Automatic dispatching and efficient sorting system for silicon wafer offcut based on AGV

Abstract

The application provides a kind of based on AGV's crystalline silicon edge material automatic scheduling and efficient sorting system, through multi-sensor data fusion, dynamic path planning and intelligent sorting optimization, significantly improve the automation level, efficiency and resource utilization of edge material processing.Laser scanning, spectral analysis and weight sensor are used to obtain the size, material and weight information of the edge material, forming a comprehensive attribute information;The path planning algorithm generates a dynamic scheduling path, and adjusts the optimal path in real time according to the obstacle position and transportation priority to ensure efficient transportation;Through accurate allocation of edge material to sorting area and monitoring of abnormal data by sensor, the sorting path is dynamically adjusted;Through space optimization algorithm and mechanical arm motion trajectory planning, efficient stacking of edge material in storage area is realized.

Description

Technical Field

[0001] This invention relates to the field of crystalline silicon edge material sorting technology, and in particular to an AGV-based automatic scheduling and efficient sorting system for crystalline silicon edge material. Background Technology

[0002] Currently, a key technical challenge in automated scrap material sorting systems is achieving efficient and accurate sorting and storage of scrap materials in complex and ever-changing factory environments. The system needs to handle scrap materials of various types, sizes, and weights simultaneously, while obstacle distribution and storage area occupancy are constantly changing within the factory environment. A critical challenge lies in how to integrate multi-source data such as laser scanning, spectral analysis, and weight sensing in real time, and dynamically generate optimal scheduling paths based on this data. Simultaneously, the system needs to consider the position, speed, and turning angle of the automated guided vehicle (AGV), as well as anomalies in the sorting area. Furthermore, efficient utilization of storage space is also a significant challenge, requiring the orderly stacking of scrap materials within a limited space. More specifically, a major technical challenge in this field is how to organically combine technologies such as real-time environmental perception, multi-dimensional attribute analysis, dynamic path planning, machine learning optimization, and precise robotic arm control to build an intelligent sorting and storage system capable of adapting to complex working conditions and possessing self-optimization capabilities. Summary of the Invention

[0003] This invention provides an AGV-based automatic scheduling and efficient sorting system for crystalline silicon edge scraps. It is designed for business scenarios involving the efficient sorting, transportation, and storage of crystalline silicon edge scraps in complex environments. The system is capable of adapting to complex working conditions and possesses self-optimization capabilities for intelligent sorting and storage of crystalline silicon edge scraps. Through multi-sensor data fusion, dynamic path planning, and intelligent sorting optimization, it significantly improves the automation level, efficiency, and resource utilization of edge scrap processing.

[0004] This invention provides an AGV-based automatic scheduling and high-efficiency sorting system for crystalline silicon edge material, comprising:

[0005] Obtain the physical information of the edge material, and obtain the attribute information of the edge material based on the physical information;

[0006] The system acquires the current location and environmental data of the automated guided vehicle (AGV), and generates a dynamic scheduling path by combining the edge material attribute information; it identifies the location and type of obstacles in the environment, and calculates the transportation priority of the edge material by combining the attribute information and generates the optimal path.

[0007] The driving speed and steering angle of the automated guided vehicle are determined based on the dynamic scheduling path, and the path is adjusted based on real-time sensor data to obtain the updated dynamic scheduling path.

[0008] The classification priority of the edge material is obtained based on the attribute information. After the classification priority is optimized, it is assigned to the corresponding sorting area. The sensor data of the sorting area is judged for anomalies. If there are anomalies, the optimized classification priority is adjusted and the sorting path of the edge material is updated.

[0009] Obtain the remaining space data of the storage area, determine the target storage location based on the attribute information and the remaining space data, plan the movement trajectory of the robotic arm, stack the edge materials, and update the occupancy status of the storage area.

[0010] Preferably, the length, width, and height data of the edge material are obtained to determine the size information of the edge material; the spectral reflectance curve of the edge material is obtained based on a spectral analyzer; and the material type of the edge material is determined based on a preset material spectral library.

[0011] Obtain the weight of the edge material and calculate its volume based on the size information. The volume is the product of length, width, and height. Calculate the density based on the volume and weight, and integrate the size information, material type, weight, and density to obtain the complete attribute information of the edge material.

[0012] Preferably, the current position and environmental data of the automated guided vehicle are obtained, including obstacle positions and storage area occupancy status; vehicle speed attributes and vehicle load attributes are extracted from the current position to obtain preliminary path constraints;

[0013] Based on preliminary path constraints and environmental data, it is determined whether the location of obstacles overlaps with the occupancy status of storage areas; if they overlap, the path nodes are adjusted and an obstacle avoidance and detour scheme is determined; based on the obstacle avoidance and detour scheme and vehicle load attributes, a dynamic scheduling path is generated, which is used for real-time navigation to the free storage area.

[0014] Preferably, obstacle features are obtained from environmental data, and the location and type of the obstacle are determined by an image segmentation algorithm; wherein, the image segmentation algorithm is a semantic segmentation process based on a convolutional neural network, which takes obstacle features as input and outputs the segmented location coordinates and type labels;

[0015] The transportation priority of the edge material is calculated using a weighted summation algorithm to obtain the priority value. The weighted summation algorithm is to assign weights to the weight factor and urgency factor in the attribute information and then sum them to obtain the priority value.

[0016] The optimal path is generated based on obstacle location and transportation priority. The optimal path is calculated as follows: the obstacle location is input as an avoidance constraint and the priority value is used as the path cost weight. The output is the optimal path from the starting point to the destination.

[0017] Preferably, obstacle location and traffic congestion monitoring information are extracted, and LiDAR and camera data are integrated using a weighted average method to determine path deviation correction requirements;

[0018] Based on the path deviation correction requirements, calculate the path energy consumption based on the battery consumption rate, check the distance threshold, calculate the speed parameters and adjust the steering angle.

[0019] Based on the determined speed parameters and the adjusted steering angle, the priority is adjusted according to the real-time delay, and the dynamic path generation is adjusted to obtain the updated dynamic scheduling path.

[0020] Preferably, a set of rules for classifying defects based on size thresholds and density ranges is used to obtain a preliminary classification priority; the defect types of the edge material based on size and material density are then obtained.

[0021] Construct a random forest model and train it on historical sorting data; specifically, integrate multiple decision tree classification models, input initial classification priority and color uniformity, and output adjusted priority levels;

[0022] The initial classification priority and color uniformity analysis were used to determine the thickness deviation and the adjusted priority level.

