A feeding method and system based on a sheep feeding intelligent robot

Abstract

The present application provides a kind of based on sheep feeding intelligent robot feeding method and system, belong to the technical field of feeding robot, including: according to the sheep feed feeding quantity database and the different life cycle stages of each sheep pen sheep quantity, determine the feeding task and bunker material quantity of each sheep pen;Based on the feeding task and bunker material quantity of each sheep pen, determine feeding time and generate task flow;By laser radar scanning sheep field area, generate three-dimensional digital map;Based on three-dimensional digital map to determine the bunker and the location of each sheep pen, bunker is used as starting point and end point, each sheep pen is used as way point, with shortest path as target to carry out feeding path planning, control robot to automatically execute task.The method can realize automatic material taking, path planning and accurate feeding, help farmers to improve the feeding efficiency of sheep, reduce the labor intensity of farmers.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of feeding robots, in particular to a feeding method, system and medium based on a sheep feeding intelligent robot. BACKGROUND

[0002] With the change of livestock production mode from extensive to intensive development, the requirement for production efficiency is continuously improved. However, the problems of low labor productivity and labor shortage have become the bottleneck restricting the rapid development of livestock industry. In rural areas, breeding is one of the important economic sources of farmers. However, the traditional manual feeding method has many shortcomings, such as low feeding efficiency, serious feed waste, high labor cost, etc. In addition, with the reduction of rural labor force and the aggravation of the aging problem, the difficulty and cost of manual feeding are further increased. Therefore, the development of AI sheep feeding intelligent robot to realize automatic and intelligent feeding management is of great significance to improve the efficiency of rural breeding and reduce labor cost.

[0003] Most of the existing feeding robots can only feed according to the preset feeding plan, lack flexible feeding strategies for different growth stages and different breeds of sheep, and cannot meet the needs of fine breeding, and the degree of automation is poor. SUMMARY

[0004] To solve the above problems, the present application provides a feeding method based on a sheep feeding intelligent robot, which can realize automatic material taking, path planning and accurate feeding, help farmers improve the feeding efficiency of sheep and reduce the labor intensity of farmers.

[0005] To achieve the above purpose, the present application provides the following technical solutions.

[0006] A feeding system based on a sheep feeding intelligent robot, comprising the following steps:

[0007] According to the sheep feeding amount database and the number of sheep in different life cycle stages of each sheep pen, the feeding task and the material taking amount of each sheep pen are determined;

[0008] Based on the feeding task and the material taking amount of each sheep pen, the feeding time is determined and the task flow is generated;

[0009] The sheep farm area is scanned by laser radar to generate a three-dimensional digital map;

[0010] Based on the three-dimensional digital map, the position of the material bin and each sheep pen is determined, the material bin is taken as the starting point and the end point, each sheep pen is taken as the passing point, and the shortest path is taken as the target for feeding path planning;

[0011] According to the dynamic algorithm, the current vehicle load weight is dynamically calculated based on the bunker material taking amount and the real-time weighing data of the robot scale;

[0012] Waiting for the robot to actively request to obtain the feeding task, obtaining the feeding task and the bunker material taking amount, obtaining the task flow, determining the bunker, the sheep shed and the feeding route according to the feeding path planning result, judging the current vehicle load weight of the robot according to the real-time weighing data of the robot scale, and controlling the robot to automatically execute the task.

[0013] Preferably, the scanning of the sheep farm area by the laser radar and the generation of the three-dimensional digital map comprise the following steps:

[0014] Using a laser radar scanning device, the sheep farm area is scanned by a laser beam and the reflected signals are recorded to obtain point cloud data;

[0015] The point cloud data is filtered to obtain denoised point cloud data;

[0016] The RANSAC algorithm is used to fit the plane of the point cloud data to identify the structure of the ground, walls and railings, and to remove point cloud data that is not on the ground or not important;

[0017] Based on the geographical information of the sheep farm, the processed point cloud data is collected according to the spatial coordinate system to generate a three-dimensional data set, so that each point cloud data has corresponding spatial coordinates and additional information of color and intensity;

[0018] According to the three-dimensional data set, the point cloud data is converted into a grid, surface or volume representation to construct a complete three-dimensional digital map.

