ExoLaN: Physics-Consistent Context-Aware Dynamics Learning for Exoskeletons

2026-09-25

AI Summary

This study addresses the challenges of user-specific adaptation and intermittent contact in exoskeleton control by proposing ExoLaN, a framework built upon physics-consistent Lagrangian networks. By integrating temporal context with plantar force measurements, ExoLaN learns human-exoskeleton coupled dynamics through joint inverse and forward dynamics modeling, inferring latent embeddings to estimate generalized contact torques. Its core contributions include enabling training-free user adaptation, explicitly modeling intermittent contacts, and capturing task information via unsupervised latent contexts. Experimental results demonstrate that for new users, ExoLaN reduces torque estimation mean squared error by 7% and acceleration prediction error by 59%, while decreasing long-horizon position and velocity errors by 60% and 93%, respectively.

Abstract

Task-agnostic assistive exoskeleton control based on human intention offers greater flexibility than conventional approaches that rely on predefined tasks or motion patterns. Human joint torque estimation enables task-agnostic assistance by characterizing user actions. Physics-consistent methods such as Deep Lagrangian Networks (DeLaN) have been applied to estimate the human torques in multi-user settings, but existing approaches cannot adapt to a specific user without retraining, and do not account for intermittent contacts during locomotion. We propose ExoLaN, a Context-Aware DeLaN for human-exoskeleton interaction that learns the full coupled system dynamics while adapting to changes in interaction context. ExoLaN combines temporal context with partial contact-force measurements from force-sensitive insoles to infer latent dynamics embeddings and estimate generalized contact torques. On seven unseen users performing 21 unseen tasks, ExoLaN reduces torque estimation MSE by 7% compared to a black-box baseline. Beyond inverse dynamics, ExoLaN serves as a unified model that also enables accurate forward prediction: training with a multi-step prediction loss reduces acceleration MSE by 59% and long-horizon position and velocity errors by 60% and 93%, respectively, compared with a single-step loss. Moreover, the learned latent context captures task information without explicit task labels, making it a promising signal for task-aware assistive control.

Problem

• Exoskeleton control
• Human joint torque estimation
• Physics-consistent dynamics learning
• Intermittent contact
• Context-aware adaptation

Innovation

• Context-Aware Dynamics Learning
• Deep Lagrangian Networks
• Human-Exoskeleton Interaction
• Task-Agnostic Control
• Latent Dynamics Embeddings