Joint Optimization of Sensor Poses and 3D Structure Boosts Robotics and Automation Performance

Strategic Insights -

Recent research by Poznan University of Technology's Institute of Robotics and Machine Intelligence has introduced a framework for simultaneous optimization of sensor poses and 3D maps, represented as surfels, in robotics and automation. The study, funded by the Polish National Agency for Academic Exchange and others, proposes a generalized LiDAR uncertainty model to address less reliable measurements in varying scenarios. Experimental results on public datasets demonstrate improved performance over most comparable state-of-the-art methods, with the system provided as open-source software to support further research.

Key Takeaways:

  • The joint optimization of sensor poses and 3D structure is fundamental for state estimation in robotics and related fields, with current LiDAR systems often prioritizing pose optimization over structure refinement.
  • The proposed framework for simultaneous optimization of sensor poses and 3D maps, represented as surfels, has been experimentally demonstrated to improve performance over most comparable state-of-the-art methods.
  • The study introduces a generalized LiDAR uncertainty model to address less reliable measurements in varying scenarios, enhancing the robustness of the system.
  • The system is provided as open-source software to support further research and development in the field of robotics and automation.
  • The research conclusions were verified through peer review, ensuring the validity and reliability of the findings.

Statistics:

  • The study was funded by the Polish National Agency for Academic Exchange (NAWA) under the STER program and the PNRR MUR Project, with additional support from the FESR Lazio 2021-2027 Program and the PUT internal fund.
  • The research was conducted by a team of researchers from the Poznan University of Technology's Institute of Robotics and Machine Intelligence, including Krzysztof Cwian, Piotr Skrzypczynski, Luca Di Giammarino, Simone Ferrari, Thomas Ciarfuglia, and Giorgio Grisetti.
  • The system provides improved performance over most comparable state-of-the-art methods, with experimental results on public datasets demonstrating enhanced accuracy and robustness.

Sources:

  • Mad-ba: 3d Lidar Bundle Adjustment - From Uncertainty Modelling To Structure Optimization, IEEE Robotics and Automation Letters, 2025;10(7):7254-7261.
  • Poznan University of Technology, Institute of Robotics and Machine Intelligence.
  • Polish National Agency for Academic Exchange (NAWA).
  • FESR Lazio 2021-2027 Program.
  • PNRR MUR Project.