| During the feeding process of cows,there will be a kind of feed offset phenomenon due to partial feeding,which has a direct impact on milk yield of cows and pasture feeding cost.It is of great significance to improve the modern and intelligent feeding level of livestock farms to automatically complete the task of overturning grass by robots instead of humans.For the traditional manual or traditional mechanical operation mode,there are some problems such as inadequate mixing of coarse and fine materials,high labor intensity and high cost when pushing grass.If the robot is used to push materials,the maintenance cost of magnetic navigation in the robot navigation mode is high,and the requirement of visual navigation on environmental light is high,which is difficult to meet in the pasture environment.In this study,based on Simultaneous localization and mapping(SLAM)technology,an intelligent grass removal robot in pasture is designed to realize the function of grass removal by autonomous navigation in pasture environment.It plays an active role in improving the level of modern,intelligent and large-scale development of aquaculture farms.Its main research contents and conclusions are as follows:(1)Aiming at sowing the problems of low intelligence and high labor intensity of feed pushing machinery and equipment for dairy cows in pasture,an autonomous navigation overturning grass robot based on laser SLAM was designed.The Cartographer algorithm was used to create the pasture environment map,AMCL algorithm was used for positioning,and Dijkstra was used for route planning to realize the autonomous navigation of the robot.The experiment shows that the mapping and navigation accuracy of the robot can meet the requirements of overturning the grass in the pasture environment and improve the intelligent level of dairy cattle raising in the pasture.(2)Aiming at sowing the problems of deviation between Li DAR and IMU sensor data when the robot Cartographer algorithm creates a complex pasture environment map,a mapping method that integrates odometer information is proposed,and the data of Li DAR,odometer and IMU are fused to create a map to improve the accuracy of the map.The test results show that the maximum deviation of the horizontal and longitudinal dimensions of the pasture environment map established by the robot in the state of loading odometer information is 0.02 m and 0.14 m,respectively,and the maximum deviation of the horizontal and longitudinal dimensions of the map built without odometer information is 0.06 m and2.18 m,respectively,indicating that the accuracy of the map created when the robot loads the odometer information is significantly higher than that created when the odometer information is not loaded,which meets the requirements of map accuracy during the robot navigation operation.(3)Aiming at sowing the problem of errors in odometer data caused by tire and ground slippage when the robot overturns grass operation,a navigation operation method that does not load odometer information is proposed,and only Li DAR and IMU data are used to navigate to improve the navigation accuracy of the robot during operation.The experimental results show that the maximum values of lateral deviation,longitudinal deviation and heading declination during navigation are less than 0.040 m,0.100 m and 11.0°,and the standard deviation is less than 0.030 m,0.025 m and 3.0°,respectively.The navigation accuracy is significantly higher than that when loading odometer information,which meets the requirements of navigation accuracy when the robot overturns the grass operation.(4)In the overturning grass robot,the screw conveyor and side sweeping are respectively designed and developed to overthrow grass and sweeping devices to realize the mixing,conveying and sweeping functions of forage.The robot uses the main power supply and the backup power supply to improve the endurance time of the robot,and meet the needs of the robot for 24 hours of uninterrupted operation by replacing the standby power supply.(5)Based on the distributed communication framework in the ROS robot operating system,the host PC remote control system and mobile APP remote control software were designed.The grass robot is remotely controlled by the host PC or mobile APP,and the final field test proves that the system can realize the basic functions such as overturning grass task delivery and camera monitoring in real time and accurately. |
