Comparitive Analysis of Deep Learning Techniques for Assistive Soft Wearable Robots

A new innovation called wearable assistive robotics has the potential of assisting those with sensorimotor disabilities in doing routine tasks. A lot of research is done on soft robots because of its adaptability, deformability, and flexibility. In contrast to soft robots and rigid robots, face obstacles in control, calibration, and modelling because the properties of soft materials which result in complex behaviors due to hysteresis and non-linearity. The use of deep learning techniques in real-time poses additional challenges for researchers when it comes to accuracy and equipment cost. Recent research has used various deep learning algorithms to address these constraints. This paper gives deep insight and analysis of existing deep learning techniques in the area of assistive soft wearable robotics and classifies their applicability in various soft robotic applications. The current constraints in the study field, along with an analysis of various deep learning models with regard to various types of assistive soft wearable robot applications, are provided, followed by a description and implementation of the available deep learning techniques for assistive soft wearable robotics.