A combined neural network design and application

The functions of different neural networks are different. According to the specific problem, rational combination of different neural networks to solve problems should be a meaningful research way. In this paper, a new type of neural network model of CMAC (Cerebellar Model Articulation Controller) neural network and BP (Back Propagation) neural network combination is introduced. The combined network uses the output of CMAC network as the input of BP network, this is equivalent to the addition of an “extra layer” in front of BP network input layer, CMAC network input and BP network input are the same. CMAC network output is connected with the nodes of BP network input through the adjustable weights. The combined neural network retained their expertise and at the same time with the advantage of fast learning speed and good generalization and so on. And used it solving the issue of robot obstacle avoidance, it is a very good solution to the issue of robot obstacle avoidance under the unknown complex environment.