CN112547528A - Logistics sorting method and system based on classification identification - Google Patents

Logistics sorting method and system based on classification identification Download PDF

Info

Publication number
CN112547528A
CN112547528A CN202110223018.8A CN202110223018A CN112547528A CN 112547528 A CN112547528 A CN 112547528A CN 202110223018 A CN202110223018 A CN 202110223018A CN 112547528 A CN112547528 A CN 112547528A
Authority
CN
China
Prior art keywords
information
classification
sorting
robot
processed
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110223018.8A
Other languages
Chinese (zh)
Other versions
CN112547528B (en
Inventor
张倩
潘志明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huapengfei Co ltd
Original Assignee
Huapengfei Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huapengfei Co ltd filed Critical Huapengfei Co ltd
Priority to CN202110223018.8A priority Critical patent/CN112547528B/en
Publication of CN112547528A publication Critical patent/CN112547528A/en
Application granted granted Critical
Publication of CN112547528B publication Critical patent/CN112547528B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C3/00—Sorting according to destination
    • B07C3/003—Destination control; Electro-mechanical or electro- magnetic delay memories
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C3/00—Sorting according to destination
    • B07C3/003—Destination control; Electro-mechanical or electro- magnetic delay memories
    • B07C3/006—Electric or electronic control circuits, e.g. delay lines
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C3/00—Sorting according to destination
    • B07C3/10—Apparatus characterised by the means used for detection ofthe destination
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C3/00—Sorting according to destination
    • B07C3/18—Devices or arrangements for indicating destination, e.g. by code marks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
    • G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
    • G06K7/10009—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation sensing by radiation using wavelengths larger than 0.1 mm, e.g. radio-waves or microwaves
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C2301/00—Sorting according to destination
    • B07C2301/0066—Check for destination address in a database or list

Abstract

The embodiment of the invention discloses a logistics sorting method and a logistics sorting system based on classification identification, wherein the method comprises the following steps: acquiring information to be classified of an article to be processed through an information sensor; determining 2 classification recognition results corresponding to the to-be-processed article based on different neural network models and classification features corresponding to the to-be-classified information/to-be-classified information, and determining a final classification recognition result according to the 2 classification recognition results; determining sorting information corresponding to the classification recognition result, wherein the sorting information comprises distributed sorting robot information corresponding to the classification recognition result and robot path planning information; and determining a sorting robot according to the sorting information, and controlling the sorting robot to sort the to-be-processed articles to a target sorting position corresponding to the robot path planning information according to the robot path planning information so as to finish logistics sorting corresponding to the to-be-processed articles. By adopting the embodiment of the invention, the accuracy of classification and identification of the logistics system and the sorting efficiency can be improved.

