Learning Possibilities in CPN

Abstract

In the flexible lines, when it is frequently necessary to change the sequence of technological process, the use of mobile robots is preferred. Training of the mobile robots in known surrounding is required for the efective functionality of control systems. Training proces is very long, so a suggestion emerges to use robots learning in known surrounding for this problem. For purposes of training and learning, the colored Petri nets are suitable. There are used global variables, it is possible to form, to accumulate and to preserve data. Authors created a special system for mobile robot learning. It uses scanner and known parameters of the surrounding. Learning of the system is implemented in the model, the search and the conflict situations are solved in the graphic subsystem of the software. Learning data is captured to the database, which is common to all mobile robots; it is in the declarations of the colored Petri nets. Optimization of the traces solves during the search or the trace. It allows us to achieve good learning results quickly. Ill. 2, bibl. 13 (in English; summaries in English, Russian and Lithuanian).