Event cameras have received extensive research interest because of their advantages over conventional cameras in high-speed motion and high-dynamic-range (HDR) environments. PA-EVIO is a polarity-aided event-visual-inertial odometry system that leverages these advantages by integrating an adaptive time-surface generation module, a robust feature processing module, and a system state estimator. The system is designed to provide accurate and robust state estimation in challenging environments.

This project is an implementation related to our research works:

1. (T-IM 2026) PA-EVIO: Polarity-aided Event-Visual-Inertial Odometry with Adaptive Event Representation
2. (IROS 2024) Monocular Event-Inertial Odometry with Adaptive decay-based Time Surface and Polarity-aware Tracking

• Polarity-aided Tracking: Effectively utilizes event polarity to enhance feature tracking robustness.
• Adaptive Time Surface: Employs an adaptive decay strategy for optimal event representation.
• Robust State Estimation: Supports optimization of event, visual, and inertial measurements in V-I, E-I, and E-V-I configurations.

1. Create a workspace (if not exists):

   mkdir -p ~/pa_evio_ws/src
   cd ~/pa_evio_ws/src

2. Clone the repository:

   git clone https://github.com/APRIL-ZJU/PA-EVIO.git

3. Install Dependencies:

   sudo apt update
   sudo apt install python3-vcstool
   vcs import < dependencies.yaml

4. Build the project:

   cd ~/pa_evio_ws
   catkin_make
   source devel/setup.bash

Configuration files are organized as follows:

• System Definitions: Located in msckf-evio/config/ and msckf-evio/ov_msckf/launch/. Please verify that topic names and calibration parameters correspond to your sensor setup.
• Time Surface Parameters: Located in ts-ros/cfg.

./scripts/1run_script_davis240c.sh

@article{pa-evio2026tang,
  author={Tang, Kai and Lang, Xiaolei and Ma, Yukai and Huang, Yuehao and Gu, Yaqing and Li, Laijian and Ren, Jie and Liu, Yong and Lv, Jiajun},
  journal={IEEE Transactions on Instrumentation and Measurement}, 
  title={PA-EVIO: Polarity-aided Event-Visual-Inertial Odometry with Adaptive Event Representation}, 
  year={2026},
  volume={},
  number={},
  pages={1-1},
  keywords={Cameras;Odometry;Tracking;Accuracy;Visualization;Event detection;Robustness;Kernel;Simultaneous localization and mapping;Visual odometry;Event Camera;Sensor Fusion;State Estimation;Time-Surface;Visual-Inertial Odometry},
  doi={10.1109/TIM.2026.3652739}}

@inproceedings{meio2024tang,
  author={Tang, Kai and Lang, Xiaolei and Ma, Yukai and Huang, Yuehao and Li, Laijian and Liu, Yong and Lv, Jiajun},
  booktitle={2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}, 
  title={Monocular Event-Inertial Odometry with Adaptive decay-based Time Surface and Polarity-aware Tracking}, 
  year={2024},
  volume={},
  number={},
  pages={12544-12551},
  keywords={Power demand;Tracking;Dynamics;Cameras;Feature extraction;Robustness;Surface texture;High dynamic range;Odometry;Intelligent robots},
  doi={10.1109/IROS58592.2024.10802605}}

The implementation of this project is based on the OpenVINS framework. We also thank everyone who has helped with this work.

Thank you for your attention and support regarding this project. If you have any questions, please contact us via the following email: kaitang [at] zju [dot] edu [dot] cn