Description

Logistics sorting method and system based on classification identification
Technical Field
The invention relates to the technical field of intelligent logistics, in particular to a logistics sorting method and system based on classification and identification.
Background
The logistics industry has developed to be an indispensable part in people's life, and express delivery or other logistics transportation bring great convenience to people's life. In the process of logistics transportation and management, each transported item needs to be subjected to path planning and sorting so as to complete the transportation from the departure place to the destination. However, during most logistics transportation, the sorting and sorting are mainly performed by manpower, which results in low overall efficiency.
Disclosure of Invention
Based on this, it is necessary to provide a logistics sorting method and system based on classification identification in order to solve the above problems.
In a first aspect of the present invention, there is provided a method for sorting logistics based on classification recognition, comprising:
acquiring information to be classified of the article to be processed through an information sensor, wherein the information sensor at least comprises one of an image sensor, a weight sensor and a radio frequency sensor, and the information to be classified at least comprises one of order information, image information, historical logistics information and article identification information;
determining a classification recognition result corresponding to the to-be-processed article based on a neural network model and the to-be-classified information;
determining sorting information corresponding to the classification recognition result, wherein the sorting information comprises distributed sorting robot information corresponding to the classification recognition result and robot path planning information;
determining a sorting robot according to the sorting information, and controlling the sorting robot to sort the to-be-processed articles to a target sorting position corresponding to the robot path planning information according to the robot path planning information so as to finish logistics sorting corresponding to the to-be-processed articles;
wherein the step of determining the classification recognition result corresponding to the article to be processed based on the neural network model and the information to be classified further comprises:
dividing information to be classified into a plurality of classification groups, wherein in each classification group: respectively extracting at least one classification group characteristic in information to be classified under each classification group according to a preset classification characteristic extraction algorithm, inputting the at least one classification group characteristic into a first neural network model corresponding to the classification group, and acquiring an output result as a first classification identification result corresponding to the article to be processed; the first neural network model corresponding to each classification group is constructed according to the classification group characteristics contained in the classification group;
inputting the information to be classified under each classification group into a second neural network model corresponding to each classification group, and acquiring an output result as a second classification identification result corresponding to the article to be processed; the second neural network model corresponding to each classification group is constructed according to the information to be classified contained in the classification group;
and determining a classification recognition result corresponding to the article to be processed according to the first classification recognition result and the second classification recognition result.
Optionally, after the step of controlling the sorting robot to sort the to-be-processed article to the target sorting position corresponding to the robot path planning information according to the robot path planning information to complete the logistics sorting of the to-be-processed article, the method further includes: and after the to-be-processed article reaches the target sorting position, acquiring a subsequent logistics processing strategy corresponding to the to-be-processed article, and carrying out logistics transportation or management on the to-be-processed article based on the acquired subsequent logistics processing strategy.
Optionally, the step of obtaining information to be classified of the article to be processed by the classification information sensor further includes: acquiring image information of the object to be processed through an image sensor, identifying the image information, and acquiring corresponding appearance information and/or order information; acquiring weight information of an article to be processed through a weight sensor; reading radio frequency information corresponding to a radio frequency device on the object to be processed through a radio frequency sensor, and searching order information and historical information corresponding to the radio frequency information in a preset logistics database according to the radio frequency information; the step of determining a classification recognition result corresponding to the article to be processed based on the neural network model and the information to be classified further includes: and taking the information to be classified as the input of the neural network model, and acquiring an output classification recognition result.
Optionally, the step of extracting at least one classification group feature in the information to be classified under each classification group according to a preset classification feature extraction algorithm further includes: according to a preset feature structure construction method, constructing classification structure features corresponding to information to be classified under a current classification group; calculating classification characteristics corresponding to each information to be classified under the current classification group according to a preset characteristic calculation method; determining classification group characteristics corresponding to information to be classified under the current classification group according to the classification structure characteristics and the classification characteristics; after the step of extracting at least one classification group feature in the information to be classified under each classification group according to a preset classification feature extraction algorithm, the method further comprises the following steps: and screening each classification group according to the classification group characteristics, and taking the screened classification group characteristics as the input of the first neural network model corresponding to the classification group.
Optionally, the first neural network model and the second neural network model respectively include a first neural network submodel, a second neural network submodel and a third neural network submodel; the step of inputting the at least one classification characteristic into a first neural network model and obtaining an output result as a first classification recognition result corresponding to the article to be processed, and the step of inputting the information to be classified into a second neural network model and obtaining an output result as a second classification recognition result corresponding to the article to be processed further include: inputting the model input information into a first neural network submodel to obtain an output first classification result; inputting the model input information and a first classification result output by the first neural network submodel into a second neural network submodel to obtain an output second classification result; inputting the model input information, a first classification result output by the first neural network submodel and a second classification result output by the second neural network submodel into a third neural network submodel to obtain an output third classification result; taking the first classification result, the second classification result and the third classification result as model output results; the model input information is at least one classification characteristic or information to be classified; the model output information is a first classification recognition result and a second classification recognition result.
Optionally, the method further includes: inputting model input information into the first neural network submodel, and performing convolution processing on the classification characteristics to obtain an output first classification result; performing fusion processing on model input information and the first classification result according to a preset Viterbi algorithm, inputting the fused features into the second neural network sub-model, performing convolution processing on the model input information and the first classification result, and acquiring an output second classification result; and fusing model input information and the first classification result and the second classification result according to a preset Viterbi algorithm, inputting the fused features into the third neural network sub-model, and performing deconvolution processing on the model input information and the first classification result and the second classification result to obtain an output third classification result.
Optionally, the determining the sorting information corresponding to the classification recognition result, where the sorting information includes allocated sorting robot information and robot path planning information corresponding to the classification recognition result, further includes: acquiring information to be classified and classified corresponding to a plurality of articles to be processed, a classification recognition result, positioning information of a plurality of sorting robots, front path information and robot state information; and distributing a sorting robot to each article to be processed according to a preset path planning algorithm, and distributing robot path planning information corresponding to each sorting robot, wherein the robot path planning information is used for controlling the sorting robot to carry out transportation on the article to be processed according to the path and speed information included in the robot path planning information.
Optionally, the step of determining a sorting robot according to the sorting information, and controlling the sorting robot to sort the to-be-processed article to a target sorting position corresponding to the robot path planning information according to the robot path planning information further includes: acquiring motion information of the sorting robot through a sensor arranged on the sorting robot, wherein the motion information comprises at least one of a motion position, a motion speed and a motion direction; judging whether the sorting robot has route deviation or not according to the motion information and the robot path planning information; determining a specific offset parameter in the case of a route offset; re-planning a path according to the offset parameter and the robot path planning information, determining updated path planning information and updating the robot path planning information according to the robot path planning information; and controlling the sorting robot to sort the to-be-processed articles to the target sorting position through the updated path planning information.
