GaTech 2025 (Domain Adaptation)¶
Overview¶
- Short Code: GT25D
- Year: 2025
- Institution: EPIC Lab (Georgia Tech)
Dataset from "Deep Domain Adaptation Eliminates Costly Data Required for Task-Agnostic Wearable Robotic Control" (Science Robotics 2025). Contains biomechanical data from 8 able-bodied adults performing 6 activity categories while wearing a hip-knee exoskeleton with 3 domain adaptation model variants (baseline, 4t4s, 0task). Includes joint kinematics, kinetics, ground reaction forces, exoskeleton torques, and IMU-derived segment angles at 200 Hz.
Downloads¶
Dataset Information¶
Subjects and Tasks¶
- Number of Subjects: 8
- Tasks Included: Backward Walking, Level Walking, Sit To Stand, Stand To Sit
Task Catalog¶
| Task | Task ID | Task Info |
|---|---|---|
| backward_walking | backward_0.6ms | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:0.6,incline_deg:0.0 |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:0.6,incline_deg:0.0 | ||
| backward_0.8ms | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:0.8,incline_deg:0.0 | |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:0.8,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:0.8,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:0.8,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:0.8,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:0.8,incline_deg:0.0 | ||
| backward_1.0ms | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:1.0,incline_deg:0.0 | |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:1.0,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:1.0,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:1.0,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:1.0,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:1.0,incline_deg:0.0 | ||
| level_walking | level_0.6ms | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:0.6,incline_deg:0.0 |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:0.6,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:0.6,incline_deg:0.0 | ||
| level_1.2ms | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:1.2,incline_deg:0.0 | |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:1.2,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:1.2,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:1.2,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:1.2,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:1.2,incline_deg:0.0 | ||
| level_1.8ms | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:1.8,incline_deg:0.0 | |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:1.8,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:1.8,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:0task,speed_m_s:1.8,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:4t4s,speed_m_s:1.8,incline_deg:0.0 | ||
| leg:r,exo_state:powered,exo_joints:hip_knee,model:baseline,speed_m_s:1.8,incline_deg:0.0 | ||
| sit_to_stand | sit_to_stand | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,incline_deg:0.0 |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,incline_deg:0.0 | ||
| stand_to_sit | stand_to_sit | leg:l,exo_state:powered,exo_joints:hip_knee,model:0task,incline_deg:0.0 |
| leg:l,exo_state:powered,exo_joints:hip_knee,model:4t4s,incline_deg:0.0 | ||
| leg:l,exo_state:powered,exo_joints:hip_knee,model:baseline,incline_deg:0.0 |
Feature Availability by Task¶
Coverage computed from converted_datasets/gtech_2025_da_phase_dirty.parquet.
Ground Reaction Forces¶
| Feature | Backward Walking | Level Walking | Sit To Stand | Stand To Sit |
|---|---|---|---|---|
grf_anterior_contra_BW |
✔ | ✔ | ✔ | ✔ |
grf_anterior_ipsi_BW |
✔ | ✔ | ✔ | ✔ |
grf_lateral_contra_BW |
✔ | ✔ | ✔ | ✔ |
grf_lateral_ipsi_BW |
✔ | ✔ | ✔ | ✔ |
grf_vertical_contra_BW |
✔ | ✔ | ✔ | ✔ |
grf_vertical_ipsi_BW |
✔ | ✔ | ✔ | ✔ |
Joint Angles¶
| Feature | Backward Walking | Level Walking | Sit To Stand | Stand To Sit |
|---|---|---|---|---|
ankle_dorsiflexion_angle_contra_rad |
✔ | ✔ | ✔ | ✔ |
ankle_dorsiflexion_angle_ipsi_rad |
✔ | ✔ | ✔ | ✔ |
foot_sagittal_angle_contra_rad |
✔ | ✔ | ✔ | ✔ |
foot_sagittal_angle_ipsi_rad |
✔ | ✔ | ✔ | ✔ |
hip_flexion_angle_contra_rad |
✔ | ✔ | ✔ | ✔ |
hip_flexion_angle_ipsi_rad |
✔ | ✔ | ✔ | ✔ |
knee_flexion_angle_contra_rad |
✔ | ✔ | ✔ | ✔ |
knee_flexion_angle_ipsi_rad |
✔ | ✔ | ✔ | ✔ |
shank_sagittal_angle_contra_rad |
✔ | ✔ | ✔ | ✔ |
shank_sagittal_angle_ipsi_rad |
✔ | ✔ | ✔ | ✔ |
thigh_sagittal_angle_contra_rad |
✔ | ✔ | ✔ | ✔ |
thigh_sagittal_angle_ipsi_rad |
✔ | ✔ | ✔ | ✔ |
Joint Moments¶
| Feature | Backward Walking | Level Walking | Sit To Stand | Stand To Sit |
|---|---|---|---|---|
ankle_dorsiflexion_assistance_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
ankle_dorsiflexion_assistance_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
ankle_dorsiflexion_biological_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
