NEURAL-NETWORK-BASED CONTROL SCHEMES FOR FLEXIBLE-LINK MANIPULATORS -SIMULATIONS AND EXPERIMENTS

This paper presents simulation and experimental results on the perform ance of neural network-based controllers for tip position tracking of flexible-link manipulators. The controllers are designed by utilizing the modified output re-definition approach. The modified output re-def inition approach requires only a priori knowledge about the linear mod el of the system and no a priori knowledge about the payload mass. Fou r different neural network schemes are proposed. The first two schemes are developed by using a modified version of the 'feedback-error-lear ning' approach to learn the inverse dynamics of the flexible manipulat or. Both schemes require only a linear model of the system for definin g the new outputs and for designing conventional PD-type controllers. This assumption is relaxed in the third and fourth schemes. In the thi rd scheme, the controller is designed based on tracking the hub positi on while controlling the elastic deflection at the tip. In the fourth scheme which employs two neural networks, the first network (referred to as the 'output neural network') is responsible for specifying an ap propriate output for ensuring minimum phase behavior of the system. Th e second neural network is responsible for implementing an inverse dyn amics controller. The performance of the four proposed neural network controllers is illustrated by simulation results for a two-link planar flexible manipulator and by experimental results for a single flexibl e-link test-bed. The networks are all trained and employed as online c ontrollers and no off-line training is required. (C) 1998 Elsevier Sci ence Ltd. All rights reserved.