Generative design of a mechanical metamaterial-based physical interface for an assistive wearable robot

Abstract

Wearable robots augment human motor abilities in various scenarios. Some systems assist healthy users by enhancing abilities and reducing fatigue, while others provide rehabilitation for patients with neurological conditions or aid in daily activities. The applicability of wearable robots is broad, but developing these systems is complex due to challenges in the human-robot interface.

Most solutions to the design challenges of wearable robots use either rigid or soft physical Human-Robot Interfaces (pHRI). Rigid solutions provide accurate control and transmit greater mechanical power but require precise alignment with the user's joints and tend to be bulky. Soft interfaces, being conformable and compliant, eliminate alignment issues but introduce nonlinearities that complicate accurate control. A hybrid mechanism that combines features of both rigid and soft interfaces can address these challenges.

To develop a systematic design methodology for a wearable robot using a hybrid mechanism like the metamaterial, this thesis aims to: (1) Model the behavior of the mechanical metamaterial that can bend and stretch to conform to desired postures and trajectories, (2) Develop a modular wearable robotic system using the conformable metamaterial as the pHRI to assist wrist flexion-extension, and (3) Model the interaction between the pHRI and the user’s wrist to develop a control model.

To address the first aim of this thesis, we developed a kinematic model of the metamaterial. We studied the influence of design variables on its behavior and determined the ranges of interest. An optimization approach was developed to tune the metamaterial to specific postures and trajectories, enabling rapid customization for specific applications.

Building on this, we designed a modular wearable device for wrist flexion-extension assistance, adaptable for different users in settings like rehabilitation clinics. The metamaterial links were created as a parametric CAD model for quick customization and fabrication. We developed Bowden-cable-based actuation for the optimized metamaterial and created a demonstrator for bimodal Mirror Therapy, using mechanical actuation of the hemiplegic wrist by the healthy hand. This was evaluated for conformance to posture, trajectory, and force transmission.

The metamaterial model was then used to develop an interaction model between the wearable robot and the user's limb, implemented in Simulink to simulate dynamic behavior. A viscous damping model for tremor suppression was developed, increasing biomechanical load to attenuate tremors. This was evaluated using a mannequin hand and an actuated metamaterial, with tremor simulated by an eccentric mass exciter. Suppression was achieved using the actuated metamaterial pHRI, triggered by kinematic data from the mannequin hand.

The first contribution of this thesis is in the development of the model and optimization approach for the metamaterial. This enabled the creation of customised metamaterial pHRIs for a specific user's wrist to account for their variations. The second contribution of this thesis is in the development of a modular wearable robot using the optimized metamaterial as the pHRI to assist the user's wrist flexion-extension. Finally, we have also described a model for the interaction between the optimized metamaterial pHRI and the adjoining user's limb. This description provides a basis for the development of a multitude of control strategies dependent upon the specific application. Overall the results of this thesis show potential and warrant future engineering and development to enable augmentation of humans using the metamaterial as the pHRI in wearable robots.