Navigating the Kinematic Maze: Analyzing, Standardizing and Unifying XR Motion Datasets

This paper addresses the critical importance of standards and documentation in kinematic research, particularly within Extended Reality (XR) environments. We focus on the pivotal role of motion data, emphasizing the challenges posed by the current lack of standardized practices in XR user motion datasets. Our work involves a detailed analysis of 8 existing datasets, identifying gaps in documentation and essential specifications such as coordinate systems, rotation representations, and units of measurement. We highlight how these gaps can lead to misinterpretations and irreproducible results. Based on our findings, we propose a set of guidelines and best practices for creating and documenting motion datasets, aiming to improve their quality, usability, and reproducibility. We also created a web-based tool for visual inspection of motion recordings, further aiding in dataset evaluation and standardization. Furthermore, we introduce the XR Motion Dataset Catalogue, a collection of the analyzed datasets in a unified and aligned format. This initiative significantly streamlines access for researchers, allowing them to download partial or entire datasets with a single line of code and without the need for additional alignment efforts. Our contributions enhance dataset integrity and reliability in kinematic research, paving the way for more consistent and scientifically robust studies in this evolving field.