Implementation Of Data Acquisition And Parallel Processing System Applied To Insect Robot Platform

With the rapid development of electronic manufacturing technology and biomimetic neurological research,it is possible for biorobots under controlled stimuli to rely on their excellent biological body functions to complete the instructions of the controller,and to replace humans in complex scenarios to perform tasks.Different from traditional biological robots,insect robots have more imaginable applications due to their excellent flying ability,ultra-small size,and strong concealment.In the research on the relevance of the flight attitude and trajectory of the insect robot under stimulation,the system platform constructed according to the experimental requirements needs to have the following functions: 1)The platform can support insect flight experiments in a wide range of scenes.2)The platform has the ability to monitor and collect insects in a full-scene twodimensional plane.3)The system can perform image fusion on multi-viewpoint behavior data.4)The system can have good storage and efficient processing capabilities for massive image data.In the field of insect robots,this subject takes the bee controlled body as the research object,and aims to realize the image collection and efficient data batch processing of the insect robot platform,and aims at the problem of image and video parallel processing under Hadoop technology,and constructs a horizontally scalable The full-scene acquisition system and the improved Map Reduce parallel batch processing platform based on image compatibility.The main work content of the paper is as follows:1.Designed a platform scheme applied to the insect robot platform acquisition system.The camera array structure based on the two-dimensional panoramic data collection from the bird’s-eye view,combined with the insect wireless control system,realizes the closed-loop behavior recording of the insect robot.Later,for the complete unity of panoramic data,the image mosaic scene fusion of multi-viewpoint data sources was designed,and a set of platform upper computer software interface was developed to integrate all the functions of the existing system,which is convenient for experimenters to operate and use.2.Constructed a parallel processing platform for massive image data of insect robots based on Hadoop technology.Use video splicing technology and ffmpeg to process video streams to achieve precise frame division.Based on the Map Reduce model,a parallel processing extension class that matches insect images is designed,and for massive fragmented small file image data,the storage is optimized through the combined fragmentation processing model structure.Design the Mapper layer adaptation process for edge detection of Scharr operator to realize batch processing of images of insect robots.3.The experiment is combined with the video and image data sources in 1 to verify,and the image binarization preprocessing algorithm is used to test and compare the performance of the platform framework parallel processing.The results show that compared with the traditional singlenode serial processing process,this The system has the ability to process the massive experimental data from the acquisition system more efficiently.Finally,the speedup and efficiency of the parallel system are analyzed by the parallel processing program of Scharr operator edge detection.After analyzing the indicators,it is shown that the platform can meet the high-efficiency batch processing requirements of massive insect robot platform experimental data.