Cooperative control for trajectory tracking of robotic fish

This paper is concerned with cooperative control for trajectory tracking of multiple biomimetic robotic fish using neural network based sliding mode control method. An experiment system is set up for multiple robotic fish cooperation, in which the information of robotic fish and the target points of the planned trajectory are sent to each robotic fish. Based on the received information, robotic fish can make decisions autonomously to track the planned trajectory in a decentralized way. Considering the difficulties in modeling the system due to the high nonlinearity of robotic fish and the complex hydro-environment, radial basis function neural networks are invoked to approximate dynamics of the biomimetic robotic fish with a gradient descent algorithm to optimize the network parameters. Furthermore, to deal with the model uncertainty, a discrete sliding mode control approach based on the neural networks, along with a target planning method, is applied to control the robotic fish to achieve cooperative trajectory tracking task. Experimental results are included to show the effectiveness of the proposed approach.