A collection device and method, and storage medium

Abstract

The embodiment of the application discloses a collection device and method, and a storage medium. The collection device comprises a camera, a computing power unit, an electromagnetic buffer device and a current modulator. The camera in the collection device is used to acquire image information of a first target and a second target. The first target comprises the second target. The second target is wrapped with a metal coil bag. The metal coil bag is composed of three groups of mutually perpendicular metal coils. The computing power unit is used to determine a current value of the electromagnetic buffer device according to the image information of the first target and the second target. The current modulator is used to supply power to the electromagnetic buffer device according to the current value. The electromagnetic buffer device is used to collect and process a third target falling in the second target in an unfolded state. The electromagnetic buffer device comprises an unfolded state and a folded state.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and more particularly to a collection device and method, and a storage medium. Background Technology

[0002] Currently, fruit drop collection machines can be mainly divided into mechanical cushioning systems and net-bag cushioning systems. Mechanical cushioning systems primarily utilize springs and hydraulic mechanisms to cushion the falling fruit, relying on spring deformation to absorb the gravitational potential and kinetic energy of the fruit, or using belt friction to slow the falling speed. Net-bag cushioning systems, on the other hand, use flexible devices such as nets to collect the fallen fruit, using the deformation of the net to provide good cushioning and shock absorption. However, both of these methods are contact-based cushioning, requiring direct contact with the fruit for cushioning and shock absorption, which can easily cause physical damage to the fruit. Summary of the Invention

[0003] This application provides a collection device and method, as well as a storage medium, which can achieve contactless buffered collection and effectively reduce physical damage to falling objects.

[0004] The technical solution of this application embodiment is implemented as follows:

[0005] In a first aspect, embodiments of this application provide a collection device, which includes a camera, a computing unit, an electromagnetic buffer device, and a current modulator; wherein...

[0006] The camera is used to acquire image information of a first target and a second target; wherein the first target includes the second target; the second target is wrapped with the metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils;

[0007] The computing unit is used to determine the current value of the electromagnetic buffer device based on the image information of the first target and the second target.

[0008] The current modulator is used to supply power to the electromagnetic buffer device according to the current value;

[0009] The electromagnetic buffer device is used to collect and process a third target falling from the second target when it is in an unfolded state; wherein the electromagnetic buffer device includes an unfolded state and a folded state; the second target includes the third target.

[0010] Secondly, embodiments of this application provide a collection method, which is applied to the collection device described above; the method includes:

[0011] Image information of a first target and a second target is acquired; wherein the first target includes the second target; the second target is wrapped in a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils;

[0012] The current value of the electromagnetic buffer device is determined based on the image information of the first target and the second target;

[0013] The electromagnetic buffer device is deployed and powered according to the current value to collect and process the falling third target in the second target; wherein the electromagnetic buffer device includes an deployed state and a folded state; the second target includes the third target.

[0014] Thirdly, embodiments of this application provide a computer-readable storage medium having a program stored thereon, which is applied in a collection device. When the program is executed by a processor, it implements the collection method described above.

[0015] This application provides a collection device and method, and a storage medium. The collection device includes a camera, a computing unit, an electromagnetic buffer device, and a current modulator. The camera is used to acquire image information of a first target and a second target. The first target includes a second target. The second target is wrapped with a metal coil bag. The metal coil bag is composed of three sets of mutually perpendicular metal coils. The computing unit is used to determine the current value of the electromagnetic buffer device based on the image information of the first and second targets. The current modulator is used to supply power to the electromagnetic buffer device based on the current value. The electromagnetic buffer device is used to collect a third target falling from the second target when it is in an unfolded state. The electromagnetic buffer device includes an unfolded state and a folded state. Therefore, in this application, the collection device can first acquire image information of the first and second targets through a camera. The second target is covered with a metal coil bag. Then, the computing unit determines the current value of the electromagnetic buffer device based on the image information of the first and second targets. When collecting the third target falling from the second target, the electromagnetic buffer device is deployed, and the current regulator supplies power to the electromagnetic buffer device according to the current value, so that the electromagnetic buffer device can generate an induced magnetic field. When the third target covered with the metal coil bag falls into the electromagnetic buffer device, an induced current is generated. The third target is subjected to magnetic force in the magnetic field generated by the electromagnetic buffer device, which reduces the falling speed of the third target and buffers the third target. This achieves contactless buffer collection and effectively reduces physical damage to the falling object. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 1 ;

[0017] Figure 2 This is a schematic diagram of the composition structure of the metal coil bag proposed in the embodiments of this application. Figure 1 ;

[0018] Figure 3 This is a schematic diagram of the composition structure of the metal coil bag proposed in the embodiments of this application. Figure 2 ;

[0019] Figure 4 This is a schematic diagram of the composition structure of the metal coil bag proposed in the embodiments of this application. Figure 3 ;

[0020] Figure 5 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 2 ;

[0021] Figure 6 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 3 ;

[0022] Figure 7 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 4 ;

[0023] Figure 8 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 5 ;

[0024] Figure 9 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 6 ;

[0025] Figure 10 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 7 ;

[0026] Figure 11 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 1 ;

[0027] Figure 12 This is a schematic diagram illustrating the implementation of the collection method proposed in the embodiments of this application. Figure 1 ;

[0028] Figure 13 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 2 ;

[0029] Figure 14 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 3 ;

[0030] Figure 15 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 4 ;

[0031] Figure 16 This is a schematic diagram illustrating the implementation of the collection method proposed in the embodiments of this application. Figure 2 . Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the relevant application and not for limiting the application. Furthermore, it should be noted that, for ease of description, only the parts related to the relevant application are shown in the accompanying drawings.

[0033] Currently, fruit-falling harvesting machines can be mainly divided into mechanical cushioning systems and net-bag cushioning systems. Mechanical cushioning systems primarily utilize springs and hydraulic mechanisms to buffer falling fruit, relying on spring deformation to absorb the gravitational potential and kinetic energy of the fruit, or using belt friction to slow the falling speed. Net-bag cushioning systems, on the other hand, use flexible devices such as nets to collect the fallen fruit, using the deformation of the net to provide good cushioning and shock absorption. However, both of these systems are contact-based cushioning methods, requiring direct contact with the fruit for cushioning and shock absorption. This can cause minor abrasion to the fruit skin, or even fruit breakage. Furthermore, as the number of falling fruits increases, collisions between them increase the overall damage rate, easily causing physical damage to the fruit. In addition, the success rate of fruit-picking machinery largely depends on the accuracy of target recognition and the control precision of the robotic arm, making it susceptible to influences from ambient light, weather, and obstructions.

