Wearable soft pneumatic self-sensing artificial muscles for rehabilitation

This thesis describes the design and fabrication of self-sensing soft pneumatic actuators and their application for wearable rehabilitation systems. It discusses the obtained results and outlines remaining challenges, as well as future research directions for possible applications. Wearable mechatronic systems for powered orthoses, exoskeletons, and prostheses require advanced soft actuation technologies acting as “artificial muscles”, capable of large strains, high stresses, fast response, and self-sensing ability, while ensuring electrical safety, low specific weight, and high compliance. Among the various soft actuation technologies under investigation, pneumatic devices have attracted renewed interest over the past decades as intrinsically compliant artificial muscles, driven by advances in soft robotics. However, only a limited number of solutions currently exist that provide linear actuation with integrated self-sensing while remaining compatible with low-cost fabrication and off-the-shelf materials. This thesis presents self-sensing pneumatic actuators functioning as “inverse artificial muscles,” which elongate rather than contract upon pressurisation. The actuators consisted of an elastomeric tube surrounded by a plastic coil that constrained radial expansion. Self-sensing capabilities were achieved through a piezoresistive stretch sensor integrated within the tube. The developed actuator was used to implement in an active dynamic hand splint. Current state-of-the-art dynamic hand splints typically rely on elastic bands or springs that provide passive resistance to finger movements. Replacing these elements with soft pneumatic actuators enables real-time controllability of rehabilitation exercises, allowing finger movements against adjustable loads and thus transforming passive devices into active dynamic splints. The ideal actuators for such systems should provide large displacements at moderate forces, maintain compact size and low weight, and ensure electrical safety. The actuator was mounted on a forearm brace, connecting one end to a finger via a tendon and the other to an on-board load cell to monitor applied force. The actuator’s behaviour within the splint was evaluated through bench testing simulating finger motion between extension and flexion. Additionally, a finite element model of the actuator was developed to perform quasi-static simulations of force generation under varying displacement and pressure conditions. The comparison between numerical simulations and experimental results showed good agreement, validating the actuator performance. To assess the interaction with the human hand, static and dynamic psychophysical tests were conducted. These evaluated perceptual differences in force, the final force in the flexed position, and sensitivity to low forces. Results showed that participants could reliably perceive force variations corresponding to pressure changes. During a secondment at Queen Mary University of London, a key objective was the development of a piezoresistive strain sensor directly integrated into the actuator. This enabled the development of new actuator prototypes, including bending and coiling self-sensing pneumatic actuators. These designs offered simple and low-cost fabrication, elongation capability, and bidirectional motion, depending on the actuated chamber. The coiling actuator incorporating the self-sensing functionality represented the first example of its kind. Both actuator prototypes were characterised through pressurisation and depressurisation cycles to measure changes in resistance alongside corresponding geometric changes. The coiling actuator was characterised through blocking force measurement and load-lifting experiment. The coiling actuator (17 g, 500 mm length) was capable of lifting approximately 44% of its own weight and a reduction in length up to 32.8% under pressures up to 1.5 bar, while the bending actuator achieved angular changes exceeding 200°. Furthermore, the thesis introduces three ongoing research directions. The first focused on the textilisation of the dynamic hand splint to improve wearability. The second one explored pneumatically active knitted compression garments for applications such as blood flow restriction in rehabilitation. The third one introduced a combination of kinaesthetic and tactile feedback for investigating softness perception in virtual reality systems, aiming to enhance perceptual accuracy through multimodal stimulation. In conclusion, this thesis introduces multiple configurations of self-sensing soft pneumatic actuators, including elongating, bending, and coiling types. These were characterised both mechanically and electrically to establish the relationship between pressure, deformation, and sensor response. The elongating actuator was successfully applied in an active dynamic hand splinting, while the coiling actuator demonstrated potential for grasping applications. Additionally, alternative designs addressed wearability and textile integration. The applicability of these actuators was further extended to compression systems and multimodal feedback devices, highlighting their versatility across rehabilitation and emerging application domains.