Spatial separation belt differential control algorithm based on 3D package classification

Abstract

The invention provides a spatial separation belt differential control algorithm based on 3D package classification, and belongs to the technical field of logistics, and the algorithm comprises the following steps: 1, erecting four 3D cameras above a conveying belt; 2, obtaining and aligning an RGB image, a depth image and point cloud data of the package; 3, processing the aligned data, fusing the data of different 3D cameras, and unifying coordinate axes, thereby obtaining the coordinates and types of each parcel; 4, processing is started from the last section of belt, and a corresponding execution task is distributed to the last section of belt; 5, sequentially judging each belt forwards to distribute and execute tasks; 6, assigning tasks to the first section of belt; 7, calculating the speed value of the corresponding belt according to a formula, and sending the speed value to the PLC for execution; the 3D camera is adopted to identify the parcels, the accuracy is improved, the overlapped parcels are classified to adopt corresponding separation tasks, and the separation success rate is increased.

Description

Technical Field

[0001] The present invention belongs to the field of logistics technology, and more specifically, particularly relates to a spatial separation belt differential speed control algorithm based on 3D package classification. Background Art

[0002] In modern logistics systems, single-piece separation technology plays a vital role, especially in improving the efficiency and accuracy of package handling. At the single-piece separation level, spatial separation is an important part of logistics automation, which can separate stacked or mixed packages one by one, facilitating subsequent scanning, weighing and sorting operations.

[0003] In recent years, with the advancement of technology, the spatial separation technology in single-piece separation technology has gradually transformed from photoelectric solutions to visual solutions; this transformation is mainly because the visual solution can more accurately detect and separate each package, especially when dealing with flat packages that are difficult to detect with photoelectric solutions. The visual solution shows a better separation effect; however, the existing separation technology still has the problem of separation failure or restacking when facing some packages that are difficult to separate or easy to roll; most of the existing methods are based on planar vision of RGB images or 3D vision of RGBD, and use the same separation strategy for all packages, which makes some packages that are difficult to separate or easy to roll fail to separate or stack again.

[0004] The present invention is based on 3D package classification technology. On the basis of obtaining the package location, the type information of the object is obtained and analyzed, so that different separation strategies can be adopted for different packages, and these separation strategies are realized through a belt differential control algorithm. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides parcel classification to solve the technical problem in the prior art that the spatial separation strategy has different effects on different parcels during logistics sorting, resulting in affected efficiency.

[0006] The purpose and effect of the package classification of the present invention are achieved by the following specific technical means:

[0007] A spatial separation belt differential speed control algorithm based on 3D package classification includes the following steps:

[0008] Step 1: Set up four 3D cameras directly above the conveyor belts at the logistics production site;

[0009] Step 2: Obtain the RGB image, depth map and point cloud data of the package and align them through the image acquisition and coordinate transformation program;

[0010] Step 3: Process the aligned RGB image, depth image and point cloud data, and fuse the data from different 3D cameras and unify them on the same coordinate axis to obtain the coordinates and types of each package;

[0011] Step 4: The belt execution tasks are divided into four categories, namely separation tasks, acceleration tasks, deceleration tasks and stop tasks. The process starts from the last section of the belt and the corresponding execution tasks are assigned to it;

[0012] Step 5: After the last section of the belt is judged, judge forward in sequence. If there are overlapping packages at the belt interface, perform the separation task, which is divided into many small categories according to the types of overlapping packages; otherwise, perform other tasks;

[0013] Step 6: Determine and assign tasks in sequence until the first section of the belt. The first section of the belt does not perform separation tasks, but is responsible for caching packages and issuing run or stop instructions to the loop according to the set value;

[0014] Step 7: After all belt tasks are assigned, the speed value of the corresponding belt is calculated according to the formula and sent to the PLC for execution.

[0015] In a preferred embodiment, in step three, there are four types of packages, namely: envelope, carton, soft bag and mail bag.

[0016] In a preferred embodiment, in step five, the separation task is divided into a number of subcategories according to the types of overlapping packages, including:

[0017] If the overlapping package contains the envelope, it is separation task 1;

[0018] If the overlapping package contains the soft package, it is separation task 2;

[0019] If the overlapping packages are cartons and cartons, cartons and mail bags, or mail bags and mail bags, it is separation task 3.

[0020] In a preferred embodiment, the otherwise performing other tasks includes:

[0021] When there are large gaps between packages, an acceleration task is performed to arrange the packages closely but without overlapping;

[0022] When the package on the rear belt section is too close to the package on the front belt section, the deceleration task is executed, and if there is a risk of overlap, the process stops.