[0019] Preferably, the construction of the sheep feed feeding amount database comprises the following steps:

[0020] The life cycle stage of the sheep is determined, including the young sheep stage, the young sheep stage and the adult sheep stage; the young sheep stage is from birth to 6 months; the young sheep stage is from 6 months to 12 months; the adult sheep stage is more than 12 months;

[0021] The required feeding amount of each sheep at different body weights in different life cycle stages is determined to establish a sheep feed feeding amount database.

[0022] Preferably, the feeding task of each sheep pen and the bunker material taking amount are determined according to the sheep feed feeding amount database and the number of sheep in different life cycle stages in each sheep pen, comprising the following steps:

[0023] The monitoring camera is configured to collect monitoring images of each sheep pen at regular intervals; the monitoring images are pre-processed by denoising, cropping and standardization;

[0024] Using computer vision technology, a pre-trained convolutional neural network model is used to identify pre-processed monitoring images. Based on the sheep's body size and color characteristics, the number of sheep and their corresponding life cycle stages are determined.

[0025] The weight of sheep is obtained by ground pressure sensors to verify the number of sheep at different life stages;

[0026] Based on the sheep feed quantity database and the number of sheep at different life stages in each pen, the feeding tasks and feed hopper retrieval quantities for each pen are determined.

[0027] Preferably, the step of determining the location of feed silos and sheep pens based on a three-dimensional digital map, using feed silos as starting and ending points and sheep pens as waypoints, and planning the feeding path with the shortest path as the objective, includes the following steps:

[0028] Based on a 3D digital map, the locations of feed silos and sheep pens within the sheep farm area are marked.

[0029] Convert coordinate information in a 3D digital map into a 2D coordinate system;

[0030] Based on a two-dimensional coordinate system, and considering obstacles and site constraints, Dijkstra's algorithm is used to calculate the shortest planar path from the silo to each sheep pen.

[0031] Based on the location and number of sheep pens, a greedy algorithm is used to optimize the order of visiting sheep pens to ensure the shortest path.

[0032] Preferably, when the intelligent sheep feeding robot along the planned feeding path performs feed retrieval from the hopper and feeding in each sheep pen, the amount of feed retrieved and the amount of feed fed are determined by the pressure sensor data of the intelligent sheep feeding robot's storage area.

[0033] A feeding system based on a sheep feeding intelligent robot, the system comprising:

[0034] The data acquisition module is used to acquire three-dimensional point cloud data of the sheep farm area, collect monitoring images of each sheep pen in the sheep farm area, and collect the weight data of the sheep in the sheep pens.

[0035] The Raspberry Pi main controller is used to construct a 3D digital map of the sheep farm based on 3D point cloud data; it is also used to establish a database of sheep feed amounts at different life stages, identify the number of sheep at different life stages through monitoring images, and determine the feed hopper dispensing amount and feeding tasks for each sheep pen based on the sheep feed amount database, according to the number and weight data of sheep at different life stages in each pen; it is also used to determine the location of feed hoppers and sheep pens based on the 3D digital map, using feed hoppers as starting and ending points and sheep pens as waypoints, and to plan feeding paths with the shortest path as the objective; and it is used to dynamically calculate the current weight of the load based on the feed hopper dispensing amount, the feed motor speed, and the real-time weighing data of the robot scale.

[0036] The robot is used to obtain feeding tasks and feed hopper retrieval amounts, acquire task flow, determine feed hoppers, sheep pens, and feeding routes based on feeding path planning results, determine the current weight of the material on the robot vehicle based on real-time weighing data from the robot's scale, and control the robot to automatically execute tasks.

[0037] Preferably, it further includes:

[0038] The cloud server is used to store the feed hopper retrieval volume, feeding tasks for each sheep pen, and feeding path planning results, and to perform data analysis to determine feeding strategies.

[0039] The mobile app and the farmer's PC are used to obtain the feed hopper retrieval volume, feeding tasks and feeding path planning results for each sheep pen, as well as the feeding strategy, stored on the cloud server via wireless communication.

