Multimodal Gait Dataset: Synchronized Sensorized-Insole (Pressure + IMU) and OptiTrack Motion Capture Recordings

Basic Information

This repository contains a multimodal dataset for human movement analysis combining instrumented insole plantar-pressure recordings and OptiTrack motion-capture data. The dataset is organized at take level so that each unit corresponds to one already segmented acquisition containing synchronized left and right insole recordings together with the associated motion-capture export for the same participant and movement.

The current release contains data from 16 participants (P1-P16) and 165 take folders in total. Most participants include ten movement takes labeled M1 to M10. In addition, five repaired OptiTrack takes are provided as interpolated variants identified with the suffix ".I".

The dataset is intended for research on gait and movement analysis, plantar-pressure processing, lower-limb kinematics, multimodal signal alignment, and benchmarking of methods that combine wearable sensing with optical motion capture.

The organization of the experimental sessions, the storage of the instrumented-insole data, and their subsequent inspection and analysis were carried out using the MVP-GAIT software platform:
https://github.com/MarcosRM02/MVP-GAIT

Experimental Overview

The dataset includes 16 adult participants identified with pseudonymous internal study IDs (P1, P2, etc.). Participant-level metadata are provided in participants.csv, including height, shoe size, weight, physical activity level, age, gender, and country. 

Participant summary:

• ⁠participants: 15
• ⁠age range: 19-30 years (young adults)
• mean age: 25.5 years
• height range: 157-180 cm
• ⁠weight range: 50-100 kg
• ⁠shoe size range: EU 37-43
• ⁠gender distribution: 7 female, 8 male

Each participant folder contains movement takes identified as M1, M2, ..., M10. The dataset preserves these original movement identifiers as used during acquisition and internal documentation. Some participants also include interpolated replacements for OptiTrack recordings affected by localized gaps, marked with the suffix ".I".

All subjects in this dataset are healthy individuals who have not been diagnosed with any gait disorders.

Dataset Structure

The dataset is organized hierarchically by participant and take:

File Contents

• participants.csv: participant-level metadata including participant code, height in centimeters, EU shoe size, weight in kilograms, physical activity level, age, gender, and country.
• Preassure_L.csv: left insole recording for one take.
• Preassure_R.csv: right insole recording for one take.
• Motion.csv: OptiTrack motion-capture export for the same take.
• Info.txt: data-quality notes describing repaired and unrepaired OptiTrack incidents.

The insole CSV files contain 32 pressure channels (PressureSensor 0 to PressureSensor 31), with pressure values ranging from 0 to 4096, along with additional acquisition variables including accelerometer and gyroscope data, center of pressure coordinates (copX, copY), and total pressure (sumP).

The motion-capture CSV files contain lower-body skeletal motion information exported from OptiTrack, including bone rotations, bone positions, and bone-marker positions for the pelvis, thighs, shanks, feet, toes, and associated markers, using the Baseline Lower marker set model: https://docs.optitrack.com/markersets/lower/baseline-lower-20.

Additionally, the root directory contains reference images illustrating the OptiTrack Baseline Lower marker configuration used for motion capture, together with images describing the physical arrangement and sensor indexing of the instrumented insole pressure sensors.

Data Quality Notes

The file info.txt documents known acquisition incidents affecting OptiTrack data.

Takes with localized OptiTrack gaps repaired by interpolation (bones and bone markers only):

• P3 M5 -> P3 M5.I
• ⁠P3 M6 -> P3 M6.I
• ⁠P13 M8 -> P13 M8.I
• ⁠P13 M9 -> P13 M9.I
• P15 M3 -> P15 M3.I

Takes with non-repairable issues:

• ⁠P3 M8: OptiTrack data contain errors that could not be corrected through localized interpolation.
• P13 M1: insole data are desynchronized due to a hardware failure.

These notes should be considered when selecting takes for analysis. When both an original take and an interpolated ".I" version are present, the ".I" folder corresponds to the repaired OptiTrack variant derived from the original acquisition.

Identifier Policy

Participant identifiers are preserved as internal pseudonymous study IDs and remain consistent with the original acquisition and internal documentation. Movement identifiers are also preserved in their original coded form (M1-M10). Interpolated repaired takes are explicitly distinguished with the suffix ".I" to maintain traceability to the original affected acquisition.

Funding

This work was funded by the Ministerio de Ciencia, Innovación y Universidades through project PDC2020-133457-I00 (SSITH: Self-recharging Sensorized Insoles for Continuous Long-Term Human Gait Monitoring) and project PID2022-142388OA-I00 (Just Move!: Early Detection of Mild Cognitive Impairment through the Analysis of Human Movement in Everyday Life), as well as by the University of Castilla-La Mancha (UCLM) through the 2022-PRED-20651 predoctoral contract and the R&D Group Support Project 2025-GRIN-38428, “STEP2MOTION: Análisis cinemático integral de la marcha humana mediante reconstrucción tridimensional del esqueleto a partir de datos de presión plantar”, and was additionally supported by the MAmI research group.