Study On The Method Of Preventing And Avoiding The Safety Of Intelligent Grain Transporters Based On Multi-sensor Fusion

As the total national grain production increases year by year,a series of problems arising from grain storage and transshipment have become increasingly prominent.Traditional grain transfer vehicles transferring grain in grain depots have problems such as low transfer efficiency,high dependence on manpower,long waiting time for vehicles and serious pollution.At the same time,there are vehicles,pedestrians and other non-fixed obstacles in the grain depot environment,and there are safety problems such as collisions and cuts in the transfer process.In order to solve the above problems,this project takes the intelligent grain transfer vehicle as the research object,builds the hardware system of millimeter wave radar and vision sensor target detection,and also filters the millimeter wave radar raw data,focuses on the target detection of images by YOLOv5,the fusion of millimeter wave radar and vision sensor and the obstacle avoidance method of intelligent grain transfer vehicle,and realizes the intelligent grain transfer vehicle in the grain depot scene The purpose of obstacle avoidance of the intelligent grain transfer vehicle in grain storage scenario is achieved.Firstly,by comparing the advantages and disadvantages of commonly used obstacle detection sensors at home and abroad,and analyzing the application scenarios of intelligent grain transfer vehicle,an obstacle detection hardware system combining millimeter wave radar and vision sensors is built.By analyzing the problem of safety targets and false alarm targets in the millimeter wave radar raw data,the millimeter wave radar raw data is filtered by using the lateral longitudinal distance threshold and life cycle method,and the millimeter wave radar raw data and the filtering effect are compared and tested,and the test results showed that the method can effectively filter the safety targets and false alarm targets in the raw data.By comparing and analyzing the advantages and disadvantages of the image target detection algorithm,YOLOv5 is used to detect the target in the image of the vision sensor,and a training set is constructed based on part of the KITTI dataset and adding self-built data,and the weights of YOLOv5 are trained again,and the experimental results showed that YOLOv5 could effectively detect the obstacle in the image under both insufficient and sufficient light conditions.The experimental results show that YOLOv5 can effectively detect obstacles in images under both low and high illumination conditions.Second,to address the problem of spatially unsynchronized millimeter wave radar and vision sensor,a spatial synchronization model is constructed by sensor calibration and coordinate system conversion to realize the spatial synchronization of millimeter wave radar and vision sensor.For the problem that the millimeter-wave radar and vision sensor are not synchronized with the vision sensor in time,a time synchronization model is constructed by interpolation method to realize the time synchronization of millimeter-wave radar and vision sensor.Based on the temporal synchronization and spatial synchronization of millimeter-wave radar and vision sensors,an obstacle detection information fusion model of millimeter-wave radar and vision sensors is constructed by decision-level fusion method and target matching strategy,and a millimeter-wave radar and vision sensor fusion test is conducted based on this model,and the test results show that the model can effectively fuse the obstacle detection information of the two sensors.Finally,an improved artificial potential field method obstacle avoidance model is constructed for static obstacles in the grain depot environment and simulated in matlab based on this model,which shows that the intelligent grain transfer vehicle can effectively avoid static obstacles in the grain depot environment.A graded braking strategy obstacle avoidance model is constructed for dynamic obstacles in the grain storage environment,and a real vehicle test is conducted based on this model,which showed that the intelligent grain transfer vehicle could effectively avoid dynamic obstacles in the grain storage environment.