HRII Visual Odometry (HRII-VO)

This repository provides tools for visual odometry experiments and teleoperation of mobile collaborative robots (MCR). The system supports multiple cameras and teleoperation interfaces, including setups involving Raspberry Pi 4 for wireless communication.

Overview

The HRII-VO framework is designed for conducting experiments with RGB-D and stereo cameras, focusing on visual odometry and remote teleoperation. It integrates rtabmap_ros for SLAM and visual-inertial odometry (VIO) algorithms, with additional configurations for ZED2i and RealSense D435i cameras.

GIF Demo

A quick GIF demonstration is also available:

Key Features

• UDP-based image transfer protocol for efficient communication.
• Support for wired and wireless camera configurations.
• Integration of VIO with rtabmap_ros for teleoperation.
• Compatibility with both stereo and RGB-D cameras.
• Experiments for odometry accuracy, comparison, and data logging.

Installation

1. Import repositories:

   git_import_repos vo_repos.yaml
   cdcb

2. Install rtabmap_ros: Follow the installation guide at rtabmap_ros GitHub.

3. For ZED2i camera support:

   • Install CUDA.
   • Follow the setup instructions in ZED ROS Wrapper.
   • Add zed-ros-wrapper to your Catkin workspace.

Usage

Launch different experiments and setups using the provided launch files:

1. Franka Motion:

   roslaunch hrii_vo vo_comparison_experiment.launch

2. Odometry Test (RGB-D and ZED2i Cameras):

   roslaunch hrii_vo All_exp_VO.launch

3. ZED2i Odometry Test with RTAB-Map:

   roslaunch hrii_vo zed_test.launch

4. RealSense D435i Odometry Test with RTAB-Map:

   roslaunch hrii_vo realsense_test.launch

5. Data Logging:

   roslaunch hrii_vo visual_connector.launch

Note:

If the system cannot handle the All_exp_VO launch file, you can launch the components individually:

roslaunch hrii_vo realsense_test.launch
roslaunch hrii_vo zed_test.launch
roslaunch hrii_vo visual_connector.launch

Teleoperation with Raspberry Pi 4

For Raspberry Pi 4 setups (running Raspbian OS), execute the scripts located in the rpi_files folder. Ensure that IP addresses and communication settings are correctly configured for UDP-based image transfer.

Experimental Scenarios

• Odometry Comparison: Run tests with both stereo and RGB-D cameras to compare odometry accuracy.
• Teleoperation Setup: Evaluate VIO performance using ZED2i and RealSense cameras in wired and wireless configurations.
• Home-Care Application: Utilize the VIO-based teleoperation interface in real-world scenarios.

References

The repository includes source code for the experiments, accessible at GitHub HRII-VO.

Citation

Citation

If you use this repository or its components in your research, please cite:

H. Raei, J. M. Gandarias, E. De Momi, P. Balatti and A. Ajoudani, "A Multipurpose Interface for Close- and Far-Proximity Control of Mobile Collaborative Robots," 2024 10th IEEE RAS/EMBS International Conference for Biomedical Robotics and Biomechatronics (BioRob), Heidelberg, Germany, 2024, pp. 457-464. doi: 10.1109/BioRob60516.2024.10719833