wolfcanli / ra-slam

2022 Real-time Semantic 3D Reconstruction for High-Touch Surface Recognition for Robotic Disinfection

Real-time Semantic Reconstruction

This project is built for the perception system of an autonomous disinfection robot. The core components of this module is a real-time 3D semantic reconstruction algorithm, with an efficient GPU implementation of Voxel Hashing which supports semmantic integration.

Official implementation of the perception system described in our IROS2022 paper Real-time Semantic 3D Reconstruction for High-Touch Surface Recognition for Robotic Disinfection pdf.

If you find this work helpful in your project, please kindly cite our work at

@inproceedings{qiu2022-real,
  title={Real-time Semantic 3D Reconstruction for High-Touch Surface Recognition for Robotic Disinfection},
  author={Qiu, Ri-Zhao and Sun, Yixiao and Marques, Joao Marcos Correia and Hauser, Kris},
  booktitle={IROS},
  year={2022},
  organization={IEEE}
}

Instllation

Please refer to INSTALL.md

For convenience, the installation instruction we wrote assumes installation to /usr/local with root access. In reality, to avoid collision with existing system packages, it is recommended to install all packages (including the ones that OpenVSLAM depends on) into a local directory such as ~/.local instead of the system path /usr/local. This can be easily achieved by adding -DCMAKE_INSTALL_PREFIX=$HOME/.local option to a CMake command. Another benifit of this is that sudo privilege is not needed for the installation, which may be desirable on multi-user servers.

Running Reconstruction

This is still WIP, but try running

./main/offline_eval

and you should see argparser working and sending you a list of arguments to put it.

TODO

• Add logic to not update TSDF when tracking is lost
• Update segmentation inference example to take in custom images
• Add rotation arthimetics (e.g. rotation matrices interpolation)
• TensorRT for saving GPU memory
• Measure optimal intervals to run semantic segmentation frames