[0023] In a preferred embodiment, in step 4, the assigning corresponding execution tasks to it includes:

[0024] The last section of the belt is allowed to enter when plane separation is being performed and runs at full speed, otherwise the separation task will not be performed;

[0025] If access is denied, the acceleration task will be executed, the package closest to the interface will be delivered to the interface location and then stop.

[0026] In a preferred embodiment, in step seven, the speed value of the corresponding belt is calculated according to the formula:

[0027] V=K 1 ·F+K 2 ·D+K 3 ·T

[0028] Wherein, V represents the speed value of the belt, F represents the influencing factor corresponding to the separation task type of the current package, and different values ​​are assigned to separation task 1, separation task 2, separation task 3 and non-separation task respectively; K 1 represents the weight coefficient related to the separation task type, and D represents the distance between packages. When the distance between packages is large, D = d 1 , when the distance is moderate, D = d 2 , when the distance is too close and there is a risk of overlap, D = d 3 ; K 2 represents the weight coefficient of the package distance, T represents the value corresponding to different package types, K 3 Indicates the weight coefficient of the package type.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1. Use multi-angle 3D cameras to comprehensively scan logistics packages, which can not only obtain the appearance image of the package, but also obtain the three-dimensional spatial information of the package. Compared with traditional two-dimensional image recognition and traditional 2D models, 3D models can accurately identify packages on the belt, and will not misidentify even in the face of some debris and wear; this enables the system to more accurately obtain the structure and characteristics of overlapping packages and classify them. Whether it is an envelope, carton, soft package or mail bag, it can be identified and classified; this allows the system to be widely used in different types of environments, adapt to various complex environments, and provide reliable technical support for logistics sorting.

[0031] 2. Thanks to the high-accuracy information obtained from the 3D model, this algorithm can achieve more precise control; different strategies can be adopted according to different package types, such as performing separation task 1 for overlapping packages containing envelopes, performing separation task 2 for packages containing soft packages, and performing separation task 3 for packages between cartons, cartons and mail bags, or mail bags and mail bags. This enhances the success rate of separation and solves the problem of separation failure or restacking when traditional spatial separation strategies are used to deal with packages that are difficult to separate or easy to roll, thereby reducing the time and cost of repeated operations due to separation failures. At the same time, in the case of large gaps between packages, acceleration tasks are taken to shorten the spacing; conversely, if the packages are too close and there is a risk of overlapping, deceleration tasks are taken to maintain an appropriate gap. This differentiated separation and speed control strategy ensures that the packages can be handled smoothly and improves the overall efficiency of logistics sorting.

[0032] 3. While improving the separation success rate, this algorithm can also protect the motor equipment from wear caused by frequent starts and stops. The traditional separation strategy adopts the same separation strategy for different packages, which easily leads to frequent starts and stops of the motor. However, this algorithm can allocate the speed and task of the belt according to the specific situation of the package, avoiding unnecessary frequent starts and stops, and improving the service life and safety of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a step flow chart of a spatial separation belt differential speed control algorithm based on 3D package classification of the present invention. DETAILED DESCRIPTION

[0034] In order to further understand the present invention, preferred embodiments of the present invention are described below in conjunction with the accompanying drawings and examples. However, it should be understood that these descriptions are only for further illustrating the features and advantages of the present invention, rather than limiting the claims of the present invention.

[0035] Example:

[0036] Please refer to Figure 1 As shown, the present invention provides a spatial separation belt differential speed control algorithm based on 3D package classification, comprising the following steps:

[0037] Step 1: In the logistics sorting system, obtaining the three-dimensional spatial information of the package is the key to achieving refined control. For this purpose, four 3D cameras are set up directly above the conveyor belt at the logistics production site for scanning; these four 3D cameras are arranged at different angles, which can perform stereoscopic photography and imaging of the packages on the conveyor belt. Through multi-angle capture, a complete three-dimensional model can be generated, and the appearance image of the package can be obtained, which has higher accuracy than traditional single-view two-dimensional image recognition;

[0038] Step 2: Based on the stereo data collected by four high-performance 3D cameras, the RGB image, depth map and point cloud data of the package are obtained and aligned through image acquisition and coordinate transformation programs; First, the RGB camera is used to capture the color image of the package's appearance, which provides basic data for subsequent appearance recognition and category judgment. At the same time, the matching depth camera measures the three-dimensional distance information of the package surface and generates a depth map; and by performing point cloud processing on these depth data, the complete three-dimensional geometric model of the package is further obtained; in some complex environments, such as when there are debris or wear on the package surface, the 3D camera can also accurately identify without misjudgment, providing reliable technical support for the formulation and implementation of subsequent differentiated separation control strategies;

[0039] Step 3: After acquiring the 3D visual data of each package, the algorithm further processes and fuses this information. First, the image processing unit is used to align and correct the RGB images, depth images, and point cloud data collected by the four 3D cameras so that they are unified in the same 3D coordinate system to ensure data consistency. Then, by analyzing and processing these 3D visual data, the coordinate position and type of each package are obtained. At the same time, the image recognition technology based on machine learning can also identify the type of package. According to the needs of actual applications, the algorithm divides the package types into four categories, namely envelopes, cartons, soft bags, and mail bags, which provides a basis for the subsequent differentiated separation control strategy.

