Design And Analysis Of Path Planning For Robotic Fish Based On Neural Dynamics Model

Most of the robotic fishes relying on tail fins in swimming have severe kinematic restrains to turn while maintaining the same position. In this respect, it is difficult to plan an optimal path for robotic fish to complete the given tasks, such as searching or tracking. Therefore, in this paper, we explored the kinematic characteristic of a robotic fish with tail fins for the purpose of deriving optimal paths. At first, the effects of the frequency, swing amplitude and deflection of the caudal fin on the forward velocity and turning radius were tested. Subsequently, we applied a neural dynamics-based path planning method to assist the robot to complete prescribed tasks. In the path planning, we converted real-time maps to pixels. Moreover, we defined a single pixel or a collection of multiple pixel points as a neuron, the initial target as an excitatory input, and the obstacles as inhibitory inputs. Then the adjacent neuron values were iteratively passed. For obstacle avoidance, this method is tested to avoid the intersection of the pixel points that are covered by the robotic fish and the obstacles. On the basis of this, a path planning method was proposed and was demonstrated to be feasible. The theoretical optimal path can be computed in various motion patterns by the method, which is adaptable and of wide range of application.