An Adaptive Direct-Discretized RNN With Fuzzy Factor for Handling Future Distinct-Layer Inequality and Equation System Applied to Robot Manipulator Control

This paper proposes an adaptive direct-discretized recurrent neural network (ADD-RNN) algorithm with fuzzy factor to address the problem of future distinct-layer inequality and equation system (FDLIES). For comparison purposes, a traditional indirect-discretized RNN (TID-RNN) algorithm is also developed to solve the same problem. The superior performance of the proposed ADD-RNN algorithm is validated through numerical experiment. The applicability and meliority of the proposed ADD-RNN algorithm are further demonstrated by applying it to tackle a motion planning and control problem of a robot manipulator with an additional constraint, which is formulated as a specific FDLIES problem.