[0040] Step 4: Based on the parcel location and type information obtained in the previous steps, a differentiated control strategy is then executed for the parcel on the last section of the conveyor belt. Specifically, the task execution of this belt section mainly includes four modes: separation task, acceleration task, deceleration task and stop task. When processing the parcel on the last section of the belt, the system will first determine whether the parcel meets the separation access conditions, that is, plane separation. If it does, the belt will run at the highest speed to ensure that the parcel can be smoothly separated to the designated location. However, if the parcel does not meet the separation access standards, it will immediately switch to the acceleration task, accelerate the parcel closest to the interface to the interface position, and then stop the belt immediately. This targeted acceleration and stopping action can effectively avoid package dislocation or collision.

[0041] Step 5: After differential control of the packages on the last section of the conveyor belt, the status of the packages on the previous sections of the belt will be judged and processed in turn; first, it will detect whether there is package overlap at each belt interface. If overlap is found, separation tasks will be taken for processing; and for different types of overlapping packages, corresponding separation strategies will be adopted; otherwise, other tasks will be executed;

[0042] Step 6: Check the parcels on the previous belt sections in turn, and assign corresponding tasks according to actual needs until the first belt section. For the first belt section near the conveyor entrance, its main task is not to directly separate the parcels, but to cache and adjust the parcels that are about to enter the system; according to the preset parameters, the operation or stop of the belt section will be controlled in time to ensure that the parcels can enter the subsequent sorting links in an orderly manner; this front-end buffering and adjustment can effectively avoid the accumulation and blockage of parcels between belts, making the operation of the entire sorting line smoother;

[0043] Step 7: After completing the task allocation section by section, calculate the speed value of the corresponding belt according to the formula based on the actual work requirements of each belt section. Finally, these speed instructions will be sent to the on-site PLC controller for execution to achieve belt control.

[0044] While ensuring the success rate of separation, the algorithm can also protect the motor equipment from wear caused by frequent starting and stopping. Traditional sorting strategies usually take a unified separation action for different types of packages, which easily leads to frequent starting and stopping of motor equipment, thereby accelerating its wear and aging. This algorithm uses the acquired package information to develop differentiated control schemes based on the specific characteristics of different packages, avoiding unnecessary frequent starting and stopping, and creating a safer and more stable environment for the entire sorting operation.

[0045] In the specific separation task allocation process, the algorithm sets up multiple separation strategies for the characteristics of different types of packages; for those overlapping packages containing envelopes, the algorithm will classify them as "separation task 1", and the separation action of this task is the strongest; and for those overlapping packages containing soft packages, the algorithm will classify them as "separation task 2", and this separation task is the second strongest; in addition, for some overlapping packages of cartons and cartons, cartons and mail bags, or mail bags and mail bags, the algorithm will classify them as "separation task 3", and this separation action is the weakest, in order to deal with problems such as rolling that may occur during the separation process; this approach of classifying according to package characteristics and adopting differentiated separation strategies for different situations enhances the success rate of the entire separation process; compared with the traditional simple spatial separation method, this algorithm can more flexibly deal with various complex package combinations, effectively avoid repeated operations caused by separation failures, and thus save time costs; at the same time, its targeted separation method also helps to reduce damage or overturning of packages during the separation process, further improving the stability of the entire sorting system.

[0046] In addition to the differentiated processing for overlapping packages, in step five, other tasks are performed, including: when a large gap is detected between packages, an acceleration task is performed to allow the packages to be arranged more compactly, thereby reducing unnecessary waste of space; on the contrary, when the package on the rear belt is too close to the package on the front belt, a deceleration task is performed, and an emergency stop is made when necessary to avoid the risk of overlap or collision; this differentiated control strategy that autonomously adjusts the belt speed according to actual conditions not only ensures the smooth progress of the entire sorting operation, but also improves the overall efficiency of logistics sorting; on the one hand, it avoids the waste of space caused by excessive gaps between packages; on the other hand, it promptly prevents possible overlap or collision accidents and ensures the safety of the operation; at the same time, this dynamic regulation not only improves the sorting efficiency, but also reduces the wear of related equipment, and to a certain extent extends their service life.

