CN118080359A - Warehouse article letter sorting arm with AI discernment - Google Patents

Abstract

The invention discloses a warehouse article sorting mechanical arm with AI identification, which comprises: the device comprises a conveyor belt device, a front scanning module, a sorting mechanical arm and an intelligent clamping mechanism; the two sides of the conveying belt device are respectively provided with a conveying area and a transferring area, and the sorting mechanical arm is arranged at one side of the transferring area of the conveying belt device and is used for transferring and storing articles on the conveying belt device; a front scanning module is arranged above one side of the conveyor belt device, which is close to the conveying area, and the front scanning module respectively adopts a high-definition camera and a bar code reader to acquire article image data and label data, and comprehensively analyzes and processes the article image data and the label data to acquire article characteristic image data; the article image data is image data for articles to be sorted.

Description

A warehouse item sorting robot with AI recognition

Technical Field

The present invention belongs to the technical field of intelligent sorting, and in particular is a storage article sorting robot arm with AI recognition.

Background technique

At present, with the continuous development of intelligent technology, the application of intelligent cargo sorting equipment is becoming more and more widespread. Common intelligent sorting manipulators generally use manipulators on conveyor belts to perform sorting operations. This sorting method has a higher conveying and sorting efficiency because the conveyor belt needs to be transported and sorted during the sorting process. For example, the invention patent with patent number CN107352206B uses a digital control unit that can intelligently identify the top surface posture information of the cargo packaging box, obtain the position deviation, correct the posture of the manipulator body, and then grab it. Although it realizes intelligent recognition and grabbing actions, it has the following defects in use:

1. The grabbing action used is single, which makes it difficult to sort and grab different types of items;

2. Unable to stably grasp irregular objects;

3. Items cannot be sorted based on their center of gravity, and safety cannot be guaranteed during grabbing and transportation.

Therefore, it is necessary to provide a warehouse item sorting robot arm with AI recognition to solve the problems raised in the above background technology.

Summary of the invention

To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a storage article sorting robot arm with AI recognition, comprising: a conveyor belt device, a front scanning module, a sorting robot arm and an intelligent clamping mechanism; the two sides of the conveyor belt device are respectively set as a delivery area and a transfer area, and the sorting robot arm is set on the transfer area side of the conveyor belt device, and is used to transfer and store the articles on the conveyor belt device; a front scanning module is installed on the upper side of the conveyor belt device close to the delivery area, and the front scanning module uses a high-definition camera and a barcode reader to obtain article image data and label data, and comprehensively analyzes and processes the article image data and label data to obtain article feature image data; the article image data is image data for the article to be sorted;

The conveyor belt device is externally provided with a data analysis system, and the data analysis system uses a preset classification database model and combines the label data to perform feature recognition on the feature image data of the object and obtains a recognition result;

The sorting robot arm is provided with a sorting module, and the sorting module indexes the corresponding storage location information and the location information of the center of gravity of the current item based on the above recognition result. The intelligent gripping mechanism analyzes the optimal gripping action of the current item based on the location information of the center of gravity of the item and in combination with the item image data, and adopts the optimal gripping action to sort and transfer the item;

The sorting robot arm is equipped with a high-definition scanner, which obtains the spatial position information of the items to be sorted on the current conveyor belt device and outputs the precise coordinates of the center of gravity of the items;

The optimal clamping action includes a contact positioning action and a moving action.

Further, as a preferred embodiment, the intelligent clamping mechanism comprises: an upper machine plate, four connecting rods are vertically arranged below the upper machine plate, a positioning plate is fixed below the connecting rod, a machine base is vertically slidably arranged between the upper machine plate and the positioning plate, and the machine base is slidably connected to each of the connecting rods;

A plurality of connecting arms are symmetrically distributed on the positioning plate, one end of each of the connecting arms is rotatably connected to the positioning plate, and a plurality of adjusting rods corresponding to the connecting arms are distributed on the machine base, one end of each of the adjusting rods is rotatably connected to the connecting arm;

A hydraulic rod is fixed on the upper machine plate, and one end of the hydraulic rod is connected to the machine base;

An L-shaped driving arm is also fixed on each of the connecting arms, and the L-shaped driving arm is driven by a hydraulic cylinder to fold and deform;

A clamping assembly is fixed to the end of the L-shaped driving arm.

