Tele-rehabilitation with online skill transfer and adaptation in $\mathbb{R}^3 \times \mathit{S}^3$

TL;DR

This work tackles remote, therapist-supported rehabilitation by introducing a tele-teaching framework that links therapist and patient robots via bilateral teleoperation and encodes rehabilitation trajectories in $\mathbb{R}^3 \times \mathit{S}^3$ using 6-DoF Dynamical Movement Primitives. It integrates a leader–follower control architecture, a motion generator that blends Euclidean translations with $S^3$ rotations through periodic DMPs and a Riemannian DMP, and an autonomy-allocation scheme that gradually transfers control from therapist to patient as learning converges. Key contributions include online learning with Recursive Least Squares for DMP weights, a decoupled learning/autonomy scheme ($\mu$ and $\eta$), and demonstrated rapid adaptation and generalization across multiple translational and rotational rehabilitation tasks on two 7-DoF robots. The approach enables remote, personalized rehabilitation with continuous therapist oversight and scalable autonomy, potentially expanding access to therapy and reducing clinician burden.

Abstract

This paper proposes a tele-teaching framework for the domain of robot-assisted tele-rehabilitation. The system connects two robotic manipulators on therapist and patient side via bilateral teleoperation, enabling a therapist to remotely demonstrate rehabilitation exercises that are executed by the patient-side robot. A 6-DoF Dynamical Movement Primitives formulation is employed to jointly encode translational and rotational motions in $\mathbb{R}^3 \times \mathit{S}^3$ space, ensuring accurate trajectory reproduction. The framework supports smooth transitions between therapist-led guidance and patient passive training, while allowing adaptive adjustment of motion. Experiments with 7-DoF manipulators demonstrate the feasibility of the approach, highlighting its potential for personalized and remotely supervised rehabilitation.

Figures (4)

• Figure 1: A multi-modal tele-rehabilitation scenario: The patient and robot are in contact through the distal part with an appropriate gamified interface on the screen. On the remote side, the therapist can directly interact with the patient, adjust and observe the parameters and indices, and teach a new exercise through the robot and the provided web-based interface.
• Figure 2: The experiment setup. The user (Therapist) demonstrate the rehabilitation skill on the TR, and the PR holds the patient's hand.
• Figure 3: Five sample motions with both translational and rotational parts. The blue trajectories are the therapist's demonstrations, and the orange trajectories with the orientation axis attached are the DMP trajectories learned online from demonstrations. For the "line" motions, the corresponding robot poses at the farthest points are marked.
• Figure 4: Experimental results of the proposed tele-rehabilitation approach. A therapist demonstrates and modifies different periodic rehabilitation skills to the PR through TR. a), b) and c) include PR's end-effector position $\boldsymbol{p}$ and DMP reference trajectory $\boldsymbol{p}^{des}$, d), e), f) and g) PR's end-effector position $\boldsymbol{Q}$ and DMP reference trajectory $\boldsymbol{Q}_{des}$, h) Autonomy level $\eta$ and learning level $\mu$. Convergence of $\mu$ to one indicates learning is finished. Once learning is finished with no further demonstration, $\eta$ rises to one gradually to switch to the autonomous mode.