Onboard Dynamic Object Detection and Tracking for Autonomous Mobile Robots

I. Introduction

This repository contains the implementation of Dynamic Obstacle Detection and Tracking (DODT) algorithm which aims at detecting and tracking dynamic obstacles for robots with extremely constraint computational resources.

If you find this work helpful, kindly show your support by giving us a free ⭐️. Your recognition is truly valued.

This repo can be used as a standalone package and also comes as a module of our autonomy framework.

The related paper can be found on:

Zhefan Xu*, Xiaoyang Zhan*, Yumeng Xiu, Christopher Suzuki, Kenji Shimada, "Onboard dynamic-object detection and tracking for autonomous robot navigation with RGB-D camera”, IEEE Robotics and Automation Letters (RA-L), 2024. [paper] [video].

*The authors contributed equally.

introduction.mp4

II. Installation

This package has been tested on Ubuntu 18.04/20.04 LTS with ROS Melodic/Noetic on Intel Realsense D435i and NVIDIA Jetson Xavier NX, Orin NX and Intel NUC. Make sure you have installed the compatible ROS version.

# this package needs ROS vision_msgs package
sudo apt install ros-noetic-vision-msgs

cd ~/catkin_ws/src
git clone https://github.com/Zhefan-Xu/onboard_detector.git
cd ..
catkin_make

III. Run DEMO

a. Run on dataset

Please download the rosbag file from this link:

rosbag play -l multiple-objects.bag
roslaunch onboard_detector run_detector.launch

Example with single dynamic object:

single-object.mp4

Example with multiple dynamic objects:

multi-objects.mp4

b. Run on your device

Please adjust the configuration file under cfg/detector_param.yaml of your camera device. Also, change the color image topic name in scripts/yolo_detector/yolo_detector.py

From the parameter file, you can find that the algorithm expects the following data from the robot:

Depth image: /camera/depth/image_rect_raw
Robot pose: /mavros/local_position/pose
Robot odom (optional): /mavros/local_position/odom
Color image (optional if YOLO is applied): /camera/color/image_rect_raw
Aligned depth image (optional): /camera/aligned_depth_to_color/image_raw

# Launch your device first. Make sure it has the above data.
roslaunch onboard_detector run_detector.launch

IV. Issue

For ImportError: No module named yaml on Ubuntu 20.04, please run:

sudo ln -sf /usr/bin/python3 /usr/local/bin/python

V. Citation and Reference

If you find this work useful, please cite the paper:

@article{xu2023onboard,
  title={Onboard dynamic-object detection and tracking for autonomous robot navigation with RGB-D camera},
  author={Xu, Zhefan and Zhan, Xiaoyang and Xiu, Yumeng and Suzuki, Christopher and Shimada, Kenji},
  journal={IEEE Robotics and Automation Letters},
  volume={9},
  number={1},
  pages={651--658},
  year={2023},
  publisher={IEEE}
}