Human Locomotion Databases. A Systematic Review

• Published in: IEEE journal of biomedical and health informatics Vol. 28; no. 3; pp. 1 - 17
• Main Authors: David, Pinto-Fernandez, David, Rodriguez-Cianca, Juan, Moreno C., Diego, Torricelli
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.03.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Artificial Intelligence; Availability; Big Data; Biometrics; Cameras; Datasets; Gait; Healthcare; Human Locomotion; Inertial sensing devices; Legged locomotion; Locomotion; Motion Capture; Motor ability; Motor skill; Recognition; Sensors; Sociology; Statistics; Systematics; Technical literature
• ISSN: 2168-2194; 2168-2208
• DOI: 10.1109/JBHI.2023.3311677

The analysis of human locomotion is highly dependent on the quantity and quality of available data to obtain reliable evidence, due to the great variability of gait characteristics between subjects. Researchers usually have to make significant efforts to generate well-structured and trustworthy datasets. This situation is aggravated when patients are involved, due to experimental, privacy, and safety constraints. The availability of public datasets can facilitate this process. In this work, we systematically review the scientific and technical literature to identify the human locomotion databases publicly available nowadays. Within the 93 datasets identified, we observed that the most basic motor skills, e.g., flat or sloped walking, are well covered, whereas many other daily-life motor skills are poorly represented. The most common sensors used to record gait are optical motion capture systems, followed by RGB cameras and inertial sensors. We observed a lack of consistency in the data formats and limited sample size in most reviewed datasets. These issues hinder researchers from systematically standing on previous research results and represent a major barrier to using Artificial Intelligence and Big Data algorithms. With this work, we aim to provide the scientific community with a comprehensive, critical, and efficient guide to human locomotion datasets across different application domains.