Semantic Shannon Mutual Information (SSMI)

S2MI is a library for autonomous robot exploration using a stream of depth and semantic segmentation images as the input to build a semantically annotated OctoMap in real-time.

S2MI is implemented as two ROS packages, which can be built on x86-64 and ARM-based processors:

• Semantic OctoMap implementation for building probabilistic multi-class maps of an environment (SSMI-Mapping)
• Autonomous exploration using Semantic Shannon Mutual Information (SSMI-Planning)

Citation

If you found this work useful, we would appreciate if you could cite our work:

• [1] A. Asgharivaskasi, N. Atanasov. Active Bayesian Multi-class Mapping from Range and Semantic Segmentation Observations. IEEE Int. Conf. on Robotics and Automation (ICRA), 2021.

@InProceedings{Asgharivaskasi-ICRA21,
  author={Asgharivaskasi, Arash and Atanasov, Nikolay},
  booktitle={IEEE International Conference on Robotics and Automation (ICRA)}, 
  title={Active Bayesian Multi-class Mapping from Range and Semantic Segmentation Observations}, 
  year={2021},
  pages={1-7}

• [2] A. Asgharivaskasi, N. Atanasov. Semantic OcTree Mapping and Shannon Mutual Information Computation for Robot Exploration. arXiv:2112.04063 .

@InProceedings{Asgharivaskasi-journal,
 title = {Semantic OcTree Mapping and Shannon Mutual Information Computation for Robot Exploration},
 author = {A. Asgharivaskasi, N. Atanasov},
 year = {2021},
 booktitle = {arxiv},
 pdf = {https://arxiv.org/pdf/2112.04063.pdf}
}

Acknowledgments

We gratefully acknowledge support from ARL DCIST CRA W911NF-17-2-0181 and NSF FRR CAREER 2045945.

License

BSD License