Design And Implementation Of Intelligent Search Robot For Radioactive Source Based On Edge Computing

With the rapid development of food,medicine,industry,biology,national defense and aerospace,the extensive application of radiation science has been brought.While enjoying the revolution brought by nuclear technology,the issue of nuclear safety has also been raised to a new height.How to effectively supervise radioactive sources,how to quickly search for lost or unknown radioactive sources,and how to eliminate the dangerous and time-consuming defects brought by manual search of radioactive sources are the basic prerequisites for further development of radiation technology.The traditional method of searching for radioactive sources is to put on heavy protective suits for experts to detect radioactive sources on site.However,radiation exposure is still unavoidable for those person.The technology of robot search has been monopolized by foreign countries,and has been slow to develop at home.Only a few universities and colleges have developed remote control search robots and unmanned aerial vehicles(UAVs).But they all have a problem,which is that you have to do it manually,not automatically.Therefore,in order to solve this pain point,this paper designs an intelligent radioactive source search robot to realize automatic and intelligent search of radioactive sources and fill the gap in China.This paper mainly uses the edge acceleration technology of machine learning and robot technology to study the location and search of radioactive sources.Firstly,the hardware and software of the robot are studied systematically.Secondly,the mechanical structure with stronger practicability and the hardware circuit with better expansibility are designed.Thirdly,the ROS(Robot Operating System)transplantation application has been completed.Then,the research and design of node algorithms including servo motor driver,lidar,gamma camera,SLAM(Simultaneous Localization and Mapping),depth camera and edge accelerator are also provided.By applying machine learning technology to the edge end,the edge acceleration technology of image recognition is researched and applied in portable devices.And according to the needs for remote control,an Android APP is developed.The portable,automatic and intelligent radioactive source intelligent search robot is realized eventually.This paper innovatively uses interdisciplinary technology to solve the security search problem of radioactive sources.By using the MA24xx-HornedSungem edge acceleration chip,the YOLO model is run on an ordinary ARM processor to achieve the image detection of up to 15 frames,meanwhile,the millisecond control speed of the robot is realized.The heat map of the radioactive source is output through the Hent33-013 a portable gamma camera of Beijing high energy new technology co.,LTD.The self-trained Caffe source finding network model was used in conjunction with HornedSungem for triangulation.The gradient descent of radial source intensity is innovatively combined with SLAM obstacle avoidance navigation algorithm.Finally,the radiating source positioning accuracy and the autonomous obstacle avoidance search speed can meet the practical application requirements.For a radioactive source with a strength of 15 mCi,the self-seeking time can be as short as 120 seconds at a distance of 10 m.