Corridor scene classification method based on FPGA mobile robot pulse neural network

Abstract

The invention discloses an FPGA-based mobile robot pulse neural network corridor scene classification method, and the method comprises the steps: measuring the environment information of the surroundings of a mobile robot in different corridor scenes in real time through employing an ultrasonic sensor disposed on the mobile robot; inputting information measured by the ultrasonic sensor into a pulse neural network corridor scene classifier realized by utilizing an FPGA (Field Programmable Gate Array), and performing pulse coding and scene identification on the information measured by the ultrasonic sensor by a pulse neural network arranged in the pulse neural network corridor scene classifier; and a corridor scene where the mobile robot is located is judged through a pulse output mode of the pulse neural network. According to the method, the pulse neural network corridor scene classifier based on the FPGA is adopted to classify the corridor scene where the mobile robot is located, the method is not affected by the illumination condition, the needed sensor measurement information is little, the classification speed is high, the accuracy is high, and the environment perception ability of the mobile robot can be improved.

Description

technical field [0001] The invention relates to a corridor scene classification method based on FPGA mobile robot pulse neural network realized by hardware, belonging to the technical field of pulse neural network. Background technique [0002] Artificial Neural Networks (ANNs) is an important aspect of intelligent technology. At present, most applications based on neural networks are in the stage of computer simulation. Many application algorithms are not implemented by hardware and cannot be used in practical applications to take advantage of the parallel and high-speed computing of neural networks. Therefore, the hardware implementation of neural networks is based on neural network-related applications. The key link of practicality. At present, the hardware implementation method of neural network has gradually become a research hotspot. [0003] So far, some scholars have divided the development of artificial neural networks into three generations, among which the third...

AI Technical Summary

Problems solved by technology

At present, mobile robots generally recognize corridor scenes based on vision. This method is greatly affected by lighting conditions, and there are many sensor measurement data, which seriously affects the recognition speed and accuracy of scenes, and reduces the accuracy of mobile robots. environmental awareness

Images