Model-free based adaptive BackStepping-Super Twisting-RBF neural network control with α-variable for 10 DOF lower limb exoskeleton

Lower limb exoskeletons play a pivotal role in augmenting human mobility and improving the quality of life for individuals with mobility impairments. In light of these pressing needs, this paper presents an improved control strategy for a 10-degree-of-freedom lower limb exoskeleton, with a particular focus on enhancing stability, precision, and robustness. To simplify the intricate dynamic model of the exoskeleton, our approach leverages a more manageable 2nd order ultra-local model. We employ two radial basis function (RBF) neural networks to accurately estimate both lumped disturbances and non-physical parameters associated with this ultra-local model. In addition, our control strategy integrates the backstepping technique and the super twisting algorithm to minimize tracking errors. The stability of the designed controller is rigorously established using Lyapunov theory. In the implementation phase, a virtual prototype of the exoskeleton is meticulously designed using SolidWorks and then exported to Matlab/Simscape Multibody for co-simulation. Furthermore, the desired trajectories are derived from surface electromyography (sEMG) measured data, aligning our control strategy with the practical needs of the user. Comprehensive experimentation and analysis have yielded compelling numerical findings that underscore the superiority of our proposed method. Across all 10 degrees of freedom, our controller demonstrates a significant advantage over alternative controllers. On average, it exhibits an approximately 45% improvement compared to the Adaptive Backstepping-Based -RBF Controller, a 74% improvement compared to the Model-Free Based Back-Stepping Sliding Mode Controller, and an outstanding 74% improvement compared to the Adaptive Finite Time Control Based on Ultra-local Model and Radial Basis Function Neural Network. Furthermore, when compared to the PID controller, our approach showcases an exceptional improvement of over 80%. These significant findings underscore the effectiveness of our proposed control strategy in enhancing lower limb exoskeleton performance, paving the way for advancements in the field of wearable robotics.