Furthermore, as a preference, a locator is provided in the middle of the upper plate on the intelligent gripping mechanism, and the locator collects and outputs the precise coordinates of the upper plate in three-dimensional space in real time. The sorting module on the sorting robot arm plans the optimal motion trajectory of the sorting robot arm based on the coordinate difference between the acquired spatial coordinates of the center of gravity of the item and the spatial coordinates of the upper plate, so that the intelligent gripping mechanism is suspended directly above the center of gravity of the item before sorting the item.

Further, as a preference, the adjustment arm adopts a two-stage telescopic adjustment mechanism, each of the adjustment rods is independently controlled, and each of the L-shaped driving arms is independently controlled;

The contact positioning action of the intelligent clamping mechanism is analyzed as follows: the adjusting rods on each of the connecting arms perform a telescopic action so that the clamping assembly contacts the surface of the object, and the hydraulic cylinder on the L-shaped driving arm performs a telescopic action adjustment. At this time, the clamping assembly contacts and positions the comprehensive optimal clamping point on the surface of the object, and the hydraulic rod controls the sliding adjustment of the driving base and applies contact pressure to the object through the clamping assembly.

Furthermore, as a preference, a pressure plate is provided in the middle portion of the lower end surface of the positioning plate through an electric telescopic rod, and the front scanning module compares the image data and label data of adjacent items on the conveyor belt device. When the image data and label data of adjacent items are the same, the intelligent clamping mechanism adopts the optimal clamping action to post-stack the items on one side close to the transfer area, and subsequently performs multiple simultaneous transfers.

Further, as a preference, the clamping assembly comprises: an outer clamping frame, inside which is rotatably connected to a main shaft via a bearing, one end of the main shaft being sleeved with a transmission tooth, a micro motor being mounted on the outer clamping frame, an output end of the micro motor being meshed with the transmission tooth for transmission; a guide plate being coaxially fixed on the main shaft, a plurality of abutment brackets being laterally equidistantly distributed on the guide plate, and an anti-slip lining being covered on the guide plate.

Further, as a preference, an annular pressure chamber is provided in the guide disc, hydraulic oil is stored in the pressure chamber, a feeler rod is slidably connected to the inner sealing of the abutment bracket, a sleeve is fixed to the end of the feeler rod, a sealing joint is connected to one side of the abutment bracket, the sealing joint is embedded and fixed on the guide disc, and is connected to the pressure chamber;

A guide cavity is also provided in the guide plate, and the guide cavity is connected with the pressure cavity through an inner channel. A piston is slidably connected in the guide cavity through a fine-tuning rod, and a support spring is connected to one side of the piston.

Further, as a preference, a displacement sensor is also provided at the end of each of the touch rods, and a signal receiver is provided on the guide plate, and the signal receiver obtains displacement information of the touch rod during operation;

The intelligent gripping mechanism is provided with a stroke monitoring module, the signal receiver converts the displacement information into an electrical signal, and sends the signal to the stroke monitoring module through a built-in data interface, the stroke monitoring module collects the first displacement characteristic data of each feeler rod on the guide plate in the current working state, obtains the mapping point corresponding to each of the feelers, and constructs a first contact surface three-dimensional model;

The travel monitoring module includes an early warning model, which comprehensively analyzes the stress points of the three-dimensional model of the first contact surface in each of the clamping components through a preset early warning model and outputs a first risk level.

Furthermore, preferably, when the first risk level is higher than a threshold value, the micromotor drives the guide plate to deflect, and the stroke monitoring module collects the real-time displacement characteristic data of each touch rod on the guide plate, and constructs a real-time three-dimensional model of the contact surface, and comprehensively analyzes the force points of the real-time three-dimensional model of the contact surface in each clamping component through a preset early warning model and outputs the real-time risk level.

