1: Instalizing DV software libraries:

sudo add-apt-repository ppa:inivation-ppa/inivation
sudo apt update
sudo apt install dv-processing dv-runtime-dev gcc-10 g++-10

• Step 2: It is build using catkin tools, run the following commands from your catkin workspace:

mkdir -p ~/catkin_ws_dvs/src
cd ~/catkin_ws_dvs
catkin config --init --mkdirs --extend /opt/ros/noetic --merge-devel --cmake-args -DCMAKE_BUILD_TYPE=Release`
cd ~/catkin_ws_dvs/src

And then copy the code from our link.

• Step 3: Modifying your .bashrc file, add the following codes in it:

source ~/catkin_dvs_ws/devel/setup.bash

alias dvsbuild='cd ~/catkin_dvs_ws && catkin build dv_ros_accumulation dv_ros_capture dv_ros_imu_bias dv_ros_messaging dv_ros_runtime_modules dv_ros_tracker dv_ros_visualization -DCMAKE_BUILD_TYPE=Release --cmake-args -DCMAKE_C_COMPILER=gcc-10 -DCMAKE_CXX_COMPILER=g++-10'

alias dvsrun='cd ~/catkin_dvs_ws/src/Event_based_VO-VIO-SLAM/driver_code/dv-ros-master/script && sh run.sh'

• Step 4: the user can directly run the following command dvsbuild or dvsrun in the terminal to build the project and run your event camera, respectively.

Tips: Users need to adjust the lens of the camera, such as the focal length, aperture. Filters are needed for avoiding the interfere from infrared light under the motion capture system. For the dvxplorer, the sensitive of event generation should be set, e.g. bias_sensitivity. Users can visualize the event streams to see whether it is similiar to the edge map of the testing environments, and then fine-tune it.

In order to launch ESVIO on your own hardware setup, you need to have a carefully calibration of the extrinsic among Event, Image and IMU. We recommend you using the following the link (DVS-IMU Calibration and Synchronization) to kindly calibrate your sensors.
Following that, you can execute the provided roslaunch file to initialize the stereo event cameras and commence our ESVIO:

roslaunch esvio_estimator stereo_davis_open.launch
roslaunch esviio_estimator esvio.launch

In order to validate the robustness and computational efficiency of ESVIO in actual robots, we build a quadrotor which can carry stereo event cameras, as shown in below.

This work was supported by General Research Fund under Grant 17204222, and in part by the Seed Fund for Collaborative Research and General Funding Scheme-HKU-TCL Joint Research Center for Artificial Intelligence. We use (VINS-Mono) as our baseline code. Thanks Dr. Qin Tong, Prof. Shen, etc. very much.

The source code is released under GPLv3 license. We are still working on improving the code reliability. If you are interested in our project for commercial purposes, please contact Dr. Peng LU for further communication.