Mobile robot control system design based on fuzzy neural networks, страница 3

It is easy to see that both fuzzy networks are similar. The distinction consists in that the TSK network result is represented by a first order polynomial, in the Wang-Mendel network the result is represented by a constant. Thus, the Wang-Mendel network is the special case of the TSK network from the functional viewpoint.

EXAMPLE OF APPLICATION

The training of the designed neural networks had been performed for the case of the hybrid algorithm with the teacher. This algorithm is based on minimization of the objective function, which can be computed by [5]

                                                                                                                                                 (6)

where p is number of the training pairs (x, d). According to this method the determination of the unknown parameters is performed in two steps. At the first step the certain values of the membership function parameters are fixed and the TSK polynomial linear parameters  are calculated from the linear equations system. At the second step the linear parameters values are fixed and the network actual output signals  for i = 1, 2, …, p are computed, then the error vector  is calculated. The error signals are applied to the network input. The objective function gradient is calculated with respect to the parameters . The gradient of the objective function is used for the network training. The training process is being continuing to the stabilization of the parameters.

CONCLUSION

In this paper a mobile robot control system design was presented when desired motion trajectory is not well defined. It was shown the solution of the problem is based on use of neural networks and fuzzy logic. The hierarchical structure of control system with the fuzzy neural controller is obtained. It allows to use methods of the coordinating control for the design of motion algorithms at the low level of the control system. 

REFERENCES

[1] A.V. Arzhanik, O.E. Vashenkov, A.V. Lyamin, V.I. Shtefan, "The mobile robot "Neva arrow", Mechatronics, automation, control, Moskow: Novye tehnologii, no. 2, 2004, pp. 23–26 (in Russian).

[2] Mobile robots and mechatronic systems: Materials of workshop, Moskow: Moskow University, 2000 (in Russian).

[3] A.V. Arzhanik, "Technical vision system of the mobile robot", The bulletin of conference of young scientists of St.Petersburg State University of Information Technologies, Mechanics and Optics. St.Petersburg: St.Petersburg State University of Information Technologies, Mechanics and Optics, vol. 2, 2004, pp. 97–102 (in Russian).

[4] A.V. Arzhanik, "Design of algorithms of motion control of mobile robot on the base of fuzzy logic for gate-problem", The bulletin of II conference of young scientists of St.Petersburg State University of Information Technologies, Mechanics and Optics. St.Petersburg: St.Petersburg State University of Information Technologies, Mechanics and Optics, vol. 2, 2005, pp.112–120 (in Russian).

[5] S. Osowski, Neural networks for processing the information, Moskow: Finances and statistics, 2004 (in Russian).