CN116184892A - AI identification control method and system for robot object taking - Google Patents

Abstract

The present invention provides an AI recognition control method and system for robot fetching objects, wherein the method includes: obtaining fetching instructions; analyzing the fetching instructions to determine the information of the object to be fetched; obtaining a perimeter image; Determine the position of the target object; determine the gripping point based on the position and the structure of the robot's pick-up device; control the robot to pick up the object based on the gripping point. The AI recognition control method for robot picking objects of the present invention realizes the grasping of different types of objects.

Description

AI recognition control method and system for robot fetching objects

technical field

The present invention relates to the technical field of intelligent control, in particular to an AI identification control method and system for picking up objects by a robot.

Background technique

At present, robots have been widely used in automated factories, unmanned storage and other fields; however, existing robots generally only grasp a single type of object when fetching objects, and cannot accurately grasp multiple types of different objects.

Contents of the invention

One of the objectives of the present invention is to provide an AI recognition control method for a robot to pick up objects, so as to realize the grabbing of different types of objects.

An AI recognition control method for a robot to pick up objects provided by an embodiment of the present invention includes:

Obtain fetch instruction;

Analyze the fetching instruction to determine the information of the object to be fetched;

Get the perimeter image;

Determine the position of the target object from the perimeter image based on the AI recognition model;

Determine the gripping point based on the position and the structure of the robot's fetching device;

Based on the gripping point, the robot is controlled to pick up the object.

Preferably, obtaining fetch instructions includes:

Obtain fetch request information from the server;

Analyze the fetching request information and generate fetching instructions;

or,

Obtain the first audio collected by the audio collection module arranged in the fetching area;

Analyzing the first audio to determine whether to enter the fetch command generation mode;

When entering the fetching instruction generation mode, acquire the first image in the fetching area collected by the first image acquisition module set in the fetching area;

Analyzing the first image to determine whether there is a first type of item in the first image;

When there is a first type of item, extracting a first area image of the first type of item from the first image and generating a retrieval instruction based on the extracted first area image;

Wherein, the first category of items includes: a form or an identifier for authorizing the retrieval of items.

Preferably, when parsing the first image, when it is determined that the first type of item does not exist in the first image, it also includes:

determine whether a second class item is present;

When there is a second type of item, build a list of items to be picked up and output the first inquiry information through the touch screen arranged in the pick-up area; the first inquiry information includes: whether to pick up the item and whether to perform identity verification;

When receiving the affirmative feedback of the corresponding inquiry information input by the pick-up personnel through the touch screen, the second image of the pick-up personnel collected by the second image acquisition module arranged in the pick-up area;

Based on the second image, verify the authority of the person taking the object;

When the verification is passed, the second inquiry information is output through the touch screen; the second inquiry information includes: the quantity of each item to be picked up on the list of items to be picked up;

The input information corresponding to the query information input by the picker through the touch screen is received, and the pick-up instruction is generated based on the input information and the list of items to be picked up.

Preferably, the position of the target object is determined from the perimeter image based on the AI recognition model, including:

Determining a plurality of standard images corresponding to the object to be obtained from a preset standard image library based on the information of the object to be obtained;

performing edge extraction and segmentation on the perimeter image, and determining a plurality of second region images;

Input the standard image and the second area image into the AI recognition model to determine whether the two match;

determining a matching standard image and a second region image;

Taking the object corresponding to the second region image matching the standard image as the target object;

determining a first direction vector corresponding to the center of the perimeter image containing the target object;

determining a second direction vector corresponding to the center of the second area image based on the position of the second area image in the perimeter image and the first direction vector;

Determine the distance and pose of the target object based on the object distance and pose recognition library corresponding to the second area image and the preset standard image;

Based on the second direction vector, the distance and the pose, the position of the target object is determined.

Preferably, based on the position and the structure of the robot's fetching device, determine the gripping point, including:

Calling the library for determining the gripping point corresponding to the structure of the fetching device;

Quantify the position based on a preset first quantization model, and determine a first quantization parameter

Quantify the information of the object to be acquired based on a preset second quantization model, and determine a second quantization parameter;

determining a parameter set based on the first quantization parameter and the second quantization parameter;

Determine the library based on the parameter set and the gripping point, and determine the gripping point.

