User-Tailored Learning to Forecast Walking Modes for Exosuits
Gabriele Abbate, Enrica Tricomi, Nathalie Gierden, Alessandro Giusti, Lorenzo Masia, Antonio Paolillo
AI summary
Problem
Lightweight exosuits lack sufficient onboard sensors to perceive diverse user gaits, making adaptive control difficult. Existing perception methods rely on complex hardware or require extensive labeled data for each new user.
Approach
The authors deploy a temporal convolutional network on two thigh-mounted IMUs to classify level ground, stair ascent, and stair descent modes across a past-present-future time window. They enable automatic personalization through a self-supervised procedure that generates pseudo-labels from the model's own hindsight estimates to fine-tune the classifier online.
Key results
- TCN achieves 0.962 AUROC, outperforming random forest baselines
- Forecasting future gait states improves control responsiveness over current-only estimates
- Self-supervised fine-tuning successfully adapts the model to new users without external labels
- Closed-loop integration validates real-time walking mode classification in a single-subject experiment
Why it matters
This approach enables lightweight, adaptive assistive exosuits to personalize control for diverse users in real-world settings without heavy hardware or manual calibration.
Abstract
Assistive robotic devices, like soft lower-limb ex- oskeletons or exosuits, are widely spreading with the promise of helping people in everyday life. To make such systems adaptive to the variety of users wearing them, it is desirable to endow exosuits with advanced perception systems. However, exosuits have little sensory equipment because they need to be light and easy to wear. This paper presents a perception module based on machine learning that aims at estimating 3 walking modes (i.e., ascending or descending stairs and walking on level ground) of users wearing an exosuit. We tackle this perception problem using only inertial data from two sensors. Our approach provides an estimate for both future and past timesteps that supports control and enables a self-labeling procedure for online model adaptation. Indeed, we show that our estimate can label data acquired online and refine the model for new users. A thorough analysis carried out on real-life datasets shows the effectiveness of our user-tailored perception module. Finally, we integrate our system with the exosuit in a closed-loop controller, validating its performance in an online single-subject experiment.