Development of a control system for core-sampling USV that combines sliding mode control and neural network

Morito, Makoto*; Fujii, Shun*; Yoshimura, Koki*; Sanada, Yukihisa   ; Baba, Shoichiro*; Matsunaga, Hiroshi*; Mori, Takami*; Sato, Keiichiro*; Tahara, Junichiro*

This study proposes a method that combines sliding mode control and neural networks as a control method for unmanned surface vehicle to measure radiation in submarine soil. Sliding mode control is used as the base of control input, and corrective inputs are added using a neural network controller so that one of the parameters of sliding mode control, , and its rate of change are reduced. After producing a control system using this method, we conducted simulation tests and sea area tests to evaluate whether the survey could be conducted using this method.