The present invention also provides an AI recognition control system for a robot to pick up objects, including:

The first obtaining module is used to obtain the retrieval instruction;

The analysis module is used to analyze the fetching instruction and determine the information of the object to be fetched;

The second acquisition module is used to acquire the perimeter image;

The first determination module is used to determine the position of the target object from the perimeter image based on the AI recognition model;

The second determination module is used to determine the gripping point based on the position and the structure of the object fetching device of the robot;

The control module is used to control the robot to pick objects based on the gripping point.

Preferably, the first acquiring module acquires the fetching instruction and performs the following operations:

Obtain fetch request information from the server;

Analyze the fetching request information and generate fetching instructions;

or,

Obtain the first audio collected by the audio collection module arranged in the fetching area;

Analyzing the first audio to determine whether to enter the fetch command generation mode;

When entering the fetching instruction generation mode, acquire the first image in the fetching area collected by the first image acquisition module set in the fetching area;

Analyzing the first image to determine whether there is a first type of item in the first image;

When there is a first type of item, extracting a first area image of the first type of item from the first image and generating a retrieval instruction based on the extracted first area image;

Wherein, the first category of items includes: a form or an identifier for authorizing the retrieval of items.

Preferably, when parsing the first image, when it is determined that the first type of item does not exist in the first image, the first acquisition module also performs the following operations:

determine whether a second class item is present;

When there is a second type of item, build a list of items to be picked up and output the first inquiry information through the touch screen arranged in the pick-up area; the first inquiry information includes: whether to pick up the item and whether to perform identity verification;

When receiving the affirmative feedback of the corresponding inquiry information input by the pick-up personnel through the touch screen, the second image of the pick-up personnel collected by the second image acquisition module arranged in the pick-up area;

Based on the second image, verify the authority of the person taking the object;

When the verification is passed, the second inquiry information is output through the touch screen; the second inquiry information includes: the quantity of each item to be picked up on the list of items to be picked up;

The input information corresponding to the query information input by the picker through the touch screen is received, and the pick-up instruction is generated based on the input information and the list of items to be picked up.

Preferably, the first determination module determines the position of the target object from the perimeter image based on the AI recognition model, and performs the following operations:

Determining a plurality of standard images corresponding to the object to be obtained from a preset standard image library based on the information of the object to be obtained;

performing edge extraction and segmentation on the perimeter image, and determining a plurality of second area images;

Input the standard image and the second area image into the AI recognition model to determine whether the two match;

determining a matching standard image and a second region image;

Taking the object corresponding to the second region image matching the standard image as the target object;

determining a first direction vector corresponding to the center of the perimeter image containing the target object;

determining a second direction vector corresponding to the center of the second area image based on the position of the second area image in the perimeter image and the first direction vector;

Determine the distance and pose of the target object based on the object distance and pose recognition library corresponding to the second area image and the preset standard image;

Based on the second direction vector, the distance and the pose, the position of the target object is determined.

Preferably, the second determining module determines the gripping point based on the position and the structure of the object picking device of the robot, and performs the following operations:

Calling the library for determining the gripping point corresponding to the structure of the fetching device;

Quantify the position based on a preset first quantization model, and determine a first quantization parameter

Quantify the information of the object to be acquired based on a preset second quantization model, and determine a second quantization parameter;

determining a parameter set based on the first quantization parameter and the second quantization parameter;

Determine the library based on the parameter set and the gripping point, and determine the gripping point.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

The technical solutions of the present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

Description of drawings

The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the description, and are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the attached picture:

FIG. 1 is a schematic diagram of an AI recognition control method for a robot to fetch objects in an embodiment of the present invention;

Fig. 2 is a schematic diagram of an AI recognition control system for a robot to pick up objects in an embodiment of the present invention.

Detailed ways

The preferred embodiments of the present invention will be described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described here are only used to illustrate and explain the present invention, and are not intended to limit the present invention.

An embodiment of the present invention provides an AI recognition control method for a robot to fetch objects, as shown in FIG. 1 , including:

Step S1: Obtaining an object fetching instruction;

Step S2: Analyzing the fetching instruction to determine the information of the object to be fetched;

Step S3: Acquire the perimeter image;

Step S4: Determine the position of the target object from the perimeter image based on the AI recognition model;

Step S5: Determine the gripping point based on the position and the structure of the object fetching device of the robot;

Step S6: Based on the gripping point, control the robot to pick up the object.