[0047] In step seven, the speed value of the corresponding belt is calculated according to the formula:

[0048] V=K 1 ·F+K 2 ·D+K 3 ·T

[0049] Where V represents the speed of the belt, which is determined by multiple factors; F represents the influencing factor corresponding to the separation task type of the current package, and different values ​​are assigned to separation task 1, separation task 2, separation task 3 and non-separation task. Different separation task types reflect different states and requirements of packages. For example, separation task 1 corresponds to overlapping packages containing envelopes, which requires a larger speed adjustment to achieve effective separation, while separation task 3 corresponds to overlapping packages of cartons and cartons, cartons and mail bags, or mail bags and mail bags, and the speed adjustment requirement is relatively weak; K 1 represents the weight coefficient related to the separation task type, and D represents the distance between packages. When the distance between packages is large, D = d 1 , the belt running speed needs to be increased to reduce the distance between packages and improve sorting efficiency; when the distance is moderate, D = d 2 , the belt speed can remain relatively stable; when the distance is too close and there is a risk of overlap, D = d 3 , the belt speed needs to be slowed down or even stopped to avoid package overlap. This dynamic adjustment not only improves the operating efficiency, but also ensures the stability of the entire sorting process; K 2 represents the weight coefficient of the package distance, T represents the value corresponding to different package types, K 3 Indicates the weight coefficient of the package type.

[0050] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A spatial separation belt differential speed control algorithm based on 3D package classification, characterized by: The following steps are included: Step 1: Set up four 3D cameras directly above the conveyor belts at the logistics production site; Step 2: Obtain the RGB image, depth map and point cloud data of the package and align them through the image acquisition and coordinate transformation program; Step 3: Process the aligned RGB image, depth image and point cloud data, and fuse the data from different 3D cameras on the same coordinate axis to obtain the coordinates and types of each package; Step 4: The belt execution tasks are divided into four categories, namely separation tasks, acceleration tasks, deceleration tasks and stop tasks. The process starts from the last section of the belt and the corresponding execution tasks are assigned to it; Step 5: After the last section of the belt is judged, judge forward in sequence. If there are overlapping packages at the belt interface, perform the separation task, which is divided into many small categories according to the types of overlapping packages; otherwise, perform other tasks; Step 6: Determine and assign tasks in sequence until the first section of the belt. The first section of the belt does not perform separation tasks, but is responsible for caching packages and issuing run or stop instructions to the loop according to the set value; Step 7: After all belt tasks are assigned, the speed value of the corresponding belt is calculated according to the formula and sent to the PLC for execution.

2. The spatial separation belt differential speed control algorithm based on 3D package classification according to claim 1 is characterized by: In step three, there are four types of packages, namely: envelope, carton, soft bag and mail bag.

3. The spatial separation belt differential speed control algorithm based on 3D package classification according to claim 2 is characterized by: In step 5, the separation task is divided into a number of subcategories according to the types of overlapping packages, including: If the overlapping package contains the envelope, it is separation task 1; If the overlapping package contains the soft package, it is separation task 2; If the overlapping packages are cartons and cartons, cartons and mail bags, or mail bags and mail bags, it is separation task 3.

4. The spatial separation belt differential speed control algorithm based on 3D package classification according to claim 3 is characterized by: Otherwise performing other tasks includes: When there are large gaps between packages, an acceleration task is performed to arrange the packages closely but without overlapping; When the package on the rear belt section is too close to the package on the front belt section, the deceleration task is executed, and if there is a risk of overlap, the process stops.

5. The spatial separation belt differential speed control algorithm based on 3D package classification according to claim 1 is characterized by: In step 4, the corresponding execution tasks are assigned to it, including: The last section of the belt is allowed to enter when plane separation is being performed and runs at full speed, otherwise the separation task will not be performed; If access is denied, the acceleration task will be executed, the package closest to the interface will be delivered to the interface location and then stop.

6. The spatial separation belt differential speed control algorithm based on 3D package classification according to claim 1 is characterized by: In step seven, the speed value of the corresponding belt is calculated according to the formula: V=K1·F+K2·D+K3·T Wherein, V represents the speed value of the belt, F represents the influencing factor corresponding to the separation task type of the current package, and different values ​​are assigned to separation task 1, separation task 2, separation task 3 and non-separation task respectively; K1 represents the weight coefficient related to the separation task type, and D represents the distance between packages. When the distance between packages is large, D = d1, when the distance is moderate, D = d2, and when the distance is too close and there is a risk of overlap, D = d3; K2 represents the weight coefficient of the package distance, T represents the value corresponding to different package types, and K3 represents the weight coefficient of the package type.

Images