Dynamical optimal training for interval type-2 fuzzy neural network (T2FNN)
Recent works in fuzzy logic systems (FLSs) have demonstrated the superiority of the higher order FLSs (type-2 FLs) compared to their traditional counterparts (Type-1 FLS) in terms of handling uncertainties. However, due to the complexity and the huge computational time of the type-reduction process, their applications in real-time is limited to simple cases. In this paper, we propose a hybrid approach to decrease the computational time of the type-reduction process. A modified interval type-2 fuzzy neural network (MIT2FNN) is developed for the navigation of mobile robots in unstructured and dynamic environments. The MIT2FNN includes a type-2 fuzzy linguistic process as the antecedent part, and a two-layer neural network as the consequent part. The back propagation algorithm is utilized to adjust the parameters of the MIT2FNN controller. The experimental results obtained using an omnidrive mobile robot named Robotino validate the effectiveness and reliability of the proposed approach.