University of Michigan LocoLab

Data & Code

Open datasets, controllers, and hardware designs

Our experimental data and designs are released publicly so that other groups can reproduce our analyses and build on our work. Each record below links to the repository and to the article it supports.

Datasets

Datasets

Human and device data from our experimental protocols, released with the article each dataset supports. Institutional Review Board approval details for the human-subjects studies appear in the individual records.

A participant wearing motion-capture markers walks on an instrumented treadmill during the ankle exoskeleton study.

Optimal Energy Shaping and Force Amplification Framework for Task-Agnostic, Biomimetic Ankle Exoskeletons

This dataset contains joint kinematics and kinetics together with ankle exoskeleton control signals (where applicable) from an experimental protocol approved by the Institutional Review Board at the University of Michigan. These data were collected to evaluate the ankle exoskeleton control biomimicry and the lower-limb joint biomechanics when receiving assistance from the powered ankle exoskeleton compared to without the device. The dataset includes time-normalized strides across activities of daily living, including level, incline, and decline walking, stair ascent and descent, sit-stand, and crouching.

Associated publicationK. Walters, G. Thomas, and R. Gregg, "Optimal Energy Shaping and Force Amplification Framework for Task-Agnostic, Biomimetic Ankle Exoskeletons," IEEE Transactions on Robotics, 2026.
A conventional microprocessor-controlled knee prosthesis compared with the Össur Power Knee.

The clinical effects of the Össur Power Knee with phase-based and default control during sitting, standing, and walking

The dataset was collected to assess the clinical benefits of the latest generation Össur Power Knee with and without phase-based control compared to traditional passive microprocessor-controlled knees. The study included seven above-knee amputee participants and examined performance across a series of sit/stand and walking experiments. Data captured includes event logs, processed inverse kinematics and dynamics data, scaled OpenSim skeletons for each session, and raw motion capture and force plate data for various test conditions (HKIC, OSSR, PRES). All data is text-based and can be viewed in any compatible text editor. Data can be visualized in OpenSim. The dataset is organized by subject and test condition. Each folder includes a scaled subject OpenSim skeleton (*_skeleton.osim), a marker file for a range of motion trial (*_range_of_motion.trc), and three test folders each containing: Event Logs (*.event_log.csv): Time-stamped records of experimental events. Inverse Kinematics Data (*_IK.mot): Joint angle trajectories. Inverse Dynamics Data (*_ID.mot): Joint moment trajectories. Raw Marker Data (*_markers.csv): 3D coordinates of motion capture markers. Ground Reaction Force Data (*_GRF.mot): Force plate measurements of ground reactions during movement tasks.

Associated publicationT. K. Best, C. A. Seelhoff, J. Wensman, and R. Gregg, "The clinical effects of the Össur Power Knee with phase-based and default control during sitting, standing, and walking," J NeuroEngineering and Rehabilitation, 2025.
A participant wearing the versatile knee exoskeleton climbs outdoor steps.

A versatile knee exoskeleton mitigates quadriceps fatigue in lifting, lowering, and carrying tasks

Data collected from 10 able-bodied participants performing fatiguing (S1) and non-fatiguing (S2) lifting-lowering-carrying (LLC) tasks with (exo condition) and without (bare condition) a bilateral knee exoskeleton. LLC tasks consist of squat lifting-lowering (LL), ramp ascent (RA), ramp descent (RD), stairs ascent (SA), stairs descent (SD), and level walking (LW). Ramp incline is 15 degrees and step height of stairs is 7 inches. dataset_S1 consists performance and posture measurements from fatiguing squat LL, and perceptual measurements from fatigued LLC tasks. dataset_S2 consists of electromyography, kinematics, torque, and foot sensor data from non-fatiguing LLC tasks.

Associated publicationN. Divekar, G. Thomas, A. Yerva, H. Frame, and R. Gregg, "A versatile knee exoskeleton mitigates quadriceps fatigue in lifting, lowering, and carrying tasks," Science Robotics, 2024.
Diagram of the activity states captured in the dataset (sit, walk, run, stairs) and the transitions between them.