In a second aspect of the present invention, there is provided a logistics sorting system based on classification recognition, the system comprising a cloud server, a logistics sorting platform, and a plurality of sorting robots;
the logistics sorting platform is used for bearing and transporting the articles to be processed; one or more information sensors are arranged on the logistics sorting platform, wherein the information sensors at least comprise one of an image sensor, a weight sensor and a radio frequency sensor;
the cloud server is used for receiving to-be-classified information of the to-be-processed articles sent by the logistics sorting platform, classifying and identifying the to-be-processed articles based on the received to-be-classified information to determine corresponding classification and identification results, and determining corresponding sorting robot information and robot path planning information according to the classification and identification results; the information to be classified at least comprises one of order information, image information, historical logistics information and article identification information;
wherein the cloud server is further configured to: dividing information to be classified into a plurality of classification groups, wherein in each classification group: respectively extracting at least one classification group characteristic in information to be classified under each classification group according to a preset classification characteristic extraction algorithm, inputting the at least one classification group characteristic into a first neural network model corresponding to the classification group, and acquiring an output result as a first classification identification result corresponding to the article to be processed; the first neural network model corresponding to each classification group is constructed according to the classification group characteristics contained in the classification group; inputting the information to be classified under each classification group into a second neural network model corresponding to each classification group, and acquiring an output result as a second classification identification result corresponding to the article to be processed; the second neural network model corresponding to each classification group is constructed according to the information to be classified contained in the classification group; determining a classification recognition result corresponding to the article to be processed according to the first classification recognition result and the second classification recognition result;
the cloud server is further used for generating a control instruction for the sorting robot according to the determined information of the sorting robot and the robot path planning information and sending the control instruction to the sorting robot;
the sorting robot is used for receiving a control instruction issued by the cloud server and transporting the to-be-processed articles positioned on the logistics sorting platform to a target sorting position corresponding to the robot path planning information according to the sorting robot information and the robot path planning information contained in the control instruction according to the control instruction;
the sorting robot is also used for acquiring the motion information of the sorting robot and sending the motion information to the cloud server;
the cloud server is further used for judging whether the sorting robot has route deviation according to the received motion information and the robot path planning information, determining a specific deviation parameter under the condition that the route deviation exists, then performing path re-planning according to the deviation parameter and the robot path planning information, determining updated path planning information, updating the robot path planning information according to the robot path planning information, generating an updated control command and issuing the updated control command to the sorting robot.
Optionally, the logistics sorting platform includes a plurality of logistics sorting channels, and the plurality of logistics sorting channels can perform classification, identification and sorting on the plurality of articles to be processed in parallel; the cloud server also comprises a plurality of sorting robots for planning the paths of the articles to be processed and the sorting robots according to the articles to be processed on the logistics sorting channels and the sorting robots included in the system, so as to obtain robot path planning information corresponding to each article to be processed.
Optionally, the sorting robot is further configured to acquire, by a sensor disposed on the sorting robot, motion information of the sorting robot, where the motion information includes at least one of a motion position, a motion speed, and a motion direction; the cloud server is further used for judging whether the sorting robot has route deviation or not according to the motion information and the robot path planning information; determining a specific offset parameter in the case of a route offset; re-planning a path according to the offset parameter and the robot path planning information, and determining updated path planning information; the cloud server is further used for sending the updated path planning information to the sorting robot; the sorting robot is also used for sorting the articles to be processed to the target sorting position.
The embodiment of the invention has the following beneficial effects:
after the logistics sorting method and the logistics sorting system based on classification recognition are adopted, for the to-be-sorted articles needing to be sorted on the logistics sorting platform, the to-be-sorted information of the to-be-sorted articles is acquired through one or more information sensors arranged on the logistics sorting platform, the to-be-sorted information at least comprises one of order information, image information, historical logistics information and article identification information, and the information sensors at least comprise one of image sensors, weight sensors and radio frequency sensors; then, based on a neural network model and the information to be classified, determining a classification recognition result corresponding to the article to be processed, and determining sorting information corresponding to the classification recognition result, wherein the sorting information comprises distributed sorting robot information corresponding to the classification recognition result and robot path planning information; and determining a sorting robot according to the sorting information, and controlling the sorting robot to sort the to-be-processed articles to a target sorting position corresponding to the robot path planning information according to the robot path planning information so as to finish the logistics sorting corresponding to the to-be-processed articles. In the process of determining the classification recognition result corresponding to the object to be processed, the classification feature obtained after feature extraction is performed on the information to be classified according to the first neural network model is used as input to obtain the corresponding first classification recognition result, the information to be classified is directly used as input according to the second neural network model to obtain the corresponding second classification recognition result, and then the first classification recognition result and the second classification recognition result are comprehensively considered to determine the final classification recognition result.
That is to say, in this embodiment, in the case of performing logistics sorting on each article to be processed, the articles to be processed need to be identified through the information sensors on the plurality of logistics sorting platforms, so as to improve the accuracy of subsequent classification and identification. The method comprises the steps of classifying the articles to be processed based on the neural network model, then determining the sorting robot according to the classification result and determining the robot path planning information obtained by path planning of the articles to be processed, so that the accuracy of classification recognition and path planning of the articles to be processed can be improved, and the efficiency of logistics sorting and logistics management can be improved. In addition, in the embodiment, in the process of classifying the articles to be processed, the corresponding classification recognition results are calculated based on the classification information of different processing layers according to two different neural network models, so that the situation that the accuracy of training of a single neural network model is possibly insufficient is avoided, the accuracy of classification recognition of the articles to be processed can be further improved, and the efficiency of logistics sorting and logistics management is further improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Wherein:
FIG. 1 is a schematic diagram of a logistics sorting system based on classification identification according to one embodiment;
FIG. 2 is a schematic flow chart illustrating a method for sorting logistics based on classification recognition according to an embodiment;
FIG. 3 is a diagram illustrating a neural network model according to one embodiment;
FIG. 4 is a flow diagram that illustrates the processing of the classification recognition based on neural network models, according to one embodiment;
FIG. 5 is a flow diagram that illustrates the processing of the classification recognition based on the neural network model in one embodiment.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a logistics sorting method based on classification and identification, which is realized based on a logistics sorting system, wherein the logistics sorting system is the logistics sorting system based on the classification and identification. By using the logistics sorting method and system based on classification and identification, the accuracy of logistics sorting can be improved, the efficiency of logistics transportation and management can be improved, and the user experience can be improved.
Specifically, as shown in fig. 1, the logistics sorting system 10 includes a cloud server 102, and a logistics sorting platform 104 and a plurality of sorting robots 106 communicatively connected to the cloud server 102. The logistics sorting platform is configured to receive an article to be processed, identify the article to be processed, upload corresponding identification data to the connected cloud server 102, perform classification calculation by the cloud server 102, and issue corresponding instructions to the logistics sorting platform 104 and the sorting robot 106, so that the logistics sorting platform 104 and the sorting robot 106 perform corresponding logistics sorting operations.
