Controller Design For Upper Limb Exoskeleton Robot With Motion Constraint And Input Constraint

This paper investiaged some kinds of problem usually encountered in exoskeleton robot system, such as deadzone input, saturation input, time-varying delay and state constraint. In order to solve these problems, we combine adaptive control with other intelligent control and propose the corresponding control algorithm based on Lyapunov direct theory. By employing adaptive control and Fuzzy Logic System(FLS) and Neural Neworks(NNs) theory, the problem of model uncertainty can be solved. Disturbance observer is exerted into the robotic exoskeleton system via feedforward loops to estimate and counteract the compounded disturbance term, i.e., the superposition of exernal disturbance, input nonlinearity term and approximation residual of model uncertainty. Barrier Lyapunov Function(BLF) as an useful mathematical tool can be used to deal with the problem of state constraint in a simple manner. To make the proposed controller more practical, it is necessary to extend the state feedback controller and propose output feedback controller based on high gain observer, in which only the output state information is measurable. The main research contents of this paper are summarized as:(i) Considering deadzone input and state time-varying delay simultaneously existing in the system, we investigated the tracking control for a class of parameterizable MIMO system and proposed an adaptive control approach;(ii) For the MIMO mechanical system, we designed an adaptive fuzzy controller to cope with the problem of deadzone and saturation input;(iii) Only by using the output information, we developed an output feedback controller based on high gain observer for the unmodelled dynamic nonlinear system;(iv) Under the BLF framework, by utilizing the NNs approximation theory we solved the problem of saturation input and state constraint at the same time and the control performance of closed-loop system was guaranteed;(v) We verified the developed control algorithms by simulation and experiment on upper limb exoskeleton robot platform, and comparative experiment results were analysed to illustrate the superiority of these proposed methods.