Research On Visual Control System Of Forest Fire Prevention Robot Based On FPGA

In the complex environment of the forest,the forest fire prevention robot completes the effective ranging and obstacle avoidance of the surrounding trees and obstacles in the process of patrol progress,which is the basis for completing real-time fire image recognition and alarm.In addition,accurate and timely identification and early warning of forest fires in the early stage of occurrence,rapid extinguishing and elimination,and effectively avoiding the largescale spread and development of forest fires are urgent problems that need to be solved at present.In view of the current problems faced by forest fire prevention robots,because FPGA(Field Programmable Gate Array)uses parallel processing and has the advantages of fast processing speed in image processing,this paper proposes the forest fire prevention robot visual control system based on FPGA.The binocular stereoscopic vision subsystem and flame image recognition subsystem are designed to form a forest fire prevention robot visual control system,with the help of FPGA parallel assembly line processing to reduce the delay of system image processing,improve the real-time performance of the system,the system integrates image processing algorithms such as color space conversion,image denoising filter preprocessing and flame recognition algorithm on the FPGA control core,and cooperates with the host computer to realize the ranging and obstacle avoidance and flame recognition of forest fire prevention robot.The main research contents and results of the paper are as follows:(1)Carry out the research of binocular stereoscopic vision ranging,use the binocular camera and FPGA as the image acquisition and pre-processing platform of the system,and combine the Gigabit Ethernet module to transmit the binocular image to the PC.Then MATLAB is used to complete the bi-target setting of the system,so as to obtain the internal and external parameters of the binocular camera.Then based on the VS2022 platform,stereoscopic matching is carried out through SGBM(Semi-Global Block Matching)algorithm and BM(Block Matching)algorithm to obtain parallax plots and distance information,and then the current distance between trees and obstacles is obtained through the three-dimensional coordinate information of key points of trees and obstacles.Experimental tests show that in terms of error rate,the relative error range of the distance measurement results of trees and obstacles of the SGBM algorithm is 0.62%～1.79%,and the relative error range of the distance measurement results of trees and obstacles of the BM algorithm is 0.62%～1.88%,and the accuracy and speed of the measurement distance of trees and obstacles by the algorithm in this paper meet the actual needs.(2)To study the binocular stereo vision obstacle avoidance algorithm,first through the discussion of the basic principles of the traditional artificial potential field method,point out its problem defects,and then improve the artificial potential field method obstacle avoidance algorithm by modifying the gravitational potential field function and repulsive potential field function of the artificial potential field method,and finally simulate the improved artificial potential field method through MATLAB.The simulation results show that the improved artificial potential field method achieves 95.83% of the power to achieve,and the obstacle avoidance ability meets expectations,which effectively overcomes the shortcomings of the traditional artificial potential field method and realizes obstacle avoidance of trees and obstacles.(3)Design the flame image recognition system,realize the color space conversion and grayscale processing of color image data with the help of the parallel processing structure of the FPGA,use Erosion algorithm and Dilation algorithm to achieve the elimination of environmental noise,design the flame recognition algorithm according to the flame color characteristics and motion characteristics,carry out flame image recognition,and send the realtime image data to the remote monitoring and control platform based on QT5 of the PC computer through the Gigabit Ethernet transmission module to complete the fire alarm operation.The experimental results show that the flame recognition effect and denoising ability of the system meet the expectations,and the Gigabit Ethernet transmission module can send the image to the PC for real-time display through two resolution formats of 1280×720 or 640×480,which meets the expected requirements.The visual control system of forest fire prevention robot based on FPGA designed in this paper can realize the ranging and obstacle avoidance of trees and obstacles around the forest fire prevention robot,realize the real-time identification of flame and remote transmission of on-site environmental images,and meet the actual needs of forest fire prevention robot for ranging obstacle avoidance and flame identification,and has a wide range of application value.