Modular Weightless Neural Network Architecture for Intelligent Navigation

The standard multi layer perceptron neural network (MLPNN)type has various drawbacks, one of which is training requiresrepeated presentation of training data, which often results in verylong learning time. An alternative type of network, almost unique, isthe Weightless Neural Network (WNNs) this is also called n-tuplenetworks or RAM based networks. In contrast to the weighted neuralmodels, there are several one-shot learning algorithms for WNNswhere training takes only one epoch. This paper describes WNNs forrecognizes and classifies the environment in mobile robot using asimple microprocessor system. We use a look-up table to minimizethe execution time, and that output stored into the robot RAMmemory and becomes the current controller that drives the robot.This functionality is demonstrated on a mobile robot using a simple,8 bit microcontroller with 512 bytes of RAM. The WNNs approach iscode efficient only 500 bytes of source code, works well, and therobot was able to successfully recognize the obstacle in real time.