The cloud server 102 processes the data, so that the requirement on the computing capacity of the logistics sorting platform or the sorting robot can be reduced, and the computing capacity of the cloud platform is fully utilized.
The sorting robot 106 may be an Automated Guided Vehicle (AGV) cart or other robot that can automatically transport the objects to be processed, and is not limited herein. In this embodiment, in order to improve the sorting efficiency, a sensor is further disposed on the sorting robot 106 for acquiring one or more of a moving position, a moving speed and a moving direction of the sorting robot 106, so that the sorting and transportation of the to-be-processed articles by the sorting robot 106 can be monitored.
The logistics sorting platform 104 is an intermediate platform for sorting a plurality of articles to be processed, and the logistics sorting platform can be divided into a plurality of transportation channels, and can simultaneously classify, identify and sort the plurality of articles to be processed through the plurality of transportation channels, so as to improve the logistics sorting efficiency. In a specific embodiment, the plurality of moving channels may be operated simultaneously, and the cloud server may control the plurality of sorting robots to perform logistics sorting corresponding to the plurality of transporting channels simultaneously, so as to further improve logistics sorting efficiency.
Further, as shown in fig. 2, a flow chart of a logistics sorting method based on classification recognition is provided. Specifically, the logistics sorting method based on classification and identification includes the following steps as shown in fig. 2:
step S102: and acquiring information to be classified of the articles to be processed through an information sensor.
Step S102 is performed based on the logistics sorting platform, and one or more sensors are disposed on the logistics sorting platform for acquiring information related to the articles to be processed. Wherein, the sensor arranged on the logistics sorting platform comprises one or more of an image sensor, a weight sensor and a radio frequency sensor. For example, an image of the article to be processed, which is transferred onto the logistics sorting platform, is acquired by an image sensor disposed above the logistics sorting platform, and then recognition is performed according to the acquired image to acquire appearance information of the article to be processed, such as size, shape and the like. Further, the order information of the article to be processed may be obtained by identifying the image, specifically, the corresponding order information may be obtained by an order label attached to the article to be processed, such as an express delivery order, and information such as a posting address, a receiving address, a type of the article living in the order information corresponding to the article to be processed may be obtained according to the order information.
The weight information of the articles to be processed is acquired through the weight sensor arranged in the logistics sorting platform.
The method comprises the steps that radio frequency information corresponding to-be-processed articles is obtained through a radio frequency sensor arranged in a logistics sorting platform, the radio frequency information is obtained through radio frequency tags arranged on the to-be-processed articles when logistics are started, namely, each to-be-processed article is provided with a radio frequency tag so as to obtain, read and store corresponding information. In this step, the information of the rf tag disposed on the article to be processed is read by the rf sensor to obtain corresponding rf information, and the logistics information corresponding to the article to be processed, such as the aforementioned order information and logistics information, can be obtained through the rf information.
In this embodiment, the information to be classified is not limited to the image information, the appearance information, the order information, the weight information, the radio frequency information, and the like given above, and any information that can be obtained in this step as the corresponding information to be classified is relevant to the classification of the article to be processed.
Step S104: and determining a classification recognition result corresponding to the to-be-processed article based on a neural network model and the to-be-classified information.
In this step, the to-be-classified information corresponding to the to-be-processed item acquired in step S102 is used to identify the to-be-processed item in a classified manner, so as to determine a corresponding classification identification result. It should be noted that, in this embodiment, the classification and identification result includes, but is not limited to, a category corresponding to the object to be processed, for example, whether the object is a fresh object or an aviation object; in addition, the classification and identification result also comprises the classification of the logistics transportation process, the classification result of the reliability of the identification result and the like.
For example, in one particular embodiment, the classification recognition result of the item to be processed may include, but is not limited to: non-fresh express, aviation, and the classification of sending addresses and receiving addresses corresponding to the classification of the logistics transportation process.
In a specific embodiment, the information to be classified acquired in step S102 is used as input, the corresponding classification recognition result is used as output, and the corresponding output result is acquired through the neural network model. In a specific embodiment, the neural network model comprises a convolutional neural network submodel and a classifier submodel, the information to be classified is subjected to feature extraction and convolution calculation through the convolutional neural network submodel to obtain corresponding features to be classified, and then corresponding classification recognition results are obtained through the classifier submodel.
In a specific embodiment, the neural network model described above is further described.
Referring to fig. 3, a schematic diagram of the neural network model is shown. The neural network model comprises a first neural network submodel, a second neural network submodel and a third neural network submodel. The process of obtaining the classification recognition result according to the application network model is specifically as follows. Firstly, processing information to be classified to obtain the processed information to be classified; then inputting the processed information to be classified into a first neural network submodel to obtain an output first classification result; inputting the processed information to be classified and a first classification result output by the first neural network submodel into a second neural network submodel to obtain an output second classification result; inputting the processed information to be classified, a first classification result output by the first neural network submodel and a second classification result output by the second neural network submodel into a third neural network submodel to obtain an output third classification result; and taking the first classification result, the second classification result and the third classification result as classification recognition results.
The information to be classified is first processed to further optimize the information to be classified, for example, the information to be classified is preprocessed to remove repeated information and incomplete information in the information to be classified. Further, the processing of the information to be classified further includes extracting the classification features of the information to be classified according to a preset feature extraction algorithm, wherein the classification features of the information to be classified can be determined according to specific features to be classified. For example, when the information to be classified is information such as an appearance and a shape corresponding to the object to be processed, the corresponding image feature may be obtained according to a preset image processing algorithm, and in a specific embodiment, the mel-frequency spectrum feature corresponding to the information to be classified is extracted according to a preset mel-frequency spectrum extraction algorithm.
After the classification features of the information to be classified are extracted, the classification features can be further processed through 3 sub-network models included in the neural network model to calculate corresponding classification recognition results.
Specifically, as mentioned above, the neural network model includes 3 sub-network models connected to each other, where the first neural network sub-model is used to process the classification features of the information to be classified, where the first neural network sub-model is a convolutional neural network, and the first classification result is obtained by performing convolutional calculation on the classification features to obtain corresponding features; to obtain a corresponding first classification result. Then, the first classification result and the classification feature are fused, for example, the first classification result and the classification feature are merged according to a preset feature merging algorithm, and then the merged feature is used as an input of a second neural network submodel, which is specifically shown in fig. 3. Further, in this embodiment, the feature fusion algorithm may be a Viterbi algorithm, and the first classification result and the classification feature are subjected to splicing processing to obtain a corresponding splicing feature as a feature after the merging processing.
And further, inputting the first classification result and the classification features or the corresponding combined features into a second neural network submodel to obtain a corresponding second classification result. The second neural network submodel is a deconvolution neural network, and the second classification result is obtained by performing deconvolution calculation on the classification characteristic and the first classification result to obtain a corresponding deconvolution characteristic as the second classification result. Then, the first classification result, the second classification result, and the classification features are fused, for example, the first classification result, the second classification result, and the classification features are merged according to a preset feature merging algorithm, and the merged features are used as an input of a second neural network sub-model, which is specifically shown in fig. 4. Further, in this embodiment, the feature fusion algorithm may be a Viterbi algorithm, and the first classification result, the second classification result, and the classification feature are subjected to splicing processing to obtain a corresponding splicing feature as a feature after the merging processing.