[0034] To address the problems existing in current collection methods, this application provides a collection device and method, as well as a storage medium. The collection device includes a camera, a computing unit, an electromagnetic buffer device, and a current modulator. The camera is used to acquire image information of a first target and a second target; wherein the first target includes the second target; the second target is wrapped in a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils; the computing unit is used to determine the current value of the electromagnetic buffer device based on the image information of the first and second targets; the current modulator is used to supply power to the electromagnetic buffer device according to the current value; the electromagnetic buffer device is used to collect a third target falling from the second target when in an unfolded state; wherein the electromagnetic buffer device includes an unfolded state and a folded state; the second target includes the third target. This achieves contactless buffer collection, effectively reducing physical damage to falling objects.

[0035] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0036] Example 1

[0037] This application provides a collection method. Figure 1 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 1 ,like Figure 1 As shown, the collection device 10 includes a camera 11, a computing unit 12, a current modulator 13, and an electromagnetic buffer device 14.

[0038] Camera 11 is used to acquire image information of a first target and a second target; wherein the first target includes the second target; the second target is wrapped with a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils.

[0039] It should be noted that, in the embodiments of this application, the first target may include a second target; for example, the first target is a fruit tree, and the second target is the fruit in the fruit tree.

[0040] Furthermore, in the embodiments of this application, the image information of the first target and the second target can be collected first using the camera 11; specifically, the method of collecting image information can be to use the camera 11 to capture video information of the first target and the second target, and then perform image frame extraction on the video information to obtain the image information of the first target and the second target.

[0041] For example, in an embodiment of this application, the collecting device 10 enters the orchard and uses the camera 11 in the collecting device 10 to collect video information of the fruit trees corresponding to the fruit to be collected. Then, the video information is framed, and the extracted fruit tree image information is marked with rectangular boxes. The rectangular boxes of the fruits with complete shapes in the fruit trees are also marked, while the obscured or incomplete fruits are ignored, thereby obtaining the image information of the fruit trees and the image information of the fruits on the fruit trees.

[0042] It should be noted that in the embodiments of this application, the second target is wrapped with a metal coil bag. In order to ensure that the second target is subjected to uniform force in all directions when it falls, the metal coil bag is composed of three sets of mutually perpendicular metal coils. This ensures that when the second target wrapped with the metal coil bag falls at any different angle, the force is balanced and the trajectory remains unchanged.

[0043] For example, in the embodiments of this application, Figure 2 This is a schematic diagram of the composition structure of the metal coil bag proposed in the embodiments of this application. Figure 1 ,like Figure 2 The image shows the first winding direction of the metal coils in a bag of metal coils, among three sets of mutually perpendicular metal coils. Figure 3 This is a schematic diagram of the composition structure of the metal coil bag proposed in the embodiments of this application. Figure 2 ,like Figure 3 The image shows a metal coil bag containing three sets of mutually perpendicular metal coils, with... Figure 2 Different, and perpendicular to Figure 2 The second winding direction in the winding direction; Figure 4 This is a schematic diagram of the composition structure of the metal coil bag proposed in the embodiments of this application. Figure 3 ,like Figure 4 The image shows a metal coil bag containing three sets of mutually perpendicular metal coils, with... Figure 2 and Figure 3 Different, and perpendicular to Figure 2 and Figure 3 The third winding direction in the winding direction.

[0044] The computing unit 12 is used to determine the current value of the electromagnetic buffer device 14 based on the image information of the first target and the second target.

[0045] In the embodiments of this application, the current value determined based on the image information of the first target and the second target enables the electromagnetic buffer device 14 to generate an induced magnetic field that adapts to the falling third target in the second target after being energized according to the current value, thereby playing a suitable buffering role.

[0046] In some embodiments of this application, the computing unit 12 can perform target detection processing on the image information of the first target and the second target according to the target detection algorithm to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target; then, based on the trained Bayesian neural network model, the height information of the first target is obtained according to at least one first image, and the size information of the second target is obtained according to at least one second image; finally, the current value is determined according to the height information of the first target and the size information of the second target.

[0047] Current modulator 13 is used to power electromagnetic buffer device 14 according to the current value.

[0048] It should be noted that, in the embodiments of this application, the current modulator 13 can be used to select an appropriate current level according to the current value, thereby supplying power to the electromagnetic buffer device 14 according to the set current level.

[0049] The electromagnetic buffer device 14 is used to collect and process the falling third target from the second target when it is in the deployed state; wherein the electromagnetic buffer device includes an deployed state and a folded state.

[0050] It should be noted that, in the embodiments of this application, the electromagnetic buffer device 14 is composed of at least one electromagnetic ring, and each electromagnetic ring is composed of a wound coil; the at least one electromagnetic ring is arranged sequentially in a preset direction, and after the electromagnetic buffer device is powered, the magnetic field strength corresponding to the at least one electromagnetic ring gradually increases along the preset direction; wherein, the preset direction is the direction perpendicular to the ground.

[0051] Furthermore, in the embodiments of this application, the magnetic field strength corresponding to the electromagnetic ring is determined based on the magnitude of the current value corresponding to the electromagnetic ring. Therefore, the current value corresponding to at least one electromagnetic ring gradually increases along a predetermined direction; wherein, the current value corresponding to at least one electromagnetic ring can increase linearly, or increase sequentially according to a quadratic relationship; for example, the current value corresponding to at least one electromagnetic ring can be expressed by the following formula:

[0052]

[0053]

[0054] Where 'a' is a preset proportional coefficient, and '△' is a preset interval coefficient. For the next electromagnetic loop, This is the previous electromagnetic loop.

[0055] Furthermore, in the embodiments of this application, the electromagnetic buffer device 14 may include an unfolded state and a folded state; when the electromagnetic buffer device 14 is working, it is in the unfolded state, and when the electromagnetic buffer device 14 is not working, it may be in the folded state, with at least one electromagnetic ring in the electromagnetic buffer device 14 folded together, thereby ensuring the portability of the collection device 10.

[0056] It should be noted that, in the embodiments of this application, the electromagnetic buffer device 14, as the core device of the collection device 10, can provide magnetic fields of different intensities according to different current levels set by the current modulator 13, thereby realizing the collection of different types of second targets among different first targets.

[0057] For example, Figure 5 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 2 ,like Figure 5 The image shows the unfolded state of the electromagnetic buffer device 14. It can be seen that the electromagnetic buffer device 14 can be composed of at least one electromagnetic ring, and at least one electromagnetic ring can be unfolded. Figure 6 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 3 ,like Figure 6 The image shows the folded state of the electromagnetic buffer device 14. It can be seen that at least one electromagnetic ring in the electromagnetic buffer device 14 can also be folded.

[0058] It should be noted that, in the embodiments of this application, the third target refers to the falling target among the second targets; for example, if the second target is the fruit on the fruit tree, then the third target is the fruit that has fallen among all the fruits.

