Research And Implementation Of A High-Precision Integrated Navigation System In Agricultural Machinery

In recent years,with the continuous promotion of precision agriculture,China’s agricultural machinery is developing towards automation,informatization,and intelligence.As one of the core technologies of precision agriculture,agricultural machinery navigation technology is at the forefront and hot spot of precision agriculture research at home and abroad.It is also an important practical way to achieve future intelligent agricultural operations and has been widely used in various production processes of agriculture.The agricultural machinery integrated navigation system combines visual positioning technology and satellite positioning technology in navigation technology,which can provide high-precision navigation services for agricultural machinery in complex farmland environments.It is of great significance to achieve precise agricultural production,reduce cultivation costs,improve yield and quality,and promote the process of agricultural modernization.To solve the problem of low image segmentation accuracy and low navigation positioning accuracy in most agricultural machinery integrated navigation systems that combine visual processing and satellite positioning,this thesis designs and implements a high-precision agricultural machinery integrated navigation system.The system uses a high-precision image segmentation model for visual image segmentation processing and improves the navigation accuracy of the system through a data fusion-based navigation positioning algorithm.The main work of this thesis is as follows:(1)To address the problem of low image segmentation accuracy in complex farmland environments,this thesis proposes an image segmentation model based on PCNN and multiscale even convolution attention.The model is used to segment the farmland boundary image to fit the navigation baseline,so that the agricultural machinery can obtain high-precision visual positioning.The model is based on the lightweight network U-Net and introduces a PCNN preprocessing module for image denoising and image enhancement.In the encoding end,the multi-scale even convolution module is used to replace the original convolution module for feature extraction to obtain comprehensive image information.In addition,by introducing an attention mechanism,the model weights different position features in the image from both channel and spatial dimensions to focus more on the farmland boundary area and reduce the computational overhead of useless features,thus improving the accuracy and robustness of image segmentation.Finally,this thesis compares the segmentation accuracy and computational cost of different image segmentation models through simulation experiments on two public image datasets,and the results show that the proposed model improves image segmentation accuracy without significantly increasing computational overhead.(2)To solve the problem that most agricultural machinery can only travel along crop boundaries inaccurately by relying solely on satellite positioning navigation,resulting in low navigation accuracy,this thesis introduces visual positioning and proposes a navigation positioning algorithm based on data fusion.First,the satellite positioning data and visual positioning data are transformed into the same coordinate system,and an integrated navigation model of agricultural machinery is established,combined with an unscented Kalman filter to filter and locate the agricultural machinery.Secondly,a point set sampling strategy based on immune particle concentration adjustment is proposed to improve the problem of accuracy reduction of the unscented Kalman filter.Finally,the navigation positioning accuracy under different navigation modes is compared through simulation experiments.The results show that the proposed navigation positioning algorithm based on data fusion effectively improves the navigation accuracy.(3)Design and implement a high-precision agricultural machinery navigation system.The system is divided into hardware and application layers in its overall architecture.The hardware layer implements data collection and transmission for various sensors,as well as motor driving for the execution mechanism based on relevant instructions.The main functional modules of agricultural machinery navigation are implemented through mobile software development technology in the application layer.Additionally,in the application layer,the system applies image segmentation models based on PCNN and multi-scale even convolution attention to improve the segmentation accuracy of the visual image processing module,in order to obtain higher-precision visual navigation points.At the same time,a navigation and positioning algorithm based on data fusion is applied to achieve the fusion processing of visual data and satellite positioning data,effectively improving the positioning accuracy of the agricultural machinery navigation system.