A public dataset of treadmill walking kinematics and kinetics at steady speeds in healthy individuals Authors: L Le Goff, F Chorin, N Reneaud, D Fontaine, E Piche Published: 2026 ABSTRACT AbstractHuman motion analysis is widely used in clinical practice to characterise motor function, gait impairments, and fall risk. However, available gait datasets are predominantly acquired at self-selected walking speeds and rarely include standardised slower speeds, despite their relevance for clinical populations. This article describes a publicly available dataset of three-dimensional gait kinematics and kinetics collected in healthy adults under controlled treadmill conditions.Forty-five participants, including young and older adults, performed walking trials at six predefined speeds ranging from 0.6 to 1.6 m·s⁻¹. Data acquisition was conducted using an optoelectronic motion capture system combined with an instrumented treadmill. The dataset includes three-dimensional marker trajectories, pelvis and lower limb joint kinematics in the sagittal, frontal, and transverse planes, as well as ground reaction forces, joint moments, and joint power. Raw data are provided in C3D format, and processed data are available as MATLAB-compatible files containing time-normalised gait cycles with associated mean values and variability metrics.The dataset is structured to allow comparisons across walking speeds and age groups, with consistent acquisition protocols and standardised processing pipelines. Supplementary files include statistical descriptors and detailed biomechanical variables across the gait cycle. In addition, a composite metric derived from gait kinematic data across multiple speeds is provided to support integrative analyses of gait patterns.This dataset can be reused for methodological development, validation of biomechanical models, and comparative analyses in both research and clinical contexts. The inclusion of standardised slow walking speeds and age-stratified data enables its use as a reference framework for evaluating gait patterns under conditions commonly encountered in clinical populations, including longitudinal assessments and comparisons with altered locomotor function.Keywords: Gait analysis, Motion capture, Ground reaction forces, Locomotion biomechanics, Normative reference, Longitudinal assessment Specifications Table SubjectHealth Sciences, Medical Sciences & PharmacologySpecific subject areaThree-dimensional gait biomechanics at six standardised walking speeds including slow locomotionType of dataTable; Figure; Graph;Raw; Processed; Filtered;C3D files; MAT files; XLS filesData collectionData were acquired using a 9-camera optoelectronic motion capture system (OptiTrack Motive 2.2.0, NaturalPoint, USA) at 100 Hz and an instrumented dual-belt treadmill (Bertec, USA) at 1000 Hz. Forty-five healthy adults walked barefoot at six predefined speeds (0.6–1.6 m·s⁻¹). Marker trajectories were filtered (6 Hz), ground reaction forces (20 Hz), and processed in MATLAB and OpenSim 4.1 using a scaled musculoskeletal model. Exclusion criteria included neurological or orthopedic disorders.Data source locationUniversity Côte d’Azur, Nice, FranceData accessibilityRepository name: Mendeley DataData identification number: 10.17632/3dmymnb65h.1Direct URL: https://data.mendeley.com/datasets/3dmymnb65h/1Instructions for accessing these data: The dataset is available via the DOI link provided above.The code used for data processing and biomechanical analysis is publicly available on GitHub: https://github.com/LEGOFFL/Code-public-dataset-of-treadmill-walking-at-steady-speeds-in-healthy-individualsRelated research articleNone Value of the Data •This dataset provides a comprehensive collection of three-dimensional gait kinematics and kinetics in healthy adults across six standardised treadmill walking speeds (0.6 to 1.6 m·s⁻¹), including slower speeds that are underrepresented in existing datasets. It includes both young and older individuals, along with raw (.c3d) and processed (.mat) data, enabling detailed biomechanical analyses across the full gait cycle. This dataset provides a comprehensive collection of three-dimensional gait kinematics and kinetics in healthy adults across six standardised treadmill walking speeds (0.6 to 1.6 m·s⁻¹), including slower speeds that are underrepresented in existing datasets. It includes both young and older individuals, along with raw (.c3d) and processed (.mat) data, enabling detailed biomechanical analyses across the full gait cycle. •The dataset includes gait data collected at six predefined walking speeds, allowing comparisons under controlled and reproducible conditions. Given the strong influence of walking speed on gait biomechanics, the availability of multiple standardised speeds enables more appropriate matching between reference and individual data, particularly when self-selected walking speeds differ across subjects. The dataset can be reused for the development and validation of computational models, including musculoskeletal simulations, machine learning algorithms, and signal processing methods. The availability of synchronised kinematic and kinetic data supports advanced modelling approaches and methodological developments in biomechanics. The dataset includes gait data collected at six