[0059] For example, in an embodiment of this application, the fruit tree (first target) includes fruit (second target), each fruit is wrapped in a metal coil bag. Besides buffering collection, the metal coil bag protects the fruit from damage by other organisms and reduces the impact of natural disasters on the fruit. When collecting falling fruit (third target), an electromagnetic buffer device 14 can be deployed. Powering the electromagnetic buffer device 14 generates a magnetic field from at least one electromagnetic ring. When the fruit wrapped in the metal coil bag falls into the electromagnetic buffer device 14, it is equivalent to an inductive element falling into the induced magnetic field generated by the device. According to the law of electromagnetic induction, an induced current is generated, which in turn generates an induced magnetic field. Since the magnetic field polarity of the metal coil bag is opposite to that of the magnetic field in the electromagnetic buffer device 14, it slows down the falling fruit, thus achieving buffered collection and reducing physical damage.

[0060] Furthermore, the computing unit 12 is specifically used to perform target detection processing on the image information of the first target and the second target according to the target detection algorithm, so as to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target.

[0061] It should be noted that, in the embodiments of this application, the initial detection algorithm can be trained to obtain the object detection algorithm. For example, the initial detection algorithm can be the YOLO real-time object detection algorithm that has been trained using the general object dataset COCO. Then, the YOLO real-time object detection algorithm is fine-tuned by using a labeled dataset of fruits to be tested. During the training process, the number of fruit types that the algorithm needs to output can also be adjusted according to specific needs. Thus, after fine-tuning the training, the object detection algorithm is obtained.

[0062] Furthermore, in the embodiments of this application, a target detection algorithm can be used to perform target detection processing on the image information of the first target and the second target to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target; wherein, at least one first image can be an image of the first target from different angles, and at least one second image can be an image of the second target from different angles.

[0063] For example, in the embodiments of this application, the YOLO object detection algorithm (object detection algorithm) is used to perform object detection processing on the fruit tree (first object) and the fruit (second object), and the detected fruit tree and fruit images are cropped based on the OpenCV library to obtain at least one fruit tree image block (at least one first image) and at least one fruit image block (at least one second image).

[0064] Furthermore, after the computing unit 12 performs target detection processing on the image information of the first target and the second target according to the target detection algorithm to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target, it can obtain the height information of the first target based on the at least one first image and the size information of the second target based on the at least one second image, according to the trained Bayesian neural network model.

[0065] It should be noted that, in the embodiments of this application, an initial Bayesian neural network model can be trained to obtain a trained Bayesian neural network model. This can be achieved by designing an initial Bayesian neural network model with a two-layer structure. Since the weights of the initial Bayesian neural network model are not fixed but are sampled within a certain distribution, and based on the central limit theorem, assuming the weight distribution satisfies a Gaussian prior, the Bayesian network can represent the weights using mean and variance parameters. The mean and variance of the initial Bayesian neural network model are optimized through backpropagation until they meet preset mean and variance parameters, thereby obtaining the trained Bayesian neural network model.

[0066] Furthermore, in the embodiments of this application, the trained Bayesian neural network model is able to represent the random uncertainty of the data, and has stronger interpretability and robustness.

[0067] For example, in an embodiment of this application, a target detection algorithm is used to perform target detection processing on the collected orchard dataset to obtain data on target fruit trees and fruits. After cropping, at least one image of the target fruit trees and fruits is obtained. The at least one image of the target fruit trees and fruits is input into an initial Bayesian neural network model for training to obtain a trained Bayesian neural network model.

[0068] Furthermore, in embodiments of this application, based on a trained Bayesian neural network model, at least one first image is estimated to obtain the height information of the first target, and at least one second image is estimated to obtain the size information of the second target.

[0069] In some embodiments of this application, the computing unit 12 can input each image in at least one first image into a trained Bayesian neural network model according to a first preset number of times to obtain multiple first target height estimates equal to the first preset number of times, and calculate the average of the multiple first target height estimates to obtain the height information of the first target; wherein, the first preset number of times is an integer greater than or equal to 2; and input each image in at least one second image into a trained Bayesian neural network model according to a second preset number of times to obtain multiple second target size estimates equal to the second preset number of times, and calculate the average of the multiple second target size estimates to obtain the size information of the second target; wherein, the second preset number of times is a positive integer greater than or equal to 2.

[0070] Furthermore, after obtaining the height information of the first target based on at least one first image and the size information of the second target based on at least one second image, the computing unit 12 can determine the current value based on the height information of the first target and the size information of the second target, according to the trained Bayesian neural network model.

[0071] In some embodiments of this application, the computing unit 12 can first calculate the initial velocity estimate of the second target based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device 14 and the ground, and determine the mass of the second target based on the size information of the second target and the preset density of the second target, and then determine the current value based on the preset dynamic model, preset boundary conditions, initial velocity estimate and the mass of the second target.

[0072] In the embodiments of this application, the computing unit 12 can calculate the initial velocity estimate of the second target based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, and determine the mass of the second target based on the size information of the second target and the preset density of the second target.

[0073] For example, in an embodiment of this application, the initial velocity estimate corresponding to the second target can be calculated according to the following formula:

[0074]

[0075] Wherein, H0 is the height information of the first target, H is the distance between the vertex of the electromagnetic buffer device 14 and the ground, v0 is the initial velocity estimate, g is the gravitational acceleration, and t0 is the time from when the second target starts falling from the first target to when it falls into the electromagnetic buffer device 14.

[0076] It is understood that, in the embodiments of this application, the initial velocity estimate is the prediction of the velocity of the second target when it falls into the electromagnetic buffer device 14.

[0077] Furthermore, in the embodiments of this application, the preset density of the second target can be measured in advance; for example, the second target is approximated as a sphere, so that the volume of the second target is obtained according to the size information of the second target, and finally the mass of the second target is obtained by multiplying the volume of the second target and the preset density of the second target.

[0078] Furthermore, after the computing unit 12 calculates the initial velocity estimate of the second target based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, and determines the mass of the second target based on the size information of the second target and the preset density of the second target, it can determine the current value based on the preset dynamic model, preset boundary conditions, initial velocity estimate and the mass of the second target.

