Designing the 6-DOF Massive Parallel Arrays with Artificial Intelligence Control

Binary-Discrete State Manipulators (b-DSMs) are force regulated manipulators that undergo continuous motions despite being commanded through a finite number of states only. Designing a real-time control of such systems requires fast and efficient methods for solving their inverse static analysis (ISA), which is a challenging problem of this paper. In particular, an artificial intelligence method based on neuro-fuzzy method is proposed to investigate the on-line computation and the generalization error of ISA problem of a class of b-DSMs featuring two-state force actuators and six degree of freedom. The main advantages of a neuro-fuzzy system for b-DSMs are: it interprets IF-THEN rules from input-output relations (orientation, moment and binary state) and focuses on accuracy of the output network and offers efficient time consumption for on-line computation. The paper proposed two architectures which are based on the Neuro-Fuzzy Takagi-Sugeno (NFTS) inference scheme with Gaussian membership functions. They are NFTS and the Look-Up Table version of NFTS, which is called as NFLUT . Both structures are with multivariate input and multi-state outputs, such as orientations and moments as input networks and binary state of the b-DSMs as output networks. The learning procedure uses an accelerated LMA with optimal training parameters with at least half-million iterations with different 10 membership functions, employ 12 of the input-output correspondences from the known input-output dataset. For experimental database,the NF structure is tested using 1024 dataset. The optimized membership function (N) after two weeks searching time using Hill Climbing (HC)procedure is N = 17 for the 10-binary Massive Parallel Robots (MPRs). Regarding model performances for the ISA solution, the NFLUT features better generalization ability compared to the NFTS model but requires a rather larger computational time during on-line testing phase.