Skip to main content Skip to main navigation

Publication

CHAD TSDF - An Associative and Discretized Data Structure for TSDF SLAM

Jan Kuhlmann; Thomas Wiemann
In: Proceedings of the 2025 European Conference on Mobile Robots. European Conference on Mobile Robots (ECMR-2025), IEEE, 2025.

Abstract

The Cluster-Hashed Associative and Discretized data structure for TSDF SLAM (CHAD TSDF) is a novel representation for Truncated Signed Distance Fields (TSDFs). It is designed to maintain the association between scan poses and their corresponding TSDF values in a global map to support posegraph optimization in SLAM applications. Such an association is crucial for such applications but currently not supported by established methods. In terms of memory efficiency and access time, CHAD TSDF’s performance is comparable to the state of the art. We evaluate CHAD TSDF on the MulRan sequences and a custom large-scale dataset that features many loop closures and compare it with VDBFusion and Voxblox as references.