The working principle and beneficial effects of the above-mentioned technical scheme are:

First of all, the robot obtains the fetching command from the outside world. The fetching command can be sent by the user through the terminal, or it can be automatically generated by the robot according to the external situation; through the analysis of the fetching command, the object to be fetched corresponding to the fetching command can be determined , mainly through the information of the object to be fetched (for example: three-dimensional information such as length, width, height, color, and picture, etc.); through the AI recognition model, according to the information of the object to be fetched, it is determined whether there is an object to be fetched from the perimeter image of the robot. When it is determined that it exists , determine the object to be fetched as the target object, and then further determine the position of the target object (area occupied in space); according to the position of the target object and the specific structure of the fetching device configured by the robot, determine the gripping point; then control the alignment of the grabbing point and the gripping point on the fetching device to realize the robot fetching; wherein, when determining the gripping point based on the position and the structure of the fetching device, the gripping point for objects of different shapes The picking points are different, for example: the picking points of a spherical object are three points evenly distributed on the circumference of the horizontal plane; the picking points of a rectangular object are two symmetrical points distributed on two symmetrical planes . The AI recognition control method of the robot fetching object of the present invention accurately realizes the grasping of many kinds of different objects.

In one embodiment, obtaining the retrieval instruction includes:

Obtain fetch request information from the server;

Analyze the fetching request information and generate fetching instructions;

or,

Obtain the first audio collected by the audio collection module arranged in the fetching area;

Analyzing the first audio to determine whether to enter the fetch command generation mode;

When entering the fetching instruction generation mode, acquire the first image in the fetching area collected by the first image acquisition module set in the fetching area;

Analyzing the first image to determine whether there is a first type of item in the first image;

When there is a first type of item, extracting a first area image of the first type of item from the first image and generating a retrieval instruction based on the extracted first area image;

Wherein, the first category of items includes: a form or an identifier for authorizing the retrieval of items.

The working principle and beneficial effects of the above-mentioned technical scheme are:

This embodiment provides two ways to obtain fetching instructions, one is to obtain from the server, which is mainly used in the remote control of the user and the remote automatic control of the server. The user communicates with the server through the mobile terminal and selects the desired command on the fetching interface According to the item selected by the user, the server generates a fetching request information based on the item selected by the user and then sends it to the robot side. The control module of the robot generates a fetching instruction by analyzing the fetching request information. The other is to automatically generate pick-up instructions based on actual situation analysis, which is mainly used in unmanned warehouse pick-up scenarios, for example: the pick-up personnel hold the authorized pick-up form (the pick-up list approved by the superior ), the pick-up personnel put the pick-up list in the pick-up area, and then say the corresponding offending keywords such as "pick-up", and the robot collects the first word containing the trigger keyword through the audio collection module set in the pick-up area. During the audio, it enters the fetch command generation mode. In this mode, the robot collects the first image in the fetch area through the first image acquisition module of the fetch area, and analyzes the fetch list in the first image to determine the fetch order. Extract the text on the list, and then determine the information of the items to be picked up (item number, shape, name, quantity, etc.), and then automatically generate the pick-up instruction according to the information of the items to be picked up; in addition, the items placed in the pick-up area can also be It is an identifier, and the identifier corresponds to a unique item; for example: a red long sign represents item A, a yellow long sign represents item B..., and the number of long signs represents the number of items to be retrieved. Furthermore, the identifier includes: a first display area and a second display area; the first display area displays a two-dimensional code generated from the information of the item to be picked up; the second display area displays a two-dimensional code generated from the information of the person picking up the object ; The first display area and the second display area are photographed by the first image acquisition module, the information of the items to be picked up and the information of the pick-up personnel are extracted, and the pick-up information is associated with the pick-up personnel to ensure unmanned storage and other management specifications and safety.

