Environment modeling of AS-R robot based on BP neural network

There are some limitations when environment model of AS-R mobile robot is obtained by only using a separate CCD camera or ultrasonic sensor array. The BP neural network is designed to fuse information of the two sensors to obtain the robot´s environment model in this paper. The input of BP neural network is the intercept and slope of the edge line obtained CCD camera and ultrasonic sensor arrays, which is processed by through the global coordinates of coordinate transformation. The output of BP network is the fused intercept and slope of the straight edge. Experiment shows that environment model is feasible for the AS-R mobile robot and the environment modeling method has more reliability and accuracy.