Lower-limb Kinematics and Kinetics During Continuously Varying Human Locomotion

This dataset reports the lower-limb kinematics and kinetics of ten able-bodied subjects walking at multiple inclines (+/- 0, 5, and 10 deg) and speeds (0.8 m/s, 1 m/s, and 1.2 m/s), running over level-ground at multiple speeds (1.8 m/s, 2 m/s, 2.2 m/s, and 2.4 m/s), walking and running with constant acceleration and deceleration (+/- 0.2 m/s2, and 0.5 m/s2), and stair ascent/descent with multiple stair inclines (+/- 20, 25, 30, and 35 deg). This dataset also includes sit-stand transitions, walk-run transitions, and walk-stairs transitions. Data were recorded by a Vicon motion capture system and, for applicable tasks, a Bertec instrumented treadmill. This dataset can aid in the development of kinematic models of multi-activity human locomotion and the design and control of agile wearable robots.

Associated publicationE. Reznick, K. Embry, R. Neuman, E. Bolivar, N. Fey, and R. Gregg, "Lower-Limb Kinematics and Kinetics During Continuously Varying Human Locomotion," Scientific Data, 2021.
A participant with a powered knee-ankle prosthesis and motion-capture markers walks on a treadmill.

Effects of a Powered Knee-Ankle Prosthesis on Amputee Hip Compensations: A Case Series

This dataset was collected to investigate the effects of a robotic prosthetic leg on amputee hip compensations during level-ground treadmill walking experiments. The dataset specifically contains data relating to Vicon (human lower-limb kinematics and kinetics), EMG, and robotic prosthetic data from an experimental protocol approved by the Institutional Review Board at the University of Texas at Dallas. Three subjects each walked on a Bertec instrumented treadmill for approximately 60 seconds with their day-to-day passive prosthesis and the powered prosthesis at their self-selected slow, normal, and fast walking speeds, resulting in a total of 6 walking trials per subject. During each trial, a 10-camera Vicon motion capture system recorded leg kinematics, while force plates in the Bertec treadmill recorded ground reaction forces, and a Delsys Trigno EMG system recorded muscle activation. The collected data has been divided into strides, and is interpolated to contain 100 points per stride. Strides begin at heel contact. Gait kinetics have been normalized by user mass.

Associated publicationT. Elery, S. Rezazadeh, E. Reznick, L. Gray, and R. Gregg, "Effects of a Powered Knee-Ankle Prosthesis on Amputee Hip Compensations: A Case Series," IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2020.
A participant with motion-capture markers and blue joint wraps walks on the treadmill used for the perturbation trials.

Human Leg Kinematics, Kinetics, and EMG During Quasi-random, Phase-shifting, Treadmill Perturbations at Varying Inclines

This dataset contains joint kinematics, kinetics, and EMG activity from an experimental protocol approved by the Institutional Review Board at the University of Texas at Dallas. This data was collected to evaluate the robustness of different parameterization variables during perturbations for application in robotic prosthetic legs. Ten able-bodied subjects self-selected a comfortable speed for walking on level (0 degree), +5 degree, and -5 degree inclines. Subjects walked at the self-selected speed for a minute without perturbations to produce a control dataset of unperturbed kinematics. From this control data, the average stance time was calculated for each subject to define a normalized time window between 0 and 80% of stance. Then, 100 uniformly distributed times were sampled from this window to determine the perturbation onset times, i.e., the amount of delay between heel strike and perturbation onset. For perturbation trials, subjects walked at the same speed for 20-25 minutes, broken into 5 sets of 4-5 minutes for each slope. During each trial, a 10-camera Vicon motion capture system recorded leg kinematics, while force plates in a Bertec split-belt treadmill recorded ground reaction forces, and a Delsys Trigno EMG system recorded muscle activation of the rectus femoris, biceps femoris, tibialis anterior, and gastrocnemius.

Associated publicationR. Macaluso, K. Embry, D. Villarreal, R. Gregg, "Parameterizing Human Locomotion Across Quasi-Random Treadmill Perturbations and Inclines," IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2021.
Motion-capture marker trajectories overlaid on a participant walking on the instrumented treadmill.