predefined walking speeds, allowing comparisons under controlled and reproducible conditions. Given the strong influence of walking speed on gait biomechanics, the availability of multiple standardised speeds enables more appropriate matching between reference and individual data, particularly when self-selected walking speeds differ across subjects. The dataset can be reused for the development and validation of computational models, including musculoskeletal simulations, machine learning algorithms, and signal processing methods. The availability of synchronised kinematic and kinetic data supports advanced modelling approaches and methodological developments in biomechanics. •The inclusion of both young and older adults allows the dataset to be used for age-stratified analyses. Researchers can explore differences in gait patterns across age groups and use these data as reference values for studies involving ageing populations or age-related functional changes. The inclusion of both young and older adults allows the dataset to be used for age-stratified analyses. Researchers can explore differences in gait patterns across age groups and use these data as reference values for studies involving ageing populations or age-related functional changes. •The dataset may be used as a reference framework for longitudinal and interventional analyses, including pre- and post-intervention comparisons. The availability of multiple walking speeds allows researchers and clinicians to select appropriate reference conditions when evaluating changes in gait performance over time. The dataset may be used as a reference framework for longitudinal and interventional analyses, including pre- and post-intervention comparisons. The availability of multiple walking speeds allows researchers and clinicians to select appropriate reference conditions when evaluating changes in gait performance over time. Background Quantitative gait analysis is increasingly used in clinical and research settings to characterise locomotor function using three-dimensional kinematic and kinetic measurements. These analyses rely on motion capture systems and force platforms to provide objective biomechanical data across the gait cycle and are widely used to evaluate gait impairments, monitor functional changes, and support clinical decision-making [[1], [2], [3]]. However, the interpretation of gait data relies on the availability of appropriate normative reference datasets. Existing gait datasets reported in the literature differ considerably according to the characteristics of the studied population (age, healthy individuals, or patients with specific conditions) and the methodological approaches used for gait assessment. These differences include walking conditions (overground or treadmill walking), walking speed (self-selected or imposed), distance covered, and the motion capture technologies or sensor systems used to record lower-limb movements. Consequently, the comparison and interpretation of gait data across studies remain challenging, underscoring the need for well-characterised normative reference datasets [4]. Walking speed is a major determinant of gait biomechanics, influencing lower-limb kinematics, joint moments, joint powers, and ground reaction forces. Consequently, meaningful comparisons between individuals or populations require consideration of walking speed. Nevertheless, many publicly available gait datasets rely on self-selected walking speeds, which vary across individuals and may reduce the reproducibility of biomechanical comparisons. Although several high-quality open-access gait databases have recently become available [[5], [6], [7], [8], [9], [10], [11]], few provide data acquired at identical predefined walking speeds for all participants. Furthermore, datasets including several standardised slow walking speeds remain scarce, despite the fact that reduced walking speed is commonly observed in older adults and in individuals with pain or locomotor disorders. Standardised speed conditions may therefore facilitate more meaningful comparisons between normative and clinical populations. Age is another important factor influencing gait biomechanics. Older adults commonly exhibit alterations in joint mobility, propulsion, postural control, and medio-lateral stability compared with younger adults [12]. Consequently, reference datasets including both young and older adults are useful for distinguishing age-related adaptations from pathological gait deviations. While recent open-access datasets have substantially expanded the availability of normative gait data [[5], [6], [7], [8], [9], [10], [11]], datasets combining age-stratified populations with multiple predefined walking speeds, including clinically relevant slow speeds, remain limited. This dataset provides raw and processed three-dimensional gait kinematics, kinetics, and ground reaction force data from healthy young and older adults walking barefoot on an instrumented treadmill at six predefined speeds ranging from 0.6 to 1.6 m·s⁻¹. By using identical speed conditions for all participants, including multiple slow walking speeds, the dataset provides a reproducible reference framework for speed-matched comparisons in both young and older healthy adults. Data Description The dataset is organised into