Abstract— Due to the complexity of its dynamic equations, parallel robot control has become an active research area in recent years. In this application domain, model-based approaches are often time-consuming, and model-free controllers generally lack the desired performance for controlling such a fast machine. Hence, a new adaptive intelligent controller is proposed here for 3PSP parallel robots based on an auto-structuring fuzzy neural architecture. This control approach aims to reach high performance while maintaining overall stability. The proposed Dynamic Growing Fuzzy-Neuro Control (DGFNC) approach adds new rules more conservatively, hence pruning mechanism is omitted. Instead, an adaptive controller ‘adapts’ the system to parameter variations. Furthermore, a sliding mode-based nonlinear controller ensures system stability. This hybrid approach leads to less computation and faster response. The merits of DGFNC are illustrated by simulation results as applied to the 3PSP robot. I. INTRODUCTION raditional controllers are usually complex in design; and relatively time-consuming because of involving complex mathematical equations. Moreover, if the plant is changed even insignificantly, these controllers must be designed anew. Thus in recent decades, scientists have taken an interest in developing intelligent controllers, and the new controllers are progressing at a phenomenal speed. The internal obscurity of the plant and external disturbance both lead to an inefficient control, so the adaptive controllers are the only choice to deal with these issues [1-3]. Neural networks have a special place in modeling the plant’s nonlinear dynamics, in both adaptive and online approaches [4, 5]. On the other hand, fuzzy systems are used in handling uncertainties in modeling, control and decision making [6, 7]. With the advanced progression of learning algorithms, Fuzzy Neural Network structure (FNN) has become one of This paper was submitted at 2011-09-30 for review. Mohsen Jalaeian-F. and Mohammad-R. Akbarzadeh-T, Cognitive Computing Lab, Center for Applied Research on Intelligent Systems and Soft Computing, Dept. of Electrical Eng., Ferdowsi University of Mashhad, Iran (E-mails: m.jalaeian@yahoo.com , akbarzadeh@ieee.org ). Alireza Akbarzadeh Tootoonchi, Center for Applied Research on Intelligent Systems and Soft Computing, Dept. of Mechanical Eng., Ferdowsi University of Mashhad, Iran (e-mail: ali_akbarzadeh_t@yahoo.com). Damoon Azarpazhooh, Center for Applied Research on Intelligent Systems and Soft Computing, Dept. of Computer Eng., Ferdowsi University of Mashhad, Iran (email: damoon.azarpazhooh@live.com ) the most efficient hybrid methods for adaptive control for its many qualities; namely fuzzy reasoning, neuron-learning and universal approximation (in FNN based structure) [8, 9]. FNN offers a favorable structure for approximation with finite nodes, when assisted by an appropriate learning algorithm. This research, studies a new intelligent controller, and we apply an appropriate modification on this controller by considering the constraints of the plant, finally, the modified controller is used to control 3-PSP parallel robot. The introduced controller divides learning algorithm to structure-learning and parameter-learning algorithms. The purpose of structure-learning algorithm is finding the optimal nodes quantity ( * R ) and it serves this purpose with two strategies node-adding and node-pruning, which modified by some changes to the dynamic node-adding. The end of Parameter-Learning Algorithm is finding the optimal value of FNN’s parameters which includes fuzzy system parameters ( , m  ) and neural network’s parameter (  ) in relation to the optimal determined structure [1]. This task is carried out using back-propagation learning algorithm. Finally, auto-structuring mechanism produces an optimal structure for FNN and adjusts its parameters at the same time, by considering the value and differentiation of error [1, 7]. The main advantages of this hybrid controller include relief from the complexity and hardships of designing the controller, offering a self-organizing controller, an optimal structure that provides optimal calculations, retaining high performance control, overall stability ensured by the supervisory controller- chosen as sliding-mode-, and finally being adaptive and model-free thereby eliminating the effect of modeling uncertainties on controller’s efficiency [1, 3]. Based on our extended experience with this robot control simulations we had the impression that the node-pruning mechanism is time-consuming and it fails to enhance the controller. This method merely balances the quantity of nodes along with the adding-mechanism, and if removed “The Curse of Dimensionality” (node redundancy) occurs. To cope with this problem node-adding must be done very strictly, that is node-adding criterion ( th  ) must change to a dynamic threshold which can be applied by making th  a function of R . Based on our experience, it can be realized Dynamic-Growing Fuzzy-Neural Controller, as Control System of Special 3PSP Parallel Robot Mohsen Jalaeian-F., Mohammad-R. Akbarzadeh-T., Alireza Akbarzadeh, Damoon Azarpazhooh T