Dynamic model and control with artificial intelligence algorithms of multi legged robots with pronking gait

The legged robots have received great attention in robotics literature because the most of the studies on mobil robots have been related to legged robots in the last years. Therefore, the tendency for studying in this area has gradually increased because the mobility of legged robots in the widest variety of terrain conditions is better than that of most existing mobile robots such as wheeled, tracked or railed robots. The most important desired features of mobil robots in today technology are the high capacity of mobility and multi-functionalities.Dynamic model, simulation and control of quadrupedal and hexapod robots have been realized for a pronking gait in this thesis. In order to obtain the numerical simulation of the equation of motion for pronking gait, the different dynamic structures have been solved in a sequential closed loop (stance, flight). Spring Loaded Inverted Pendulum (SLIP) has been used as a dynamic model to simplify the simulation of the system and ease to understand the pronking gait. Different from the studies in the literature, the quadrupedal robot and hexapod robot with pronking gait have been controlled by Fuzzy Logic and Artificial Neural Based Fuzzy Logic control algorithms. Although there have been a lot of control algorithms used to perform different gaits of legged-robots, the proposed fuzzy logic control (FLC) and Artificial Neural Based Fuzzy Logic(ANFIS) control algorithms for pronking gait have never been utilized to control gaits of legged-robots in the literature in the limit of authors knowledge. It has been seen that the simulation results of the quadrupedal and hexapod robots controlled by Fuzzy Logic and Artificial Neural Based Fuzzy Logic control algorithms are better than those of the existing studies in the literatures in terms of energy consumption and stability of motion.In addition, the number of the performed control actions during the phases of pronking gait in this thesis is smaller than those of the studies in the literature. As a result, the proposed control methods in this study makes both the system control easy and the system performance increase by decreasing the run time for each loop.In order to apply the designed control algorithms and analyze the performances of these controllers, four-legged quadrupedal robot and six-legged hexapod robot models have been considered. In addition to that, three different practical control algorithms have been applied for each robot model. These control algorithms called Fuzzy Logic and Artificial Neural Based Fuzzy Logic control algorithms have most commonly been used in robotics applications. The mathematical models of the considered systems were obtained then they were transformed in the state-space forms. For simulations, the MATLAB code was prepared and ANFIS Toolbox in MATLAB was used for the Artificial Neural Based Fuzzy Logic Controller. The obtained results were presented in graphical and table forms and examined. As a result, it has been seen that the performances and applicability of Fuzzy Logic and Neural Based Fuzzy Logic controller for pronking gait of guadrupedal robot and hexapod robot are much better than PD controller.Keywords: Legged Robots, Dynamic Model, Four-Legged Robot, Quadrupedal Robot, Six-Legged Robot, Hexapod Robot, Artificial Intelligence, Fuzzy Control, ANFIS Control.