Manoeuvring planar snake robot in uncertain underwater condition using adaptive neural network sliding mode control

In this work, the adaptive radial basis function neural network sliding mode control (RBF-NN SMC) is proposed for the motion control of the underwater snake robot by estimating uncertain environments. The virtual holonomic constraints are defined such that the autonomous motion of the snake robot can be generated by designing the controller on a reduced-order dynamical system. Generally, an adaptive SMC is designed with lumped uncertainties and disturbances to deal with uncertain environmental conditions. However, in the proposed control approach, the environmental uncertainties are estimated using RBF-NN. It estimates the system dynamics by using the feedback of the system response without prior knowledge of the uncertainties. The Lyapunov stability analysis is performed to identify the controller design parameters that stabilize the dynamical system. Moreover, the performance of the proposed control scheme is verified by introducing the measurement noise. The effectiveness of the proposed control method results in a reduction in the chattering effect and error response compared to the existing adaptive SMC method. The computational complexity in tuning the neural network hyper-parameters is the challenge that can be addressed in future.