ankle_dorsiflexion_biological_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
ankle_dorsiflexion_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
ankle_dorsiflexion_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
hip_flexion_assistance_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
hip_flexion_assistance_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
hip_flexion_biological_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
hip_flexion_biological_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
hip_flexion_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
hip_flexion_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
knee_flexion_assistance_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
knee_flexion_assistance_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
knee_flexion_biological_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
knee_flexion_biological_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
knee_flexion_moment_contra_Nm_kg |
✔ | ✔ | ✔ | ✔ |
knee_flexion_moment_ipsi_Nm_kg |
✔ | ✔ | ✔ | ✔ |
Joint Velocities¶
| Feature | Backward Walking | Level Walking | Sit To Stand | Stand To Sit |
|---|---|---|---|---|
ankle_dorsiflexion_velocity_contra_rad_s |
✔ | ✔ | ✔ | ✔ |
ankle_dorsiflexion_velocity_ipsi_rad_s |
✔ | ✔ | ✔ | ✔ |
foot_sagittal_velocity_contra_rad_s |
✔ | ✔ | ✔ | ✔ |
foot_sagittal_velocity_ipsi_rad_s |
✔ | ✔ | ✔ | ✔ |
hip_flexion_velocity_contra_rad_s |
✔ | ✔ | ✔ | ✔ |
hip_flexion_velocity_ipsi_rad_s |
✔ | ✔ | ✔ | ✔ |
knee_flexion_velocity_contra_rad_s |
✔ | ✔ | ✔ | ✔ |
knee_flexion_velocity_ipsi_rad_s |
✔ | ✔ | ✔ | ✔ |
shank_sagittal_velocity_contra_rad_s |
✔ | ✔ | ✔ | ✔ |
shank_sagittal_velocity_ipsi_rad_s |
✔ | ✔ | ✔ | ✔ |
thigh_sagittal_velocity_contra_rad_s |
✔ | ✔ | ✔ | ✔ |
thigh_sagittal_velocity_ipsi_rad_s |
✔ | ✔ | ✔ | ✔ |
Other Features¶
| Feature | Backward Walking | Level Walking | Sit To Stand | Stand To Sit |
|---|---|---|---|---|
cop_anterior_contra_m |
✔ | ✔ | ✔ | ✔ |
cop_anterior_ipsi_m |
✔ | ✔ | ✔ | ✔ |
cop_lateral_contra_m |
✔ | ✔ | ✔ | ✔ |
cop_lateral_ipsi_m |
✔ | ✔ | ✔ | ✔ |
cop_vertical_contra_m |
✔ | ✔ | ✔ | ✔ |
cop_vertical_ipsi_m |
✔ | ✔ | ✔ | ✔ |
Data Structure¶
- Format: Phase-normalized (150 points per gait cycle)
- Sampling: Phase-indexed from 0-100%
- Variables: Standard biomechanical naming convention
Validation Snapshot¶
- Status: ⚠️ Partial (90.5%)
- Stride Pass Rate: 90.5%
- Validation Ranges: Download validation ranges (source: contributor_tools/validation_ranges/default_ranges.yaml)
- Detailed Report: View validation report
Citation¶
Scherpereel, K.L. et al. Deep domain adaptation eliminates costly data required for task-agnostic wearable robotic control. Science Robotics (2025). https://doi.org/10.1126/scirobotics.ado6509
Collection Details¶
Protocol¶
Data collected from 8 able-bodied adults wearing a clothing-integrated hip-knee exoskeleton. Subjects performed 6 activity categories: level walking (0.6, 1.2, 1.8 m/s on treadmill), backward walking (0.6, 0.8, 1.0 m/s), stair ascent/descent, sit-to-stand transfers, ball tossing, and cutting maneuvers. Three domain adaptation model variants were tested per subject: baseline (semisupervised), 4t4s (semisupervised with domain adaptation), and 0task (unsupervised). Model variant is stored in task_info as model:
Processing Notes¶
Gait cycles detected using vertical GRF threshold crossing (20 N). First 2 and last 1 strides per trial excluded as transitions. IQR-based filtering removes outlier stride durations. Sit-to-stand and stand-to-sit transfers segmented using GRF state machine. Data phase-normalized to 150 points per cycle. Stair ascent/descent trials were too short for stride extraction with current filtering. All data collected with powered exoskeleton. Source data from Georgia Tech Digital Repository.
Files Included¶
converted_datasets/gtech_2025_da_phase_dirty.parquet— Phase-normalized dataset- Validation report
- Conversion script in
contributor_tools/conversion_scripts/gt25d/
Generated by Dataset Submission Tool on 2026-02-05 14:38
Report generated: 2026-02-05 14:38
Status Summary¶
| Metric | Value |
|---|---|
| Stride Pass Rate | 90.5% |
| Total Strides | 5470 |
| Passing Strides | 4948 |
Summary¶
Status: ⚠️ PARTIAL (90.5% valid)
Total Strides: 5470
Passing Strides: 4948
Task Breakdown¶
| Task | Pass Rate | Status |
|---|---|---|
| Backward Walking | 85.7% | ⚠️ |
| Level Walking | 97.6% | ✅ |
| Sit To Stand | 89.0% | ⚠️ |
| Stand To Sit | 85.5% | ⚠️ |
Validation ranges snapshot embedded below.
Validation Ranges Snapshot¶
Download the YAML snapshot used for this validation: Download
Source ranges file: contributor_tools/validation_ranges/default_ranges.yaml
Validation Plots¶
Generated from converted_datasets/gtech_2025_da_phase_dirty.parquet on 2026-02-05 14:38