Navigational Analysis for Underwater Mobile Robot based on Multiple ANFIS Approach

Multiple neuro-fuzzy inference systems using hybrid learning algorithm as an adaptation mechanism have been focused here for navigation of autonomous underwater vehicle (AUV). The underwater vehicle can be exhibited as six-dimensional nonlinear and coupled equations of motion associated with variations of hydrodynamic coefficients which are difficult to model in a realistic manner. Without earlier acquaintance, the feed-forward neuro-fuzzy controller can be directed to obtain the unknown parameters of the model which may aid motion planning strategy of underwater robot by overlooking the nonlinear effects of the AUV dynamics. By amending fuzzy membership function of neural networks, the benefits of fuzzy logic and neural network can be mingled, such as capability of FIS to deal with uncertainty, employing human perception and comprehensive approximation as well as adapting competence of neural networks. ANFIS has been trained with the hybrid-learning mechanism which employs back-propagation-based gradient descent approach and least squares estimate (LSE) to estimate parameters of the model. This approach instigates faster decision- making, obstacle avoidance and also tracking targets. The simulated analysis may authenticate that the heuristic navigational approach is able to negotiate with chaotic environment during navigation of under-water robot