two main folders corresponding to participant groups: “Young” and “Older”. Within each group folder, each participant is stored in an individual subfolder identified by a unique code (e.g., “01_LLG”, “02_NR”). Participant demographic and anthropometric information are provided in Table 1. Table 1Anthropometric information of included participants.Table 1 dummy alt textFile nameAge (year)Body mass (kg)Height (cm)Sex01_LLG3372178Male02_NR2870173Male03_RC2579184Male04_EP2660169Female05_FC3799189Male06_KS2476176Female07_LD2670179Female08_CF4655159Female09_LV2960160Female10_CF2872167Female11_YD2365182Female12_LF2350165Female13_TP3087198Male14_VP3665176Female15_CB2860172Male16_BC6358160Female17_TC6082182Male18_HJ6950158Female19_AO6954156Female20_OP6970163Female21_JMM6463172Male22_FM6665162Female23_RLL6669170Female24_RC7054162Female25_MM6658164Female26_EM5871169Female27_VL5048164Female28_AC6551162Female29_MDV6545160Female30_VD6477165Female31_SZ5156156Female32_AL2261175Male33_MH2180187Male34_AV2766168Female36_AB2161170Female35_DLG6875180Male37_SD2166173Female38_EPG2464167Female39_XB3497195Male40_VL2581185Male41_MP5257164Female42_CM5066153Female43_FD4072180Male44_PM5859158Male45_MC4854158Female Anthropometric information of included participants. Within each participant subfolder, seven raw data files in C3D format are provided. These include six walking trials corresponding to predefined treadmill speeds: “0_6.c3d”, “0_8.c3d”, “1_0.c3d”, “1_2.c3d”, “1_4.c3d”, and “1_6.c3d”, as well as one static calibration file named “static.c3d”. Each C3D file contains three-dimensional marker trajectories and synchronised ground reaction force signals. The structure of the force plate signals and analog channels is detailed in Supplementary Table 1. Processed data are provided in MATLAB (.mat) format within the same participant subfolders. For each walking speed, one file is available (e.g., “0_60norm_data.mat”). Each file contains time-normalised data over 100 points of the gait cycle (0–100%) for both right and left sides. The variables include pelvis kinematics (“TILT”, “OBLIQUITY”, “ROTATION”), lower-limb joint kinematics (hip, knee, ankle), ground reaction forces (“Fx”, “Fy”, “Fz”), force plate moments (“Mx”, “My”, “Mz”), joint moments (“M_hip”, “M_knee”, “M_ankle”), and joint powers (“P_hip”, “P_knee”, “P_ankle”). Each variable is provided with mean values and variability using “mean”, “std_plus”, and “std_less”. Table 2 lists the anatomical location of the 24 reflective markers used for data acquisition and specifies the corresponding body segment or joint. Fig. 1 illustrates the marker-set protocol based on the modified IOR Leardini model [13] and provides a visual representation of the marker locations used to generate the kinematic variables contained in the dataset. Table 2Marker locations.Table 2 dummy alt textMarker nameAnatomical locationSegment or jointLTOELATleft head fifth metatarsal boneleft footRTOELATright head fifth metatarsal boneright footLTOEMEDleft head first metatarsal boneleft footRTOEMEDright head first metatarsal boneright footLANKLELATleft lateral malleolusleft footRANKLELATright lateral malleolusright footLANKLEMEDleft medial malleolusleft footRANKLEMEDright medial malleolusright footLHEELleft heelleft footRHEELright heelright footLTIBleft tibial tuberosityleft kneeRTIBright tibial tuberosityright kneeLKNEELATleft lateral condyle of femurleft kneeRKNEELATright lateral condyle of femurright kneeLKNEEMEDleft medial condyle of femurleft kneeRKNEEMEDright medial condyle of femurright kneeLFIBleft head of fibulaleft kneeRFIBright head of fibularight kneeLHIPleft great trochanterleft hipRHIPright great trochanterright hipLASISleft anterior superior iliac spinepelvisRASISright anterior superior iliac spinepelvisLPSISleft posterior superior iliac spinepelvisRPSISright posterior superior iliac spinepelvis Marker locations. Marker-set protocol based on the modified IOR Leardini model. Supplementary Table 2 contains statistical values (p-values and Cohen’s d) for 17 kinematic and kinetic variables across the gait cycle (1–100%) and for each walking speed. Experimental Design, Materials and Methods Participants A total of 55 healthy adults were recruited. Due to a synchronisation module malfunction, complete datasets were available for 45 participants, which constitute the dataset provided in this repository. The final dataset includes 25 young adults (14 females, 11 males; mean age: 29 ± 7.5 years; mean height: 175 ± 11 cm; mean weight: 69.7 ± 12.4 kg) and 20 older adults (16 females, 4 males; mean age: 62 ± 6.8 years; mean height: 164 ± 7.5 cm; mean weight: 61.4 ± 10.2 kg) (Table 1). All participants provided written informed consent prior to participation. The study protocol was approved by the local institutional review board (approval number: 2023-BS-546) and conducted in accordance with the Declaration of Helsinki. Experimental setup Data acquisition was performed using an optoelectronic motion capture system composed of nine infrared cameras (OptiTrack, NaturalPoint, Inc., USA) sampling at 100 Hz. Ground reaction forces were recorded simultaneously using a dual-belt instrumented treadmill (Bertec, Columbus, OH, USA) at a sampling frequency of 1000 Hz. The acquisition system was controlled using Motive software (version 