Adaptive computed torque control based on RBF network for a lower limb exoskeleton

A 12 degrees of freedom (DOFs) lower limb exoskeleton for gait rehabilitation is designed in our research. Traditional computed torque control (CTC) is an effective motion control method because of its globally asymptotic stability. Despite the impossibility of obtaining the precise dynamical models of exoskeletons, effective compensative methods can still contribute to achieve satisfying motion control. Adaptive radial basis functional (RBF) neural networks is applied to estimate the unknown part and compensate for it. A comparative research illustrates the better performance and beneficial effects while RBF neural networks are applied. MATLAB/Robotics Toolbox (RTB) helps to obtain the dynamic model information, avoiding the calculating complexity and the corresponding co-simulation results in MATLAB/SimMechanics are presented to demonstrate the effectiveness and fundamental improvement of the proposed control scheme. The combined utilization of RTB and SolidWorks sufficiently indicates the feasibility of the proposed strategy in hardware system.