Benchmarking Foot Trajectory Estimation Methods for Mobile Gait Analysis

Abstract

Mobile gait analysis systems based on inertial sensing on the shoe are applied in a wide range of applications. Especially for medical applications, they can give new insights into motor impairment in, e.g., neurodegenerative disease and help objectify patient assessment. One key component in these systems is the reconstruction of the foot trajectories from inertial data. In literature, various methods for this task have been proposed. However, performance is evaluated on a variety of datasets due to the lack of large, generally accepted benchmark datasets. This hinders a fair comparison of methods. In this work, we implement three orientation estimation and three double integration schemes for use in a foot trajectory estimation pipeline. All methods are drawn from literature and evaluated against a marker-based motion capture reference. We provide a fair comparison on the same dataset consisting of 735 strides from 16 healthy subjects. As a result, the implemented methods are ranked and we identify the most suitable processing pipeline for foot trajectory estimation in the context of mobile gait analysis.

Estimation of spatio-temporal parameters of gait from magneto-inertial measurement units: multicenter validation among Parkinson, mildly cognitively impaired and healthy older adults

“…Overall, the of both temporal and spatial parameters was often below and, except NEWCA Stance Time at FW, never over 3% which is an excellent result, although a thorough comparison with the results obtained in studies proposing other methods is not straightforward [ 1 , 15 – 17 , 28 – 34 ]. Regarding the estimation of the spatial parameters, it has been shown in the study conducted by Hannink et al [ 35 ], that the OFDRI technique was the best performing among the double integration methods for mobile gait analysis tested in their study.…”

Wearable Inertial Measurement Units for Assessing Gait in Real-World Environments

“…successful approaches (Hamacher et al, 2014;Rampp et al, 2015;Benoussaad et al, 2016;Hannink et al, 2017). As a result, the sensor acceleration data a s (t) had to be expressed in global coordinates.…”

“…As a result, the sensor acceleration data a s (t) had to be expressed in global coordinates. This was achieved using a rotation matrix R WS (t) that identified how the sensor frame was oriented with respect to the global frame at every time instance, t. Orientation estimation was applied individually for each gait cycle and combined the acceleration and angular velocity measurements to obtain the rotation matrices, R WS (t) [similar to gyroscope integration (Hannink et al, 2017), described in section "3. Orientation estimation" in Supplementary Material].…”

Indoor Trajectory Reconstruction of Walking, Jogging, and Running Activities Based on a Foot-Mounted Inertial Pedestrian Dead-Reckoning System

“…A trajectory reconstruction pipeline was carried out separately for each activity of both datasets ( Figure 4). This pipeline is based on previous work by Hannink et al [26]. A type of activity classification step was included.…”