Intelligent control of three-dimensional motion of the robotic fish with central pattern generator

In nature, fish are the best swimmers with their fast swimming, instantaneous acceleration and high motion abilities. A fish-like underwater robot can swim faster, quieter and lower power consumption than a conventional propeller underwater vehicle with generated thrust force. When these features are considered, it is seen that fish have very suitable biological structure for underwater robot design with multi-link tail mechanism. In this thesis, a Carangiform type biomimetic robotic fish prototype with two-link flexible tail mechanism, which has three dimensional autonomous swimming ability, was developed and intelligent control of three-dimensional motion of the robotic fish was performed with the proposed biomimetic control structure. In order to carry out simulation studies and design control algorithms, three-dimensional 6 Degrees of Freedom (DoF) nonlinear mathematical model of the robotic fish was obtained. 6-DoF model describes the three-dimensional motion of the robotic fish. Lagrange method was used to obtain this model and hydrodynamic forces acting on the tail were calculated to derive the realistic effects of water. For the design procedure of the robotic fish prototype, morphological characteristics of a real Carangiform carp fish was adapted. Central Pattern Generator (CPG) model inspired by the spinal cord neurons of Lamprey was applied to perform locomotion control of the robotic fish. The neural Lamprey CPG model was composed of unidirectional and chain linked oscillator series. Rhythmic, oscillatory and sinusoidal output signals were obtained with this model to drive servo motors of the tail. A sensory feedback mechanism based on Lamprey neuron and two different fuzzy controllers were established to achieve the closed loop control structure of the robotic fish and Central Nervous System (CNS) was developed. In order to validate the proposed methods, open and closed loop analyzes of robotic fish prototype were performed both in the simulation environment and experimental pool. In the open loop experiments, forward and turning motions of the prototype were analyzed and characteristic features of the robotic fish were determined. In the closed loop experiments, controls of the yaw angle, autonomous swimming avoiding the obstacles and yaw angle avoiding the obstacles were performed. Thus, sensory feedback closed loop forward and turning motion controls were achieved depending on the characteristic features of the robotic fish. In addition, pitch angle was controlled to provide up/down motions with center of gravity control mechanism of the robotic fish. The effectiveness and performance of the three dimensional motions of the designed robotic fish prototype were shown with the simulation and experimental studies.