[0023] Obtain the correlation between the adjusted priority level and the hardness index, determine the optimization parameters based on the recycling value and sorting cost. If the recycling value exceeds the sorting cost, increase the optimization parameters to obtain the optimized classification priority.

[0024] Preferably, the edge material is allocated to the sorting area by a preset classification priority, the initial occupancy data of the sorting area is obtained, and the preliminary location of the edge material is determined; based on the preliminary location, the real-time signal of the edge material is collected by the sensor, the feature value is extracted from the real-time signal, and the arrival status confirmation value is obtained; if the confirmation value exceeds the preset threshold, the allocation parameters of the sorting area are adjusted to obtain the final arrival status of the edge material.

[0025] Preferably, the sorting area is monitored in real time to determine whether there are any abnormal detection situations. If there are abnormalities, the classification priority is adjusted and optimized according to the abnormal detection situation, and the adjusted priority is determined. The sorting path of the edge material is updated using the adjusted priority to obtain the updated sorting path.

[0026] Preferably, the remaining space data of the storage area is obtained, the available volume index is extracted from the remaining space data, and attribute information, including size and shape characteristics, is obtained for the edge material. Based on the available volume index and attribute information, a genetic algorithm is used to determine the target storage location.

[0027] The genetic algorithm is as follows: using available volume index and attribute information as the initial population, it iteratively optimizes the selection of crossover and mutation operations to obtain the position with the highest space utilization.

[0028] The motion trajectory of the robotic arm is planned based on the target storage location. The robotic arm is driven by the motion trajectory to grab the edge material and perform stacking operation. If the volume change after stacking exceeds the preset threshold, the trajectory parameters are adjusted. The volume change is the difference between the remaining space data before and after stacking, and the stacked edge material arrangement is obtained.

[0029] Update the occupancy status of the storage area by refreshing the remaining space data from the stacked edge material arrangement. The new remaining space data is obtained by subtracting the original remaining space data from the occupied volume after the arrangement, thus realizing dynamic management of the storage area.

[0030] The working principle and beneficial effects of this invention are as follows:

[0031] Obtain the physical information of the edge material, and obtain the attribute information of the edge material based on the physical information;

[0032] The system acquires the current location and environmental data of the automated guided vehicle (AGV), and generates a dynamic scheduling path by combining the edge material attribute information; it identifies the location and type of obstacles in the environment, and calculates the transportation priority of the edge material by combining the attribute information and generates the optimal path.

[0033] The driving speed and steering angle of the automated guided vehicle are determined based on the dynamic scheduling path, and the path is adjusted based on real-time sensor data to obtain the updated dynamic scheduling path.

[0034] The classification priority of the edge material is obtained based on the attribute information. After the classification priority is optimized, it is assigned to the corresponding sorting area. The sensor data of the sorting area is judged for anomalies. If there are anomalies, the optimized classification priority is adjusted and the sorting path of the edge material is updated.

[0035] Obtain the remaining space data of the storage area, determine the target storage location based on the attribute information and the remaining space data, plan the movement trajectory of the robotic arm, stack the edge materials, and update the occupancy status of the storage area.

[0036] In this invention, the size, material, and weight information of the edge material are acquired through laser scanning, spectral analysis, and weight sensors to form comprehensive attribute information. Combined with the environmental data of the automated guided vehicle, a path planning algorithm is used to generate a dynamic scheduling path, and the optimal path is adjusted in real time according to the location of obstacles and transportation priorities to ensure efficient transportation. Next, the classification priority is optimized based on a preset sorting algorithm and machine learning model to allocate the edge material to the sorting area, and the data anomaly is monitored by sensors to further realize the dynamic adjustment of the sorting path. Through spatial optimization algorithms and robotic arm motion trajectory planning, the edge material is efficiently stacked in the storage area.

[0037] This invention significantly improves the automated production efficiency and resource utilization of edge material processing by using multi-sensor data fusion, dynamic path planning, and intelligent sorting optimization in the automatic scheduling and efficient sorting of edge materials.

[0038] This invention is based on a series of technical means such as efficient sorting, automated transportation, and intelligent storage. It can achieve efficient sorting, automated transportation, and intelligent storage of crystalline silicon edge materials. It also has the ability to self-adapt and optimize according to the complex working conditions, which makes it highly flexible and reliable. It has great application value for the current business scenarios of efficient sorting, transportation and storage of crystalline silicon edge materials.

[0039] At the same time, it transforms the uncertainties brought about by the complex working conditions into the driving force for its self-optimization, enabling it to better adapt to the current complex working conditions and maximize its working efficiency and reliability.

[0040] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0041] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0042] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0043] Figure 1 This is a schematic diagram of the structure of the present invention; Detailed Implementation

[0044] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0045] First embodiment:

[0046] according to Figure 1 As shown, this embodiment of the invention provides an automatic scheduling and efficient sorting system for crystalline silicon edge material based on AGV, which includes:

[0047] Obtain the physical information of the edge material, and obtain the attribute information of the edge material based on the physical information;

[0048] The system acquires the current location and environmental data of the automated guided vehicle (AGV), and generates a dynamic scheduling path by combining the edge material attribute information; it identifies the location and type of obstacles in the environment, and calculates the transportation priority of the edge material by combining the attribute information and generates the optimal path.

[0049] The driving speed and steering angle of the automated guided vehicle are determined based on the dynamic scheduling path, and the path is adjusted based on real-time sensor data to obtain the updated dynamic scheduling path.

[0050] The classification priority of the edge material is obtained based on the attribute information. After the classification priority is optimized, it is assigned to the corresponding sorting area. The sensor data of the sorting area is judged for anomalies. If there are anomalies, the optimized classification priority is adjusted and the sorting path of the edge material is updated.

[0051] Obtain the remaining space data of the storage area, determine the target storage location based on the attribute information and the remaining space data, plan the movement trajectory of the robotic arm, stack the edge materials, and update the occupancy status of the storage area.

[0052] More specifically, the physical information of the edge material is obtained, and the attribute information of the edge material is obtained based on the physical information;

[0053] The system acquires the current position and environmental data of the automated guided vehicle (AGV). Based on the attribute information of the edge material and the environmental data, a dynamic scheduling path is generated. The location and type of obstacles contained in the environment are determined according to the environmental data, and the transportation priority of the edge material is calculated by combining the attribute information. An optimal path is generated according to the obstacle location and transportation priority. The driving speed and steering angle of the AGV are determined according to the dynamic scheduling path, and the optimal path is adjusted through real-time sensor data to obtain an updated dynamic scheduling path.