In the unmanned storage application scenario, after obtaining the fetch command and before obtaining the perimeter image, it is necessary to determine the point where the robot needs to move. In the warehouse, the location of the warehouse is associated with the point where the robot can stay; it is necessary to determine the location of the item to be picked up in the warehouse, and then the robot moves to the corresponding point where the robot can stay; The image acquisition module collects the perimeter image; then, based on the AI recognition model, the position of the target object is determined from the perimeter image; based on the position and the structure of the robot's object picking device, the gripping point is determined; based on the gripping point, the robot is controlled to carry out Pick-up: The picked-up goods are transported to the pick-up area by an automatic mobile trolley. Among them, the point where the robot can stay is on the aisle between storage locations.

In addition, it can also be applied to the storage scene of unmanned storage. The image acquisition module in the storage area collects the storage order placed by the staff, establishes the pick-up instruction corresponding to the storage, and the robot stores the item corresponding to the storage area. Take images, then determine the corresponding relationship between each item in the item storage area and the storage order, determine the storage location corresponding to each item, and then group the items according to the corresponding relationship between the storage location and the point where they can stay. Sort each group according to the distance between the point and the storage area of the item, the order with the longest distance is first, and the order with the shortest distance is last; the items in each group are extracted in turn, and placed in the automatic movement corresponding to each corresponding point. On the trolley, when the items in the group in the storage area are taken out or the automatic mobile trolley reaches the maximum load, it stops picking up the objects, and then each automatic mobile trolley moves to the corresponding point; the robot starts from the farthest point. Pick up items on the automatic mobile trolley and place them in the warehouse. When the items on the automatic mobile trolley at the nearest point are placed, the next round of grabbing items in the item storage area is carried out; the automatic storage of the robot is realized.

In one embodiment, when parsing the first image, when it is determined that the first type of item does not exist in the first image, it further includes:

Determine whether there is a second type of item; the second type of item is the same object or shell as the item to be picked;

When there is a second type of item, build a list of items to be picked up and output the first inquiry information through the touch screen arranged in the pick-up area; the first inquiry information includes: whether to pick up the item and whether to perform identity verification;

When receiving the affirmative feedback of the corresponding inquiry information input by the pick-up personnel through the touch screen, the second image of the pick-up personnel collected by the second image acquisition module arranged in the pick-up area;

Based on the second image, verify the authority of the picker; mainly determine whether the picker has the authority to receive the item to be picked;

When the verification is passed, the second inquiry information is output through the touch screen; the second inquiry information includes: the quantity of each item to be picked up on the list of items to be picked up;

The input information corresponding to the query information input by the picker through the touch screen is received, and the pick-up instruction is generated based on the input information and the list of items to be picked up.

The working principle and beneficial effects of the above-mentioned technical scheme are:

The main application scenario of this embodiment is when the pick-up personnel are maintenance personnel, and when the maintenance personnel are in urgent need of accessories and have not obtained the signed pick-up list or other certificates, they directly place the accessories that need to be replaced (disassembled accessories) in the pick-up box. After the identity verification and the quantity input by the picker, a pick-up instruction is generated to ensure pick-up in emergency situations.

In one embodiment, determining the position of the target object from the perimeter image based on the AI recognition model includes:

Based on the information of the object to be picked up, from the preset standard image library, determine multiple standard images corresponding to the object to be picked up; multiple standard images correspond to images of the item to be picked up from different angles; for example: from the front, side and diagonal Images of the items to be picked up in the same direction;

Perform edge extraction and segmentation on the perimeter image to determine a plurality of second area images; through edge extraction and segmentation, segment the items in the perimeter image, one item corresponds to one second area image;

Input the standard image and the second area image into the AI recognition model to determine whether the two match;

determining a matching standard image and a second region image;

Taking the object corresponding to the second region image matching the standard image as the target object;

Determine the first direction vector corresponding to the center of the perimeter image containing the target object; in general, when the perimeter image is taken, the direction vector of the central axis of the image acquisition module in the space coordinate system is the first direction vector corresponding to the center of the perimeter image a direction vector;

Based on the position of the second area image in the perimeter image and the first direction vector, determine the second direction vector corresponding to the center of the second area image; mainly according to the deviation between the center of the second area image and the center of the perimeter image, determine A deflection vector, which is added to the first direction vector to obtain the second direction vector;

