This code offers a multi-motion visual odometry system for dynamic environments. If you use it in an academic work, please cite:
@article{zhang2020iros,
title={{Robust Ego and Object 6-DoF Motion Estimation and Tracking}},
author={Zhang, Jun and Henein, Mina and Mahony, Robert and Ila, Viorela},
booktitle={IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
page={5017-5023},
year={2020},
organization={IEEE}
}
An ArXiv version of this paper is available HERE.
!!! IMPORTANT NOTES: This work has been extended to a Visual Object-aware Dynamic SLAM system (VDO-SLAM), acting as the front end with many improvements. Please refer to VDO-SLAM for detailed information. The author recommends potential users try the full VDO-SLAM system with latest updates (this repository is not guaranteed to be updated as frequently as the VDO-SLAM repository).
This code is released under a GPLv3 license. As the code is developed based on the framework of ORB-SLAM2, please check a list of all code/library dependencies (and associated licenses), see Dependencies.md.
The prerequisites are the same as ORB-SLAM2. If compiling problems met, please refer to ORB-SLAM2. We have tested the library in MacOS Mojave 10.14.6, but it should be easy to compile in other platforms.
Clone the repository:
git clone https://github.com/halajun/MultMotTracking.git MultMotTracking
We provide a script build.sh to build the Thirdparty libraries and MultMotTracking. Please make sure you have installed all required dependencies. Execute:
cd MultMotTracking
chmod +x build.sh
./build.sh
This will create libMultMotTrack.dylib at lib folder and the executables rgbd_mmt in Examples folder.
We have included a small set of sample data in kitti_sample folder.
Execute the following command to get the code run on the sample data.
./Examples/RGB-D/rgbd_mmt Vocabulary/ORBvoc.txt kitti_sample/kitti03.yaml kitti_sample