[0079] For example, in an embodiment of this application, the electromagnetic buffer device 14 includes four electromagnetic rings, and the preset dynamic model can be expressed as the following formula:

[0080]

[0081] in, When the second target passes through the j-th electromagnetic loop (j = 1, 2, 3, 4), the magnetic force experienced by the i-th coil in the metal coil bag is, correspondingly, Let the magnetic force experienced by the first coil in the metal coil bag be the magnetic force when the second target passes through the j-th electromagnetic loop. Let the magnetic force experienced by the second coil in the metal coil bag be the magnetic force when the second target passes through the j-th electromagnetic loop. The magnetic force experienced by the third coil in the metal coil bag when the second target passes through the j-th electromagnetic loop; Let be the gravity of the second target, and m be the mass of the second target. I is the acceleration of the net force acting on the second target. ij This represents the induced current generated in the i-th coil of the metal coil bag when the second target passes through the j-th electromagnetic loop. Correspondingly, B represents the induced current generated by the first coil in the metal coil bag when the second target passes through the j-th electromagnetic loop. j Let L represent the vacuum permeability of the j-th electromagnetic ring, L represent the total length of the coils in the three directions within the metal coil bag, t represent the falling time period within the electromagnetic buffer device 14, and ε represent the total permeability of the coils. ij Let be the induced electromotive force of the i-th coil in the j-th electromagnetic loop within the metal coil bag, t represent the falling time of the second target in the electromagnetic buffer device 14, μ0 is a constant used to calculate the vacuum permeability, which can be set by the user, and n is the number of coils per unit length in the electromagnetic loop. Let S be the current in the i-th electromagnetic ring of the electromagnetic buffer device 14, S be the cross-sectional area of ​​the metal coil bag surrounding the second target, and d be the diameter of the metal coil bag.

[0082] It is understandable that the current value determined based on the aforementioned preset dynamic model, preset boundary conditions, initial velocity estimate, and the mass of the second target is... Includes the current corresponding to each electromagnetic ring in the electromagnetic buffer device 14.

[0083] For example, in an embodiment of this application, based on the preset dynamic model shown in formula (4) above, the preset boundary conditions can be expressed as the following formula:

[0084]

[0085] Where v0 is the initial velocity estimate, The acceleration of the net force acting on the second objective.

[0086] Furthermore, Figure 7 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 4 ,like Figure 7 As shown, the collection device 10 also includes a gimbal 15.

[0087] The gimbal 15 is used to adjust the position and angle of the electromagnetic buffer device 14.

[0088] It should be noted that, in the embodiments of this application, the gimbal 15 may have four degrees of freedom of motion, including displacement degrees of freedom in the x and y directions, and rotational degrees of freedom of rotation along the x and y directions.

[0089] Furthermore, the electromagnetic buffer device 14 can be placed above the gimbal 15 and connected to the gimbal 15, thereby using the gimbal 15 to adjust the position and angle of the electromagnetic buffer device 14.

[0090] Furthermore, the computing unit 12 is specifically used to input each image in at least one first image into the trained Bayesian neural network model according to a first preset number of times, to obtain multiple first target height estimates equal to the first preset number of times, and to calculate the average of the multiple first target height estimates to obtain the height information of the first target; wherein, the first preset number of times is an integer greater than or equal to 2.

[0091] For example, in an embodiment of this application, there are three first images (images of fruit trees) a, b, and c. Image a is input three times (a first preset number of times) to a trained Bayesian neural network model to obtain three first target height estimates corresponding to a; image b is input three times to a trained Bayesian neural network model to obtain three first target height estimates corresponding to b; image c is input three times to a trained Bayesian neural network model to obtain three first target height estimates corresponding to c; finally, the average of the three first target height estimates corresponding to a, b, and c is calculated to obtain the height information of the first target.

[0092] Furthermore, the computing unit 12 can also determine the height estimation range of the first target based on the maximum and minimum values ​​among multiple height estimates of the first target. In the subsequent calculation of the current value using the height estimates of the first target, the maximum and minimum values ​​of the current value are calculated based on the maximum and minimum values ​​in the height estimation range of the first target, thereby obtaining the estimation range of the current value.

[0093] Furthermore, after the computing unit 12 inputs each image from at least one first image into the trained Bayesian neural network model according to a first preset number of times to obtain multiple first target height estimates equal to the first preset number of times, and calculates the average of the multiple first target height estimates to obtain the height information of the first target, it can input each image from at least one second image into the trained Bayesian neural network model according to a second preset number of times to obtain multiple second target size estimates equal to the second preset number of times, and calculates the average of the multiple second target size estimates to obtain the size information of the second target; wherein, the second preset number of times is a positive integer greater than or equal to 2.

[0094] For example, in an embodiment of this application, there are two second images (images of fruit trees) d and e. d is input twice (a second preset number of times) to a trained Bayesian neural network model to obtain two second target size estimates corresponding to d; e is input twice to a trained Bayesian neural network model to obtain two first target height estimates corresponding to e; finally, the average of the two second target size estimates corresponding to d and the two first target height estimates corresponding to e is calculated to obtain the height information of the first target.

[0095] Furthermore, in the embodiments of this application, Figure 8 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 5 ,like Figure 8 As shown, the collection device 10 may also include a storage box 16, a motor 17, and an operation display platform 18.

[0096] It should be noted that, in the embodiments of this application, the storage box 16 can be pushed out after it is full, and collection can continue after the empty storage box is replaced. The various processes in the entire collection process are closely connected, which improves the collection efficiency of fallen objects.

[0097] It is understood that in the embodiments of this application, the electric motor 17 is used to drive the collection device 10; the various instructions in the collection method can be issued through the display operation display platform 18, and the various instructions and working status can be displayed.

[0098] For example, in the embodiments of this application, Figure 9 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 6 ,like Figure 9 As shown, the collection device 10 can be configured on a wheeled, pushable platform for easy movement; the devices visible from the outside of the collection device 10 include a camera 11, an electromagnetic buffer device 14, a storage box 16, and an operation display platform 18. Inside the collection device 10 are a computing unit 12, a current modulator 13, a gimbal 15, and a motor 17; wherein, the gimbal 15 is mounted on the bottom of the electromagnetic buffer device 14. Figure 10 This is a schematic diagram of the composition of the collection device proposed in the embodiments of this application. Figure 7 ,like Figure 10 As shown, the collection device 10 is mounted on a wheeled, slidable platform, and the electromagnetic buffer device 14 and storage box 16 are visible from the outside.

[0099] This application provides a collection device, which includes a camera, a computing unit, an electromagnetic buffer device, and a current modulator. The camera is used to acquire image information of a first target and a second target. The first target includes a second target. The second target is wrapped with a metal coil bag. The metal coil bag is composed of three sets of mutually perpendicular metal coils. The computing unit is used to determine the current value of the electromagnetic buffer device based on the image information of the first and second targets. The current modulator is used to supply power to the electromagnetic buffer device based on the current value. The electromagnetic buffer device is used to collect a third target falling from the second target when it is in an unfolded state. The electromagnetic buffer device includes an unfolded state and a folded state. Therefore, in this application, the collection device can first acquire image information of the first and second targets through a camera. The second target is covered with a metal coil bag. Then, the computing unit determines the current value of the electromagnetic buffer device based on the image information of the first and second targets. When collecting the third target that has fallen from the second target, the electromagnetic buffer device is deployed, and the current regulator supplies power to the electromagnetic buffer device according to the current value, so that the electromagnetic buffer device can generate an induced magnetic field. When the third target covered with the metal coil bag falls into the electromagnetic buffer device, an induced current is generated. The third target is subjected to magnetic force in the magnetic field generated by the electromagnetic buffer device, which reduces the falling speed of the third target and buffers the third target. This achieves contactless buffer collection and effectively reduces physical damage to the falling object.