[0040] The beneficial effects of this invention are:

[0041] This invention proposes a feeding method based on an intelligent sheep feeding robot. This method acquires millimeter-level 3D point cloud data of the sheep farm area through laser scanning, accurately reproducing the spatial distribution of terrain, buildings, and obstacles, thus constructing a realistic environmental foundation for path planning. A 3D digital map is constructed based on a point cloud classification algorithm, automatically distinguishing functional areas such as feed silos and sheep pens, and generating a queryable semantic spatial index, significantly improving the efficiency of path planning decisions. A feed quantity database is established to dynamically match the nutritional needs of sheep at different life stages. This invention integrates monitoring images and weight sensor data, employs deep learning algorithms to identify the sheep's life stage, and dynamically adjusts the feeding amount based on real-time weight data, effectively reducing feed waste. This method achieves conflict-free path allocation and task scheduling through 3D path planning, providing an intelligent and precise feeding solution for large-scale sheep farms. Attached Figure Description

[0042] Figure 1 This is a flowchart of a method according to an embodiment of the present invention;

[0043] Figure 2is a system architecture diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0045] Embodiment 1

[0046] This embodiment proposes a feeding method based on a sheep feeding intelligent robot, and the specific steps are as shown in Figure 1 , which include:

[0047] S1: According to the sheep feeding amount database and the number of sheep in different life cycle stages of each sheep pen, determine the feeding task of each sheep pen and the bunker material quantity.

[0048] S2: Based on the feeding task of each sheep pen and the bunker material quantity, determine the feeding time and generate the task flow.

[0049] S3: Scan the sheep farm area by laser radar to generate a three-dimensional digital map.

[0050] S4: Based on the three-dimensional digital map, determine the bunker and the location of each sheep pen, take the bunker as the starting point and the end point, and take each sheep pen as the passing point. The shortest path is taken as the target for feeding path planning.

[0051] S5: According to the dynamic algorithm, based on the bunker material quantity, the motor speed and the real-time weighing data of the robot scale are dynamically calculated to determine the current vehicle load weight.

[0052] S6: Wait for the robot to actively request to obtain the feeding task, obtain the feeding task and the bunker material quantity, obtain the task flow, determine the bunker, the sheep house and the feeding route according to the feeding path planning result, determine the current vehicle load weight of the robot scale according to the real-time weighing data of the robot scale, and control the robot to automatically execute the task.

[0053] The sheep feeding intelligent robot used in this embodiment adopts a double-arm carrying type mobile robot, and is equipped with vision, radar and gravity sensing devices.

[0054] Further, S1 in which the three-dimensional digital map of the sheep farm is constructed according to the three-dimensional point cloud data includes the following steps:

[0055] S1.1: Use a laser radar scanning device to scan the sheep farm area by laser beam and record the reflected signal to obtain point cloud data.

[0056] S1.2: Filter the point cloud data to obtain denoised point cloud data.

[0057] S1.3: Perform plane fitting on the point cloud data through the RANSAC algorithm to identify the structure of the ground, walls, and railings, and remove the point cloud data that is not on the ground or not important.

[0058] S1.4: Based on the geographical information of the sheep farm, the processed point cloud data is collected according to the spatial coordinate system to generate a three-dimensional data set, so that each point cloud data has corresponding spatial coordinates and additional information of color and intensity.

[0059] S1.5: According to the three-dimensional data set, the point cloud data is converted into a grid, surface or volume representation to construct a complete three-dimensional digital map.

[0060] Further, the present application determines the life cycle stage of the sheep, including the young sheep stage, the young sheep stage and the adult sheep stage; the young sheep stage is from birth to 6 months; the young sheep stage is from 6 months to 12 months; the adult sheep stage is more than 12 months.

[0061] Determine the feeding amount required for each sheep at different life cycle stages, and establish a sheep feed feeding amount database.

[0062] According to the number and weight data of sheep at different life cycle stages in each pen, the present application determines the feed bin feeding amount and the feeding task of each pen based on the sheep feed feeding amount database, including the following steps:

[0063] S2.1: Configure a monitoring camera to collect monitoring images of each pen at regular intervals; and pre-process the monitoring images by denoising, cropping and standardizing.

