Learning control algorithms for robot contact task using feedforward neural networks

摘要

The major concern of this paper is the application of connectionist architectures for fast online learning of robot dynamic uncertainties which are used at the executive hierarchical control level in the case of robot contact tasks. The connectionist structures are integrated in the nonlearning control laws for contact tasks which enable simultaneous stabilization and good tracking performance of position and force. It has been shown that the problem of tracking a specified reference trajectory and specified force profile with a preset quality of their transient response can be efficiently solved by means of application of the four-layer perceptron. The four-layer perceptron as part of hybrid learning control algorithms through the process of synchronous training use fast learning rules and available sensor informations in order to improve robotic performance progressively for minimal possible number of learning epochs. Some simulation results of deburring process with robot MANUTEC r3 are shown to verify effectiveness of the proposed control learning algorithms.