[0100] Example 2

[0101] Based on the above embodiments, in another embodiment of this application, a collection method is proposed. Figure 11 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 1 ,like Figure 11 As shown, the collection method of the collection device includes the following steps:

[0102] Step 101: Obtain image information of the first target and the second target; wherein the first target includes the second target; the second target is wrapped in a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils.

[0103] In embodiments of this application, the collecting device can acquire image information of a first target and a second target; wherein the first target includes the second target; the second target is wrapped with a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils.

[0104] It should be noted that, in the embodiments of this application, the first target may include a second target; for example, the first target is a fruit tree, and the second target is the fruit in the fruit tree.

[0105] Furthermore, in the embodiments of this application, the image information of the first target and the second target can be collected first using the camera in the collection device; specifically, the method of collecting image information can be to use the camera to capture video information of the first target and the second target, and then perform image frame extraction on the video information to obtain the image information of the first target and the second target.

[0106] For example, in an embodiment of this application, the collecting device enters the orchard, uses the camera in the collecting device to collect video information of the fruit trees corresponding to the fruit to be collected, then extracts frames from the video information, marks the extracted fruit tree image information with rectangular boxes, and marks the rectangular boxes of the fruits with complete shapes in the fruit trees, ignoring the obscured or incomplete fruits, thereby obtaining the image information of the fruit trees and the image information of the fruits on the fruit trees.

[0107] It should be noted that in the embodiments of this application, the second target is wrapped with a metal coil bag. In order to ensure that the second target is subjected to uniform force in all directions when it falls, the metal coil bag is composed of three sets of mutually perpendicular metal coils. This ensures that when the second target wrapped with the metal coil bag falls at any different angle, the force is balanced and the trajectory remains unchanged.

[0108] Step 102: Determine the current value of the electromagnetic buffer device based on the image information of the first target and the second target.

[0109] In the embodiments of this application, after acquiring image information of the first target and the second target, the collecting device can determine the current value of the electromagnetic buffer device based on the image information of the first target and the second target.

[0110] In some embodiments of this application, the computing unit in the collection device can be used to perform target detection processing on the image information of the first target and the second target according to the target detection algorithm to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target; then, based on the trained Bayesian neural network model, the height information of the first target is obtained according to at least one first image, and the size information of the second target is obtained according to at least one second image; finally, the current value is determined according to the height information of the first target and the size information of the second target.

[0111] It should be noted that, in the embodiments of this application, the electromagnetic buffer device is composed of at least one electromagnetic ring, and each electromagnetic ring is composed of a wound coil; the at least one electromagnetic ring is arranged sequentially from top to bottom according to a preset direction; wherein, the preset direction is a vertical direction pointing to the ground; and after the electromagnetic buffer device is powered, the magnetic field strength of the at least one electromagnetic ring gradually increases from top to bottom.

[0112] Step 103: Deploy the electromagnetic buffer device and supply power to the electromagnetic buffer device according to the current value to collect and process the falling third target in the second target; wherein, the electromagnetic buffer device includes an deployed state and a folded state.

[0113] In the embodiments of this application, after the collecting device determines the current value of the electromagnetic buffer device based on the image information of the first target and the second target, it can unfold the electromagnetic buffer device and supply power to the electromagnetic buffer device according to the current value to collect the falling third target in the second target; wherein, the electromagnetic buffer device includes an unfolded state and a folded state.

[0114] It should be noted that, in the embodiments of this application, the current modulator in the collection device can be used to select an appropriate current level according to the current value, thereby supplying power to the electromagnetic buffer device according to the set current level.

[0115] Furthermore, in the embodiments of this application, the electromagnetic buffer device may include an unfolded state and a folded state; when the electromagnetic buffer device is working, it is in the unfolded state, and when the electromagnetic buffer device is not working, it may be in the folded state, with at least one electromagnetic ring in the electromagnetic buffer device folded together vertically, thereby ensuring the portability of the collection device.

[0116] It should be noted that, in the embodiments of this application, the third target refers to the falling target among the second targets; for example, if the second target is the fruit on the fruit tree, then the third target is the fruit that has fallen among all the fruits.

[0117] For example, in an embodiment of this application, the fruit tree (first target) includes fruit (second target), each fruit wrapped in a metal coil bag. Besides buffering collection, the metal coil bag protects the fruit from damage by other organisms and reduces the impact of natural disasters. When collecting falling fruit (third target), an electromagnetic buffer device can be deployed. Powering the device generates a magnetic field in at least one electromagnetic ring. When the fruit wrapped in the metal coil bag falls into the device, it's equivalent to an inductive element falling into the induced magnetic field generated by the device. According to the law of electromagnetic induction, an induced current is generated, which in turn generates an induced magnetic field. Since the magnetic field polarity of the metal coil bag is opposite to that of the magnetic field in the electromagnetic buffer device, it slows down the falling fruit, thus achieving buffered collection and reducing physical damage. For example... Figure 12 This is a schematic diagram illustrating the implementation of the collection method proposed in the embodiments of this application. Figure 1 ,like Figure 12The image shows a fruit falling into an electromagnetic buffer device. Because the fruit is encased in a metal coil bag, magnetic forces F1, F2, and F3 are generated in three directions within the electromagnetic buffer device. B G is the resultant force of the three magnetic forces, and G is the gravity of the fruit.

[0118] Figure 13 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 2 ,like Figure 13 As shown, the method by which the collecting device determines the current value of the electromagnetic buffer device based on the image information of the first target and the second target, i.e., the method proposed in step 102, may include the following steps:

[0119] Step 102a: Perform target detection processing on the image information of the first target and the second target according to the target detection algorithm to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target.

[0120] In some embodiments of this application, the collecting device determines the current value of the electromagnetic buffer device based on the image information of the first target and the second target; in some embodiments of this application, the collecting device may first perform target detection processing on the image information of the first target and the second target according to the target detection algorithm to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target.

[0121] It should be noted that, in the embodiments of this application, the initial detection algorithm can be trained to obtain the object detection algorithm. For example, the initial detection algorithm can be the YOLO real-time object detection algorithm that has been trained using the general object dataset COCO. Then, the YOLO real-time object detection algorithm is fine-tuned by using a labeled dataset of fruits to be tested. During the training process, the number of fruit types that the algorithm needs to output can also be adjusted according to specific needs. Thus, after fine-tuning the training, the object detection algorithm is obtained.