2.2.0, NaturalPoint, Inc.). A modified IOR Leardini marker set was used, consisting of 24 reflective markers placed on anatomical landmarks of the pelvis and lower limbs (Table 2) as illustrated in Fig. 1. Markers were positioned directly on the skin whenever possible or on tight-fitting clothing to minimise soft tissue artefacts. Experimental procedure Each participant completed a single acquisition session of approximately 30 min. Participants walked barefoot on the instrumented treadmill while wearing shorts and a fitted T-shirt. Prior to data collection, participants underwent a familiarisation period of at least 5 min to adapt to treadmill walking. A static calibration trial was first recorded to define the biomechanical model. Participants then performed walking trials at six predefined speeds: 0.6, 0.8, 1.0, 1.2, 1.4, and 1.6 m·s⁻¹. For each speed, participants walked for 60 s, including 30 s of familiarisation followed by 30 s of data acquisition. Participants were instructed to walk naturally without holding the treadmill handrails. System calibration and acquisition parameters The motion capture system was calibrated prior to each session according to the manufacturer’s recommendations. This included definition of the global coordinate system, dynamic camera calibration, and zeroing of force plates. Marker trajectories were recorded in three dimensions and synchronised with ground reaction force data. Data processing Raw data were exported in C3D format and processed using MATLAB (MathWorks, USA). Marker trajectories and ground reaction forces were filtered using a fourth-order Butterworth filter with cut-off frequencies of 6 Hz and 20 Hz, respectively. A threshold of 20 N was applied to ground reaction forces to remove noise associated with treadmill artefacts. Processed data were imported into OpenSim (version 4.1) for biomechanical modelling. The Rajagopal musculoskeletal model was used and scaled to each participant based on anthropometric measurements obtained during the static trial [14]. Inverse kinematics and inverse dynamics analyses were performed to compute joint kinematics, joint moments, and joint powers. Time normalisation and signal processing Gait cycles were identified based on ground reaction force thresholds. Data were segmented into individual gait cycles and time-normalised to 100 points (0–100% of the gait cycle). For each variable, mean values and variability were computed across cycles. Software and code availability Data processing and analysis were performed using MATLAB and OpenSim. Custom MATLAB scripts were used for signal processing, filtering, cycle detection, and data normalisation. These scripts are publicly available in the associated GitHub repository (https://github.com/LEGOFFL/Code-public-dataset-of-treadmill-walking-at-steady-speeds-in-healthy-individuals). Limitations For one older participant, data are not available at the highest walking speeds (1.4 and 1.6 m·s⁻¹), as these conditions could not be completed during acquisition. Data were collected during treadmill walking under controlled laboratory conditions, which differ from overground walking environments. Although both young and older adults are represented, the sample size within each group remains limited. Data acquisition and processing were performed using a specific motion capture system and musculoskeletal modelling framework, which may influence compatibility with datasets generated using different systems or modelling approaches. Ethics Statement This study involved human participants. All participants provided written informed consent prior to data collection. The study protocol was approved by the local ethics committee (approval number: 2023-BS-546) and conducted in accordance with the principles of the Declaration of Helsinki. CRediT Author Statement Ludovic Le Goff: Conceptualization, Methodology, Data curation, Formal analysis, Investigation, Writing - original draft preparation. Frédéric Chorin: Methodology, Supervision, Writing - review & editing. Nicolas Reneaud: Software, Data curation, Formal analysis. Denys Fontaine: Supervision, Writing - review & editing. Elodie Piche: Conceptualization, Methodology, Writing - original draft. Declaration of AI use in the Writing Process During the preparation of this work, the authors used ChatGPT (OpenAI) to assist with language editing. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article. Acknowledgments The authors would like to thank all participants for their time and involvement in this work. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Footnotes Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.dib.2026.112943. Supplementary materials Data Availability Mendeley DataA public dataset of treadmill walking kinematics and kinetics at steady speeds in healthy individuals (Original data) Mendeley DataA public dataset of treadmill walking kinematics and kinetics at steady speeds in healthy individuals (Original data) References Associated Data Supplementary Materials Data Availability Statement Mendeley DataA public dataset of treadmill walking kinematics and kinetics at steady speeds in healthy individuals (Original data) Mendeley DataA public dataset of treadmill walking kinematics and kinetics at steady speeds in healthy individuals (Original data)