Adaptive Iterative Learning Control of Fluidic Muscle Driven Parallel Manipulators for Force Control With Sliding Mode Technique

Abstract

Parallel manipulator has received much attention in the last few decades due to the special advantages such as high rigidity, high accuracy and great carrying payload capability. In many applications such as robot assisted surgery, the interaction force between the end-effector and environment should be limited or controlled. Therefore, the requirement of force control is increasing. Fluidic muscle is well suited for parallel manipulator due to it has a high power/weight ratio, high strength, high tension force and compactness. And since fluidic muscle is virtually free of stick-slip effects, it is a perfect driver for force control.

For the fluidic muscle driven parallel manipulator with required force control, conventional PID control gradually cannot meet the increasing performance requirements. In this paper, an adaptive iterative learning control method combined with sliding mode technique is proposed to improve the force control performance for repeating tasks of fluidic muscle driven parallel manipulator. Since the accurate model of the fluidic muscle driven parallel manipulator is difficult to obtain, approximate models of both fluidic muscle and parallel manipulator are used in the proposed method. Then, by using the input and output data of the fluidic muscle and the end-effector, the parameters of both the fluidic muscle and the parallel manipulator are updated via iterative learning control. Sliding mode control is used to deal with parameter uncertainties, unknown nonlinearities and external disturbance to improve the robustness. Since the exact upper bounds of system uncertainties and external disturbance are difficult to obtain and the gain of sliding mode control usually need to be larger than the sum of the upper bounds of uncertainties and disturbance, the selection of control gain is conservative in majority of existing sliding mode control. Conservative gain results in larger control chattering. To reduce the control chattering, the gain of the sliding mode control is also iteratively updated by using iterative learning control in the proposed method.

The close-loop stability of the proposed method is analyzed. Moreover, convergence of the proposed method along iterative axis is proved by using composite energy function. Simulation study is performed by using a two degrees of freedom planar parallel manipulator driven by fluidic muscles. Performance comparisons are conducted among sliding mode control, a pre-existing sliding mode iterative learning control method in which the control input is composed of sliding mode control and iterative learning control, and the proposed method. Simulation results demonstrate that the proposed method can achieve better force control performance and robustness.

Keywords: force control, sliding mode iterative learning control, fluidic muscle, parallel manipulator.