[0122] Furthermore, in the embodiments of this application, a target detection algorithm can be used to perform target detection processing on the image information of the first target and the second target to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target; wherein, at least one first image can be an image of the first target from different angles, and at least one second image can be an image of the second target from different angles.

[0123] For example, in the embodiments of this application, the YOLO object detection algorithm (object detection algorithm) is used to perform object detection processing on the fruit tree (first object) and the fruit (second object), and the detected fruit tree and fruit images are cropped based on the OpenCV library to obtain at least one fruit tree image block (at least one first image) and at least one fruit image block (at least one second image).

[0124] Step 102b: Based on the trained Bayesian neural network model, obtain the height information of the first target from at least one first image, and obtain the size information of the second target from at least one second image.

[0125] In the embodiments of this application, after the collection device performs target detection processing on the image information of the first target and the second target according to the target detection algorithm to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target, it can obtain the height information of the first target based on the at least one first image and the size information of the second target based on the at least one second image, according to the trained Bayesian neural network model.

[0126] It should be noted that, in the embodiments of this application, an initial Bayesian neural network model can be trained to obtain a trained Bayesian neural network model. This can be achieved by designing an initial Bayesian neural network model with a two-layer structure. Since the weights of the initial Bayesian neural network model are not fixed but are sampled within a certain distribution, and based on the central limit theorem, assuming the weight distribution satisfies a Gaussian prior, the Bayesian network can represent the weights using mean and variance parameters. The mean and variance of the initial Bayesian neural network model are optimized through backpropagation until they meet preset mean and variance parameters, thereby obtaining the trained Bayesian neural network model.

[0127] Furthermore, in the embodiments of this application, the trained Bayesian neural network model is able to represent the random uncertainty of the data, and has stronger interpretability and robustness.

[0128] For example, in an embodiment of this application, a target detection algorithm is used to perform target detection processing on the collected orchard dataset to obtain data on target fruit trees and fruits. After cropping, at least one image of the target fruit trees and fruits is obtained. The at least one image of the target fruit trees and fruits is input into an initial Bayesian neural network model for training to obtain a trained Bayesian neural network model.

[0129] Furthermore, in embodiments of this application, based on a trained Bayesian neural network model, at least one first image is estimated to obtain the height information of the first target, and at least one second image is estimated to obtain the size information of the second target.

[0130] In some embodiments of this application, the computing unit in the collection device can input each image in at least one first image into a trained Bayesian neural network model according to a first preset number of times to obtain multiple first target height estimates equal to the first preset number of times, and calculate the average of the multiple first target height estimates to obtain the height information of the first target; wherein, the first preset number of times is an integer greater than or equal to 2; and input each image in at least one second image into a trained Bayesian neural network model according to a second preset number of times to obtain multiple second target size estimates equal to the second preset number of times, and calculate the average of the multiple second target size estimates to obtain the size information of the second target; wherein, the second preset number of times is a positive integer greater than or equal to 2.

[0131] Step 102c: Determine the current value based on the height information of the first target and the size information of the second target.

[0132] In the embodiments of this application, after the collecting device obtains the height information of the first target based on at least one first image and the size information of the second target based on at least one second image, it can determine the current value based on the height information of the first target and the size information of the second target.

[0133] In some embodiments of this application, the computing unit in the collection device can first calculate the initial velocity estimate of the second target based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, and determine the mass of the second target based on the size information of the second target and the preset density of the second target, and then determine the current value based on the preset dynamic model, preset boundary conditions, initial velocity estimate and the mass of the second target.

[0134] Figure 14 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 3 ,like Figure 14 As shown, the method by which the collecting device determines the current value based on the height information of the first target and the size information of the second target, i.e., the method proposed in step 102c, may include the following steps:

[0135] Step 102c1: Based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, calculate the estimated initial velocity value of the second target, and determine the mass of the second target based on the size information of the second target and the preset density of the second target.

[0136] In some embodiments of this application, the collecting device determines the current value based on the height information of the first target and the size information of the second target. In some embodiments of this application, the collecting device may first calculate the initial velocity estimate of the second target based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, and determine the mass of the second target based on the size information of the second target and the preset density of the second target.

[0137] For example, in an embodiment of this application, the initial velocity estimate corresponding to the second target can be calculated according to the aforementioned formula (3).

[0138] It is understood that, in the embodiments of this application, the initial velocity estimate is the prediction of the velocity of the second target when it falls into the electromagnetic buffer device.

[0139] Furthermore, in the embodiments of this application, the preset density of the second target can be measured in advance; for example, the second target is approximated as a sphere, so that the volume of the second target is obtained according to the size information of the second target, and finally the mass of the second target is obtained by multiplying the volume of the second target and the preset density of the second target.

[0140] Step 102c2: Determine the current value based on the preset dynamic model, preset boundary conditions, initial velocity estimate, and the mass of the second target.

[0141] In the embodiments of this application, after the collecting device calculates the initial velocity estimate of the second target based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, and determines the mass of the second target based on the size information of the second target and the preset density of the second target, it can determine the current value based on the preset dynamic model, preset boundary conditions, the initial velocity estimate and the mass of the second target.

[0142] For example, in an embodiment of this application, the electromagnetic buffer device 14 includes four electromagnetic rings, and the preset dynamic model can be expressed as in the aforementioned formula (4).

[0143] For example, in an embodiment of this application, based on the preset dynamic model shown in the above formula (4), the preset boundary conditions can be expressed as the aforementioned formula (5).

[0144] Furthermore, in embodiments of this application, before the collecting device deploys the electromagnetic buffer device and supplies power to the electromagnetic buffer device according to the current value to collect the falling third target from the second target, i.e. before step 103, the following steps may be included:

[0145] Step 104: Adjust the position and angle of the electromagnetic buffer device.

[0146] In the embodiments of this application, before the collecting device deploys the electromagnetic buffer device and supplies power to the electromagnetic buffer device according to the current value to collect the falling third target from the second target, the position and angle of the electromagnetic buffer device can be adjusted.

[0147] It should be noted that, in the embodiments of this application, the collecting device can adjust the position and angle of the electromagnetic buffer device through the gimbal; the gimbal can have 4 degrees of freedom of motion, including displacement degrees of freedom in the x and y directions, and rotational degrees of freedom of rotation along the x and y directions.

[0148] Figure 15 This is a schematic diagram of the implementation process of the collection method proposed in the embodiments of this application. Figure 4 ,like Figure 15 As shown, the method by which the collection device obtains the height information of the first target based on at least one first image and the size information of the second target based on at least one second image, i.e., the method proposed in step 102b, may include the following steps:

[0149] Step 102b1: Input each image in at least one first image into the trained Bayesian neural network model according to a first preset number of times to obtain multiple first target height estimates equal to the first preset number of times, and calculate the average of the multiple first target height estimates to obtain the height information of the first target; wherein, the first preset number of times is an integer greater than or equal to 2.

