The Positioning Prediction System Of Agricultural Vehicle Autonomous Navigation Based On Multisensor Fusion

In the autonomous navigation of agricultural vehicles,accurate positioning of vehicles is particularly important.Accurate positioning data can not only provide the vehicle with the absolute position information of the operating environment and itself,but also accurately determine the route,calculate the transverse deviation and achieve high-precision path planning and so on.However,the satellite positioning system with high precision and high frequency output is expensive and difficult to be popularized.Although the low frequency satellite positioning system is cheap,the dynamic positioning accuracy is poor.At the same time,navigation satellite signal is weak,easy to be occluded and lost.In view of the above problems,based on the realization of the autonomous navigation of the vehicle,this subject makes a prediction of the positioning information of the future moments in the vehicle driving process and provides the precise predicted positioning information for vehicles.Firstly,in this paper,the dongfanghong SG250 tractor was modified and a multi-sensor-based intelligent mobile experimental platform was designed.RS485 bus serial port communication,automatic control and other technologies are used to design the electronic control hydraulic steering module and wireless corner signal acquisition module,further enhance the intelligence of agricultural vehicles.Secondly,the overall design of agricultural vehicle autonomous navigation and positioning prediction system based on multi-sensor fusion is presented.The design is mainly divided into two steps:realizing the autonomous navigation of agricultural vehicles and completing the positioning prediction of agricultural vehicles in the autonomous navigation.The overall design of autonomous navigation is mainly composed of the data acquisition module and the control module of the mobile test platform.The data of all kinds of sensors are collected and processed in a multi-threaded way,and relevant data are input into the autonomous navigation control module for vehicle corner control to complete the autonomous navigation function.On this basis,aiming at the slow refresh frequency of RTK-GPS positioning data and easy occlusion,two interpolation-gray prediction algorithms based on INS and vision were proposed to complete the autonomous navigation positioning prediction function of agricultural vehicles by using the method of multi-sensor data fusion positioning.Finally,the autonomous navigation and positioning prediction system of agricultural vehicles is tested and analyzed.Experimental results show that:(1)According to different routes,manned location prediction of agricultural vehicle INS interpolation-grey test:when vehicles are driven on the general route,interpolation-positioning forecast data of grey theory in the east direction error mean and mean square error is 0.0071 m,0.1082 respectively,due north direction error mean and mean square error were 0.0318 m,0.1081;When the vehicle is driving on a special route,the mean error and mean error variance of the positioning prediction data of interpolation-gray theory are 0.0452m and 0.4503 respectively in the positive east direction,and 0.0034m and 0.323 respectively in the positive north direction.The errors of the general gray prediction location and interpolation-gray prediction method vary with different driving routes.At the same time,under different routes,the error mean and error mean variance of interpolation-gray prediction are both smaller than the traditional gray prediction,and its accuracy is improved by about 52%.(2)In view of the different driving manual and autonomous navigation,the location prediction of vision interpolation-grey test:when the vehicles running on the control mode,interpolation-positioning forecast data of grey theory in the east direction error mean and mean square error is 0.1179 m,0.2151 respectively,due north direction error mean and mean square error were 0.1628 m,0.9334;When the vehicle is driving in the autonomous navigation mode,the mean error and mean error variance of the positioning prediction data of interpolation-gray theory are 0.0082m and 0.0038 respectively in the positive east direction,and 0.0039m and 0.0018 respectively in the positive north direction.The errors of the two prediction methods vary with different driving routes.In different routes,the error mean and error mean variance of interpolation-gray prediction are both smaller than traditional gray prediction,and their accuracy is improved by about 57%.