GIRA: Gaussian Mixture Models for Inference and Robot Autonomy

Robotics: Science and Systems XIX
Daegu, Republic of Korea

Daegu Exhibition and Convention Center (EXCO)
July 14, 2023
1330 - 1700 KST

Room: 321 or Virtual


Half-Day Tutorial
Organizers: Kshitij Goel and Wennie Tabib


[ Home | Paper | Schedule | Tutorials | Github ]


Session Time (KST) Additional Resources Speaker
Welcome (slides)
  • Opening remarks and tutorial overview
 1330 - 1340 Kshitij Goel
Introduction and Getting Started (slides)
  • An introduction to using Gaussian mixture models for environment modeling.
  • State-of-art in point cloud modeling.
  • Overview of the GIRA API.
  • GIRA: installation and setup.
 1340 - 1415 Wennie Tabib
Short Break: 5 minutes
Information-theoretic learning for adaptive point cloud compression and environment modeling (slides)
  • Using GIRA for adaptive point cloud compression via the Principle of Relevant Information (PRI).
  • GIRA on SWaP-constrained systems.
  • GIRA tools for comparing reconstruction performance.
 1420 - 1500 Kshitij Goel
Coffee Break: 30 minutes
Pose Estimation using GIRA (slides)
  • Registration techniques using GIRA.
  • Simultaneous Localization and Mapping (SLAM) with GIRA.
  • Demo: Frame-to-frame registration using GMMs.
  • Demo: Loop closure for SLAM using GMMs
 1530 - 1610 Wennie Tabib
Short Break: 5 minutes
Occupancy Modeling and Uncertainty Representation using GIRA (slides)
  • Variable-resolution occupancy modeling from GMMs.
  • Uncertainty representation using GIRA.
  • Informative planning using the occupancy models from GMMs.
  • Demo: occupancy modeling using GIRA.
 1615 - 1650 Kshitij Goel
Wrap Up (slides)
  • Roadmap for future GIRA development.
  • Further resources.
  • Closing remarks.
 1650 - 1700 Wennie Tabib