Information Fusion Technology Used In The Intelligent Mobile Robot Navigation

In order to achieve the accurate understanding about the object and makecorrect decisions the paper studies the information fusion technology on how to usethe additional information effectively to reduce the vague incomplete information.Using multi-sensor information fusion technology solves the problems of intelligentmobile robot navigation in complex and unknown dynamic environments effectively.However, the selected information fusion algorithm is not the same for a differentrobot control system or in a specific environment. Because each information fusionalgorithm has inherent drawbacks, there’s not a fusion algorithm which is suitablefor all robot control systems and operation enviroments. So, we should further studythe improvement of information fusion algorithm.In this paper, it mainly studies on the neural network and fuzzy logicinformation fusion algorithm and the optimization algorithms which are used inrobot navigation. Because the weights convergence of BP neural network is veryslow, this paper uses the learning rate adjustment factor to adjust the learning rateadaptively and reduce the training time. The two-tier BP neural network model isdesigned for the identification of obstacles. The process of information fusion isdivided into the feature level and decision-making level fusion, which improves therobustness and flexibility of the system. Simulation results show that the type ofobstacles can be identified effectively and the convergence speed of the weight isvery fast.This paper combines the advantages of neural network and fuzzy logicalgorithm to construct a fuzzy neural network control model to achieve real-timenavigation of mobile robot in an unknown enviroment. The neural network learningalgorithm is used to adjust the parameters of membership functions, designing alearning algorithm to reduce the redundant fuzzy rules automatically, using the statememory solutions to solve the problem of "dead ends" in robot navigation.Through the simulation verifies the validity of the model.