Exoskeleton-Mediated Physical Human-Human Interaction for a Sit-to-Stand Rehabilitation Task

Authors: Lorenzo Vianello, Emek Barış Küçüktabak, Matthew Short, Clément Lhoste, Lorenzo Amato, Kevin Lynch, Jose Pons · ICRA 2024

Licences

Open Access
Paper: CC BY 4.0

Everything below draws on the full paper under its CC BY 4.0 licence; its figure is credited to the authors.

TL;DR

Couples two X2 lower-limb exoskeletons through a virtual spring-damper so a therapist and patient share a sit-to-stand task, with QP-based whole-body control adding assistance and balance limits.

Why it matters

The set-up lets a therapist feel and guide a patient across several joints at once while the exoskeletons enforce safety limits and record performance over many repetitions. Because the coupling is virtual, the authors note the same infrastructure could support tele-rehabilitation with users in different places. Balance constraints matter here because the X2 ankle is not actuated.

Problem

Sit-to-stand training is a core part of stroke rehabilitation, but assisting patients by hand is physically demanding for therapists and can lead to work-related injuries. Assistive devices often make the patient passive, which reduces motor learning, and remove the therapist's direct physical guidance. To the authors' knowledge, no earlier framework had used overground lower-limb exoskeletons to mediate physical interaction between therapist and patient for this task.

Proposed solution

Each user wears an X2 exoskeleton modelled in double stance (backpack, hip and knee angles), and a virtual-mass controller tracks a desired interaction torque. The two exoskeletons are linked by a virtual spring-damper either joint by joint or between the vertical positions of the two centres of mass, and a real-time quadratic program (OSQP) solves for motor torques and joint accelerations under torque, power, velocity, joint-limit and balance constraints. Balance is kept by constraining the divergent component of motion inside the sagittal-plane support region, and a scalable vertical assistance term (in kg) delays fatigue.

Key insights

Linking the two users through the vertical position of the centre of mass leaves freedom of movement in the null space, while joint-space coupling transmits feedback at every joint.
Constraining the divergent component of motion, plus position-dependent limits on CoM height and backpack angle, compensates for the unactuated ankle by steering the user to a safe posture when the forward limit is violated.
Vertical assistance changed how users moved: they quickened upward movements and delayed downward ones, so the authors judge it better suited to discrete-target tasks.
Each tracking reference has a trade-off: a single sine is least fatiguing but too predictable to study learning, a multi-sine adds variability but is fatiguing, and discrete targets reduce both problems but are harder to score.

Main results

For one pilot dyad, tracking error (RMS against the visual target) was lower when the two users were coupled in joint space than when each moved alone, and the improvement appeared related to the coupling stiffness.
Joint-space coupling was tested at a soft setting of 30 Nm/rad and 4 Nms/rad and a stiff setting of 70 Nm/rad and 10 Nms/rad per joint; CoM coupling used 1000 N/m and 50 Ns/m.
No clear difference was observed between task-space (CoM) and joint-space coupling, which the authors attribute to the small variability of hip and knee motion in sit-to-stand.
Each trial alternated 30 s solo, 30 s coupled and 30 s rest, with 8 trials per condition, using single-sine, multi-sine (0.50, 0.20 and 0.16 Hz components) and discrete-step references.
Users reported difficulty telling the vertical assistance apart from the virtual connection.

Limitations

The study is exploratory: results come from one pilot dyad, and the authors state it is not their intent to validate the findings for lower-limb exoskeletons.
The framework has not yet been tested with therapists and patients; the authors plan healthy-population tests first.
Only a limited stiffness range was tried; higher stiffness values and users with different body properties are left to future experiments.
The ankle joint is not actuated and all actuation and balance constraints act only in the sagittal plane.

Field impact

The paper won the ICRA 2024 Best Paper Award in Medical Robotics and was a finalist for the ICRA 2024 Best Conference Paper and Best Student Paper Awards. It has 10 citations, and the licensed copy gives no code or data release.

Abstract

Sit-to-Stand (StS) is a fundamental daily activity that can be challenging for stroke survivors due to strength, motor control, and proprioception deficits in their lower limbs. Existing therapies involve repetitive StS exercises, but these can be physically demanding for therapists while assistive devices may limit patient participation and hinder motor learning. To address these challenges, this work proposes the use of two lower-limb exoskeletons to mediate physical interaction between therapists and patients during a StS rehabilitative task. This approach offers several advantages, including improved therapist-patient interaction, safety enforcement, and performance quantification. The whole body control of the two exoskeletons transmits online feedback between the two users, but at the same time assists in movement and ensures balance, and thus helping subjects with greater difficulty. In this study we present the architecture of the framework, presenting and discussing some technical choices made in the design.

Open Source Robotics