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

1. The intelligent gripping mechanism used in the present invention can adjust the gripping work in a targeted manner according to the identified feature image data of the object, and use the optimal gripping action to sort the objects, thereby improving the safety of the objects during transportation;

2. The gripping assembly mainly used in the present invention can stably contact the irregular surface of the object and judge the risk of the goods falling, so as to adopt appropriate movement actions to store and stack the objects according to the risk of falling, with high stability.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a schematic diagram of the structure of the present invention;

FIG2 is a schematic diagram of the structure of the intelligent clamping mechanism of the present invention;

FIG3 is a cross-sectional view of the intelligent clamping mechanism of the present invention;

FIG4 is a schematic diagram of the structure of the clamping assembly of the present invention;

FIG5 is an enlarged schematic diagram of point A in FIG4 ;

FIG6 is a schematic diagram of the structure of the guide plate of the present invention;

FIG7 is a flowchart of AI recognition and sorting of the present invention;

In the figure: 1. Sorting robot arm; 11. Conveyor belt device; 12. Front scanning module; 2. Intelligent clamping mechanism; 21. Upper plate; 22. Positioning plate; 23. Connecting rod; 24. Machine base; 25. Connecting arm; 26. Adjusting rod; 27. Hydraulic rod; 28. L-shaped driving arm; 29. Pressure plate; 3. Clamping assembly; 31. External clamping frame; 32. Spindle; 33. Anti-slip lining; 4. Guide plate; 41. Pressure chamber; 42. Guide chamber; 43. Inner channel; 44. Support spring; 45. Fine-tuning rod; 5. Resistance bracket; 51. Touch rod; 52. Connecting sleeve; 53. Sealing joint.

Detailed ways

Please refer to Figures 1-7. In an embodiment of the present invention, a storage article sorting robot arm with AI recognition includes: a conveyor belt device 11, a front scanning module 12, a sorting robot arm 1 and an intelligent clamping mechanism 2; the two sides of the conveyor belt device 11 are respectively set as a delivery area and a transfer area, and the sorting robot arm 1 is set on the transfer area side of the conveyor belt device 11, and is used to transfer and store the articles on the conveyor belt device 11; a front scanning module 12 is installed above the side of the conveyor belt device 11 close to the delivery area, and the front scanning module 12 uses a high-definition camera and a barcode reader to obtain article image data and label data, and comprehensively analyzes and processes the article image data and label data to obtain article feature image data; the article image data is image data for the article to be sorted; wherein, by reading the barcode or QR code data on the article, these codes contain the identity recognition information of the article, so as to facilitate more accurate positioning of the article, judge its attributes and plan the best sorting path, thereby improving the automation and efficiency of the entire sorting process and reducing the error rate;

The conveyor belt device 11 is externally provided with a data analysis system, which uses a preset classification database model and combines the label data to perform feature recognition on the feature image data of the item and obtains the recognition result; the preset classification database model is constructed based on a large amount of item storage training data and machine learning algorithms, and has the ability to efficiently recognize the features of various items, that is, the extracted image features are combined with the label data, and feature matching and recognition are performed through the classification database model. The model will compare the shape, color, size and label coding information of the item with the feature templates of various items pre-stored in the database to determine the specific category of the item;

The sorting robot arm 1 is provided with a sorting module, which indexes the corresponding storage location information and the center of gravity location information of the current item based on the above recognition result, and the intelligent gripping mechanism 2 analyzes the optimal gripping action of the current item based on the center of gravity location information of the item and in combination with the item image data, and adopts the optimal gripping action to sort and transfer the item;

The sorting robot arm 1 is equipped with a high-definition scanner, which obtains the spatial position information of the items to be sorted on the current conveyor belt device 11 and outputs the precise coordinates of the position of the center of gravity of the items; wherein, the optimal clamping action is mainly based on the division of the corresponding clamping points based on the center of gravity of the items, and the principle of selecting the clamping points is to ensure that the clamping points are close to or just near the center of gravity of the items, so that during the grasping and handling process, the items can maintain a good balance state and are not easy to roll or fall, thereby ensuring the stability and safety of the sorting operation, and at the same time, according to factors such as the shape, texture and weight distribution of the items, combined with the preset mechanical model and algorithm, the clamping force and clamping method are dynamically adjusted;

The optimal clamping action includes a contact positioning action and a moving action.

In this embodiment, the intelligent clamping mechanism 2 includes: an upper machine plate 21, four connecting rods 23 are vertically arranged below the upper machine plate 21, a positioning plate 22 is fixed below the connecting rod 23, a base 24 is vertically slidably arranged between the upper machine plate 21 and the positioning plate 22, and the base 24 is slidably connected to each of the connecting rods 23;

A plurality of connecting arms 25 are symmetrically distributed on the positioning plate 22, one end of each of the connecting arms 25 is rotatably connected to the positioning plate 22, and a plurality of adjusting rods 26 corresponding to the connecting arms 25 are distributed on the base 24, one end of each of the adjusting rods 26 is rotatably connected to the connecting arm 25;

A hydraulic rod 27 is fixed on the upper machine plate 21, and one end of the hydraulic rod 27 is connected to the machine base 24;

An L-shaped driving arm 28 is also fixed to each of the connecting arms 25, and the L-shaped driving arm 28 is driven by a hydraulic cylinder to be folded and deformed;

A clamping assembly 3 is fixed to the end of the L-shaped driving arm 28 .

