As part of intelligent fish farm, the use of the underwater cleaning robot to clean aquaculture cages is of great significance to the future intelligent development of the aquaculture industry. It is a useful method to let the cleaning robot perform cleaning tasks along the preset path to improve the cleaning efficiency of aquaculture cages, and to solve the problem of path tracking, an adaptive controller combined with backstepping sliding mode and RBF neural network is designed. In addition, considering engineering applicability, a thrust allocation method based on the combination of normalization and pseudo-inverse method is adopted in this paper. The simulation results show that the designed controller is of excellent robustness to uncertain disturbances.
With the increasing demand for seafood in human society, the aquaculture industry has developed rapidly, and it has further promoted the continuous expansion of the scale of fish farms around the world (Ola, B, Sm, C, Sam, D, and Nk, A, 2021). Fish farming in aquaculture cages using marine resources not only improves the quality of fish but also creates a large and sustainable economic value for society. However, the aquaculture cage is attached by marine organisms easily, easily dirty, and blocked, which leads to poor water exchange and fish disease, so it needs to be cleaned regularly (Ouyang, B, Wills, PS, Tang, Y, Hallstrom, JO, and Ouden, C,2021). The traditional manual cleaning of cages is labor-intensive, inefficient, and expensive. Therefore, with the support of modern science and technology, it will speed up the modernization of the aquaculture industry by using underwater cleaning robots.
In recent years, the research of underwater cleaning robot has attracted extensive attention (Song, C, and Cui, W, 2020; Wu, X, Wang, S, Wang, X, and He, G,2021). It becomes necessary to consider how to use various advanced technologies to achieve accurate path tracking of underwater robot (Xu, R, Tang, G, Han, L, and Xie, D, 2019). In the complex marine environment, the underwater robot has characteristics of time-varying hydrodynamic, strong motion coupling, highly nonlinear model, irregular ocean current disturbance force, and the hydrodynamic parameters that are hard to measure accurately lead to difficulty to establish a dynamic model accurately. These factors will lead to the deterioration of the path tracking performance of the underwater robot (Chao, Yang, Feng, Yao, MingJun, and Zhang, 2018). In order to overcome the internal and external uncertain disturbances of the underwater robot during the task execution and meet the actual accuracy requirements of the path tracking, the designed controller is required to be insensitive to parameter perturbations and have good robustness. Nowadays, a series of methods applicable to the motion control of underwater robots have been proposed, such as PID control, adaptive control, fuzzy control, sliding mode control, neural network control, and some integrated control methods, some of which have been applied to the actual environment. In addition, some advanced control strategies have also been considered for use in the field of marine equipment (Sun, Y, Zhang, C, Zhang, G, Xu, H, and Ran, X, 2019; Yang, Y, Tu, H, Song, L, Chen, L, and Sun, J, 2021).
