Obstacles pattern recognition based on BP neural network

Adopted theory of BP neural network to realize pattern identification of obstacle. Constructed a kind of neural network classifier based on BP algorithm, according to the characteristics of omnidirectional mobile robot. Took the distance of obstacle acquired by ultrasound sensor as input, classification pattern expected as output, then, trained the parameter of BP neural network classifier through training the data in sample library. Simulated the process of training with Matlab software. During training, this pattern classification could gain less error and gain correct result by adopting testing sample to test classifier. Showed that this kind of BP neural network classifier can execute effectively pattern classification to obstacles around mobile robot.