The Effect of Walking Incline and Speed on Human Leg Kinematics, Kinetics, and EMG

This dataset contains leg joint kinematics, kinetics, and EMG activity from an experimental protocol approved by the Institutional Review Board at the University of Texas at Dallas. Ten able-bodied subjects walked at steady speeds and inclines on a Bertec instrumented treadmill for one minute per trial. Each subject walked at every combination of the speeds 0.8 m/s, 1.0 m/s, and 1.2 m/s and inclines from -10 degrees to +10 degrees at 2.5 degree increments, for a total of 27 trials. During each trial, a 10-camera Vicon motion capture system recorded leg kinematics, while force plates in the Bertec treadmill recorded ground reaction forces, and a Delsys Trigno EMG system recorded muscle activation of the rectus femoris, biceps femoris, tibialis anterior, and gastrocnemius. This data can be used to test many hypotheses and models of human locomotion at varying speeds and inclines. This kinematic data has been used to train a predictive model that represents gait kinematics as a continuous function of gait cycle percentage, incline, and speed:

Associated publicationK. Embry, D. Villarreal, R. Macaluso, and R. Gregg, "Modeling the Kinematics of Human Locomotion over Continuously Varying Speeds and Inclines," IEEE Transactions on Neural Systems and Rehabilitation Engineering, 26(12): 2342-2350, 2018.
A participant walking on the perturbation platform, which translates fore-aft beneath the stance foot.

A Perturbation Mechanism for Investigations of Phase-Dependent Behavior in Human Locomotion

The dataset contains sagittal-plane leg joint kinematics, ground reaction forces, and EMG activity from one stride starting at heel strike on the perturbation platform, which randomly moved the stance foot in the anterior or posterior direction. These perturbations effected a phase shift in the gait cycle without substantially deviating from the nominal kinematic orbits of the leg joints. Only the right leg of each subject was perturbed, but both legs' kinematics are recorded. This data can be used to test different hypotheses and models of human locomotor control.

Associated publicationD. Villarreal, D. Quintero, and R. Gregg, "A Perturbation Mechanism for Investigations of Phase-Dependent Behavior in Human Locomotion," IEEE Access, 4: 893-904, 2016.
Diagram of the instrumented ankle-foot setup, showing the electrogoniometer, force plate, and ground reaction force.

Evidence for a Time-invariant Phase Variable in Human Ankle Control

The data contained herein includes kinetic and kinematic data acquired during walking trials, while a subject walked a across a walkway. A perturbation device was recessed into the walkway and randomly applied a two or five degree incline perturbation to the ankle as subjects walked across the platform. The perturbations occurred at one of four timing intervals, namely 100, 225, 350 and 475 ms following heel strike. During each trial, subject ankle angle and ground reaction force data were acquired, including location of the center of pressure for each subject.

Associated publicationR. D. Gregg, E. J. Rouse, L. J. Hargrove, and J. W. Sensinger, "Evidence for a Time-Invariant Phase Variable in Human Ankle Control," PLoS ONE, 9(2): e89163, 2014.

Software

Software

Controller releases and simulation code.

A powered knee-ankle prosthesis descending a stair rig, one of the activities this controller supports.A participant in a fall-protection harness testing the sit-to-stand mode of the controller beside an instrumented staircase.

Data-Driven Variable Impedance Control of a Powered Knee-Ankle Prosthesis for Variable Activities

This release contains a simplified version of our multi-activity knee-ankle prosthesis controller compiled to run on the Open-Source Leg. In automatic mode, the controller allows the user to sit, stand, and walk across level ground while automatically switching between walk and sit/stand modes. Configurable walking speed and ground incline inputs allow the controller to handle walking speeds between 0.6 and 1.4 m/s and ramps between -10 and 10 deg. In manual mode, the controller also enables stair ascent and descent on staircases with slope between 19 and 30 deg. We provide a starter python script, allowing anyone with an Open Source Leg to run our controller with minimal upfront effort. Please follow the link in the heading to be redirected to Michigan's Innovation Partnerships website to accept licensing terms and download the release. The release includes a main python file, a compiled library file, and a detailed getting started guide.

Associated publications
Stick-figure diagram of the simulated biped walking model, showing stance and swing leg configuration.

MATLAB Code for Biped Walker Simulation

This MATLAB package contains a simulation of an autonomous biped walker. This walker has mass and dimensions proportions similar to that of a human. It models the physics of the system using 2nd order differential equations. The derivation of the equations in Mathematica are included. There is a set-point Proportional-Derivative control (i.e., joint impedance) that drives the joints of the biped so that it walks stably down a shallow slope. This software can be useful for researchers looking for simulating controllers for autonomous bipeds, prosthetic legs, and/or exoskeletons.

Hardware designs

Hardware designs

Bills of materials and CAD models for our open-source exoskeleton hardware.

Citing these releases

Please cite the associated article listed with each record. Where a dataset carries its own digital object identifier (DOI), cite the dataset as well. Questions about a specific release can go to rdgregg@umich.edu.