As a preferred embodiment, a locator (not shown in the figure) is provided in the middle of the upper plate on the intelligent clamping mechanism 2. The locator collects and outputs the precise coordinates of the upper plate 21 in three-dimensional space in real time. The sorting module on the sorting robot 1 plans the optimal motion trajectory of the sorting robot based on the coordinate difference between the acquired spatial coordinates of the center of gravity of the item and the spatial coordinates of the upper plate 21, so that the intelligent clamping mechanism 2 is suspended above the center of gravity of the item before sorting the item, thereby realizing intelligent positioning and clamping. The sorting method based on the center of gravity of the item can maximize the stability of the clamping process, reduce the risk of swaying, tilting or falling of the item due to the shift of the center of gravity during the clamping and moving process, thereby improving the efficiency and safety of the sorting operation.

In this embodiment, the adjustment rod 26 adopts a two-stage retractable adjustment mechanism, and each of the adjustment rods 26 is independently controlled by each other, and each of the L-shaped driving arms 28 is independently controlled by each other; thereby, the clamping assembly on each L-shaped driving arm can be freely adjusted to contact the object according to the clamping point analyzed according to the optimal clamping action; specifically, the clamping assembly will adjust its own positioning according to the analyzed clamping point position before contacting the object, ensuring that it can reach the ideal clamping position near the center of gravity when contacting the object.

The contact positioning action of the intelligent clamping mechanism 2 is analyzed as follows: the adjusting rod 26 on each of the connecting arms 25 performs a telescopic action so that the clamping assembly 3 contacts the surface of the object, and the hydraulic cylinder on the L-shaped driving arm 28 performs a telescopic action adjustment. At this time, the clamping assembly 3 contacts and positions the comprehensive optimal clamping point on the surface of the object, and the hydraulic rod 27 controls the sliding adjustment of the driving base 24, and applies contact pressure to the object through the clamping assembly 3.

In this embodiment, a pressure plate 29 is provided in the middle of the lower end surface of the positioning plate 22 through an electric telescopic rod, and the front scanning module 12 compares the image data and label data of adjacent items on the conveyor belt device 11. When the image data and label data of adjacent items are the same, the intelligent clamping mechanism 2 adopts the optimal clamping action to post-stack the items close to the transfer area, and subsequently performs multiple simultaneous transfers to further improve the sorting efficiency. When performing post-stack, the pressure plate can rest against the upper surface of the relatively upper object, and the clamping component contacts the side of the relatively lower object.

In this embodiment, the clamping assembly 3 includes: an outer clamping frame 31, which is rotatably connected to a main shaft 32 via a bearing, one end of the main shaft 32 is sleeved with a transmission tooth, and a micro motor (not shown in the figure) is installed on the outer clamping frame 31, and the output end of the micro motor is meshed with the transmission tooth for transmission; a guide plate 4 is coaxially fixed on the main shaft 32, and a plurality of abutment brackets 5 are laterally equidistantly distributed on the guide plate 4, and an anti-slip lining 33 is covered on the guide plate 4. The anti-slip lining is made of a material with a high friction coefficient, such as rubber, silicone, a special anti-slip coating, etc. It has a certain flexibility and adaptability and can provide sufficient friction under different contact pressures.

As a preferred embodiment, an annular pressure chamber 41 is provided in the guide disc 4, and hydraulic oil is stored in the pressure chamber 41. A feeler rod 51 is sealingly and slidably connected inside the abutment bracket 5, and a sleeve 52 is fixed to the end of the feeler rod 51. A sealing joint 53 is connected to one side of the abutment bracket 5. The sealing joint 53 is embedded and fixed on the guide disc 4 and communicated with the pressure chamber 41.