[0150] In embodiments of this application, the collection device obtains the height information of a first target based on at least one first image and the size information of a second target based on at least one second image, using a trained Bayesian neural network model. In some embodiments of this application, the collection device can input each image from at least one first image into the trained Bayesian neural network model a first preset number of times to obtain multiple height estimates of the first target equal to the first preset number of times, and calculate the average of the multiple height estimates of the first target to obtain the height information of the first target. The first preset number of times is an integer greater than or equal to 2.

[0151] For example, in an embodiment of this application, there are three first images (images of fruit trees) a, b, and c. Image a is input three times (a first preset number of times) to a trained Bayesian neural network model to obtain three first target height estimates corresponding to a; image b is input three times to a trained Bayesian neural network model to obtain three first target height estimates corresponding to b; image c is input three times to a trained Bayesian neural network model to obtain three first target height estimates corresponding to c; finally, the average of the three first target height estimates corresponding to a, b, and c is calculated to obtain the height information of the first target.

[0152] Furthermore, the height estimation interval of the first target can be determined based on the maximum and minimum values ​​among multiple height estimates of the first target. In the subsequent calculation of the current value using the height estimates of the first target, the maximum and minimum values ​​of the current value can be calculated based on the maximum and minimum values ​​in the height estimation interval of the first target, thereby obtaining the estimation interval of the current value.

[0153] Step 102b2: Input each image in at least one second image into the trained Bayesian neural network model according to a second preset number of times to obtain multiple second target size estimates equal to the second preset number of times, and calculate the average of the multiple second target size estimates to obtain the size information of the second target; wherein, the second preset number of times is an integer greater than or equal to 2.

[0154] In embodiments of this application, the collection device obtains the height information of a first target based on at least one first image and the size information of a second target based on at least one second image, using a trained Bayesian neural network model. In some embodiments of this application, the collection device can input each image from at least one second image into the trained Bayesian neural network model a second preset number of times to obtain multiple second target size estimates equal to the second preset number of times, and calculate the average of the multiple second target size estimates to obtain the size information of the second target. The second preset number of times is an integer greater than or equal to 2.

[0155] For example, in an embodiment of this application, there are two second images (images of fruit trees) d and e. d is input twice (a second preset number of times) to a trained Bayesian neural network model to obtain two second target size estimates corresponding to d; e is input twice to a trained Bayesian neural network model to obtain two first target height estimates corresponding to e; finally, the average of the two second target size estimates corresponding to d and the two first target height estimates corresponding to e is calculated to obtain the height information of the first target.

[0156] In summary, for example, Figure 16This is a schematic diagram illustrating the implementation of the collection method proposed in the embodiments of this application. Figure 2 ,like Figure 16 As shown, when implementing the collection method of this application using the collection device, the collection device can first be moved to the area below the fruit tree (first target) to be collected, and image acquisition can be performed to obtain images of the fruit tree and the fruit on the tree (second target). Then, the detection results are obtained through a target detection algorithm. The detection results include at least one first image (fruit tree image) corresponding to the fruit tree and at least one second image (fruit image) corresponding to the fruit. Next, the trained Bayesian neural network model is used to predict and estimate at least one fruit tree image and at least one fruit image to obtain the height information of the fruit tree (height information of the first target) and the size information of the fruit (size information of the second target). Then, using a preset dynamic model and preset boundary conditions, the current value required to power the electromagnetic buffer device is calculated by combining the height information of the fruit tree and the size information of the fruit. Then, the current level of the current modulator can be set according to the current value, and the collection device can be moved to a suitable position to deploy the electromagnetic buffer device, thereby realizing non-contact, buffered collection of the falling fruit (third target).

[0157] Therefore, the embodiments of this application can realize non-contact buffer collection of falling objects, which can reduce physical damage to falling objects compared with the collection methods in the prior art. At the same time, it can make adaptive adjustments for any different types of first and second targets, and adaptively realize non-contact buffer collection, which has a wider range of applications. For example, in addition to fruit collection in the agricultural field, it can also be applied to other scenarios such as deceleration of objects to be tested and device vibration reduction.

[0158] This application provides a collection method applied to a collection device. The collection device includes a camera, a computing unit, an electromagnetic buffer device, and a current modulator. The camera is used to acquire image information of a first target and a second target. The first target includes a second target. The second target is wrapped with a metal coil bag. The metal coil bag consists of three sets of mutually perpendicular metal coils. The computing unit is used to determine the current value of the electromagnetic buffer device based on the image information of the first and second targets. The current modulator is used to supply power to the electromagnetic buffer device based on the current value. The electromagnetic buffer device is used to collect a third target falling from the second target when it is in an unfolded state. The electromagnetic buffer device includes an unfolded state and a folded state. Therefore, in this application, the collection device can first acquire image information of the first and second targets through a camera. The second target is covered with a metal coil bag. Then, the computing unit determines the current value of the electromagnetic buffer device based on the image information of the first and second targets. When collecting the third target falling from the second target, the electromagnetic buffer device is deployed, and the current regulator supplies power to the electromagnetic buffer device according to the current value, so that the electromagnetic buffer device can generate an induced magnetic field. When the third target covered with the metal coil bag falls into the electromagnetic buffer device, an induced current is generated. The third target is subjected to magnetic force in the magnetic field generated by the electromagnetic buffer device, which reduces the falling speed of the third target and buffers the third target. This achieves contactless buffer collection and effectively reduces physical damage to the falling object.

[0159] Furthermore, in this embodiment, the functional modules can be integrated into one analysis unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional module.

[0160] If the integrated unit is implemented as a software functional module and is not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the method of this embodiment. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0161] This application provides a collection device and method, and a storage medium. The collection device includes a camera, a computing unit, an electromagnetic buffer device, and a current modulator. The camera is used to acquire image information of a first target and a second target. The first target includes a second target. The second target is wrapped with a metal coil bag. The metal coil bag is composed of three sets of mutually perpendicular metal coils. The computing unit is used to determine the current value of the electromagnetic buffer device based on the image information of the first and second targets. The current modulator is used to supply power to the electromagnetic buffer device based on the current value. The electromagnetic buffer device is used to collect a third target falling from the second target when it is in an unfolded state. The electromagnetic buffer device includes an unfolded state and a folded state. Therefore, in this application, the collection device can first acquire image information of the first and second targets through a camera. The second target is covered with a metal coil bag. Then, the computing unit determines the current value of the electromagnetic buffer device based on the image information of the first and second targets. When collecting the third target falling from the second target, the electromagnetic buffer device is deployed, and the current regulator supplies power to the electromagnetic buffer device according to the current value, so that the electromagnetic buffer device can generate an induced magnetic field. When the third target covered with the metal coil bag falls into the electromagnetic buffer device, an induced current is generated. The third target is subjected to magnetic force in the magnetic field generated by the electromagnetic buffer device, which reduces the falling speed of the third target and buffers the third target. This achieves contactless buffer collection and effectively reduces physical damage to the falling object.

