Research On Autonomous Navigation System Of Agricultural Robot

As the agriculture constantly develops in the direction of mechanization,automation and intelligence,an inevitable trend of future agricultural development is embodied in smart agriculture.When it comes to the intelligent monitoring of farmland,it belongs to a significant application field of smart agriculture.The single method is adopted by most of the farmland monitoring at this stage.To achieve the comprehensive monitoring and management of farmland,the combination of the internal monitoring of the farmland with the peripheral monitoring of the farmland is necessary.However,the key of farmland internal and external monitoring is that agricultural robots performing different inspection tasks are able to make automatic navigation in line with the predetermined path.GPS navigation cannot meet the accuracy requirements since the distance between adjacent crop rows in the farmland is close,so the inter-row inspection of crops is conducted by using visual navigation.Due to the complex peripheral environment of the farmland,the visual navigation is difficult to adapt to complex scenes,so the GPS navigation is used to patrol the periphery of farmland.In view of the navigation tasks,the research on two navigation methods is carried out,and low-cost autonomous navigation systems for agricultural robots are designed and developed.The following is the concrete research content:（1）Design of visual navigation system between crop rows.Through the contrast of the processing effects of various color spaces,the image processing is conducted by the acomponent of the Lab color space,which is less affected by shadows and lighting and has better segmentation effect.Meanwhile,the image segmentation is made by using the Mini Batch Kmeans clustering algorithm to improve image processing speed.To reduce the influence of abnormal navigation points on the fitted navigation lines,the RANSAC robust regression is adopted to fit the navigation lines more accurately.（2）Design of GPS navigation system on the periphery of farmland.The GPS receiver and attitude sensor are adopted to acquire the robot’s pose information.Through the contrast with the preset path,it controls the robot to move towards the target position.In the process of the movement,the 4G communication module is used to upload the position information in real time.For the purpose of realizing the path planning and real-time position display of the robot,the Baidu map API is used to make the development of the remote terminal system.（3）The test of navigation function.For the implementation of the experimental test on the quadruped robot platform,the experimental results indicate that the navigation systems are able to realize relevant functions and satisfy the requirements of autonomous navigation tasks in farmland.The feasibility of the navigation systems is verified and a certain theoretical and practical value is embodied.