And then, inputting the first classification result, the second classification result and the classification characteristic or the corresponding combined characteristic into a third neural network submodel to obtain a corresponding third classification result.
The fusion, combination and splicing processing between the classification result and the classification feature can be simple to splice and combine the features, and the features can be further processed according to a preset feature processing algorithm, so that the fusion between the features is more suitable for the calculation of a subsequent neural network submodel.
Each neural network sub-model is used for extracting features of information to be classified in different degrees, and all the features are not processed by one large neural network model, so that the accuracy of each feature can be improved, and the accuracy of a classification recognition result is further improved. Furthermore, in the process of training the neural network model, the training of the model is also processed by a plurality of network submodels, so that the accuracy of the subsequent calculation of the neural network model can be improved.
Here, the first classification result obtained by the first neural network submodel is a classification result corresponding to the first class of features, for example, a classification result corresponding to basic logistics information corresponding to whether the article is an aviation piece, whether the article is fresh or fresh, and an order information waiting for processing. The second classification result obtained by the second neural network submodel is a classification result corresponding to the second type of feature, and may be, for example, further classification of the article to be processed, for example, whether an abnormality exists or not, or classification according to the position of the current logistics node in the logistics link and subsequent logistics nodes; the second neural network sub-model is mainly used for further classifying the articles to be processed, so that the articles to be processed are classified more finely on the basis of the basic information, and the logistics sorting confirmation and path planning can be performed more accurately in the follow-up process. The third classification result obtained by the third neural network submodel is a classification result corresponding to the third type of feature, and here, mainly refers to a subsequent logistics path determined according to the order information and the logistics information, including a sorting position corresponding to sorting, information and a path of the sorting robot, and a planned subsequent logistics path.
Since there is a mutual correlation between a plurality of classification results, it is not possible to perform calculation by a plurality of submodels alone, but it is necessary to further input the output result of the previous layer into the next layer to obtain the output of the more accurate classification result of the next layer.
Further, in order to improve the accuracy of the classification recognition, in this embodiment, the output of the classification recognition result may also be determined by dual classification recognition.
Specifically, a first classification recognition result corresponding to the first neural network model is output through the first neural network model, a second classification recognition result corresponding to the second neural network model is output through the second neural network model, and then a final classification recognition result is determined according to the first classification recognition result and the second classification recognition result. The structure of the first neural network model and the second neural network model can be described by the structure and calculation of the neural network model. However, in the present embodiment, there is also a certain difference between the first neural network model and the second neural network model, for example, the inputs of the first neural network model and the second neural network model are different.
Specifically, the calculation process of the first classification recognition result is as follows.
In the process of determining the first classification recognition result, a plurality of information to be classified used for classification recognition are firstly divided into a plurality of classification groups according to a preset grouping method. Each classification group comprises one or more pieces of information to be classified, and at least one piece of attribute information of the one or more pieces of information to be classified in each classification group is matched.
Then, aiming at each classification group, according to a preset classification feature extraction algorithm, the classification group features corresponding to one or more pieces of information to be classified contained in each classification group are respectively extracted. The classification group features are the same and different features contained in the one or more pieces of information to be classified respectively according to a preset classification feature extraction algorithm.
In one embodiment, the classification group features of one or more pieces of information to be classified included in the classification group may be further constructed by:
according to a preset feature structure construction method, constructing classification structure features corresponding to information to be classified under a current classification group; for example, according to the information to be classified in the current classification group, the corresponding classification structure features are constructed in a structure construction manner such as a binary tree structure (including but not limited to a binary tree structure). Then, aiming at each node (namely each information to be classified) under the classification structure, calculating corresponding classification characteristics according to a preset characteristic calculation method; here, the calculation of the classification feature may be any feature value calculation method, such as an IV value, a WOE value, or any other feature value. After the feature value corresponding to each node is obtained through calculation, the classification group feature corresponding to the information to be classified under the current classification group can be determined according to the classification structure feature and the classification feature; to complete the construction of the classification group characteristics for each classification group. The classification group of features may then be used as an input to a neural network model in calculating the classification-identifying features from the neural network model.
Further, in another embodiment, in order to improve the accuracy of the calculation of the neural network model, the calculated classification group features need to be screened. And screening the classification group according to whether the classification group characteristics meet the preset conditions to determine the classification group corresponding to the screened classification group characteristics and determine the screened classification group. And then, the screening result is used as the input of the neural network model so as to reduce the calculation amount of the model and improve the classification and identification efficiency.
After the step of extracting at least one classification group feature in the information to be classified under each classification group according to a preset classification feature extraction algorithm, the method further comprises the following steps:
and then respectively constructing a corresponding neural network model (namely a first neural network model corresponding to the classification group) under each classification group, outputting a corresponding classification recognition result, and determining a first classification recognition result corresponding to the to-be-processed article according to the classification recognition result corresponding to each classification group.
Further, the second classification identification result is determined by directly taking one or more pieces of information to be classified under each classification group as the input of the neural network model (i.e., the second neural network model corresponding to the current classification group) under each classification group, then outputting the corresponding classification identification result, and then determining the second classification identification result corresponding to the object to be processed according to the classification identification result corresponding to each classification group. And then carrying out merging processing according to the first classification recognition result and the second classification recognition result to obtain a final classification recognition result.
The above-mentioned process of obtaining the first classification recognition result and the second classification recognition result and determining the corresponding classification recognition result may refer to a schematic diagram given in fig. 5.
It should be noted that, the above-mentioned obtaining process of obtaining the corresponding first classification recognition result and the second classification recognition result according to the first neural network model and the second neural network model may apply the process of obtaining the corresponding classification recognition secondary result according to the neural network model described in the above embodiments, and details are not repeated here.
Step S106: and determining sorting information corresponding to the classification recognition result, wherein the sorting information comprises distributed sorting robot information corresponding to the classification recognition result and robot path planning information.
As described above, the classification recognition result includes the relevant information of the sorting robot and the relevant information of the path planning, so that it can be determined that the sorting robot and the relevant information corresponding to the current article to be sorted are determined according to the information to be sorted in the current logistics sorting process, and the path planning information corresponding to the sorting robot is also included.
That is, in step S104, according to the current article to be processed and other articles to be processed in the current logistics system, how to sort the current article to be processed and the sorting path are planned, so as to use the corresponding information as the classification recognition result. That is to say, in this embodiment, the neural network model is used to perform path planning on the sorting robot allocation of the articles to be processed and the corresponding path, and compared with the common path planning, the accuracy of the path planning can be improved, so that the overall logistics efficiency is improved.
