X-ICP: Localizability-Aware LiDAR Registration for Robust Localization in Extreme Environments

Turcan Tuna†,‡, Julian Nubert†, Yoshua Nava‡, Shehryar Khattak†, Marco Hutter

†Robotics Systems Lab, ETH Zürich.

‡ANYbotics A.G

PDF

Additional Data (WIll be shared upon acceptance)

Supplemantary Video

Talk

System Overview

X-ICP is a localizability-aware contrained point cloud registration method. It is platform independent and does not require heuristic environment dependent parameter tuning.
Our proposed localizability awareness module enables robust and accurate pose estimation.

Localizability Module

Robust multi-category localizability detection is achieved by analyzing the information with a histogram, doing outlier detection and summerizing the information with a decision tree.

Optimization Module

Localizability-aware constrained optimization utilizes the localizability categories and forms constraints based on these categories. Later, these constraints are used in the optimization.

Results

X-ICP is tested in various environments including open-spaces, rough underground terrain and unstructured natural environments. All the results suggests improvement in accuracy and robustness over the state-of-the-art localizability detection methods.

Seemuhle Cave VLP-16

Rumlang Construction

Seemuhle Cave OS0-128

Opfikon City Park

Computational Analysis

X-ICP is real-time applicable.

Citation

Tuna, T., Nubert, J., Nava, Y., Khattak, S., & Hutter, M. (2022). X-ICP: Localizability-Aware LiDAR Registration for Robust Localization in Extreme Environments. arXiv preprint arXiv:2211.16335.

Email: tutuna@ethz.ch

Page updated

Report abuse