Upper-Limb Rehabilitation with a Dual-Mode Individualized Exoskeleton Robot: A Generative-Model-Based Solution

TL;DR

The development of a new upper-limb exoskeleton robot with a novel online generative capability that allows it to provide individualized assistance to support the rehabilitation training of stroke patients and is illustrated in experiments involving healthy subjects and stroke patients.

Abstract

Several upper-limb exoskeleton robots have been developed for stroke rehabilitation, but their rather low level of individualized assistance typically limits their effectiveness and practicability. Individualized assistance involves an upper-limb exoskeleton robot continuously assessing feedback from a stroke patient and then meticulously adjusting interaction forces to suit specific conditions and online changes. This paper describes the development of a new upper-limb exoskeleton robot with a novel online generative capability that allows it to provide individualized assistance to support the rehabilitation training of stroke patients. Specifically, the upper-limb exoskeleton robot exploits generative models to customize the fine and fit trajectory for the patient, as medical conditions, responses, and comfort feedback during training generally differ between patients. This generative capability is integrated into the two working modes of the upper-limb exoskeleton robot: an active mirroring mode for patients who retain motor abilities on one side of the body and a passive following mode for patients who lack motor ability on both sides of the body. The performance of the upper-limb exoskeleton robot was illustrated in experiments involving healthy subjects and stroke patients.

Figures (21)

• Figure 1: Dual-mode upper-limb exoskeleton rehabilitation training. In active mirroring mode, the reference trajectory of the affected side is determined by the motion intention of the unaffected side (the red dashed line). Conversely, in passive following mode, the reference trajectory is pre-programmed (the red dashed line). The generative model in active mirroring mode can predict the motion intentions of the unaffected side and utilize interactive feedback to generate an anomaly score, which is used for trajectory refinement in both modes.
• Figure 2: Overview of the developed upper-limb exoskeleton robot, which is cable-driven and consists of five active joints (Joints 1–5) and one passive joint (Joint 0).
• Figure 3: Block diagram of the upper-limb exoskeleton robot, where the red arrow represents the electrical transmission, and the black line represents the mechanical connection
• Figure 4: The operating principle of the self-developed SEA
• Figure 5: Workflow of dual-mode training. In active mirroring mode, the predicted motion intention serves as the reference trajectory. In passive following mode, the reference trajectory is generated offline by fitting ProMPs to the data obtained from the demonstrations of healthy individuals. In both modes, the overarching goal is to deliver the most personalized assistance possible. Additionally, the motion data from healthy individuals are used to train the intention predictor and anomaly detector, and the anomaly score is used to guide trajectory refinement and the adjustment of impedance parameters.