Based on the object distance and pose recognition library corresponding to the second area image and the preset standard image, determine the distance and pose of the target object; the specific operation is: determine the location of the object in the second area image and the object in the standard image The ratio of the area occupied by the object, the standard image is scaled to the same or similar size as the second area image, and then the second area image and the standard image are divided into blocks using the same block rules to determine the similarity corresponding to each block. Then arrange the similarity in order to form a similarity set sum; merge the ratio and similarity set into an analysis set; match the analysis set with the standard set in the library, and extract the distance and pose parameters corresponding to the matched standard set; The attitude parameter indicates the deflection of the object relative to the reference position defined in space; for example: define the front flat of the object as the reference position; not every object is in the front flat state when it is actually placed, and the reference position is represented by the pose The deflection of the object can more accurately represent the position of the object;

Based on the second direction vector, the distance and the pose, the position of the target object is determined. Elements of the position include a second direction vector, a distance, a pose, and the like.

In one embodiment, determining the gripping point based on the position and the structure of the object fetching device of the robot includes:

Calling the library for determining the gripping point corresponding to the structure of the fetching device;

Quantify the position based on a preset first quantization model, and determine a first quantization parameter

Quantify the information of the object to be acquired based on a preset second quantization model, and determine a second quantization parameter;

determining a parameter set based on the first quantization parameter and the second quantization parameter;

Determine the library based on the parameter set and the gripping point, and determine the gripping point.

The working principle and beneficial effects of the above-mentioned technical scheme are:

Different pick-up structures correspond to setting different pick-up point determination libraries; pick-up structures include:

Through the comprehensive analysis of the structure of the pick-up device and the shape and position of the object, etc., the pick-up point is adaptively selected to ensure the stability of the pick-up, and the robot pick-up can be adapted to the pick-up of different objects, which improves the pick-up efficiency of the robot. applicability. Wherein, quantify the parameter information of the position by the first quantization model, determine at least one first quantization parameter, quantify the information of the object to be fetched by the second quantization model, and determine at least one second quantization parameter; then construct the parameter set, Match the parameter set with the analysis set of the gripping point determination library, and extract the matching analysis set corresponding to the associated gripping point determination set; analyze the gripping point determination set to determine the deviation of the center coordinates of the object in the remaining positions, and then determine the gripping point The coordinates of the robot realize the gripping of the robot; wherein, the second quantitative parameter includes: a parameter corresponding to the weight of the object, a parameter corresponding to the three-dimensional size of the object, and the like.

The present invention also provides an AI recognition control system for a robot to pick up objects, as shown in Figure 2, including:

The first obtaining module 1 is used to obtain the retrieval instruction;

Parsing module 2, used for parsing the fetching instruction, and determining the information of the object to be fetched;

The second acquiring module 3 is used to acquire the perimeter image;

The first determination module 4 is used to determine the position of the target object from the perimeter image based on the AI recognition model;

The second determining module 5 is used to determine the gripping point based on the position and the structure of the object picking device of the robot;

The control module 6 is configured to control the robot to pick up objects based on the gripping point.

In one embodiment, the first acquiring module 1 acquires the fetching instruction and performs the following operations:

Obtain fetch request information from the server;

Analyze the fetching request information and generate fetching instructions;

or,

Obtain the first audio collected by the audio collection module arranged in the fetching area;

Analyzing the first audio to determine whether to enter the fetch command generation mode;

When entering the fetching instruction generation mode, acquire the first image in the fetching area collected by the first image acquisition module set in the fetching area;

Analyzing the first image to determine whether there is a first type of item in the first image;

When there is a first type of item, extracting a first area image of the first type of item from the first image and generating a retrieval instruction based on the extracted first area image;

Wherein, the first category of items includes: a form or an identifier for authorizing the retrieval of items.

In one embodiment, when parsing the first image, when it is determined that the first type of item does not exist in the first image, the first acquisition module 1 also performs the following operations:

determine whether a second class item is present;

When there is a second type of item, build a list of items to be picked up and output the first inquiry information through the touch screen arranged in the pick-up area; the first inquiry information includes: whether to pick up the item and whether to perform identity verification;

When receiving the affirmative feedback of the corresponding inquiry information input by the pick-up personnel through the touch screen, the second image of the pick-up personnel collected by the second image acquisition module arranged in the pick-up area;

Based on the second image, verify the authority of the person taking the object;

When the verification is passed, the second inquiry information is output through the touch screen; the second inquiry information includes: the quantity of each item to be picked up on the list of items to be picked up;

The input information corresponding to the query information input by the picker through the touch screen is received, and the pick-up instruction is generated based on the input information and the list of items to be picked up.