[0162] Specifically, the program instructions corresponding to a collection method in this embodiment can be stored on storage media such as optical discs, hard disks, and USB flash drives; when the program instructions corresponding to a collection method in the storage media are read or executed by an electronic device, the following steps are included:

[0163] Image information of a first target and a second target is acquired; wherein the first target includes the second target; the second target is wrapped in a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils;

[0164] The current value of the electromagnetic buffer device is determined based on the image information of the first target and the second target;

[0165] The electromagnetic buffer device is deployed and powered according to the current value to collect and process the falling third target in the second target; wherein the electromagnetic buffer device includes an deployed state and a folded state.

[0166] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0167] This application is described with reference to schematic and / or block diagrams of implementations of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block of the schematic and / or block diagrams can be implemented by computer program instructions, and combinations of blocks in the schematic and / or block diagrams can be implemented. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the schematic and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0168] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in the implementation flow diagram. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0169] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0170] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. A collection device, characterized in that, The collection device includes a camera, a computing unit, an electromagnetic buffer device, and a current modulator; wherein, The camera is used to acquire image information of a first target and a second target; wherein the first target includes the second target; the second target is wrapped with a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils; The computing unit is used to determine the current value of the electromagnetic buffer device based on the image information of the first target and the second target. The current modulator is used to supply power to the electromagnetic buffer device according to the current value; The electromagnetic buffer device is used to collect and process the falling third target from the second target when it is in the deployed state; wherein, the electromagnetic buffer device includes an deployed state and a folded state; The computing unit is also specifically used to calculate the initial velocity estimate of the second target based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, and to determine the mass of the second target based on the size information of the second target and the preset density of the second target; The electromagnetic buffer device consists of at least one electromagnetic ring, each of which is composed of a wound coil; the at least one electromagnetic ring is arranged sequentially in a preset direction, and after the electromagnetic buffer device is powered, the magnetic field strength corresponding to the at least one electromagnetic ring gradually increases along the preset direction; wherein, the preset direction is a direction perpendicular to the ground.

2. The collecting device according to claim 1, characterized in that, The computing unit is specifically used to perform target detection processing on the image information of the first target and the second target according to the target detection algorithm, so as to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target. Based on the trained Bayesian neural network model, the height information of the first target is obtained from the at least one first image, and the size information of the second target is obtained from the at least one second image; The current value is determined based on the height information of the first target and the size information of the second target.

3. The collecting device according to claim 1, characterized in that, The computing unit is also specifically used to determine the current value based on a preset dynamic model, preset boundary conditions, the initial velocity estimate, and the mass of the second target.

4. The collecting device according to claim 1, characterized in that, The collection device also includes a gimbal; wherein... The gimbal is used to adjust the position and angle of the electromagnetic buffer device.

5. The collecting device according to claim 2, characterized in that, The computing unit is further specifically configured to input each image in the at least one first image into the trained Bayesian neural network model according to a first preset number of times, to obtain multiple first target height estimates equal to the first preset number of times, and to calculate the average of the multiple first target height estimates to obtain the height information of the first target; wherein, the first preset number of times is an integer greater than or equal to 2; Each image in at least one second image is input into the trained Bayesian neural network model according to a second preset number of times to obtain multiple second target size estimates equal to the second preset number of times, and the average of the multiple second target size estimates is calculated to obtain the size information of the second target; wherein, the second preset number of times is an integer greater than or equal to 2.

6. A collection method, characterized in that, The method is applied to the collection device as described in any one of claims 1 to 5; the method includes: Image information of a first target and a second target is acquired; wherein the first target includes the second target; the second target is wrapped in a metal coil bag; the metal coil bag is composed of three sets of mutually perpendicular metal coils; The current value of the electromagnetic buffer device is determined based on the image information of the first target and the second target; The electromagnetic buffer device is deployed and powered according to the current value to collect and process the falling third target in the second target; wherein the electromagnetic buffer device includes an deployed state and a folded state.

7. The method according to claim 6, characterized in that, Determining the current value of the electromagnetic buffer device based on the image information of the first target and the second target includes: The image information of the first target and the second target are processed by the target detection algorithm to obtain at least one first image corresponding to the first target and at least one second image corresponding to the second target. Based on the trained Bayesian neural network model, the height information of the first target is obtained from the at least one first image, and the size information of the second target is obtained from the at least one second image; The current value is determined based on the height information of the first target and the size information of the second target.

8. The method according to claim 7, characterized in that, Determining the current value based on the height information of the first target and the size information of the second target includes: Based on the height information of the first target and the distance between the vertex of the electromagnetic buffer device and the ground, the initial velocity estimate of the second target is calculated, and the mass of the second target is determined based on the size information of the second target and the preset density of the second target. The current value is determined based on the preset dynamic model, preset boundary conditions, the initial velocity estimate, and the mass of the second target.

9. The method according to claim 8, characterized in that, The electromagnetic buffer device consists of at least one electromagnetic ring, each of which is composed of a wound coil; the at least one electromagnetic ring is arranged sequentially in a preset direction, and after the electromagnetic buffer device is powered, the magnetic field strength corresponding to the at least one electromagnetic ring gradually increases along the preset direction; wherein, the preset direction is a direction perpendicular to the ground.

10. The method according to claim 6, characterized in that, Before deploying the electromagnetic buffer device and supplying power to the electromagnetic buffer device according to the current value to collect and process the falling third target from the second target, the method includes: The position and angle of the electromagnetic buffer device are adjusted.

11. The method according to claim 7, characterized in that, The trained Bayesian neural network model obtains the height information of the first target from at least one first image and the size information of the second target from at least one second image, including: Each of the at least one first image is input into the trained Bayesian neural network model a first preset number of times to obtain multiple first target height estimates equal to the first preset number of times, and the average of the multiple first target height estimates is calculated to obtain the height information of the first target; wherein, the first preset number of times is an integer greater than or equal to 2; Each image in at least one second image is input into the trained Bayesian neural network model according to a second preset number of times to obtain multiple second target size estimates equal to the second preset number of times, and the average of the multiple second target size estimates is calculated to obtain the size information of the second target; wherein, the second preset number of times is an integer greater than or equal to 2.

12. A computer-readable storage medium having a program stored thereon, applied in a collection device, wherein the program, when executed by a processor, implements the method as described in any one of claims 6-11.

Images