Empirical Control Strategy for Autonomous Industrial Robot Based on Recognition System and Machine Learning

Abstract

The paper presents the results of machine learning of the educational industrial robot Mitsubishi Movemaster-EX and the self-developed anthropomorphic mobile robot called Don Kihot, which confirmed the validity of the scientific hypothesis. The starting hypothesis implies that machine learning mechanisms, based on artificial neural networks and the concept of artificial life, can provide the mechatronic system-robot with appropriate autonomy when performing the robotized task of manipulating recognized objects - parts in general and within the assembly based on camera system. In order to prove this hypothesis, the aforementioned robots and more specially developed software BPnet were used. In the function of experimental verification and hardware-software integration, a new control strategy of intelligent robot control based on an empirical control algorithm was developed.