[0054] The classification priority of edge leather is obtained based on attribute information, and the classification priority is optimized by a machine learning model trained based on historical sorting data, resulting in the optimized classification priority.

[0055] Based on the optimized classification priority, the edge leather is assigned to the corresponding sorting area, and sensor data of the sorting area is obtained;

[0056] The sensor data is used to determine whether there are any abnormalities in the sorting area. If there are any abnormalities, the optimized classification priority is adjusted. Based on the adjusted classification priority, the sorting path of the edge material is updated.

[0057] Obtain the remaining space data of the storage area, determine the target storage location based on the attribute information and the remaining space data; plan the movement trajectory of the robotic arm based on the target storage location and stack the edge materials, and update the occupancy status of the storage area.

[0058] In this invention, the size, material, and weight information of the edge material are acquired through laser scanning, spectral analysis, and weight sensors to form comprehensive attribute information. Combined with the environmental data of the automated guided vehicle, a path planning algorithm is used to generate a dynamic scheduling path, and the optimal path is adjusted in real time according to the location of obstacles and transportation priorities to ensure efficient transportation. Next, the classification priority is optimized based on a preset sorting algorithm and machine learning model to allocate the edge material to the sorting area, and the data anomaly is monitored by sensors to further realize the dynamic adjustment of the sorting path. Through spatial optimization algorithms and robotic arm motion trajectory planning, the edge material is efficiently stacked in the storage area.

[0059] This invention significantly improves the automated production efficiency and resource utilization of edge material processing by using multi-sensor data fusion, dynamic path planning, and intelligent sorting optimization in the automatic scheduling and efficient sorting of edge materials.

[0060] This invention is based on a series of technical means such as efficient sorting, automated transportation, and intelligent storage. It can achieve efficient sorting, automated transportation, and intelligent storage of crystalline silicon edge materials. It also has the ability to self-adapt and optimize according to the complex working conditions, which makes it highly flexible and reliable. It has great application value for the current business scenarios of efficient sorting, transportation and storage of crystalline silicon edge materials.

[0061] At the same time, it transforms the uncertainties brought about by the complex working conditions into the driving force for its self-optimization, enabling it to better adapt to the current complex working conditions and maximize its working efficiency and reliability.

[0062] Second embodiment;

[0063] Furthermore, based on the first embodiment, preferably, the length, width, and height data of the edge material are obtained to determine the size information of the edge material; the spectral reflectance curve of the edge material is obtained based on a spectral analyzer; and the material type of the edge material is determined based on a preset material spectral library.

[0064] Obtain the weight of the edge material and calculate its volume based on the size information. The volume is the product of length, width, and height. Calculate the density based on the volume and weight, and integrate the size information, material type, weight, and density to obtain the complete attribute information of the edge material.

[0065] In this embodiment, the dimensions of the crystalline silicon edge material are measured using a high-precision laser measuring instrument; the laser measuring instrument calculates the length, width, and height of the edge material by emitting a laser beam and receiving the reflected signal.

[0066] In this embodiment, a spectral reflectance curve is obtained based on a spectral analyzer. By irradiating a broadband light source and recording the intensity of the reflected light, a curve relating wavelength to reflectance is generated. The edge material is scanned using a spectrometer. For example, if the curve shows a reflectance peak of 0.8 at 500 nm, it is a typical characteristic of monocrystalline silicon.

[0067] In this embodiment, the material type is determined based on a preset material spectral library. The material type is determined by comparing the spectral curve of the edge material with the standard curve in the library. Specifically, monocrystalline silicon and polycrystalline silicon show significant differences in reflectivity in the near-infrared band. For example, the spectral library contains standard curves for monocrystalline silicon, polycrystalline silicon, and amorphous silicon; after comparison, the edge material is confirmed to be monocrystalline silicon. This material determination method can quickly distinguish material types, avoid subjective errors from manual judgment, and help optimize the material recycling process.

[0068] In this embodiment, the volume of the edge material is calculated based on dimensional data, for example, a volume of 0.0001 m³. The density is determined by combining the weight and volume of the edge material; for example, the weight of the edge material is 230 g. The density calculation result obtained by combining the volume calculation result is: 230 ÷ 0.0001 = 2300000 g / m³, or 2.3 g / cm³. This value is close to the density of monocrystalline silicon (2.33 g / cm³), thus verifying the reliability of the material determination. This helps assess material purity and guides subsequent processing or recycling.

[0069] Furthermore, by combining the density calculation results mentioned above, complete attributes such as the size and density of the edge material can be obtained. Precise size and density information can guide the improvement of the cutting process and reduce edge material waste. Spectral analysis and material judgment can ensure the high purity of recycled materials and enhance their reuse value. If the edge material is used for recycled silicon ingots, the above attributes can help determine the smelting parameters, such as the addition ratio and temperature control, thereby improving the quality of silicon ingots.

[0070] Furthermore, by combining spectral and density analysis, high-purity edge materials can be screened out, thereby reducing the impact of impurities on the performance of the final product; ensuring the integrity and consistency of edge materials from measurement to application, and providing efficient technical support for the material management of crystalline silicon edge materials.

[0071] Third embodiment;

[0072] Furthermore, based on the first embodiment, preferably, the current position and environmental data of the automated guided vehicle are obtained, including the location of obstacles and the occupancy status of storage areas; vehicle speed attributes and vehicle load attributes are extracted from the current position to obtain preliminary path constraints;

[0073] Based on preliminary path constraints and environmental data, it is determined whether the location of obstacles overlaps with the occupancy status of storage areas; if they overlap, the path nodes are adjusted and an obstacle avoidance and detour scheme is determined; based on the obstacle avoidance and detour scheme and vehicle load attributes, a dynamic scheduling path is generated, which is used for real-time navigation to the free storage area.

[0074] In this embodiment, when the Automated Guided Vehicle (AGV) needs to obtain real-time location and environmental data in the smart warehousing scenario, the current location can be obtained through LiDAR or inertial navigation system. The accuracy is determined based on actual usage requirements. For example, with centimeter-level accuracy, the vehicle's location in the warehouse coordinates is (5, 10, 0).

[0075] In this embodiment, the environmental data includes: obstacle location and storage area occupancy status; obstacle location is scanned by LiDAR. For example, when the radar identifies a pile of boxes 3 meters ahead, it marks their coordinates as (8, 10, 0); storage area occupancy status is queried through the warehouse management system. For example, by querying the target storage area, the information is that A1 is full and A2 is free.