In one embodiment, the first determination module 4 determines the position of the target object from the perimeter image based on the AI recognition model, and performs the following operations:

Determining a plurality of standard images corresponding to the object to be obtained from a preset standard image library based on the information of the object to be obtained;

performing edge extraction and segmentation on the perimeter image, and determining a plurality of second area images;

Input the standard image and the second area image into the AI recognition model to determine whether the two match;

determining a matching standard image and a second region image;

Taking the object corresponding to the second region image matching the standard image as the target object;

determining a first direction vector corresponding to the center of the perimeter image containing the target object;

determining a second direction vector corresponding to the center of the second area image based on the position of the second area image in the perimeter image and the first direction vector;

Determine the distance and pose of the target object based on the object distance and pose recognition library corresponding to the second area image and the preset standard image;

Based on the second direction vector, the distance and the pose, the position of the target object is determined.

In one embodiment, the second determination module 5 determines the gripping point based on the position and the structure of the object fetching device of the robot, and performs the following operations:

Calling the library for determining the gripping point corresponding to the structure of the fetching device;

Quantify the position based on a preset first quantization model, and determine a first quantization parameter

Quantify the information of the object to be acquired based on a preset second quantization model, and determine a second quantization parameter;

determining a parameter set based on the first quantization parameter and the second quantization parameter;

Determine the library based on the parameter set and the gripping point, and determine the gripping point.

Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims (10)