[0064] S2.2: Use computer vision technology to identify the pre-processed monitoring images through a pre-trained convolutional neural network model, and determine the number of sheep and the corresponding life cycle stage according to the body shape and color feature data of the sheep.

[0065] S2.3: Obtain the weight of the sheep through the ground pressure sensor, and check the number of sheep at different life cycle stages.

[0066] S2.4: According to the sheep feed feeding amount database and the number of sheep at different life cycle stages in each pen, determine the feeding task of each pen and the feed bin feeding amount.

[0067] Further, based on the three-dimensional digital map, the positions of the feed bin and each pen are determined, the feed bin is taken as the starting point and the end point, each pen is taken as the passing point, and the feeding path planning is carried out with the shortest path as the target, including the following steps:

[0068] S3.1: Based on the three-dimensional digital map, mark the positions of the feed bin and each pen in the sheep farm area.

[0069] S3.2: Convert coordinate information in the three-dimensional digital map into a two-dimensional coordinate system.

[0070] S3.3: Based on the two-dimensional coordinate system, consider obstacles and site restrictions, and use Dijkstra algorithm to calculate the shortest plane path from the silo to each sheep pen.

[0071] S3.4: Based on the location and number of sheep pens, use the greedy algorithm to optimize the access order of the sheep pens to ensure the shortest path.

[0072] The embodiment provides a feeding system based on a sheep feeding intelligent robot, as shown in the accompanying drawings, the system comprises: Figure 2

[0073] A data acquisition module is configured to acquire three-dimensional point cloud data of a sheep farm area, acquire monitoring images of each sheep pen in the sheep farm area, and acquire weight data of sheep in the sheep pen.

[0074] A Raspberry Pi master is configured to construct a three-dimensional digital map of the sheep farm according to the three-dimensional point cloud data, establish a database of sheep feed feeding amounts at different life cycle stages, identify the number of sheep at different life cycle stages through the monitoring images, determine the silo feeding amount and the feeding task of each sheep pen based on the number of sheep at different life cycle stages and the weight data of each pen and the database of sheep feed feeding amounts, and determine the positions of the silo and each sheep pen based on the three-dimensional digital map, take the silo as a starting point and an ending point, take each sheep pen as a passing point, and plan a feeding path with the shortest path as a target.

[0075] An actuator is configured to control the sheep feeding intelligent robot to execute silo feeding and feeding of each sheep pen along the planned feeding path based on the silo feeding amount and the feeding task of each sheep pen, collect environmental information through the sheep feeding intelligent robot, execute obstacle avoidance, return to the initially set feeding path after obstacle avoidance, and execute the planned feeding path until the feeding is completed.

[0076] A cloud server is configured to store the silo feeding amount, the feeding task of each sheep pen, and the feeding path planning result, and perform data analysis to determine a feeding strategy.

[0077] A mobile APP and a farmer's PC are configured to acquire the silo feeding amount, the feeding task of each sheep pen, and the feeding path planning result, and the feeding strategy stored in the cloud server through wireless communication.

[0078] ​The system can realize precise feeding of sheep feed, real-time monitoring of sheep farm environment and optimal planning of feed feeding path, support remote monitoring and operation, greatly reduce labor cost and improve intelligent level of feeding management. The system uses Raspberry Pi 5 development board as the main control unit, uses RS485 communication protocol to control multiple servo motors, and can flexibly move and accurately feed the feed. First, the feed feeding path is planned, the geographic information of the sheep farm is measured by using point cloud map and laser three-dimensional technology, three-dimensional point cloud data is collected by using laser scanning, structured light scanning and other technologies, and data algorithm preprocessing operations such as filtering are performed, then the point cloud data set is formed and a three-dimensional digital map is formed; in the precise feeding link of the feed, first, a sheep feed feeding database of different life cycle stages is established, then according to the number of sheep in each pen, the robot is instructed to perform a series of behaviors such as taking feed from the corresponding feed bin, calculating the weight of the feed and feeding the sheep pen position; at the same time, the system integrates a high-definition camera and a laser radar to collect environmental information in real time, ensures the safe driving of the trolley in complex terrain, and makes intelligent path planning and feeding decisions according to the obtained data. The high-definition camera of the system is used for image recognition, the laser radar is used for distance measurement, the pressure sensor is used for calculating the weight of the feed, the ultrasonic sensor is used for detecting obstacles, and the voice module is used for enhancing human-computer interaction. The system also accesses a cloud server, and users can remotely monitor the system state, view feeding data and adjust feeding strategies through the cloud server on a mobile APP and a farmer's PC. It provides higher flexibility and convenience for farmers and helps precise management. Users can realize remote monitoring and data analysis, and further optimize the breeding management process

[0079] The sheep feeding intelligent robot provided by the present application can automatically store, transport and feed the feed according to the preset feeding plan, which significantly improves the automation degree of feeding. This not only reduces the time and labor required for manual feeding, but also makes the feeding process more timely and accurate. At the same time, the sheep feeding intelligent robot can provide personalized feeding according to the growth stage and needs of the sheep, improve the utilization rate of feed and the growth speed of the sheep, and thus improve the overall breeding efficiency.

[0080] The application of the sheep feeding intelligent robot provided by the present application reduces the breeding cost. On the one hand, automated feeding reduces labor costs, allowing farmers to save the cost of hiring labor. On the other hand, through precise feeding and personalized management, the sheep feeding intelligent robot can reduce feed waste and improve feed conversion rate, thereby reducing feed costs.

[0081] The intelligent management function of the sheep feeding intelligent robot makes the breeding management more fine and intelligent. The sheep feeding intelligent robot can monitor the health status and environmental parameters of the sheep in real time, and provide scientific breeding suggestions through big data analysis technology. This intelligent management method not only improves the accuracy and efficiency of breeding, but also helps the breeders to discover and handle problems in the breeding process in time, reduces the breeding risk. At the same time, the sheep feeding intelligent robot can record and analyze the breeding data, provide decision support for the breeders, and promote the sustainable development of the breeding industry.

[0082] The above is only a preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A feeding method based on an intelligent sheep feeding robot, characterized in that, Includes the following steps: Based on the sheep feed quantity database and the number of sheep at different life stages in each pen, the feeding tasks and feed hopper retrieval quantities for each pen are determined. Based on the feeding tasks and feed hopper retrieval volume of each sheep pen, the feeding time is determined and the task flow is generated. A three-dimensional digital map is generated by scanning the sheep farm area with lidar. Based on the three-dimensional digital map, the locations of feed bins and sheep pens are determined. The feed bins are used as the starting and ending points, and the sheep pens are used as the waypoints. The feeding path is planned with the shortest path as the goal. Based on the dynamic algorithm, the current weight of the material on the vehicle is dynamically calculated by judging the speed of the feeding motor and the real-time weighing data of the robot scale based on the material picking volume of the hopper. Wait for the robot to actively request a feeding task, obtain the feeding task and feed hopper amount, obtain the task flow, determine the feed hopper, sheepfold and feeding route based on the feeding path planning results, determine the current weight of the material on the robot vehicle based on the real-time weighing data of the robot scale, and control the robot to automatically execute the task. The process of scanning the sheep farm area with lidar to generate a three-dimensional digital map includes the following steps: Using lidar scanning equipment, the sheep farm area is scanned with a laser beam and the reflected signals are recorded to obtain point cloud data; The point cloud data is filtered to obtain denoised point cloud data; The RANSAC algorithm is used to perform planar fitting on point cloud data to identify the structure of the ground, walls, and railings, and to remove point cloud data of non-ground or non-important parts. Based on the geographic information of the sheep farm, the processed point cloud data is aggregated according to the spatial coordinate system to generate a three-dimensional dataset, so that each point cloud data has corresponding spatial coordinates and additional information such as color and intensity. Based on the 3D dataset, point cloud data is transformed into mesh, surface, or volume representations to construct a complete 3D digital map.

2. The feeding method based on a sheep feeding intelligent robot according to claim 1, characterized in that, The construction of the sheep feed quantity database includes the following steps: Determine the life cycle stages of sheep, including the lamb stage, the blue sheep stage, and the adult sheep stage; the lamb stage is from birth to 6 months; the blue sheep stage is from 6 months to 12 months; and the adult sheep stage is 12 months and above. Determine the required feed amount for each sheep at different life stages and weights, and establish a database of sheep feed amounts.

3. The feeding method based on a sheep feeding intelligent robot according to claim 2, characterized in that, The process of determining the feeding tasks and feed hopper retrieval amounts for each sheep pen based on a sheep feed quantity database and the number of sheep at different life stages in each pen includes the following steps: Configure surveillance cameras to periodically collect monitoring images of each sheep pen; perform noise reduction, cropping, and standardization preprocessing on the monitoring images; Using computer vision technology, a pre-trained convolutional neural network model is used to identify pre-processed monitoring images. Based on the sheep's body size and color characteristics, the number of sheep and their corresponding life cycle stages are determined. The weight of sheep is obtained by ground pressure sensors to verify the number of sheep at different life stages; Based on the sheep feed quantity database and the number of sheep at different life stages in each pen, the feeding tasks and feed hopper retrieval quantities for each pen are determined.

4. The feeding method based on a sheep feeding intelligent robot according to claim 1, characterized in that, The method of determining the location of feed silos and sheep pens based on a 3D digital map, using feed silos as the starting and ending points and sheep pens as waypoints, and planning feeding routes with the shortest path as the objective, includes the following steps: Based on a 3D digital map, the locations of feed silos and sheep pens within the sheep farm area are marked. The z-axis of the point cloud data is taken as an interval value and compressed into a two-dimensional image, converting the coordinate information in the three-dimensional digital map into a two-dimensional coordinate system; Based on a two-dimensional coordinate system, considering obstacles and site constraints, the coordinate positions of obstacles, feed troughs, and feed bins are marked, and the Dijkstra algorithm is used to calculate the shortest planar path from the feed bins to each sheep pen. Based on the location and number of sheep pens, a greedy algorithm is used to optimize the order of visiting sheep pens to ensure the shortest path.

5. The feeding method based on a sheep feeding intelligent robot according to claim 1, characterized in that, When the intelligent sheep feeding robot is controlled to perform feed retrieval from the hopper and feeding in each sheep pen along the planned feeding path, the amount of feed retrieved and fed is determined by the pressure sensor data of the feed storage area of ​​the intelligent sheep feeding robot.

6. A feeding system employing the feeding method based on a sheep feeding intelligent robot as described in any one of claims 1-5, characterized in that, The system includes: The data acquisition module is used to acquire three-dimensional point cloud data of the sheep farm area; The Raspberry Pi main controller is used to construct a 3D digital map of the sheep farm based on 3D point cloud data; it is also used to establish a database of sheep feed amounts at different life stages, identify the number of sheep at different life stages through monitoring images, and determine the feed hopper dispensing amount and feeding tasks for each sheep pen based on the sheep feed amount database, according to the number and weight data of sheep at different life stages in each pen; it is also used to determine the location of feed hoppers and sheep pens based on the 3D digital map, using feed hoppers as starting and ending points and sheep pens as waypoints, and to plan feeding paths with the shortest path as the objective; and it is used to dynamically calculate the current weight of the load based on the feed hopper dispensing amount, the feed motor speed, and the real-time weighing data of the robot scale. The robot is used to obtain feeding tasks and feed hopper retrieval amounts, acquire task flow, determine feed hoppers, sheep pens, and feeding routes based on feeding path planning results, determine the current weight of the material on the robot vehicle based on real-time weighing data from the robot's scale, and control the robot to automatically execute tasks.

7. The feeding system according to claim 6, characterized in that, Also includes: The cloud server is used to store the feed hopper retrieval volume, feeding tasks for each sheep pen, and feeding path planning results, and to perform data analysis to determine feeding strategies. The mobile app and the farmer's PC are used to obtain the feed hopper retrieval volume, feeding tasks and feeding path planning results for each sheep pen, as well as the feeding strategy, stored on the cloud server via wireless communication.

Images