Specifically, in the process of determining the sorting robot and the corresponding robot path planning information for the current article to be processed, first, information to be sorted corresponding to a plurality of articles to be processed in the logistics system, a corresponding sorting recognition result, positioning information of one or more sorting robots in the current sorting link and relevant information of a sorting task being executed, and state information, such as historical fault information and current electric quantity information, corresponding to each sorting robot need to be obtained. And then planning the path of each article to be processed according to a preset path planning algorithm, and distributing the corresponding sorting robot so as to ensure that the efficiency in the current sorting link is highest. And the robot path planning information is used for controlling the sorting robot to transport the to-be-processed articles according to the path and the speed information included in the robot path planning information.
In another embodiment, the classification identification information includes planning information corresponding to subsequent logistics paths of the to-be-processed articles, but does not include allocation and corresponding path planning of the sorting robots in the current sorting link, and the allocated sorting robots and corresponding running paths of each sorting robot can be determined for each to-be-processed article through a multi-objective path planning algorithm.
In a specific embodiment, the multi-objective path planning algorithm may be a multi-constraint optimization algorithm, and the path planning for the sorting robot is completed by agreeing a boundary condition, a remaining endurance time of the sorting robot, and a target sorting position, then establishing a first optimization target as a multi-objective random planning algorithm that transports a plurality of articles to be processed to the target sorting position within a shortest time, and a second optimization target as a multi-objective random planning algorithm that obtains a maximum utilization for the sorting robot, and then performing a fuzzy processing on the optimization target by using a fuzzy mathematic method and a neural network model, and performing a processing on the path planning algorithm by using the neural network model, and obtaining an output result, that is, an optimal method for path planning of the sorting robot.
Step S108: and determining a sorting robot according to the sorting information, and controlling the sorting robot to sort the to-be-processed articles to a target sorting position corresponding to the robot path planning information according to the robot path planning information so as to finish the logistics sorting corresponding to the to-be-processed articles.
In the embodiment, the sorting target sorting position in the current sorting link is determined by classifying the information to be classified corresponding to the articles to be processed, and subsequent logistics nodes are planned; and determining the content to be executed in the current sorting link through the distribution of the sorting robots and the planning of the sorting robots. Specifically, the sorting robot is controlled to move according to relevant control information such as a starting position, a target sorting position and a corresponding running path in the robot path planning information so as to sort the to-be-processed articles to the target sorting position, and thus, the logistics sorting of the to-be-processed articles is completed.
After the sorting robot transports the articles to be processed to the corresponding target sorting positions, the subsequent other logistics links corresponding to the current sorting link in the logistics links can be continuously completed. Specifically, the subsequent logistics processing strategies corresponding to the to-be-processed item, such as vehicle separation and assembly, distribution and the like, are acquired, and then the subsequent flow corresponding to the to-be-processed item is executed to complete the corresponding logistics transportation and management flow.
It should be noted that, in this embodiment, in the process of controlling the sorting robot to sort and transport the to-be-processed articles to the target sorting position corresponding to the robot path planning information according to the robot path planning information, it is further necessary to determine whether the sorting robot travels according to the path planning information, and to find the motion deviation of the sorting robot in time.
Specifically, set up one or more sensor on sorting robot to the motion state of monitoring sorting robot, specifically, acquire sorting robot's motion information, wherein, included and acquireed sorting robot's motion through the GPS sensor till, acquireed sorting robot's information such as movement speed and direction of motion through gyroscope etc. to can confirm sorting robot's motion state.
And then judging with the previously determined robot path planning information according to the monitored motion information so as to judge whether the sorting robot has route deviation. Specifically, whether the current movement position of the sorting robot is matched with the robot path planning information or not is judged, and therefore whether the route deviation exists or not is determined. Determining a specific offset parameter in the case of a route offset;
re-planning a path according to the offset parameter and the robot path planning information, determining updated path planning information and updating the robot path planning information according to the robot path planning information;
and controlling the sorting robot to sort the to-be-processed articles to the target sorting position through the updated path planning information.
Further, in the process of determining whether the sorting robot has the route deviation, it is also required to determine whether the speed of the sorting robot meets the requirement, that is, whether the speed of each motion position given in the path planning information meets the requirement corresponding to the speed. Specifically, the path planning information further includes the speed corresponding to the planned path, so as to avoid the sorting robot colliding with other sorting robots, and further improve the sorting efficiency.
Judging with robot path planning information according to the monitored motion information to judge whether the motion state of the sorting robot is matched with the motion speed planning information corresponding to the motion position in the motion information, if so, determining that the route has no offset, if not, determining that the route has no offset, and generating corresponding offset parameters according to the motion information and the robot path planning information, wherein the offset parameters comprise at least one of the motion information offset parameters, the motion position offset parameters and the like. And planning the path of the sorting robot again according to the offset parameter so as to determine new path planning information. The process of determining the new path planning information also needs to plan the paths of the sorting robots according to the path planning algorithm according to the relevant information of the plurality of sorting robots on the current logistics sorting platform, so that the sorting robots process the articles to be processed according to the re-planned path planning information, and sort the articles to be processed to the target sorting position.
After the logistics sorting method and the logistics sorting system based on classification recognition are adopted, for the to-be-sorted articles needing to be sorted on the logistics sorting platform, the to-be-sorted information of the to-be-sorted articles is acquired through one or more information sensors arranged on the logistics sorting platform, the to-be-sorted information at least comprises one of order information, image information, historical logistics information and article identification information, and the information sensors at least comprise one of image sensors, weight sensors and radio frequency sensors; then, based on a neural network model and the information to be classified, determining a classification recognition result corresponding to the article to be processed, and determining sorting information corresponding to the classification recognition result, wherein the sorting information comprises distributed sorting robot information corresponding to the classification recognition result and robot path planning information; and determining a sorting robot according to the sorting information, and controlling the sorting robot to sort the to-be-processed articles to a target sorting position corresponding to the robot path planning information according to the robot path planning information so as to finish the logistics sorting corresponding to the to-be-processed articles. In the process of determining the classification recognition result corresponding to the object to be processed, the classification feature obtained after feature extraction is performed on the information to be classified according to the first neural network model is used as input to obtain the corresponding first classification recognition result, the information to be classified is directly used as input according to the second neural network model to obtain the corresponding second classification recognition result, and then the first classification recognition result and the second classification recognition result are comprehensively considered to determine the final classification recognition result.
That is to say, in this embodiment, in the case of performing logistics sorting on each article to be processed, the articles to be processed need to be identified through the information sensors on the plurality of logistics sorting platforms, so as to improve the accuracy of subsequent classification and identification. The method comprises the steps of classifying the articles to be processed based on the neural network model, then determining the sorting robot according to the classification result and determining the robot path planning information obtained by path planning of the articles to be processed, so that the accuracy of classification recognition and path planning of the articles to be processed can be improved, and the efficiency of logistics sorting and logistics management can be improved. In addition, in the embodiment, in the process of classifying the articles to be processed, the corresponding classification recognition results are calculated based on the classification information of different processing layers according to two different neural network models, so that the situation that the accuracy of training of a single neural network model is possibly insufficient is avoided, the accuracy of classification recognition of the articles to be processed can be further improved, and the efficiency of logistics sorting and logistics management is further improved.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the program is executed. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims. Please enter the implementation content part.

