Research on Indoor Patrol Robot Location based on BP Neural Network

Abstract

With the development of science and technology, patrol robots play a more and more important role in indoor monitoring. It is the location of the abnormal situation that the most important task of the patrol robot is to display accurately, so studying its location is of vital significance. In the case of known indoor maps, it is indispensable for patrol robots to select the odometer which achieves accurate positioning. The traditional odometer is mainly realized by means of displacement, laser signal or multi-sensor fusion. However, the positioning often fails due to problems such as mechanical structure, error accumulation, power consumption and heating. In addition, the phenomenon that peripherals occupy limited robot resources is also prominent. To this end, this paper proposes to use BP neural network instead of sensors to obtain the odometer. This method is that the laser data, constructing the map of the robot, and the BP neural network determined are combined to achieve precise positioning and reduce development costs. Experiments have proved that this method may be used to improve the endurance of the robot battery in addition to accurate positioning.