The work of CLS doctoral fellow Julian Nubert was highlighted at the 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2022), held on October 23-27, 2022.
The paper “Learning-based Localizability Estimation for Robust LiDAR Localization” authored by Julian and colleagues was one of eleven entries on the short list for best paper award. Julian comments that, “Unfortunately we did not win the award in the end, but … I think it is still something cool, as the conference had way above 1000 papers.“
The work addresses the problem of LiDAR-based localization and mapping failing in environments like a tunnel because of insufficient environmental constraints present in the scene. Julian and co-authors propose a neural network-based estimation approach for detecting (non-)localizability during robot operation. To learn more, see this video about the paper.
Julian joined CLS in 2020 and is conducting his doctoral research supervised by Prof. Marco Hutter at the Robotic Systems Lab, D-MAVT, ETH Zurich and co-supervised by Dr. Katherine J. Kuchenbecker, Director at the Haptic Intelligence Department, Max Planck Institute for Intelligent Systems.
CLS and the Hutter team were further represented at IROS 2022 with a paper from Jonas Frey and colleagues on “Locomotion Policy Guided Traversability Learning using Volumetric Representations of Complex Environments.”
With the theme of “Embodied AI for a Symbiotic Society,” over 3,800 on-site participants joined the event at The Kyoto International Conference Center, Kyoto, Japan, (along with 873 online participants), more than half of whom were students and younger researchers. This was the first chance for the IROS community to come together after two years of online meetings.
Photo Credit: IROS 2022.