Claims (10)

1. A logistics sorting method based on classification recognition is characterized by comprising the following steps:
acquiring information to be classified of the article to be processed through an information sensor, wherein the information sensor at least comprises one of an image sensor, a weight sensor and a radio frequency sensor, and the information to be classified at least comprises one of order information, image information, historical logistics information and article identification information;
determining a classification recognition result corresponding to the to-be-processed article based on a neural network model and the to-be-classified information;
determining sorting information corresponding to the classification recognition result, wherein the sorting information comprises distributed sorting robot information corresponding to the classification recognition result and robot path planning information;
determining a sorting robot according to the sorting information, and controlling the sorting robot to sort the to-be-processed articles to a target sorting position corresponding to the robot path planning information according to the robot path planning information so as to finish logistics sorting corresponding to the to-be-processed articles;
wherein the step of determining the classification recognition result corresponding to the article to be processed based on the neural network model and the information to be classified further comprises:
dividing information to be classified into a plurality of classification groups, wherein in each classification group: respectively extracting at least one classification group characteristic in information to be classified under each classification group according to a preset classification characteristic extraction algorithm, inputting the at least one classification group characteristic into a first neural network model corresponding to the classification group, and acquiring an output result as a first classification identification result corresponding to the article to be processed; the first neural network model corresponding to each classification group is constructed according to the classification group characteristics contained in the classification group;
inputting the information to be classified under each classification group into a second neural network model corresponding to each classification group, and acquiring an output result as a second classification identification result corresponding to the article to be processed; the second neural network model corresponding to each classification group is constructed according to the information to be classified contained in the classification group;
and determining a classification recognition result corresponding to the article to be processed according to the first classification recognition result and the second classification recognition result.
2. The logistics sorting method based on classification and identification as claimed in claim 1, wherein after the step of controlling the sorting robot to sort the to-be-processed item to the target sorting position corresponding to the robot path planning information according to the robot path planning information to complete the logistics sorting of the to-be-processed item, the method further comprises:
and after the to-be-processed article reaches the target sorting position, acquiring a subsequent logistics processing strategy corresponding to the to-be-processed article, and carrying out logistics transportation or management on the to-be-processed article based on the acquired subsequent logistics processing strategy.
3. The logistics sorting method based on classification identification as claimed in claim 1, wherein the step of obtaining information to be classified of the articles to be processed by the classification information sensor further comprises:
acquiring image information of the object to be processed through an image sensor, identifying the image information, and acquiring corresponding appearance information and/or order information;
acquiring weight information of an article to be processed through a weight sensor;
and reading radio frequency information corresponding to a radio frequency device on the object to be processed through a radio frequency sensor, and searching order information and historical information corresponding to the radio frequency information in a preset logistics database according to the radio frequency information.
4. The logistics sorting method based on classification and identification as claimed in claim 1, wherein the step of extracting at least one classification group feature of the information to be classified under each classification group according to a preset classification feature extraction algorithm further comprises:
according to a preset feature structure construction method, constructing classification structure features corresponding to information to be classified under a current classification group;
calculating classification characteristics corresponding to each information to be classified under the current classification group according to a preset characteristic calculation method;
determining classification group characteristics corresponding to information to be classified under the current classification group according to the classification structure characteristics and the classification characteristics;
after the step of extracting at least one classification group feature in the information to be classified under each classification group according to a preset classification feature extraction algorithm, the method further comprises the following steps:
and screening each classification group according to the classification group characteristics, and taking the screened classification group characteristics as the input of the first neural network model corresponding to the classification group.
5. The logistics sorting method based on classification identification as claimed in claim 1, wherein the first neural network model and the second neural network model comprise a first neural network submodel, a second neural network submodel and a third neural network submodel, respectively;
the step of inputting the at least one classification characteristic into a first neural network model and obtaining an output result as a first classification recognition result corresponding to the article to be processed, and the step of inputting the information to be classified into a second neural network model and obtaining an output result as a second classification recognition result corresponding to the article to be processed further include:
inputting the model input information into a first neural network submodel to obtain an output first classification result;
inputting the model input information and a first classification result output by the first neural network submodel into a second neural network submodel to obtain an output second classification result;
inputting the model input information, a first classification result output by the first neural network submodel and a second classification result output by the second neural network submodel into a third neural network submodel to obtain an output third classification result;
taking the first classification result, the second classification result and the third classification result as model output results;
the model input information is at least one classification characteristic or information to be classified; the model output information is a first classification recognition result and a second classification recognition result.
6. The method for sorting logistics according to claim 5, wherein the method further comprises:
inputting model input information into the first neural network submodel, and performing convolution processing on the classification characteristics to obtain an output first classification result;
performing fusion processing on model input information and the first classification result according to a preset Viterbi algorithm, inputting the fused features into the second neural network sub-model, performing convolution processing on the model input information and the first classification result, and acquiring an output second classification result;
and fusing model input information and the first classification result and the second classification result according to a preset Viterbi algorithm, inputting the fused features into the third neural network sub-model, and performing deconvolution processing on the model input information and the first classification result and the second classification result to obtain an output third classification result.
7. The method for sorting logistics according to claim 1, wherein the step of determining sorting information corresponding to the result of classification recognition, the sorting information including allocated sorting robot information corresponding to the result of classification recognition and robot path planning information, further comprises:
acquiring information to be classified and classified corresponding to a plurality of articles to be processed, a classification recognition result, positioning information of a plurality of sorting robots, front path information and robot state information;
and distributing a sorting robot to each article to be processed according to a preset path planning algorithm, and distributing robot path planning information corresponding to each sorting robot, wherein the robot path planning information is used for controlling the sorting robot to carry out transportation on the article to be processed according to the path and speed information included in the robot path planning information.
8. The logistics sorting method based on classification and identification as claimed in claim 7, wherein the step of determining a sorting robot according to the sorting information, controlling the sorting robot to sort the to-be-processed item to a target sorting position corresponding to the robot path planning information according to the robot path planning information further comprises:
acquiring motion information of the sorting robot through a sensor arranged on the sorting robot, wherein the motion information comprises at least one of a motion position, a motion speed and a motion direction;
judging whether the sorting robot has route deviation or not according to the motion information and the robot path planning information;
determining a specific offset parameter in the case of a route offset;
re-planning a path according to the offset parameter and the robot path planning information, determining updated path planning information and updating the robot path planning information according to the robot path planning information;
and controlling the sorting robot to sort the to-be-processed articles to the target sorting position through the updated path planning information.
9. The logistics sorting system based on classification recognition is characterized by comprising a cloud server, a logistics sorting platform and a plurality of sorting robots;
the logistics sorting platform is used for bearing and transporting the articles to be processed; one or more information sensors are arranged on the logistics sorting platform, wherein the information sensors at least comprise one of an image sensor, a weight sensor and a radio frequency sensor;
the cloud server is used for receiving to-be-classified information of the to-be-processed articles sent by the logistics sorting platform, classifying and identifying the to-be-processed articles based on the received to-be-classified information to determine corresponding classification and identification results, and determining corresponding sorting robot information and robot path planning information according to the classification and identification results; the information to be classified at least comprises one of order information, image information, historical logistics information and article identification information;
wherein the cloud server is further configured to: dividing information to be classified into a plurality of classification groups, wherein in each classification group: respectively extracting at least one classification group characteristic in information to be classified under each classification group according to a preset classification characteristic extraction algorithm, inputting the at least one classification group characteristic into a first neural network model corresponding to the classification group, and acquiring an output result as a first classification identification result corresponding to the article to be processed; the first neural network model corresponding to each classification group is constructed according to the classification group characteristics contained in the classification group; inputting the information to be classified under each classification group into a second neural network model corresponding to each classification group, and acquiring an output result as a second classification identification result corresponding to the article to be processed; the second neural network model corresponding to each classification group is constructed according to the information to be classified contained in the classification group; determining a classification recognition result corresponding to the article to be processed according to the first classification recognition result and the second classification recognition result;
the cloud server is further used for generating a control instruction for the sorting robot according to the determined information of the sorting robot and the robot path planning information and sending the control instruction to the sorting robot;
the sorting robot is used for receiving a control instruction issued by the cloud server and transporting the to-be-processed articles positioned on the logistics sorting platform to a target sorting position corresponding to the robot path planning information according to the sorting robot information and the robot path planning information contained in the control instruction according to the control instruction;
the sorting robot is also used for acquiring the motion information of the sorting robot and sending the motion information to the cloud server;
the cloud server is further used for judging whether the sorting robot has route deviation according to the received motion information and the robot path planning information, determining a specific deviation parameter under the condition that the route deviation exists, then performing path re-planning according to the deviation parameter and the robot path planning information, determining updated path planning information, updating the robot path planning information according to the robot path planning information, generating an updated control command and issuing the updated control command to the sorting robot.
10. The logistics sorting system based on classification and identification of claim 9, wherein the logistics sorting platform comprises a plurality of logistics sorting channels, and the plurality of logistics sorting channels can perform classification and sorting on a plurality of articles to be processed respectively in parallel;
the cloud server also comprises a plurality of sorting robots, wherein the sorting robots plan paths of the to-be-processed articles and the to-be-processed articles according to the to-be-processed articles on the logistics sorting channels and the sorting robots included in the system so as to obtain robot path planning information corresponding to each to-be-processed article; the sorting robot is further used for acquiring motion information of the sorting robot through a sensor arranged on the sorting robot, and the motion information comprises at least one of a motion position, a motion speed and a motion direction;
the cloud server is further used for judging whether the sorting robot has route deviation or not according to the motion information and the robot path planning information; determining a specific offset parameter in the case of a route offset; re-planning a path according to the offset parameter and the robot path planning information, and determining updated path planning information;
the cloud server is further used for sending the updated path planning information to the sorting robot;
the sorting robot is also used for sorting the articles to be processed to the target sorting position.
CN202110223018.8A 2021-03-01 2021-03-01 Logistics sorting method and system based on classification identification Active CN112547528B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110223018.8A CN112547528B (en) 2021-03-01 2021-03-01 Logistics sorting method and system based on classification identification

