Soft tissue can absorb surprising amounts of energy during knee exoskeleton use
Data files
Nov 05, 2024 version files 63.10 MB
-
Data_S1.zip
53.78 KB
-
MATLAB_Data_Analysis.zip
63.04 MB
-
README.md
3.60 KB
Abstract
Soft tissue deformation at the human exoskeleton interface can deform under load to absorb, return, and dissipate the mechanical energy generated by the exoskeleton. These soft tissue effects are often not accounted for and may mislead researchers on the actual joint assistance an exoskeleton provides. We assessed the effects of soft tissue by quantifying the performance and energy distribution of a knee exoskeleton under different assistance strategies using a mechanical lower limb phantom. The phantom emulated knee kinematics and soft tissue deformation at the exoskeleton interface. We loaded the exoskeleton on the phantom under six different spring stiffness conditions. Motion capture marker and load cell data from the phantom-exoskeleton assembly allowed us to estimate the moments, stiffness, and energy contributions of the exoskeleton and physical interface to the total knee power. We found that soft tissue caused interface power to increase and exoskeleton power to decrease with increasing spring stiffness. Additionally, increases in exoskeleton peak moments were not proportional to the change in spring stiffness despite consistent phantom joint motion under all conditions. Our methodology improves the exoskeleton design process by estimating energy distribution and transfer for exoskeletons while accounting for the effects of soft tissue deformation before human testing.
This repository contains data pertaining to the synthetic phantom limb and exoskeleton study to assess the effects of soft tissue deformation on exoskeleton performance. We used a synthetic phantom limb to move a passive exoskeleton across 6 different spring stiffness conditions. The springs were rated at 3.4 kN/m, 6.1 kN/m, 9.2 kN/m, 13.6 kN/m, 18.0 kN/m, and 21.7 kN/m. We recorded kinematics from the exoskeleton, phantom limb, and spring with motion capture markers. Load cells attached to the bottom of the phantom limb recorded reaction forces. A load cell attached in series with the exoskeleton spring recorded spring force.
The processed data contains exoskeleton/phantom kinematics (deg), work (J), and power (W) for every spring stiffness condition.
Description of the data and file structure
This repository contains two zip files: "Data S1" and "MATLAB Data Analysis"
"MATLAB Data Analysis" contains a "Raw_Data" subfolder with raw experimental data titled "3.4 Marker", "3.4 Load Cell", "6.1 Marker", "6.1 Load Cell", "9.2 Marker", "9.2 Load Cell", "13.6 Marker", "13.6 Load Cell", "18.0 Marker", "18.0 Load Cell", "21.7 Load Cell", and "21.7 Marker". The number dictates the spring stiffness condition. The "Marker" data contains the X, Y (up), and Z positions of markers placed at the exoskeleton (thigh and shank), phantom (ankle, knee, and hip), and exoskeleton spring (upper and lower ends). Marker data was recorded at 100 Hz with units of millimeters. The "Load Cell" data contains raw data for three load cells attached to the ankle of the phantom limb recording vertical reaction forces and a spring load cell recording spring forces. Load cell data was recorded at 1000 Hz with units of Volts (converted to Newtons in the MATLAB processing file). The "Analysis.mat" file contains the processing pipeline for all the raw data. It outputs processed data pertaining to the exoskeleton knee angle, phantom knee angle, knee moments, work, and power. The remaining .mat files contain functions used to process the raw data.
The "Data S1" zip file contains Excel files with the processed data pertaining to the study. Each Excel file has the following format:
Columns - gait stride ID # (1 - 126)
Rows - spring stiffness (3.4, 6.1, 9.2, 13.6, 18.0, and 21.7 kN/m)
Phantom_do = phantom stance knee flexion (degrees)
Exo_do = exoskeleton stance knee flexion (degrees)
MM = knee moments (Nm)
Phantom_KK = phantom quasi-stiffness (Nm/rad)
Exo_KK = exoskeleton quasi-stiffness (Nm/rad)
Phantom_W_net = knee net work (J)
Phantom_W_neg = knee negative work (J)
Phantom_W_pos = knee positive work (J)
Exo_W_net = exoskeleton net work (J)
Exo_W_neg = exoskeleton negative work (J)
Exo_W_pos = exoskeleton positive work (J)
Gel_W_net = interface net work (J)
Gel_W_neg = interface negative work (J)
Gel_W_pos = interface positive work (J)
Spring_W_net = spring net work (J)
Spring_W_neg = spring negative work (J)
Spring_W_pos = spring positive work (J)
Frame_W_net = frame net work (J)
Frame_W_neg = frame negative work (J)
Frame_W_pos = frame positive work (J)
"Data S1" also contains an R file for statistical analysis of the processed data in the same folder. The statistical analysis uses the Krustal-Wallis test to assess the significance of the processed variables across spring conditions.
Code/Software
The code for data processing was written in MATLAB R2023a. Statistical analysis was done in R version 4.3.2
- Barrutia, W. Sebastian; Yumiceva, Ada; Thompson, Mai-Ly; Ferris, Daniel P. (2024). Soft tissue can absorb surprising amounts of energy during knee exoskeleton use. Journal of The Royal Society Interface. https://doi.org/10.1098/rsif.2024.0539