[0076] Furthermore, in one possible implementation, vehicle speed and load attributes are extracted to form preliminary path constraints. For example, the AGV's current speed is 1 m / s and its load is 200 kg. The speed attribute needs to account for safety issues related to sharp turns or accelerations during path planning, while the load attribute has a certain impact on power consumption and path selection. For instance, when the AGV is heavily loaded, it prioritizes flat paths to reduce energy consumption. The preliminary path constraints require avoiding obstacles and navigating to the idle storage area A2. Therefore, the preliminary path planning is a straight line from (5, 10, 0) to (10, 15, 0).

[0077] Furthermore, it is determined whether the location of the obstacle overlaps with the occupancy status of the storage area. For example, suppose A2 is located at (12, 15, 0), while the obstacle (8, 10, 0) is on the initial path, there is a risk of overlap; by analyzing the lidar point cloud data, it is confirmed that the obstacle is a fixed container, and the path nodes need to be adjusted.

[0078] Furthermore, the obstacle avoidance and detour scheme can choose to bypass the obstacle, generate a new path node (7, 12, 0), and detour to (10, 15, 0); the detour scheme takes into account the vehicle's turning radius and load stability to ensure safe navigation.

[0079] In this embodiment, a dynamic scheduling path is generated based on the obstacle avoidance and detour scheme and the vehicle load attributes. Heavy-duty AGVs need to avoid steep slopes or narrow passages; the system selects wide paths and optimizes speed distribution, such as maintaining 1 m / s on straight sections and reducing to 0.5 m / s at turns. Furthermore, during dynamic path scheduling, environmental data is updated in real time to ensure smooth navigation around temporary obstacles when reaching A2. If temporary objects are added near A2, the system dynamically adjusts to the suboptimal storage area A3 (15, 15, 0), thereby improving navigation efficiency and safety.

[0080] Understandably, dynamic scheduling paths are generated through real-time data fusion; among them, environmental data is continuously updated to enable the planned paths to adapt to dynamic changes in the warehouse in real time, such as temporary obstacles or changes in the status of storage areas.

[0081] Vehicle load attributes are used to optimize route selection, thereby extending equipment life and improving scheduling efficiency; for example, flat paths are prioritized for a 200 kg load to reduce motor wear; AGVs navigate to idle storage areas via dynamic paths, thereby enabling efficient, fast and accurate completion of transportation tasks.

[0082] Fourth embodiment;

[0083] Furthermore, based on the first embodiment, preferably, obstacle features are obtained from environmental data, and the location and type of the obstacle are determined by an image segmentation algorithm; wherein, the image segmentation algorithm is a semantic segmentation process based on a convolutional neural network, which takes obstacle features as input and outputs the segmented location coordinates and type labels;

[0084] The transportation priority of the edge material is calculated using a weighted summation algorithm to obtain the priority value. The weighted summation algorithm is to assign weights to the weight factor and urgency factor in the attribute information and then sum them to obtain the priority value.

[0085] The optimal path is generated based on obstacle location and transportation priority. The optimal path is calculated as follows: the obstacle location is input as an avoidance constraint and the priority value is used as the path cost weight. The output is the optimal path from the starting point to the destination.

[0086] In this embodiment, in a smart warehousing scenario, Automated Guided Vehicles (AGVs) achieve efficient navigation through environmental data and path planning. The following analysis and examples focus on obstacle feature acquisition, image segmentation algorithms, weighted summation algorithms, and optimal path generation, maintaining consistency with previous warehousing practices.

[0087] In this embodiment, obstacle features in the environmental data can be collected by a camera or LiDAR; for example, a camera captures images inside a warehouse, including shelves, pallets, or temporary stacks of goods. In one embodiment, the AGV is equipped with a high-definition camera that captures images of a pile of boxes in front of it, with an image resolution of 1920×1080 pixels.

[0088] In this embodiment, after the image data is transmitted to the processing unit, the obstacle features are analyzed by an image segmentation algorithm. Specifically, based on a convolutional neural network, the image is semantically segmented into a background region and an obstacle region, and the coordinates and type label of the cargo box (transfer box) are output.

[0089] Specifically, the convolutional neural network extracts features through multi-layer convolution and pooling, and combined with a pre-trained model, it distinguishes between the cargo box and the ground and outputs accurate bounding boxes and labels. Among them, the identification of obstacle type helps to determine whether the obstacle is movable. For example, if the cargo box is a fixed obstacle, it needs to be bypassed, while temporary stacked items need to wait for clearing based on the actual production situation.

[0090] In this embodiment, the weighted summation algorithm is used to calculate the transportation priority of the edge material; the transportation priority of the edge material refers to the material that needs to be moved quickly in the warehouse, such as packaging materials or semi-finished products.

[0091] In one embodiment, the AGV receives a task to transport a set of edge materials weighing 150 kg, with a high urgency level. Further, a weighted summation algorithm assigns weights to weight and urgency level, assuming a weight weight of 0.4 and an urgency weight of 0.6; 150 kg corresponds to a score of 60, and high urgency corresponds to a score of 80, with a priority value calculated as 60 × 0.4 + 80 × 0.6 = 72; another set of edge materials weighing 200 kg has a low urgency level, a score of 60, and a priority of 60 × 0.4 + 20 × 0.6 = 36. In this scheme, tasks with higher priority are scheduled first based on the priority level calculation.

[0092] It should be noted that the weighting can be adjusted according to warehouse needs, such as increasing the urgency weight when urgent orders are prioritized; the optimal path is generated by combining the obstacle location and priority. Assuming the AGV's starting point is (2, 3, 0) and the target storage area is (10, 12, 0); the obstacle is located at (5, 6, 0), which is a fixed cargo box; in this case, the path planning uses the obstacle location as an avoidance constraint, and the priority value as the cost weight.

[0093] In one embodiment, the system generates two paths: path A is a straight line from (2, 3, 0) to (10, 12, 0), and path B bypasses an obstacle, passing through node (4, 8, 0) to (10, 12, 0). Path A is shorter but passes through an obstacle, while path B is slightly longer but safer. Considering the high-priority task with a priority value of 72, the system selects path B to ensure both safety and efficiency.

[0094] It should be noted that the path cost weight in this invention can be dynamically adjusted. For example, high-priority tasks are allowed to take slightly longer paths to avoid obstacles, while low-priority tasks prioritize the shortest path.

[0095] In this scheme, obstacle feature extraction, priority calculation and path optimization are used to ensure the efficient operation of AGV in complex warehousing environments; image segmentation is used to improve the accuracy of obstacle recognition, and a weighted summation algorithm is used to balance task urgency and resource allocation; the optimal path is used to generate an automatic scheduling scheme that takes into account both safety and efficiency; thus forming a complete scheduling scheme, thereby improving the flexibility and reliability of the scheduling and sorting process of crystalline silicon edge material.

[0096] Fifth embodiment;

[0097] Furthermore, based on the first embodiment, the obstacle location and traffic congestion monitoring information are extracted, and the LiDAR and camera data are integrated by a weighted average method to determine the path deviation correction requirements;

[0098] Based on the path deviation correction requirements, calculate the path energy consumption based on the battery consumption rate, check the distance threshold, calculate the speed parameters and adjust the steering angle.

[0099] Based on the determined speed parameters and the adjusted steering angle, the priority is adjusted according to the real-time delay, and the dynamic path generation is adjusted to obtain the updated dynamic scheduling path.

[0100] In this embodiment, the warehouse environment is scanned by LiDAR to generate high-precision point cloud data and detect the coordinates of obstacles such as shelves or mobile forklifts, such as (7, 9, 0); images are captured by cameras to identify dynamic congested areas, such as channels where multiple AGVs are gathered.

[0101] In one embodiment, the AGV is equipped with a LiDAR and a 1080p camera. When the radar detects a distance of 5 meters from the shelf ahead, and the camera captures two AGVs queuing in the aisle, congestion is identified. A weighted average method integrates the two types of data, with LiDAR data having a weight of 0.6 and camera data a weight of 0.4, to comprehensively determine the precise location of obstacles and congested areas. Path deviation correction requirements are determined based on the data fusion results; if the AGV detects an actual path deviating from the planned path during its movement, such as a deviation of 0.5 meters, correction is required.

[0102] In one embodiment, the system analyzes the lidar point cloud and camera images to confirm that the deviation is caused by temporary debris, and adjusts the AGV's turning angle by 5 degrees to return to the correct path. It should be noted that deviation correction needs to consider obstacle type; fixed obstacles need to be bypassed, while temporary obstacles may require pausing and waiting for clearing. For example, battery consumption rate calculations are combined with path deviation to determine correction requirements; if the AGV's battery capacity is 1000Wh, normal driving consumes 0.5Wh / meter, and the correction path increases by 0.2Wh / meter.

[0103] In one embodiment, the planned path is 10 meters, the deviation correction increases by 2 meters, and the total energy consumption is 10×0.5+2×0.7=6.4Wh; if the distance threshold is set to 15 meters, the current path meets the requirements; based on this, the speed parameters are further calculated, such as maintaining 2 meters / second to optimize energy consumption, and adjusting the steering angle by 5 degrees to avoid obstacles; furthermore, dynamic path generation is based on the speed parameters and steering angle adjustment values, combined with real-time delay adjustment priority.

[0104] In one embodiment, the AGV transports urgent materials with an initial priority of 80. After detecting a congested area, the delay is expected to increase by 10 seconds, and the priority is dynamically reduced to 70. The system replans the path to avoid the congested area, from (2,3,0) to (10,10,0), and selects the path around node (5,8,0).

[0105] It should be noted that dynamic scheduling balances scheduling efficiency and safety by adjusting priorities and paths in real time; through multi-sensor data fusion, deviation correction, energy consumption optimization and dynamic scheduling, AGVs can efficiently schedule and sort edge materials in complex warehousing environments.

[0106] Sixth embodiment;

[0107] Furthermore, based on the first embodiment, a set of rules for classifying defects according to size thresholds and density ranges is preferred to obtain a preliminary classification priority; the defect types of edge leather based on size and material density are obtained; a random forest model is constructed, and historical sorting data is trained based on this model; specifically, a classification model integrating multiple decision trees is used, with the preliminary classification priority and color uniformity as input, and the adjusted priority level as output.

[0108] The initial classification priority and color uniformity are analyzed for thickness deviation to determine the adjusted priority level. The correlation between the adjusted priority level and hardness index is obtained. The optimization parameters are judged based on the recycling value and sorting cost. If the recycling value exceeds the sorting cost, the optimization parameters are increased to obtain the optimized classification priority.

[0109] In this embodiment, in the sorting scenario of crystalline silicon edge material, the defect classification rules based on size threshold and density range are the key first step; furthermore, crystalline silicon edge material may have defects such as cracks, missing corners or impurities due to the cutting process, and needs to be initially classified according to size and density.

[0110] In one embodiment, the size threshold is set as follows: small defects are defined as those with both length and width less than 5 cm, and large defects are defined as those with both length and width greater than or equal to 5 cm; the density range is defined as 2.3 g / cm³, with values ​​below this value being low-density defects and values ​​above this value being high-density defects.

[0111] In one embodiment, a piece of edge material measuring 4 cm × 3 cm with a density of 2.1 g / cm³ was detected and initially classified as a small, low-density defect, with a low priority; another piece of edge material measuring 6 cm × 5 cm with a density of 2.4 g / cm³ was classified as a large, high-density defect, with a high priority; the classification rules provide basic data for subsequent model training.

[0112] In this embodiment, when constructing the random forest model, multiple decision trees are integrated to improve classification accuracy. The historical sorting data includes features such as size, density, and color uniformity. It is assumed that the historical data contains 1000 records, each labeled with defect type and priority. The random forest model generates 10 decision trees through multiple random samplings, and each tree determines the defect category based on some features. For example, one decision tree may mainly rely on size and density, while another may focus more on color uniformity. After training, the model is input with the initial classification priority and color uniformity data of a piece of edge leather. If the uniformity score is 80%, the adjusted priority is output based on this.

[0113] In one embodiment, if a piece of edge material has a low initial priority but high color uniformity, the model can adjust it to a medium priority to reflect its potential recycling value. For example, thickness deviation analysis can further optimize the priority. In actual production operations, thickness deviation may be caused by uneven cutting, affecting the reuse value of the edge material. For example, if the thickness deviation of an edge material is 0.2 mm, exceeding the standard threshold of 0.1 mm, it indicates that its structural stability is poor. Combining the initial priority and color uniformity, the model may reduce the priority from high to medium to reduce sorting costs.

[0114] In one embodiment, a piece of edge material with a thickness deviation of 0.15 mm and a color uniformity of 75% is detected. After comprehensive judgment by the random forest model, its priority is adjusted to medium to ensure reasonable allocation of sorting resources. For example, the correlation analysis between hardness index and adjusted priority is used to optimize sorting parameters. The hardness index reflects the mechanical strength of the edge material and affects its recycling value. Assuming the hardness index threshold is 7, if the hardness of a certain edge material is 8, the expected recycling value is 1000 yuan, and the sorting cost is 600 yuan, which meets the optimization conditions. Further, by improving and optimizing parameters, such as increasing sorting accuracy or prioritizing the processing of this material, the priority is reduced. Conversely, if the hardness is 6, the recycling value is only 500 yuan, which is lower than the sorting cost of 700 yuan.

[0115] In one embodiment, edge material with a hardness index of 7.5 is selected for priority sorting by the system in conjunction with intermediate priority, thereby optimizing resource utilization. It can be understood that the above method, through size density classification, random forest model, thickness deviation analysis and hardness index correlation, is progressively improved to ensure a balance between sorting efficiency and recycling value. Through the mutual support of each link, a complete defect classification and optimization scheme is formed.

[0116] Seventh embodiment;

[0117] Furthermore, in a preferred embodiment based on the first embodiment, the edge material is allocated to the sorting area by a preset classification priority, the initial occupancy data of the sorting area is obtained, and the preliminary positioning of the edge material is determined; based on the preliminary positioning, the real-time signal of the edge material is collected by a sensor, feature values ​​are extracted from the real-time signal, and the arrival status confirmation value is obtained; if the confirmation value exceeds a preset threshold, the allocation parameters of the sorting area are adjusted to obtain the final arrival status of the edge material.

[0118] In this embodiment, the edge leather is assigned to the sorting area by a preset classification priority, and the sorting area is further divided according to the material characteristics. The classification priority is set based on the physical properties of the edge leather, such as weight or surface smoothness. For example, edge leather weighing less than 100 grams is assigned to the light area, 100 to 200 grams is assigned to the medium area, and more than 200 grams is assigned to the heavy area.

[0119] In one embodiment, a piece of edge material weighing 120 grams with a surface smoothness score of 85% is detected and initially allocated to the medium-sized area, occupying 10% of the area's capacity. The initial occupancy data is statistically analyzed by the area sensors to generate a report containing occupancy rate and material quantity, ensuring reasonable resource allocation in the sorting area. Specifically, after initial positioning, collecting real-time signals of the edge material using sensors is a key step. The sensors may include infrared sensors or ultrasonic sensors to detect information such as the position and movement speed of the edge material.

[0120] For example, an infrared sensor captures the position signal of the edge material on the conveyor belt, generating a real-time data stream containing coordinates and speed; feature values, such as the center point offset of the edge material or the stability of the movement trajectory, are extracted from these signals to obtain a confirmation value of the arrival status; assuming the confirmation value is set based on the offset, with a threshold of 2 cm; if the offset of an edge material is 1.5 cm, and the confirmation value does not exceed the threshold, it indicates that its arrival status is stable and it maintains its original assigned area.

[0121] Understandably, if the confirmed value exceeds the threshold, the allocation parameters of the sorting area will be adjusted. For example, if the offset of a certain piece of edge material is 2.5 cm, which exceeds the threshold, it indicates that it may be slipping on the conveyor belt, affecting the sorting accuracy. Further, the parameters will be adjusted according to the offset, such as reducing the conveyor belt speed to 0.8 m / s, or redistributing the edge material to the spare area.

[0122] In one embodiment, after adjustment, the edge material is repositioned to the spare area, and the sensor detects again that the offset has dropped to 1.8 cm, confirming that the value has returned to normal and the final state is stable; the adjusted allocation parameters are recorded by the system to form an optimized sorting log; for example, the confirmation of the final state can be combined with multi-dimensional feature value analysis.

[0123] Furthermore, in addition to detecting the offset, the sensor can also collect the vibration frequency of the edge material to assess its stability during the sorting process; assuming the vibration frequency threshold is 50 Hz, and the frequency of a certain edge material is 60 Hz, it indicates that it may be unstable due to its irregular shape.

[0124] Furthermore, the edge material is allocated to the low-speed sorting area based on the offset and vibration frequency to ensure a smooth sorting process; the sorting status of the edge material is confirmed by multi-sensor data fusion to ensure the positioning accuracy of the edge material sorting process.

[0125] Specifically, the real-time signals collected by the sensors are combined with historical data for optimized allocation; for example, historical data shows that the offset of the edge material in the heavy area often exceeds the threshold, and the system can adjust the tilt angle of the conveyor belt in that area to 5 degrees in advance, thereby reducing the risk of offset.

[0126] In one embodiment, after adjusting the tilt angle, the average offset of the edge material in the heavy area was reduced to 1.7 cm, the confirmed value was stable, and the sorting efficiency was optimized; the combination of real-time signals and historical data makes the allocation of sorting areas more dynamic and adaptable; it can be understood that the above method, through classification priority, sensor signal extraction, feature value analysis and dynamic adjustment, ensures the accuracy and stability of edge material sorting in a step-by-step manner; in this scheme, each link supports each other to form a complete area allocation and status confirmation scheme.

[0127] Eighth embodiment;

[0128] Furthermore, based on the first embodiment, the sorting area is preferably monitored in real time to determine whether there are any abnormal detection situations in the sorting area. If there are abnormalities, the optimized classification priority is adjusted according to the abnormal detection situation, and the adjusted priority is determined. The sorting path of the edge material is updated using the adjusted priority to obtain the updated sorting path.

[0129] In this embodiment, real-time monitoring of the sorting area is achieved by collecting real-time monitoring data such as the amount of edge material accumulation, the operating status of the conveyor belt, and regional environmental parameters through sensors.

[0130] The sensors include, but are not limited to, laser rangefinders and pressure sensors, which are used to detect the stacking height and weight distribution of edge materials within a region. For example, if the laser rangefinder detects that the stacking height in a medium-sized area reaches 30 cm, exceeding the set threshold of 25 cm, it indicates that there may be an abnormal stacking. If the pressure sensor detects that the weight distribution in the area is uneven, with a pressure value of 500 Newtons in a certain part, exceeding the normal range of 400 Newtons, this abnormality may be caused by improper stacking of edge materials or conveyor belt jamming.

[0131] It should be noted that the above-mentioned anomaly detection needs to be combined with multi-dimensional data analysis, such as the comprehensive evaluation of parameters like stacking height and weight distribution, in order to avoid misjudgment results caused by a single indicator evaluation.

[0132] In one possible implementation, upon detecting an anomaly, the system adjusts the classification priority based on the anomaly type. For example, for an anomaly related to stacking, the system can temporarily increase the allocation priority of the light area, guiding subsequent scrap materials to areas with less stacking. Assuming the original priority was heavy area priority, the adjustment increases the priority of the light area, and scrap materials weighing less than 100 grams are reassigned to the light area. Furthermore, the adjusted priority updates the sorting path, generating a new path plan. For example, a scrap material weighing 80 grams, originally planned to enter the medium area, is now redirected to the light area, and the conveyor belt path is switched from line B to line A to avoid congested areas. Real-time path updates through real-time calculations ensure that scrap materials can quickly reach the target area.

[0133] Specifically, in the path update, to improve the accuracy of sorting or transfer, the physical properties of the edge material are matched with the sorting area. For example, when edge material with low surface smoothness slides on a high-speed conveyor belt, it will affect the path stability. The path selection is adjusted by smoothness scoring. For example, edge material with a score below 70% is preferentially assigned to a low-speed conveyor belt with a speed set at 0.5 m / s. This reduces the deviation of edge material during transmission and ensures the accuracy of reaching the target area.

[0134] It should be noted that the path update will also be further optimized by combining historical data; for example, if the medium-sized area often causes path congestion due to accumulation in the historical data, the congestion can be reduced by adding backup paths in advance and setting diversion points to guide some edge materials to the backup area.

[0135] In one possible implementation, anomaly detection and path update form a closed-loop feedback; after adjusting the priority, the sensor detects that the accumulation height in the medium-sized area has dropped to 20 cm and the pressure distribution has recovered to 350 Newtons, indicating that the anomaly has been alleviated; through real-time monitoring and dynamic adjustment, the stability and efficiency of the sorting process are ensured.

[0136] It should be noted that the flexibility of the path update is also reflected in its ability to handle emergencies. Specifically, when a certain area becomes temporarily unavailable due to equipment failure, the system can quickly switch to a backup area to ensure the continuity of sorting. For example, when sorting and transferring a piece of edge material weighing 150 grams, if the system detects that the heavy area is temporarily closed due to equipment maintenance, it will immediately update the path and assign it to the medium area, while reducing the conveyor belt speed to 0.6 meters per second to ensure stable transmission. Through the above dynamic adjustment scheme, the adaptability and reliability of the sorting system are significantly improved.

[0137] Ninth embodiment;

[0138] Furthermore, based on the first embodiment, preferably, the remaining space data of the storage area is obtained, the available volume index is extracted from the remaining space data, and attribute information including size and shape features is obtained for the edge material. Based on the available volume index and attribute information, a genetic algorithm is used to determine the target storage location.

[0139] The genetic algorithm is as follows: using available volume index and attribute information as the initial population, it iteratively optimizes the selection of crossover and mutation operations to obtain the position with the highest space utilization.

[0140] The motion trajectory of the robotic arm is planned based on the target storage location. The robotic arm is driven by the motion trajectory to grab the edge material and perform stacking operation. If the volume change after stacking exceeds the preset threshold, the trajectory parameters are adjusted. The volume change is the difference between the remaining space data before and after stacking, and the stacked edge material arrangement is obtained.

[0141] Update the occupancy status of the storage area by refreshing the remaining space data from the stacked edge material arrangement. The new remaining space data is obtained by subtracting the original remaining space data from the occupied volume after the arrangement, thus realizing dynamic management of the storage area.

[0142] In this embodiment, the remaining space of the storage area is obtained through laser scanning or 3D imaging. For example, a storage warehouse uses a laser scanner to periodically scan the shelf area, generate 3D point cloud data, and calculate the usable volume of the remaining space as 2.5 cubic meters. This accurately reflects the actual state of the storage area and provides a reliable basis for subsequent decision-making.

[0143] It should be noted that in scenarios where edge materials frequently enter and leave the warehouse, the work status of scheduling or sorting can be dynamically updated by monitoring the remaining space data in real time, thereby avoiding sorting or scheduling failures caused by space misjudgment; for example, machine vision recognition systems can be used to obtain attribute information such as the size and shape characteristics of edge materials.

[0144] Specifically, the image information of the edge material is captured by a camera, and the length, width, and height of the edge material are further identified, providing a basic input for subsequent optimization of the genetic algorithm. It should be noted that the identification of shape features needs to take into account the material and surface characteristics of the edge material to ensure that the algorithm can adapt to different types of edge material.

[0145] In one possible implementation, when using a genetic algorithm to determine the target storage location, the available volume and edge material attributes are used as the initial population; for example, the algorithm uses 10 candidate storage locations on the shelf as the population, and the available volume and edge material size matching degree of each location are used as the fitness function; through multiple rounds of selection, crossover and mutation operations, the algorithm finally selects the third layer of shelf B as the target location because its space utilization rate reaches 85%.

[0146] Through iterative optimization, the goal is to find the optimal solution in complex storage environments, thereby improving space utilization efficiency. For example, when planning the movement trajectory of a robotic arm based on the target storage location, a collision-free path from the initial point to the target location can be generated by path planning software. Specifically, the robotic arm needs to move 2 meters along the shelf aisle and then rise 0.5 meters to accurately place the edge material. It should be noted that the trajectory planning needs to take into account the joint limitations of the robotic arm and surrounding obstacles to ensure that the grasping and stacking process is smooth and efficient.

[0147] In one possible implementation, the volume change after stacking is determined by comparing it with the remaining space data before and after stacking. For example, if the remaining space before stacking is 2.5 cubic meters and after stacking it is 2.0 cubic meters, the change of 0.5 cubic meters exceeds the preset threshold of 0.3 cubic meters, triggering trajectory parameter adjustments, such as reducing the robotic arm's descent speed or fine-tuning the placement angle. This dynamic adjustment effectively addresses potential spatial deviations during stacking. Furthermore, when updating the storage area occupancy status, the data is refreshed based on the completed edge material arrangement. For example, after adding 0.5 cubic meters of edge material, the system calculates the new remaining space to be 2.0 cubic meters and updates the database, thereby ensuring the real-time accuracy of the storage area status and providing a reliable reference for subsequent operations.

[0148] In one possible implementation, when dynamically managing storage areas, space is optimized by combining predictive models; this further improves the overall operational efficiency of the warehouse and reduces space waste; for example, by analyzing historical stacking data, the amount of scrap material entering the warehouse in the next 24 hours can be predicted, and appropriate space can be reserved.

[0149] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An AGV-based automatic scheduling and high-efficiency sorting system for crystalline silicon edge scrap, characterized in that, include: Obtain the physical information of the edge material, and obtain the attribute information of the edge material based on the physical information; The system acquires the current location and environmental data of the automated guided vehicle (AGV), and generates a dynamic scheduling path by combining the edge material attribute information; it identifies the location and type of obstacles in the environment, and calculates the transportation priority of the edge material by combining the attribute information and generates the optimal path. The driving speed and steering angle of the automated guided vehicle are determined based on the dynamic scheduling path, and the path is adjusted based on real-time sensor data to obtain the updated dynamic scheduling path. The classification priority of the edge material is obtained based on the attribute information. After the classification priority is optimized, it is assigned to the corresponding sorting area. The sensor data of the sorting area is judged for anomalies. If there are anomalies, the optimized classification priority is adjusted and the sorting path of the edge material is updated. Obtain the remaining space data of the storage area, determine the target storage location based on the attribute information and the remaining space data, plan the movement trajectory of the robotic arm, stack the edge materials, and update the occupancy status of the storage area.

2. The AGV-based automatic scheduling and efficient sorting system for crystalline silicon edge material as described in claim 1, characterized in that, The length, width, and height data of the edge material are obtained to determine its size information; the spectral reflectance curve of the edge material is obtained using a spectral analyzer; and the material type of the edge material is determined based on a preset material spectral library. Obtain the weight of the edge material and calculate its volume based on the size information. The volume is the product of length, width, and height. Calculate the density based on the volume and weight, and integrate the size information, material type, weight, and density to obtain the complete attribute information of the edge material.

3. The AGV-based automatic scheduling and efficient sorting system for crystalline silicon edge material as described in claim 1, characterized in that, The current position and environmental data of the automated guided vehicle are obtained, including obstacle positions and storage area occupancy status; vehicle speed and load attributes are extracted from the current position to obtain preliminary path constraints. Based on preliminary path constraints and environmental data, it is determined whether the location of obstacles overlaps with the occupancy status of storage areas; if they overlap, the path nodes are adjusted and an obstacle avoidance and detour scheme is determined; based on the obstacle avoidance and detour scheme and vehicle load attributes, a dynamic scheduling path is generated, which is used for real-time navigation to the free storage area.

4. The AGV-based automatic scheduling and high-efficiency sorting system for crystalline silicon edge material as described in claim 1, characterized in that, Obstacle features are obtained from environmental data, and the location and type of obstacles are determined by image segmentation algorithms. The image segmentation algorithm is a semantic segmentation process based on convolutional neural networks, which takes obstacle features as input and outputs the segmented location coordinates and type labels. The transportation priority of the edge material is calculated using a weighted summation algorithm to obtain the priority value. The weighted summation algorithm is to assign weights to the weight factor and urgency factor in the attribute information and then sum them to obtain the priority value. The optimal path is generated based on obstacle location and transportation priority. The optimal path is calculated as follows: the obstacle location is input as an avoidance constraint and the priority value is used as the path cost weight. The output is the optimal path from the starting point to the destination.

5. The AGV-based automatic scheduling and efficient sorting system for crystalline silicon edge material as described in claim 1, characterized in that, Extract obstacle location and traffic congestion monitoring information, and integrate LiDAR and camera data through a weighted average method to determine path deviation correction requirements; Based on the path deviation correction requirements, calculate the path energy consumption based on the battery consumption rate, check the distance threshold, calculate the speed parameters and adjust the steering angle. Based on the determined speed parameters and the adjusted steering angle, the priority is adjusted according to the real-time delay, and the dynamic path generation is adjusted to obtain the updated dynamic scheduling path.

6. The AGV-based automatic scheduling and efficient sorting system for crystalline silicon edge material as described in claim 1, characterized in that, Based on the set of rules for classifying defects according to size thresholds and density ranges, a preliminary classification priority is obtained; the defect types of edge skin are obtained based on size and material density. Construct a random forest model and train it on historical sorting data; specifically, integrate multiple decision tree classification models, input initial classification priority and color uniformity, and output adjusted priority levels; The initial classification priority and color uniformity analysis were used to determine the thickness deviation and the adjusted priority level. Obtain the correlation between the adjusted priority level and the hardness index, determine the optimization parameters based on the recycling value and sorting cost. If the recycling value exceeds the sorting cost, increase the optimization parameters to obtain the optimized classification priority.

7. The AGV-based automatic scheduling and efficient sorting system for crystalline silicon edge material as described in claim 1, characterized in that, By using a preset classification priority, the edge material is allocated to the sorting area, and the initial occupancy data of the sorting area is obtained to determine the preliminary location of the edge material. Based on the preliminary location, the real-time signal of the edge material is collected by the sensor, and the feature value is extracted from the real-time signal to obtain the arrival status confirmation value. If the confirmation value exceeds the preset threshold, the allocation parameters of the sorting area are adjusted to obtain the final arrival status of the edge material.

8. The AGV-based automatic scheduling and high-efficiency sorting system for crystalline silicon edge material as described in claim 1, characterized in that, The sorting area is monitored in real time to determine whether there are any abnormal detection situations. If there are abnormalities, the classification priority is adjusted and optimized according to the abnormal detection situation, and the adjusted priority is determined. The sorting path for the edge leather is updated using the adjusted priority to obtain the updated sorting path.

9. The AGV-based automatic scheduling and high-efficiency sorting system for crystalline silicon edge material as described in claim 1, characterized in that, The remaining space data of the storage area is obtained, and the usable volume index is extracted from the remaining space data. At the same time, attribute information, including size and shape characteristics, is obtained for the edge material. Based on the usable volume index and attribute information, a genetic algorithm is used to determine the target storage location. The genetic algorithm is as follows: using available volume index and attribute information as the initial population, it iteratively optimizes the selection of crossover and mutation operations to obtain the position with the highest space utilization. The motion trajectory of the robotic arm is planned based on the target storage location. The robotic arm is driven by the motion trajectory to grab the edge material and perform stacking operation. If the volume change after stacking exceeds the preset threshold, the trajectory parameters are adjusted. The volume change is the difference between the remaining space data before and after stacking, and the stacked edge material arrangement is obtained. Update the occupancy status of the storage area by refreshing the remaining space data from the stacked edge material arrangement. The new remaining space data is obtained by subtracting the original remaining space data from the occupied volume after the arrangement, thus realizing dynamic management of the storage area.

Images