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110223018.8A CN112547528B (en) 2021-03-01 2021-03-01 Logistics sorting method and system based on classification identification

Publications (2)

Publication Number Publication Date
CN112547528A true CN112547528A (en) 2021-03-26
CN112547528B CN112547528B (en) 2021-05-25

Family

ID=75034936

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110223018.8A Active CN112547528B (en) 2021-03-01 2021-03-01 Logistics sorting method and system based on classification identification

Country Status (1)

Country Link
CN (1) CN112547528B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113663931A (en) * 2021-07-30 2021-11-19 广州佳帆计算机有限公司 Article sorting method and device
CN114029243A (en) * 2021-11-11 2022-02-11 江苏昱博自动化设备有限公司 Soft object grabbing and identifying method for sorting robot hand
CN117160877A (en) * 2023-11-02 2023-12-05 启东亦大通自动化设备有限公司 Article sorting method for logistics robot
WO2024038941A1 (en) * 2022-08-17 2024-02-22 쿠팡 주식회사 Data providing method and device

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003022406A (en) * 2001-07-10 2003-01-24 Hitachi Ltd Article sorting machine, and system and method for managing use of article sorting machine
CN206897873U (en) * 2017-04-12 2018-01-19 浙江硕和机器人科技有限公司 A kind of image procossing and detecting system based on detection product performance
WO2018078613A1 (en) * 2016-10-31 2018-05-03 D.I.R. Technologies (Detection Ir) Ltd. System and method for automatic inspection and classification of discrete items
CN109102543A (en) * 2018-08-17 2018-12-28 深圳蓝胖子机器人有限公司 Object positioning method, equipment and storage medium based on image segmentation
CN109513630A (en) * 2018-11-14 2019-03-26 深圳蓝胖子机器人有限公司 Packages system and its control method, storage medium
CN109513629A (en) * 2018-11-14 2019-03-26 深圳蓝胖子机器人有限公司 Packages method, apparatus and computer readable storage medium
CN110694921A (en) * 2019-09-06 2020-01-17 北京农业智能装备技术研究中心 Vegetable sorting system and control method thereof
CN111468430A (en) * 2020-04-17 2020-07-31 无锡雪浪数制科技有限公司 Depth vision-based coal gangue separation method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003022406A (en) * 2001-07-10 2003-01-24 Hitachi Ltd Article sorting machine, and system and method for managing use of article sorting machine
WO2018078613A1 (en) * 2016-10-31 2018-05-03 D.I.R. Technologies (Detection Ir) Ltd. System and method for automatic inspection and classification of discrete items
CN206897873U (en) * 2017-04-12 2018-01-19 浙江硕和机器人科技有限公司 A kind of image procossing and detecting system based on detection product performance
CN109102543A (en) * 2018-08-17 2018-12-28 深圳蓝胖子机器人有限公司 Object positioning method, equipment and storage medium based on image segmentation
CN109513630A (en) * 2018-11-14 2019-03-26 深圳蓝胖子机器人有限公司 Packages system and its control method, storage medium
CN109513629A (en) * 2018-11-14 2019-03-26 深圳蓝胖子机器人有限公司 Packages method, apparatus and computer readable storage medium
CN110694921A (en) * 2019-09-06 2020-01-17 北京农业智能装备技术研究中心 Vegetable sorting system and control method thereof
CN111468430A (en) * 2020-04-17 2020-07-31 无锡雪浪数制科技有限公司 Depth vision-based coal gangue separation method

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113663931A (en) * 2021-07-30 2021-11-19 广州佳帆计算机有限公司 Article sorting method and device
CN113663931B (en) * 2021-07-30 2023-03-10 广州佳帆计算机有限公司 Article sorting method and device
CN114029243A (en) * 2021-11-11 2022-02-11 江苏昱博自动化设备有限公司 Soft object grabbing and identifying method for sorting robot hand
WO2024038941A1 (en) * 2022-08-17 2024-02-22 쿠팡 주식회사 Data providing method and device
CN117160877A (en) * 2023-11-02 2023-12-05 启东亦大通自动化设备有限公司 Article sorting method for logistics robot
CN117160877B (en) * 2023-11-02 2024-01-02 启东亦大通自动化设备有限公司 Article sorting method for logistics robot

Also Published As

Publication number Publication date
CN112547528B (en) 2021-05-25

Similar Documents

Publication Publication Date Title
CN112547528B (en) Logistics sorting method and system based on classification identification
CN108629531B (en) Method for transporting goods and device for transporting goods
CN111768629B (en) Vehicle scheduling method, device and system
CN107036618A (en) A kind of AGV paths planning methods based on shortest path depth optimization algorithm
CN110807236A (en) Warehouse logistics simulation system based on multiple robots
CN110097231A (en) Multiple target objects stream scheduling method and device, logistics system and computer-readable medium
CN114331257A (en) Logistics transportation loading management method, device, equipment and storage medium
Shetty et al. Route optimization for warehouse order picking operations via vehicle routing and simulation
Gotthardt et al. Digitalized milk-run system for a learning factory assembly line
CN114372758A (en) Production raw material storage method and system
CN114111830A (en) Path planning method and device based on AI model
CN114911238A (en) Unmanned mine car cooperative control method and system
KR102499502B1 (en) Scheduling optimization method of vehicles and drones using delivery position clustering of parallel delivery using vehicles and drones and the system thereof
CN116522801B (en) Layout simulation method and device for logistics system
Ruta et al. Diversified gradient boosting ensembles for prediction of the cost of forwarding contracts
CN115907374A (en) Intelligent dispatching method, device and equipment for vehicle of distribution center and storage medium
CN113687651B (en) Path planning method and device for dispatching vehicles on demand
CN115660524A (en) E-commerce logistics monitoring method and system
Xidias et al. SERobWaS: a support environment for a robot-based warehousing system
CN111445100B (en) Vehicle and goods matching method based on self-adaptive time window under limited transport capacity
CN112001646A (en) Material scheduling method and device, storage medium and electronic equipment
CN112734111A (en) AGV dynamic time estimation method for horizontal transportation task
CN115730879B (en) Commodity logistics transportation distribution method based on zero storage
Simonetto et al. Introduction to Material Feeding 4.0: strategic, tactical, and operational impact
CN117389309B (en) Control method, device, equipment and storage medium for auxiliary maintenance of unmanned aerial vehicle

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant