Развој машинског учења интелигентног мобилног робота базиран на систему вештачких неуронских мрежа

Abstract

Унутрашњи транспорт сировина, материјала и готових делова подразумева брзо, ефикасно и економично деловање постављеног транспортног задатка...

Material Handling Systems in manufacturing environment imply efficient and economical transport solutions. Automated Guided Vehicles (AGVs) are a common choice made by many companies for Material Handling in manufacturing systems. Nowadays, AGV based internal transport of raw materials, goods and parts is becoming improved with advances in technology. Demands for fast, efficient and reliable transport imply the usage of the flexible AGVs with onboard sensing and special kinds of algorithms needed for daily operation. These transport solutions can be modified and enhanced by applying advanced methods and technologies. New generation of internal transport systems should operate autonomously, without direct human control. Level of development of mobile robots insures reliability and efficiency needed for dayily operations within manufacturing environemnt. In this thesis, the implementation of mobile robots for internal transport within Material Handling System is analyzed and new solutions are proposed. Focus of research efforts is devoted to the ability to estimate position and orientation of mobile robot within manufacturing environment using newly developed algorithms and sensory information. Simultaneous localization (of the mobile robot) and mapping (of the working environment) is one of the most important problems in mobile robotics community. The soultion to this problem insures autonomous navigation and henceforth autonomous operation for transport purposes within manufacturing/industrial facility without direct human control. In this thesis, new algorithm for state estimation is proposed and analyzed; the algorithm is based on integration of Extended Kalman Filter and feedforward neural networks (Neural Extended Kalman Filter) and camera is used as exteroceptive sensor. Furhermore, to achieve intelligent behavior, the X new robotic hybrid control architecture is developed and analyzed. Finally, the new hybrid control algorithm for guidance of mobile robot is proposed. Two building blocks form the hybrid algorithm: visual servoing and position based control. Neural Extended Kalman Filter is used for state estimation of the mobile robots, and at each time instant the robot knows its position and orientation...