[Paper Review] Lower-limb kinematics and kinetics during continuously varying human locomotion

Emma Reznick, Kyle Embry+4 more|arXiv (Cornell University)|Aug 27, 2021

Prosthetics and Rehabilitation Robotics36 references142 citations

TL;DR

This paper presents a comprehensive dataset of lower-limb kinematics and kinetics from ten able-bodied participants performing continuously varying locomotion tasks, including walking at multiple speeds and inclines, running, stair climbing, and transitions between activities. The data, collected via Vicon motion capture and instrumented treadmills, enables continuous modeling of human gait for improved control of robotic prostheses and exoskeletons.

ABSTRACT

Human locomotion involves continuously variable activities including walking, running, and stair climbing over a range of speeds and inclinations as well as sit-stand, walk-run, and walk-stairs transitions. Understanding the kinematics and kinetics of the lower limbs during continuously varying locomotion is fundamental to developing robotic prostheses and exoskeletons that assist in community ambulation. However, available datasets on human locomotion neglect transitions between activities and/or continuous variations in speed and inclination during these activities. This data paper reports a new dataset that includes the lower-limb kinematics and kinetics of ten able-bodied participants walking at multiple inclines ($\pm$ 0, 5, 10 $^{\circ}$) and speeds (0.8, 1, 1.2 m/s), running at multiple speeds (1.8, 2, 2.2, 2.4 m/s), walking and running with constant acceleration ($\pm$ 0.2, 0.5 $ ext{m/s^2}$), and stair ascent/descent with multiple stair inclines (20, 25, 30, 35 $^{\circ}$). 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.

Motivation & Objective

• Address the lack of comprehensive, continuous human locomotion data that includes transitions between walking, running, stair climbing, and sit-stand activities.
• Overcome limitations in existing datasets that focus only on steady-state tasks with discrete speed and incline conditions.
• Provide a unified, high-fidelity dataset to support continuous modeling of joint kinematics and kinetics across diverse locomotion modes.
• Enable the development of adaptive control systems for robotic prostheses and exoskeletons that can handle real-world, non-steady-state locomotion.
• Facilitate cross-participant modeling and individualization of control strategies through standardized, time-normalized stride data.

Proposed method

• Collected 3D motion capture data using a Vicon system with Plug-in Gait and custom marker sets to ensure robust segment tracking.
• Recorded force plate and treadmill data during treadmill-based tasks to calculate joint kinetics via inverse dynamics.
• Implemented a custom MATLAB pipeline to parse trials into individual strides, detect heel strikes using force plates (treadmill) or manual labeling (stairs), and time-normalize strides to 150 points.
• Applied 4th-order Butterworth low-pass filtering (6 Hz cutoff) and Woltring smoothing (smoothing parameter 20) to reduce noise while preserving dynamics.
• Used symmetry and range-of-motion checks during marker calibration to ensure data quality and minimize gait asymmetry.
• Removed 8.3% of treadmill strides and 7.2% of stair strides based on outlier detection (3× median absolute deviation), non-periodicity, or data gaps to ensure data integrity.

Experimental results

Research questions

• RQ1 How do lower-limb joint kinematics and kinetics vary continuously across walking speeds (0.8–1.2 m/s), inclines (±10°), and running speeds (1.8–2.4 m/s)?
• RQ2 What are the characteristic patterns of joint motion and force during transitions between locomotion modes such as walk-run, walk-stairs, and sit-stand?
• RQ3 To what extent can continuous kinematic models trained on this dataset accurately interpolate between discrete locomotion tasks?
• RQ4 How do joint moments and pelvic orientation change during constant acceleration/deceleration walking and running?
• RQ5 Can this dataset support the development of adaptive control systems that generalize across diverse locomotion conditions in real-world ambulation?

Key findings

• The dataset includes 2006 median strides per participant across all locomotion modes, with 8.3% of treadmill strides and 7.2% of stair strides excluded due to data quality issues.
• Joint kinematics and kinetics were successfully captured during continuous variations in speed (0.8–2.4 m/s), incline (±10°), and stair incline (20°–35°), including transitions.
• Pelvic tilt, hip, knee, and ankle joint angles were recorded with high temporal resolution, enabling detailed analysis of gait dynamics.
• Joint moments were computed for treadmill-based tasks using inverse dynamics and force plate data, providing a complete kinetic profile.
• The dataset supports continuous modeling of gait through Fourier-based or basis function interpolation, enabling smoother transitions than finite-state machine approaches.
• A custom MATLAB pipeline enabled automated parsing, normalization, and quality control of 150-point time-normalized strides, ensuring consistency across participants and tasks.