Wearable environmental sensing to forecast foot center of pressure and time of impact during level-ground-to-stair-ascent transitions

Objective Computer-vision has been used for environmental classification during gait and is often used to inform control in assistive systems; however, the ability to predict how the foot will contact a changing environment is underexplored. Methods We evaluated the feasibility of forecasting the anterior-posterior foot center-of-pressure (COP) and time-of-impact (TOI) prior to foot-strike on a level-ground-to-stair-ascent transition. Eight subjects wore a camera that supplied paired grayscale stereo frames on their right shank and instrumented insoles in their shoes while performing level-ground-to-stair-ascent transitions in a controlled laboratory setting (~90 trials each). We trained one CNN-RNN base model for COP and another for TOI, then fine-tuned both for each subject to generate continuous forecasts (16.67 ms temporal resolution) within a 250 ms window prior to foot-strike, termed the forecast horizon (FH). Performance was evaluated on held-out trials from the same subject using leave-one-out-cross-validation (LOOCV). Results The COP mean absolute error (MAE) was 29.42, 26.82, and 23.72 mm, and the TOI MAE was 21.14, 20.08, and 17.73 ms, for FHs of 150, 100, and 50 ms, respectively. Torso velocity and toe-swing velocity prior to foot-strike were not significantly associated with prediction error in either the COP or TOI task. Further, more anterior foot-strikes were found to reduce COP prediction accuracy but did not affect the TOI prediction accuracy. In a separate live-camera frame-rate test, the image-preprocessing and inference pipeline ran both models simultaneously at approximately 60 processing cycles/s on a consumer-grade laptop and an edge computing device (i.e., Jetson Orin Nano). Conclusion This study demonstrates the feasibility of forecasting COP and TOI from pre-contact visual data for a level-ground-to-stair-ascent transition in a controlled laboratory setting. Significance This may have important implications for anticipatory control in legged systems.

Authors

Institutions

Publication Details

Journal
PLoS ONE
Published
2026-11-19
DOI
https://doi.org/10.1371/journal.pone.0362954
Primary Topic
Prosthetics and Rehabilitation Robotics
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Wearable environmental sensing to forecast foot center of pressure and time of impact during level-ground-to-stair-ascent transitions

James Yungjen Tung, Richard W. Nuckols, M. G. Murray
2 links
PLoS ONE
Prosthetics and Rehabilitation Robotics
article

Wearable environmental sensing to forecast foot center of pressure and time of impact during level-ground-to-stair-ascent transitions

James Yungjen Tung, Richard W. Nuckols, M. G. Murray
article en

Abstract

Objective Computer-vision has been used for environmental classification during gait and is often used to inform control in assistive systems; however, the ability to predict how the foot will contact a changing environment is underexplored. Methods We evaluated the feasibility of forecasting the anterior-posterior foot center-of-pressure (COP) and time-of-impact (TOI) prior to foot-strike on a level-ground-to-stair-ascent transition. Eight subjects wore a camera that supplied paired grayscale stereo frames on their right shank and instrumented insoles in their shoes while performing level-ground-to-stair-ascent transitions in a controlled laboratory setting (~90 trials each). We trained one CNN-RNN base model for COP and another for TOI, then fine-tuned both for each subject to generate continuous forecasts (16.67 ms temporal resolution) within a 250 ms window prior to foot-strike, termed the forecast horizon (FH). Performance was evaluated on held-out trials from the same subject using leave-one-out-cross-validation (LOOCV). Results The COP mean absolute error (MAE) was 29.42, 26.82, and 23.72 mm, and the TOI MAE was 21.14, 20.08, and 17.73 ms, for FHs of 150, 100, and 50 ms, respectively. Torso velocity and toe-swing velocity prior to foot-strike were not significantly associated with prediction error in either the COP or TOI task. Further, more anterior foot-strikes were found to reduce COP prediction accuracy but did not affect the TOI prediction accuracy. In a separate live-camera frame-rate test, the image-preprocessing and inference pipeline ran both models simultaneously at approximately 60 processing cycles/s on a consumer-grade laptop and an edge computing device (i.e., Jetson Orin Nano). Conclusion This study demonstrates the feasibility of forecasting COP and TOI from pre-contact visual data for a level-ground-to-stair-ascent transition in a controlled laboratory setting. Significance This may have important implications for anticipatory control in legged systems.

PLoS ONEVol. 21(11)
University of Massachusetts Lowell (US), University of Waterloo (CA)
University of Waterloo, Natural Sciences and Engineering Research Council of Canada
Prosthetics and Rehabilitation Robotics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon