Apple picking robot intelligent path planning system and method based on deep learning

Abstract

The application discloses an apple picking robot intelligent path planning system and method based on deep learning, relates to the field of robot path planning, and comprises the following steps: acquiring the channel distance between two adjacent hedges, the length of each hedge and the number of hedges in a picking area; picking apples on both sides of an apple tree by using a complementary apple picking mode; acquiring a dynamic delay coefficient prediction value of single-frame area apple picking of both sides of a picking robot; acquiring an adaptability function of dynamic picking of both sides of the picking robot; determining an implementation principle of dynamic picking path planning of the picking robot; establishing an intelligent path planning model, acquiring an optimal path planning of the apple picking robot; and establishing a man-machine interaction platform, storing, displaying and warning monitoring data of the apple picking robot in a picking process. The application effectively improves the picking efficiency of the apple picking robot and reduces the influence of light, and greatly increases the flexibility of the picking robot.

Description

Technical Field

[0001] This invention relates to the field of robot path planning, specifically to an intelligent path planning system and method for apple picking robots based on deep learning. Background Technology

[0002] Apple-picking robots can continuously harvest ripe apples. Compared with traditional manual harvesting, this not only greatly improves the efficiency of apple picking but also saves a lot of labor costs. It is also a core driving force for the transformation of agriculture from labor-intensive to intelligent. Planning the optimal picking path for apple-picking robots is an important means to improve their picking efficiency and ability to handle abnormal events. Therefore, improving the intelligent path planning of apple-picking robots is of great significance.

[0003] Existing apple-picking robots can pick apples from both sides or one side. However, picking from both sides is significantly more complex than picking from one side, requiring higher standards for the picking environment and the robot itself. The system's processing speed is also slower, which can easily lead to missed apples and slow efficiency during the picking process. Furthermore, due to changes in lighting, apple-picking robots are easily affected by the switching between sunlight and shade when picking apples from both sides of the tree. This increases the shadow area when the robot acquires images, resulting in poor image quality and inaccurate detection during apple picking. Summary of the Invention

[0004] To address the aforementioned technical problems, a deep learning-based intelligent path planning system and method for apple picking robots is provided. This technical solution solves the problems mentioned in the background, such as the high requirements for the picking environment and the picking robot, the slow system processing speed, which easily leads to missed picking and slow efficiency during the picking process. Furthermore, due to changes in lighting, the apple picking robot is easily affected by the switching between sunlight and shade when picking from both sides of the apple tree, resulting in an increased shadow area when the robot acquires images, leading to poor image quality and inaccurate detection during apple picking.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A deep learning-based intelligent path planning method for apple-picking robots includes:

[0007] Based on the distribution of apple planting, obtain the spacing between adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area;

[0008] A complementary apple harvesting method is adopted, which harvests apples on both sides of the apple tree. The two machines work together to improve the efficiency of apple harvesting and reduce the impact of light.

[0009] Based on the picking efficiency of the apple picking robot, the predicted dynamic delay coefficient of apple picking in a single frame area of ​​the two picking robots is obtained.

[0010] Based on the predicted value of the dynamic delay coefficient and the sudden obstacles encountered during the picking process, the adaptive function for the dynamic picking of the two picking robots is obtained.

[0011] Based on the adaptive function of the dynamic harvesting of the two harvesting robots, the implementation principle of the dynamic harvesting path planning of the harvesting robots is determined.

[0012] Based on a deep learning algorithm model, combined with the predicted value of dynamic delay coefficient and the dynamic picking adaptation function, an intelligent path planning model is established to obtain the optimal path planning for the apple picking robot.

[0013] Establish a human-machine interaction platform to store, display, and provide early warnings of monitoring data from the apple-picking robot during the picking process, as well as initial path planning.

[0014] Preferably, the step of obtaining the passage distance between two adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area based on the apple planting distribution specifically includes:

[0015] Based on the distribution of apple trees in the picking area, obtain the distance between adjacent hedges, the length of each hedge, and the number of hedges in the picking area;

[0016] Based on the number of apple hedges planted in the picking area, each hedge is numbered, and an apple planting list for the picking area is established.

[0017] Among them, the hedge numbers in the apple planting list of the picking area correspond one-to-one with the data on the distance between two adjacent hedges, the length of each hedge, and the number of hedges in the picking area.

[0018] Preferably, the complementary apple harvesting method, which harvests apples from both sides of the apple tree, improves harvesting efficiency and reduces the impact of light through dual-machine collaboration, specifically includes:

[0019] The complementary apple harvesting method specifically includes:

[0020] Two robots are set up as a group. At the beginning of apple picking, the two robots are located on opposite sides of the apple tree and move in opposite directions.

[0021] The robot located on the left side of the apple tree prioritizes picking apples from the left side of the tree, while the robot located on the right side of the apple tree prioritizes picking apples from the right side of the tree.

[0022] Based on the influence of unidirectional light, the image quality of the harvesting robots on both sides is optimized to reduce the impact of unidirectional light on the harvesting robots.

[0023] Preferably, obtaining the predicted dynamic delay coefficient of apple picking in a single frame area by the two apple picking robots based on the picking efficiency of the apple picking robots specifically includes:

[0024] Based on the area of ​​the apple tree that the apple-picking robot collects from, obtain the number of apples in the single frame area collected by the picking robot.

[0025] Based on the apple picking efficiency set by the apple picking robot and the number of apples in a single frame collection area, the effective time for the picking robot to complete apple picking in a single frame collection area is obtained.

[0026] Based on the effective time for the two picking robots to complete apple picking in a single frame of the collection area, obtain the relative value of the dynamic delay for the two picking robots to complete apple picking in a single frame of the collection area.

[0027] Based on the relative dynamic delay of apple picking in a single frame area by the two picking robots, the predicted value of the dynamic delay coefficient for apple picking in a single frame area by the two picking robots is determined.

[0028] Based on the predicted dynamic delay coefficient of apple picking in a single frame area by the two picking robots, a dynamic delay coefficient prediction value input matrix is ​​established.

[0029] The expression for the predicted dynamic delay coefficient of apple picking in a single frame region by the two-sided picking robots is as follows:

[0030] ;

[0031] In the formula, , The left and right picking robots completed the first... The effective time for apple picking in a single frame acquisition area. , These represent the number of apples collected in a single frame by the left and right picking robots, respectively. To improve the harvesting efficiency of the harvesting robot, The two harvesting robots completed the first... The relative dynamic delay of apple picking in a single frame acquisition area. This represents the predicted dynamic delay coefficient for apple picking in a single frame area by a harvesting robot. For the equilibrium constant term, This represents the total number of areas captured by the two harvesting robots in a single frame.

[0032] Preferably, the adaptive function for dynamic harvesting by the two harvesting robots, based on the predicted value of the dynamic delay coefficient and the unexpected obstacles encountered during the harvesting process, specifically includes:

[0033] Based on the length of each hedge and the distance of the single-frame acquisition area of ​​the harvesting robot, the number of units that need to be fixed for harvesting apples for each hedge is obtained;

[0034] Set up position acquisition devices for the two harvesting robots to obtain their position information.

[0035] Based on the position information of the two harvesting robots and their fixed moving speed, the displacement time of the two harvesting robots is obtained.

[0036] Based on the changes in the position information of the two harvesting robots, the relative time of danger avoidance and the relative time of collision waiting of the two harvesting robots are obtained;

[0037] Based on the predicted value of the dynamic delay coefficient and the sudden obstacles encountered during the picking process, the adaptive function for the dynamic picking of the two picking robots is obtained.

[0038] The adaptive function for the dynamic harvesting by the two-sided harvesting robots is:

[0039] ;

[0040] In the formula, The adaptive function value represents the dynamic harvesting process of the two-sided harvesting robots. The two harvesting robots completed the first... The relative displacement time of apple picking in a single frame acquisition area. Weighting of the relative waiting time when the harvesting robot collides. The relative time waiting for the harvesting robot to collide with the target robot. Weighting the relative time for avoiding danger in the harvesting robot. The relative time required for the harvesting robot to avoid danger.

[0041] Preferably, the implementation principle for determining the dynamic harvesting path planning of the harvesting robots based on the adaptive function of the dynamic harvesting of the two-sided harvesting robots specifically includes:

[0042] Based on big data, set phased thresholds for the occupancy of passageways between two adjacent hedges;

[0043] The threshold for passage occupancy between two adjacent hedges includes three stages: a small threshold, a medium threshold, and a large threshold. The small threshold range is as follows: The threshold range is: The large threshold range is: ,in, The total time spent by the picking robot that first completed the task of picking a single apple from a hedge. This is the equilibrium multiple coefficient for small thresholds. The equilibrium factor is the threshold value.

[0044] Based on the adaptive function of the dynamic harvesting of the two harvesting robots, the implementation principle of the dynamic harvesting path planning of the harvesting robots is determined.

[0045] Based on the implementation principles of dynamic picking path planning for picking robots, guidance is provided for the apple picking methods of the picking robots on both sides.

[0046] Preferably, the step of establishing an intelligent path planning model based on a deep learning algorithm model, combining dynamic delay coefficient prediction and dynamic picking adaptability function, to obtain the optimal path planning for the apple picking robot specifically includes:

[0047] Based on the predicted value of dynamic delay coefficient, adaptive function and the implementation principle of path planning, feature terms of intelligent path planning for apple picking robots are extracted.

[0048] Based on big data or test experiments, obtain training and target sample sets for intelligent path planning of apple picking robots;

[0049] Based on a deep learning algorithm model, an intelligent path planning model is established. By acquiring the predicted dynamic delay coefficient and dynamic picking adaptability function of the picking robots on both sides in real time, as well as the implementation principle judgment of the picking robot's dynamic picking path planning, the optimal path planning of the apple picking robot is obtained.

[0050] Preferably, the establishment of the human-computer interaction platform for storing, displaying, and issuing early warnings of monitoring data from the apple-picking robot during the picking process, as well as initial path planning, specifically includes:

[0051] Establish a human-machine interaction platform based on Internet of Things technology to receive and store monitoring data from both ends of the harvesting robot during the harvesting process;

[0052] Based on the human-computer interaction platform, we built the operating environment for the intelligent path planning model, and updated and improved the model's practicality by optimizing parameters.

[0053] Based on the human-computer interaction platform, the location information and abnormal situations of the two picking robots during the apple picking process are displayed and alerted.

[0054] Based on the human-computer interaction platform, the initial path planning and manual remote control of the two harvesting robots are completed through the human-computer interaction interface.

[0055] Furthermore, this solution proposes a deep learning-based intelligent path planning system for apple-picking robots, used to implement the deep learning-based intelligent path planning method for apple-picking robots described above, including:

[0056] The harvesting distribution acquisition module is used to acquire the channel spacing between two adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area based on the apple planting distribution.

[0057] The harvesting method module is used to harvest apples on both sides of the apple tree using a complementary apple harvesting method, thereby improving apple harvesting efficiency and reducing the impact of light through dual-machine collaboration.

[0058] The path planning module is used to obtain the predicted dynamic delay coefficient of apple picking in a single frame area by the apple picking robots on both sides based on the picking efficiency of the apple picking robot; obtain the adaptive function of dynamic picking by the two side picking robots based on the predicted dynamic delay coefficient and the sudden obstacle situation during the picking process; determine the implementation principle of dynamic picking path planning of the picking robot based on the adaptive function of dynamic picking by the two side picking robots; and establish an intelligent path planning model based on a deep learning algorithm model, combined with the predicted dynamic delay coefficient and the dynamic picking adaptive function, to obtain the optimal path planning of the apple picking robot.

[0059] The interactive platform module is used to establish a human-computer interaction platform for storing, displaying, and providing early warnings of monitoring data of the apple picking robot during the picking process, as well as initial path planning.

[0060] Preferably, the path planning module includes:

[0061] A dynamic delay unit is used to obtain the predicted value of the dynamic delay coefficient of apple picking in a single frame area of ​​the apple picking robots on both sides based on the picking efficiency of the apple picking robot.

[0062] An adaptive function unit is used to obtain the adaptive function of dynamic picking by the two picking robots based on the predicted value of the dynamic delay coefficient and the sudden obstacle situation during the picking process.

[0063] The implementation principle unit is used to determine the implementation principle of the dynamic picking path planning of the picking robot based on the adaptive function of the dynamic picking of the two picking robots.

[0064] The path planning unit is used to establish an intelligent path planning model based on a deep learning algorithm model, combined with the predicted value of dynamic delay coefficient and the dynamic picking adaptation function, to obtain the optimal path planning for the apple picking robot.

[0065] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0066] By employing a complementary apple-picking method, apples are harvested from both sides of the apple tree. Dual-machine collaboration improves harvesting efficiency and reduces the impact of light. Secondly, based on the area covered by the apple-picking robot, the number of apples in a single frame of the robot's collection area is determined. Furthermore, based on the harvesting efficiency of the two robots, the predicted dynamic delay coefficient for apple harvesting in a single frame is obtained. Finally, based on the predicted dynamic delay coefficient and unexpected obstacles encountered during harvesting, an adaptive function for the dynamic harvesting of the two robots is derived. Finally, the adaptive function of the dynamic harvesting of the two robots is compared with the phase... The passage between two adjacent hedges is occupied within a certain threshold range to guide the apple picking method of the picking robots on both sides, thereby determining the implementation principle of the dynamic picking path planning of the picking robots. Finally, based on a deep learning algorithm model, an intelligent path planning model is established. By acquiring the predicted dynamic delay coefficient and dynamic picking adaptability function of the picking robots on both sides in real time, as well as the judgment of the implementation principle of the dynamic picking path planning of the picking robots, the optimal path planning of the apple picking robots is obtained, thereby effectively improving the picking efficiency of the apple picking robots and reducing the influence of light, greatly increasing the flexibility of the picking robots. Attached Figure Description

[0067] Figure 1 This is a flowchart of an intelligent path planning method for an apple picking robot based on deep learning, according to the present invention.

[0068] Figure 2 The flowchart of the present invention is as follows: Based on the picking efficiency of the apple picking robot, the dynamic delay coefficient prediction value of apple picking in a single frame area of ​​the two picking robots is obtained.

[0069] Figure 3 The flowchart of the adaptive function for obtaining dynamic harvesting by the two harvesting robots based on the predicted value of the dynamic delay coefficient and the sudden obstacle situation during the harvesting process is shown in the present invention.

[0070] Figure 4 The flowchart below illustrates the implementation principle of the present invention for determining the dynamic harvesting path planning of the harvesting robot based on the adaptive function of the dynamic harvesting of the two-sided harvesting robots. Detailed Implementation

[0071] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0072] Reference Figure 1As shown, a deep learning-based intelligent path planning method for apple-picking robots includes:

[0073] Based on the distribution of apple planting, obtain the spacing between adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area;

[0074] A complementary apple harvesting method is adopted, which harvests apples on both sides of the apple tree. The two machines work together to improve the efficiency of apple harvesting and reduce the impact of light.

[0075] Based on the picking efficiency of the apple picking robot, the predicted dynamic delay coefficient of apple picking in a single frame area of ​​the two picking robots is obtained.

[0076] Based on the predicted value of the dynamic delay coefficient and the sudden obstacles encountered during the picking process, the adaptive function for the dynamic picking of the two picking robots is obtained.

[0077] Based on the adaptive function of the dynamic harvesting of the two harvesting robots, the implementation principle of the dynamic harvesting path planning of the harvesting robots is determined.

[0078] Based on a deep learning algorithm model, combined with the predicted value of dynamic delay coefficient and the dynamic picking adaptation function, an intelligent path planning model is established to obtain the optimal path planning for the apple picking robot.

[0079] Establish a human-machine interaction platform to store, display, and provide early warnings of monitoring data from the apple-picking robot during the picking process, as well as initial path planning.

[0080] This can be explained by the following: First, this solution employs a complementary apple-picking method, harvesting apples from both sides of the apple tree. This dual-machine collaboration improves harvesting efficiency and reduces the impact of light. Second, based on the area the apple-picking robot collects from the tree, the number of apples in a single frame is obtained. Third, based on the harvesting efficiency of the apple-picking robot, the predicted dynamic delay coefficient for apple harvesting in a single frame by both robots is obtained. Finally, based on the predicted dynamic delay coefficient and unexpected obstacles encountered during harvesting, an adaptive function for the dynamic harvesting of both robots is derived. Furthermore, the adaptive function of the dynamic harvesting of both robots is determined by... The range of the threshold stage for the passage occupancy between two adjacent hedges is used to guide the apple picking method of the picking robots on both sides, thereby determining the implementation principle of the dynamic picking path planning of the picking robots. Finally, based on the deep learning algorithm model, an intelligent path planning model is established. By acquiring the predicted value of the dynamic delay coefficient and the dynamic picking adaptability function of the picking robots on both sides in real time, as well as the judgment of the implementation principle of the dynamic picking path planning of the picking robots, the optimal path planning of the apple picking robots is obtained, thereby effectively improving the picking efficiency of the apple picking robots and reducing the influence of light, greatly increasing the flexibility of the picking robots.

[0081] Reference Figure 2 As shown, the step of obtaining the predicted dynamic delay coefficient of apple picking in a single frame area by the apple picking robots on both sides, based on the picking efficiency of the apple picking robots, specifically includes:

[0082] Based on the area of ​​the apple tree that the apple-picking robot collects from, obtain the number of apples in the single frame area collected by the picking robot.

[0083] Based on the apple picking efficiency set by the apple picking robot and the number of apples in a single frame collection area, the effective time for the picking robot to complete apple picking in a single frame collection area is obtained.

[0084] Based on the effective time for the two picking robots to complete apple picking in a single frame of the collection area, obtain the relative value of the dynamic delay for the two picking robots to complete apple picking in a single frame of the collection area.

[0085] Based on the relative dynamic delay of apple picking in a single frame area by the two picking robots, the predicted value of the dynamic delay coefficient for apple picking in a single frame area by the two picking robots is determined.

[0086] Based on the predicted dynamic delay coefficient of apple picking in a single frame area by the two picking robots, a dynamic delay coefficient prediction value input matrix is ​​established.

[0087] The expression for the predicted dynamic delay coefficient of apple picking in a single frame region by the two-sided picking robots is as follows:

[0088] ;

[0089] In the formula, , The left and right picking robots completed the first... The effective time for apple picking in a single frame acquisition area. , These represent the number of apples collected in a single frame by the left and right picking robots, respectively. To improve the harvesting efficiency of the harvesting robot, The two harvesting robots completed the first... The relative dynamic delay of apple picking in a single frame acquisition area. This represents the predicted dynamic delay coefficient for apple picking in a single frame area by a harvesting robot. For the equilibrium constant term, This represents the total number of areas captured by the two harvesting robots in a single frame.

[0090] The best outcome of the complementary apple picking method is that the two picking robots simultaneously complete the picking of apples on both sides of the apple tree, ensuring that they do not interfere with each other during the picking process. This improves the stability and efficiency of the apple picking robots. However, since the apple data in a single frame of the picking robot is not unique, the picking time in a single frame of the picking area will also be different under the same picking efficiency. Therefore, it is necessary to relativize the picking time of the two picking robots in a single frame of the picking area. This solution establishes a predicted dynamic delay coefficient for apple picking in a single frame of the picking robot on both sides, thereby effectively reflecting the relative difference in the time it takes for the two picking robots to complete the picking of apples in a single frame of the picking area, and thus providing data support for subsequent analysis.

[0091] Reference Figure 3 As shown, the adaptive function for obtaining the dynamic harvesting of the two harvesting robots based on the predicted value of the dynamic delay coefficient and the unexpected obstacles encountered during the harvesting process specifically includes:

[0092] Based on the length of each hedge and the distance of the single-frame acquisition area of ​​the harvesting robot, the number of units that need to be fixed for harvesting apples for each hedge is obtained;

[0093] Set up position acquisition devices for the two harvesting robots to obtain their position information.

[0094] Based on the position information of the two harvesting robots and their fixed moving speed, the displacement time of the two harvesting robots is obtained.

[0095] Based on the changes in the position information of the two harvesting robots, the relative time of danger avoidance and the relative time of collision waiting of the two harvesting robots are obtained;

[0096] Based on the predicted value of the dynamic delay coefficient and the sudden obstacles encountered during the picking process, the adaptive function for the dynamic picking of the two picking robots is obtained.

[0097] The adaptive function for the dynamic harvesting by the two-sided harvesting robots is:

[0098] ;

[0099] In the formula, The adaptive function value represents the dynamic harvesting process of the two-sided harvesting robots. The two harvesting robots completed the first... The relative displacement time of apple picking in a single frame acquisition area. Weighting of the relative waiting time when the harvesting robot collides. The relative time waiting for the harvesting robot to collide with the target robot. Weighting the relative time for avoiding danger in the harvesting robot. The relative time required for the harvesting robot to avoid danger.

[0100] It can be explained that when the passageway between two adjacent hedges is too narrow to accommodate two harvesting robots, the dynamic harvesting by the robots on both sides may be affected by environmental obstacles and mutual influences, leading to a lag in harvesting time. This could result in the robots on both sides sharing a single passageway, rendering them unable to operate. Therefore, to ensure that the robots on both sides stagger their dynamic harvesting while not sharing the same passageway, it is necessary to know the relative time relationship between their dynamic harvesting activities. This solution establishes an adaptive function for the dynamic harvesting activities of the robots on both sides. This adaptive function effectively evaluates the relative time relationship between the dynamic harvesting activities of the robots on both sides, thereby reducing environmental obstacles and mutual influences, and improving the stability and orderliness of the dynamic harvesting activities of the robots on both sides.

[0101] The specific method for determining the relative time of danger avoidance and the relative time of collision waiting for the two harvesting robots based on the changes in their position information is as follows:

[0102] When the distance between the two harvesting robots is less than the set value, and one position changes while the other remains unchanged, it indicates that the two harvesting robots are in a collision waiting state.

[0103] When the distance between the two harvesting robots is greater than the set value, if one of the robots remains in the same position or moves in the opposite direction, it indicates that the robot is avoiding danger.

[0104] The expression for the distance between the two harvesting robots is: In the formula, The distance between the two harvesting robots is [value]. Here are the position coordinates of the robot on the left. These are the position coordinates of the robot on the right.

[0105] Reference Figure 4 As shown, the implementation principles for determining the dynamic harvesting path planning of the harvesting robots based on the adaptive function of the dynamic harvesting of the two-sided harvesting robots specifically include:

[0106] Based on big data, set phased thresholds for the occupancy of passageways between two adjacent hedges;

[0107] The threshold for passage occupancy between two adjacent hedges includes three stages: a small threshold, a medium threshold, and a large threshold. The small threshold range is as follows: The threshold range is: The large threshold range is: ,in, The total time spent by the picking robot that first completed the task of picking a single apple from a hedge. This is the equilibrium multiple coefficient for small thresholds. The equilibrium factor is the threshold value.

[0108] Based on the adaptive function of the dynamic harvesting of the two harvesting robots, the implementation principle of the dynamic harvesting path planning of the harvesting robots is determined.

[0109] Based on the implementation principles of dynamic picking path planning for picking robots, guidance is provided for the apple picking methods of the picking robots on both sides.

[0110] It can be explained that the specific steps for determining the implementation principles of the dynamic harvesting path planning for the harvesting robot are as follows:

[0111] Step 1: Determine the adaptive function value of the dynamic harvesting by the two harvesting robots. Is it greater than If so, then the picking robot that is guided to complete the task of picking apples from a single hedge will start from the number [number missing]. The hedge reached the numbered [number]. If a robot is not selected, it will wait after completing the apple picking task for that single hedge. The robot will then proceed to the next hedge once the time limit is reached or another robot returns to normal. Apples are harvested from a hedge;

[0112] Step 2: Determine the adaptive function values ​​of the dynamic harvesting by the two harvesting robots. Is it greater than If so, the picking robot that has completed the picking of all the apples on one side of the hedge in the picking area will be guided to pick the apples on the other side, until the apples on the other side are picked.

[0113] Furthermore, based on the same inventive concept as the aforementioned deep learning-based intelligent path planning method for apple picking robots, this solution proposes a deep learning-based intelligent path planning system for apple picking robots, comprising:

[0114] The harvesting distribution acquisition module is used to acquire the channel spacing between two adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area based on the apple planting distribution.

[0115] The harvesting method module is used to harvest apples on both sides of the apple tree using a complementary apple harvesting method, thereby improving apple harvesting efficiency and reducing the impact of light through dual-machine collaboration.

[0116] The path planning module is used to obtain the predicted dynamic delay coefficient of apple picking in a single frame area by the apple picking robots on both sides based on the picking efficiency of the apple picking robot; obtain the adaptive function of dynamic picking by the two side picking robots based on the predicted dynamic delay coefficient and the sudden obstacle situation during the picking process; determine the implementation principle of dynamic picking path planning of the picking robot based on the adaptive function of dynamic picking by the two side picking robots; and establish an intelligent path planning model based on a deep learning algorithm model, combined with the predicted dynamic delay coefficient and the dynamic picking adaptive function, to obtain the optimal path planning of the apple picking robot.

[0117] An interactive platform module is used to establish a human-computer interaction platform for storing, displaying, and issuing early warnings of monitoring data of the apple-picking robot during the picking process, as well as initial path planning.

[0118] The route planning module includes:

[0119] A dynamic delay unit is used to obtain the predicted value of the dynamic delay coefficient of apple picking in a single frame area of ​​the apple picking robots on both sides based on the picking efficiency of the apple picking robot.

[0120] An adaptive function unit is used to obtain the adaptive function of dynamic picking by the two picking robots based on the predicted value of the dynamic delay coefficient and the sudden obstacle situation during the picking process.

[0121] The implementation principle unit is used to determine the implementation principle of the dynamic picking path planning of the picking robot based on the adaptive function of the dynamic picking of the two picking robots.

[0122] The path planning unit is used to establish an intelligent path planning model based on a deep learning algorithm model, combined with the predicted value of dynamic delay coefficient and the dynamic picking adaptation function, to obtain the optimal path planning for the apple picking robot.

[0123] In summary, the advantages of this invention are: it effectively improves the picking efficiency of apple picking robots and reduces the impact of light, greatly increasing the flexibility of the picking robots.