1. A kind of AI recognition control method of robot fetching, it is characterized in that, comprising: Obtain fetch instruction; Analyze the fetching instruction to determine the information of the object to be fetched; Get the perimeter image; Determining the position of the target object from the perimeter image based on the AI recognition model; Determine the gripping point based on the position and the structure of the fetching device of the robot; Based on the gripping point, the robot is controlled to fetch objects. 2. The AI recognition control method of robot fetching as claimed in claim 1, wherein said acquiring fetching instructions includes: Obtain fetch request information from the server; Analyzing the retrieval request information to generate the retrieval instruction; or, Obtain the first audio collected by the audio collection module arranged in the fetching area; Analyzing the first audio to determine whether to enter the fetch instruction generation mode; When entering the fetching instruction generation mode, acquire the first image in the fetching area collected by the first image acquisition module set in the fetching area; Analyzing the first image to determine whether there is a first type of item in the first image; When the first type of item exists, extracting a first area image of the first type of item from the first image and generating the retrieval instruction based on the extracted first area image; Wherein, the first type of item includes: a form or an identifier for authorizing the retrieval of the item. 3. The AI recognition control method for robot fetching objects according to claim 2, wherein when analyzing the first image, when it is determined that the first type of item does not exist in the first image, further comprising: : determine whether a second class item is present; When there is a second type of item, build a list of items to be picked up and output the first inquiry information through the touch screen arranged in the picking area; the first inquiry information includes: whether to carry out picking and whether to perform identity verification; When receiving the affirmative feedback corresponding to the inquiry information input by the pick-up personnel through the touch screen, the second image of the pick-up personnel collected by the second image acquisition module arranged in the pick-up area; Based on the second image, verify the authority of the fetching person; When the verification is passed, the second inquiry information is output through the touch screen; the second inquiry information includes: the quantity of each item to be picked up on the item list to be picked up; receiving the input information corresponding to the query information input by the pick-up personnel through the touch screen, and generating the pick-up instruction based on the input information and the list of items to be picked. 4. The AI recognition control method of robot fetching as claimed in claim 1, wherein the determining the position of the target object from the perimeter image based on the AI recognition model comprises: Determining a plurality of standard images corresponding to the object to be obtained from a preset standard image library based on the information of the object to be obtained; performing edge extraction and segmentation on the perimeter image to determine a plurality of second area images; Input the standard image and the second area image into the AI recognition model, and determine whether the two match; determining matching of the standard image and the second area image; taking the object corresponding to the second region image that matches the standard image as the target object; determining a first direction vector corresponding to the center of the perimeter image containing the target object; determining a second direction vector corresponding to the center of the second area image based on the position of the second area image in the perimeter image and the first direction vector; Determine the distance and pose of the target object based on the second area image and the preset object distance and pose recognition library corresponding to the standard image; A position of the target object is determined based on the second direction vector, the distance, and the pose. 5. The AI recognition control method for robot fetching as claimed in claim 4, wherein the determination of the gripping point based on the position and the structure of the fetching device of the robot includes: Calling the library for determining the gripping point corresponding to the structure of the fetching device; Quantify the position based on a preset first quantization model, and determine a first quantization parameter Quantify the information of the object to be acquired based on a preset second quantization model, and determine a second quantization parameter; determining a parameter set based on the first quantization parameter and the second quantization parameter; The gripping point is determined based on the parameter set and the gripping point determination library. 6. An AI recognition control system for robot fetching, characterized in that it comprises: The first obtaining module is used to obtain the retrieval instruction; The analysis module is used to analyze the fetching instruction and determine the information of the object to be fetched; The second acquisition module is used to acquire the perimeter image; The first determination module is used to determine the position of the target object from the perimeter image based on the AI recognition model; The second determination module is used to determine the gripping point based on the position and the structure of the object fetching device of the robot; A control module, configured to control the robot to pick up objects based on the gripping point. 7. The AI recognition control system for robot fetching as claimed in claim 6, wherein the first acquisition module obtains the fetching instruction and performs the following operations: Obtain fetch request information from the server; Analyzing the retrieval request information to generate the retrieval instruction; or, Obtain the first audio collected by the audio collection module arranged in the fetching area; Analyzing the first audio to determine whether to enter the fetch instruction generation mode; When entering the fetching instruction generation mode, acquire the first image in the fetching area collected by the first image acquisition module set in the fetching area; Analyzing the first image to determine whether there is a first type of item in the first image; When the first type of item exists, extracting a first area image of the first type of item from the first image and generating the retrieval instruction based on the extracted first area image; Wherein, the first type of item includes: a form or an identifier for authorizing the retrieval of the item. 8. The AI recognition control system for robot fetching objects according to claim 7, wherein when analyzing the first image, when it is determined that the first type of item does not exist in the first image, the The first acquisition module also performs the following operations: determine whether a second class item is present; When there is a second type of item, build a list of items to be picked up and output the first inquiry information through the touch screen arranged in the picking area; the first inquiry information includes: whether to carry out picking and whether to perform identity verification; When receiving the affirmative feedback corresponding to the inquiry information input by the pick-up personnel through the touch screen, the second image of the pick-up personnel collected by the second image acquisition module arranged in the pick-up area; Based on the second image, verify the authority of the fetching person; When the verification is passed, the second inquiry information is output through the touch screen; the second inquiry information includes: the quantity of each item to be picked up on the item list to be picked up; receiving the input information corresponding to the query information input by the pick-up personnel through the touch screen, and generating the pick-up instruction based on the input information and the list of items to be picked. 9. The AI recognition control system for robot fetching according to claim 6, wherein the first determination module determines the position of the target object from the perimeter image based on the AI recognition model, and performs the following operations : Determining a plurality of standard images corresponding to the object to be obtained from a preset standard image library based on the information of the object to be obtained; performing edge extraction and segmentation on the perimeter image to determine a plurality of second area images; Input the standard image and the second area image into the AI recognition model, and determine whether the two match; determining matching of the standard image and the second area image; taking the object corresponding to the second region image that matches the standard image as the target object; determining a first direction vector corresponding to the center of the perimeter image containing the target object; determining a second direction vector corresponding to the center of the second area image based on the position of the second area image in the perimeter image and the first direction vector; Determine the distance and pose of the target object based on the second area image and the preset object distance and pose recognition library corresponding to the standard image; A position of the target object is determined based on the second direction vector, the distance, and the pose. 10. The AI recognition control system for robot fetching as claimed in claim 9, wherein the second determination module determines the gripping point based on the position and the structure of the fetching device of the robot, and performs the following operations: Calling the library for determining the gripping point corresponding to the structure of the fetching device; Quantify the position based on a preset first quantization model, and determine a first quantization parameter Quantify the information of the object to be acquired based on a preset second quantization model, and determine a second quantization parameter; determining a parameter set based on the first quantization parameter and the second quantization parameter; The gripping point is determined based on the parameter set and the gripping point determination library.

Images