Data-driven and physics-informed modelling and control of soft robotic exoskeleton, prosthetic and orthoses to assist elderly people

Mobility impairments present a pressing global health issue. Millions of individuals live with locomotor disabilities caused by spinal cord injuries, strokes, and age-related decline, conditions that increasingly affect those over the age of 60. Worldwide, SCI prevalence reached 20.6 million in 2019, accounting for 6.2 million disability-adjusted life years. Notably, 46% of individuals experiencing a first-time stroke are unable to return to work immediately after post-hospitalization. Furthermore, 31% of locomotor impairments occur in individuals aged 40–60, with incidence rates doubling beyond the age of 60. These statistics highlight the urgent need for assistive technologies such as exoskeletons, prosthetics, and orthotic devices, particularly for aging populations. Although rigid exoskeletons are commercially available, their utility is limited by excessive weight, high costs, and low FDA approval rates. These figures underscore the growing urgency for advanced assistive technologies, such as exoskeletons, prosthetics, and orthotic devices, particularly for the aging population.

Rigid exoskeletons, though commercially available, often face limitations related to excessive weight, prohibitive costs, and low regulatory approval rates. In response, researchers are developing soft exoskeletons (or “exosuits”) that offer greater flexibility and reduced weight. Soft exoskeletons address these limitations by enabling additional movements, such as elbow pronation and supination. However, soft actuators struggle to generate sufficient force or torque, especially in rehabilitation scenarios, where precise human-robot interaction is critical. Signal processing and sensor fusion techniques aim to mitigate these limitations. Kinematic analysis in soft exoskeletons faces difficulties due to infinite degrees of freedom, and multibody dynamics models fail to capture effects like variable stiffness. Nonlinear behaviors, including compliance and hysteresis, hinder conventional control methods, though feedback systems are essential for accurate closed-loop control. The complexity of modelling often prevents real-time control. Reduced-order methods such as model order reduction and the Koopman operator show promise, while data-driven and physics-informed neural networks offer a balance between model complexity and real-time performance by integrating physical insights with sparse data.

This Research Topic aims to bring together cutting-edge contributions focused on the design, modelling, and control of soft robotic systems intended for mobility assistance and rehabilitation. It seeks to address the critical challenges in real-time dynamic control, human-exoskeleton interaction, and actuator limitations through multidisciplinary approaches integrating robotics, biomechanics, control systems, and artificial intelligence.

Topics of interest include, but are not limited to:

• Actuation and Mechanical Design: Design and development of soft exoskeletons for mobility assistance and rehabilitation and novel soft actuator technologies and variable stiffness mechanisms

• Control Systems and Modelling: Real-time dynamic and kinematic control strategies for soft exosuits and reduced-order and data-efficient modelling techniques

•Artificial Intelligence Integration: Physics-informed and data-driven AI methods for control and estimation and feedback systems and adaptive control techniques for accurate closed-loop performance

• Innovative approaches to human-exoskeleton interaction, including sensor integration, fusion, and feedback systems, with solutions to key challenges such as compliance, hysteresis, and limited torque output.

• Interdisciplinary research on real-world control system deployment, combining robotics, biomechanics, control theory, and artificial intelligence for clinical and daily-use applications.