[0124] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A deep learning-based intelligent path planning method for apple-picking robots, characterized in that, include: Based on the distribution of apple planting, obtain the spacing between adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area; A complementary apple harvesting method is adopted, which harvests apples on both sides of the apple tree. The two machines work together to improve the efficiency of apple harvesting and reduce the impact of light. Based on the picking efficiency of the apple picking robot, the predicted dynamic delay coefficient of apple picking in a single frame area of ​​the two picking robots is obtained. Based on the predicted value of the dynamic delay coefficient and the sudden obstacles encountered during the picking process, the adaptive function for the dynamic picking of the two picking robots is obtained. Based on the adaptive function of the dynamic harvesting of the two harvesting robots, the implementation principle of the dynamic harvesting path planning of the harvesting robots is determined. Based on a deep learning algorithm model, combined with the predicted value of dynamic delay coefficient and the dynamic picking adaptation function, an intelligent path planning model is established to obtain the optimal path planning for the apple picking robot. Establish a human-computer interaction platform to store, display, and provide early warnings of monitoring data from the apple-picking robot during the picking process, as well as initial path planning; The adaptive function for dynamic harvesting by the two harvesting robots, based on the predicted value of the dynamic delay coefficient and the unexpected obstacles encountered during the harvesting process, specifically includes: Based on the length of each hedge and the distance of the single-frame acquisition area of ​​the harvesting robot, the number of units that need to be fixed for harvesting apples for each hedge is obtained; Set up position acquisition devices for the two harvesting robots to obtain their position information. Based on the position information of the two harvesting robots and their fixed moving speed, the displacement time of the two harvesting robots is obtained. Based on the changes in the position information of the two harvesting robots, the relative time of danger avoidance and the relative time of collision waiting of the two harvesting robots are obtained; Based on the predicted value of the dynamic delay coefficient and the sudden obstacles encountered during the picking process, the adaptive function for the dynamic picking of the two picking robots is obtained. The adaptive function for the dynamic harvesting by the two-sided harvesting robots is: ； In the formula, The adaptive function value represents the dynamic harvesting process of the two-sided harvesting robots. This represents the predicted dynamic delay coefficient for apple picking in a single frame area by a harvesting robot. The two harvesting robots completed the first... The relative displacement time of apple picking in a single frame acquisition area. Weighting of the relative waiting time when the harvesting robot collides. The relative time waiting for the harvesting robot to collide with the target robot. Weighting the relative time for avoiding danger in the harvesting robot. The relative time for the harvesting robot to avoid danger. This represents the total number of data collection areas per frame for the harvesting robots on both sides. The implementation principles for determining the dynamic harvesting path planning of the harvesting robots based on the adaptive function of the dynamic harvesting of the two-sided harvesting robots specifically include: Based on big data, set phased thresholds for the occupancy of passageways between two adjacent hedges; The threshold for passage occupancy between two adjacent hedges includes three stages: a small threshold, a medium threshold, and a large threshold. The small threshold range is as follows: The threshold range is: The large threshold range is: ,in, The total time spent by the picking robot that first completed the task of picking a single apple from a hedge. This is the equilibrium multiple coefficient for small thresholds. The equilibrium factor is the threshold value. Based on the adaptive function of the dynamic harvesting of the two harvesting robots, the implementation principle of the dynamic harvesting path planning of the harvesting robots is determined. Based on the implementation principles of dynamic picking path planning for picking robots, guidance is provided for the apple picking methods of the picking robots on both sides. Step 1: Determine the adaptive function value of the dynamic harvesting by the two harvesting robots. Is it greater than If so, then the picking robot that is guided to complete the task of picking apples from a single hedge will start from the number [number missing]. The hedge reached the numbered [number]. If a robot is not selected, it will wait after completing the apple picking task for that single hedge. The robot will then proceed to the next hedge once the time limit is reached or another robot returns to normal. Apples are harvested from a hedge; Step 2: Determine the adaptive function values ​​of the dynamic harvesting by the two harvesting robots. Is it greater than If so, the picking robot that has completed the picking of all the apples on one side of the hedge in the picking area will be guided to pick the apples on the other side, until the apples on the other side are picked.

2. The intelligent path planning method for an apple-picking robot based on deep learning according to claim 1, characterized in that, The specific steps of obtaining the passage distance between two adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area based on the apple planting distribution include: Based on the distribution of apple trees in the picking area, obtain the distance between adjacent hedges, the length of each hedge, and the number of hedges in the picking area; Based on the number of apple hedges planted in the picking area, each hedge is numbered, and an apple planting list for the picking area is established. Among them, the hedge numbers in the apple planting list of the picking area correspond one-to-one with the data on the distance between two adjacent hedges, the length of each hedge, and the number of hedges in the picking area.

3. The intelligent path planning method for an apple-picking robot based on deep learning according to claim 2, characterized in that, The complementary apple harvesting method, which harvests apples from both sides of the apple tree, improves harvesting efficiency and reduces the impact of light through dual-machine collaboration. Specifically, this includes: The complementary apple harvesting method specifically includes: Two robots are set up as a group. At the beginning of apple picking, the two robots are located on opposite sides of the apple tree and move in opposite directions. The robot located on the left side of the apple tree prioritizes picking apples from the left side of the tree, while the robot located on the right side of the apple tree prioritizes picking apples from the right side of the tree. Based on the influence of unidirectional light, the image quality of the harvesting robots on both sides is optimized to reduce the impact of unidirectional light on the harvesting robots.

4. The intelligent path planning method for an apple-picking robot based on deep learning according to claim 3, characterized in that, The step of obtaining the predicted dynamic delay coefficient of apple picking in a single frame area by the apple picking robots on both sides based on the picking efficiency of the apple picking robots specifically includes: Based on the area of ​​the apple tree that the apple-picking robot collects from, obtain the number of apples in the single frame area collected by the picking robot. Based on the apple picking efficiency set by the apple picking robot and the number of apples in a single frame collection area, the effective time for the picking robot to complete apple picking in a single frame collection area is obtained. Based on the effective time for the two picking robots to complete apple picking in a single frame of the collection area, obtain the relative value of the dynamic delay for the two picking robots to complete apple picking in a single frame of the collection area. Based on the relative dynamic delay of apple picking in a single frame area by the two picking robots, the predicted value of the dynamic delay coefficient for apple picking in a single frame area by the two picking robots is determined. Based on the predicted dynamic delay coefficient of apple picking in a single frame area by the two picking robots, a dynamic delay coefficient prediction value input matrix is ​​established. The expression for the predicted dynamic delay coefficient of apple picking in a single frame region by the two-sided picking robots is as follows: ； In the formula, , The left and right picking robots completed the first... The effective time for apple picking in a single frame acquisition area. , These represent the number of apples collected in a single frame by the left and right picking robots, respectively. To improve the harvesting efficiency of the harvesting robot, The two harvesting robots completed the first... The relative dynamic delay of apple picking in a single frame acquisition area. This represents the predicted dynamic delay coefficient for apple picking in a single frame area by a harvesting robot. For the equilibrium constant term, This represents the total number of areas captured by the two harvesting robots in a single frame.

5. The intelligent path planning method for an apple-picking robot based on deep learning according to claim 4, characterized in that, The method for establishing an intelligent path planning model based on a deep learning algorithm model, combining dynamic delay coefficient prediction and dynamic harvesting adaptability function, to obtain the optimal path planning for the apple harvesting robot specifically includes: Based on the predicted value of dynamic delay coefficient, adaptive function and the implementation principle of path planning, feature terms of intelligent path planning for apple picking robots are extracted. Based on big data or test experiments, obtain training and target sample sets for intelligent path planning of apple picking robots; Based on a deep learning algorithm model, an intelligent path planning model is established. By acquiring the predicted dynamic delay coefficient and dynamic picking adaptability function of the picking robots on both sides in real time, as well as the implementation principle judgment of the picking robot's dynamic picking path planning, the optimal path planning of the apple picking robot is obtained.

6. The intelligent path planning method for an apple-picking robot based on deep learning according to claim 5, characterized in that, The establishment of a human-computer interaction platform for storing, displaying, and issuing early warnings of monitoring data from the apple-picking robot during the picking process, as well as initial path planning, specifically includes: Establish a human-machine interaction platform based on Internet of Things technology to receive and store monitoring data from both ends of the harvesting robot during the harvesting process; Based on the human-computer interaction platform, we built the operating environment for the intelligent path planning model, and updated and improved the model's practicality by optimizing parameters. Based on the human-computer interaction platform, the location information and abnormal situations of the two picking robots during the apple picking process are displayed and alerted. Based on the human-computer interaction platform, the initial path planning and manual remote control of the two harvesting robots are completed through the human-computer interaction interface.

7. A deep learning-based intelligent path planning system for apple-picking robots, characterized in that, The method for implementing the intelligent path planning method for apple picking robots based on deep learning as described in any one of claims 1-6 includes: The harvesting distribution acquisition module is used to acquire the channel spacing between two adjacent hedges, the length of each hedge, and the number of hedges in the harvesting area based on the apple planting distribution. The harvesting method module is used to harvest apples on both sides of the apple tree using a complementary apple harvesting method, thereby improving apple harvesting efficiency and reducing the impact of light through dual-machine collaboration. The path planning module is used to obtain the predicted dynamic delay coefficient of apple picking in a single frame area by the apple picking robots on both sides based on the picking efficiency of the apple picking robot; obtain the adaptive function of dynamic picking by the two side picking robots based on the predicted dynamic delay coefficient and the sudden obstacle situation during the picking process; determine the implementation principle of dynamic picking path planning of the picking robot based on the adaptive function of dynamic picking by the two side picking robots; and establish an intelligent path planning model based on a deep learning algorithm model, combined with the predicted dynamic delay coefficient and the dynamic picking adaptive function, to obtain the optimal path planning of the apple picking robot. The interactive platform module is used to establish a human-computer interaction platform for storing, displaying, and providing early warnings of monitoring data of the apple picking robot during the picking process, as well as initial path planning.

8. The intelligent path planning system for an apple-picking robot based on deep learning according to claim 7, characterized in that, The path planning module includes: A dynamic delay unit is used to obtain the predicted value of the dynamic delay coefficient of apple picking in a single frame area of ​​the apple picking robots on both sides based on the picking efficiency of the apple picking robot. An adaptive function unit is used to obtain the adaptive function of dynamic picking by the two picking robots based on the predicted value of the dynamic delay coefficient and the sudden obstacle situation during the picking process. The implementation principle unit is used to determine the implementation principle of the dynamic picking path planning of the picking robot based on the adaptive function of the dynamic picking of the two picking robots. The path planning unit is used to establish an intelligent path planning model based on a deep learning algorithm model, combined with the predicted value of dynamic delay coefficient and the dynamic picking adaptation function, to obtain the optimal path planning for the apple picking robot.

Images