Modelling inverse kinematics of a robotic manipulator through the use of artificial intelligence methods

Diplomski rad bavi se problemom modeliranja inverzne kinematike robotskog manipulatora primjenom metoda umjetne inteligencije. Prije samog rješavanja zadatka u radu je opisan uvod u robotiku, robotika kao znanost, tipovi robota, postojeće metode kinematike robota, korišteni robotski manipulator ABB IRB 120, korišteni programi za rješavanje zadatka diplomskog rada (RobotStudio, Anaconda Navigator, Spyder), zatim neuronska mreža i korišteni model neuronske mreže te samo rješavanje zadatka koji je podijeljen na dva dijela. Prvi dio zadatka odnosi se na prikupljanje skupa podataka za treniranje i testiranje neuronske mreže dok se drugi dio zadatka odnosi na modeliranje inverzne kinematike korištenjem neuronske mreže i podataka generiranih iz prvog dijela zadatka. Na kraju rada uspoređene su predviđene i stvarne vrijednosti.This graduate thesis deals with the problem of modeling the inverse kinematics of a robotic manipulator by using the method of artificial intelligence. Before solving the task, the thesis describes the introduction to robotics, robotics as a science, types of robots, existing methods of robot kinematics, used robotic manipulator ABB IRB 120, used programs for solving the thesis (RobotStudio, Anaconda Navigator, Spyder), neural network and the used neural network model and finally, the solution of the task which is divided into two parts. The first part of the task refers to the collection of data sets for training and testing neural networks, while the second part of the task refers to modeling the inverse kinematics by using neural network and data generated from the first part of the task. At the end of this thesis predicted and actual values are compared