A guide cavity 42 is also provided in the guide plate 4. The guide cavity 42 is connected to the pressure cavity 41 through an inner channel 43. A piston is slidably connected to the guide cavity 42 through a fine-tuning rod 45. A support spring 44 is connected to one side of the piston.

In this embodiment, a displacement sensor is further provided at the end of each of the touch rods 51, and a signal receiver is provided on the guide plate 4, and the signal receiver obtains the displacement information of the touch rods during the working process; wherein, the contact positioning action also includes: when the guide plate contacts the surface of the object with a certain pressure, each of the touch rods slides and contracts in the abutment bracket at different distances (the electric telescopic rod is in a non-working state), and the electric telescopic rod drives the piston to send hydraulic oil into the pressure chamber, at which time the contact pressure between each of the touch rods and the surface of the object gradually increases and reaches the rated clamping pressure value (the maximum clamping force that can be applied by the intelligent clamping mechanism under normal working conditions, and this value is set under the premise of ensuring that the object can be effectively clamped and transferred without causing damage to it);

The intelligent gripping mechanism 2 is provided with a stroke monitoring module, the signal receiver converts the displacement information into an electrical signal, and sends the signal to the stroke monitoring module through a built-in data interface, and the stroke monitoring module collects the first displacement characteristic data of each feeler rod 51 on the guide plate 4 under the current working state, obtains the mapping point corresponding to each of the feelers 51 and constructs a first contact surface three-dimensional model; thereby accurately describing the specific state of the gripping component when in contact with the object, including information such as the shape, size and force distribution of the contact surface;

The travel monitoring module includes an early warning model, which uses a preset early warning model to comprehensively analyze the stress points of the three-dimensional model of the first contact surface in each of the clamping components 3 (including but not limited to stress analysis, strain analysis, and potential failure mode analysis, etc.) and output a first risk level.

In this embodiment, when the first risk level is higher than the threshold, the micro motor drives the guide plate 4 to deflect, the travel monitoring module collects the real-time displacement characteristic data of each touch rod 51 on the guide plate 4, and constructs a real-time contact surface three-dimensional model, and comprehensively analyzes the force points of the real-time contact surface three-dimensional model in each clamping component 3 through a preset early warning model and outputs a real-time risk level, that is, in this process, the real-time displacement data of each touch rod on the guide plate are continuously collected and integrated, and these data have extremely high timeliness and can truly reflect the actual action of the touch rod under the current working conditions. Based on these real-time data, a three-dimensional model of the contact surface is dynamically constructed, so that the model can accurately reflect the state and structural changes of the contact surface at each moment, and by applying the preset early warning model, an in-depth comprehensive analysis of the force points is performed on the real-time updated three-dimensional model of the contact surface in each clamping component, and the analysis content includes key indicators such as the distribution, concentration, direction change and potential mechanical fatigue of the force. After a series of analysis steps, the system will output the adjustment plan corresponding to the current minimum risk level, so that the control program can make timely and effective decisions.

What has been described above are only preferred specific implementations of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, can make equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, which should be covered by the protection scope of the present invention.

Claims (9)

1. Warehouse article letter sorting arm with AI discernment, its characterized in that: it comprises the following steps: the device comprises a conveyor belt device (11), a front scanning module (12), a sorting mechanical arm (1) and an intelligent clamping mechanism (2); the two sides of the conveying belt device (11) are respectively provided with a conveying area and a transferring area, and the sorting mechanical arm (1) is arranged at one side of the transferring area of the conveying belt device (11) and is used for transferring and storing articles on the conveying belt device (11); a front scanning module (12) is arranged above one side, close to the conveying area, of the conveyor belt device (11), the front scanning module (12) respectively adopts a high-definition camera and a bar code reader to acquire article image data and label data, and comprehensively analyzes and processes the article image data and the label data to obtain article characteristic image data; the article image data is image data for articles to be sorted;

A data analysis system is arranged outside the conveyor belt device (11), and the data analysis system adopts a preset classification database model and combines tag data to perform feature recognition on the article feature image data, and obtains a recognition result;

the sorting mechanical arm (1) is provided with a sorting module, the sorting module indexes the corresponding storage position information of the current article and the position information of the center of gravity of the article based on the identification result, and the intelligent clamping mechanism (2) analyzes the optimal clamping action of the current article based on the position information of the center of gravity of the article and in combination with the image data of the article, and sorts and transfers the article by adopting the optimal clamping action;

The sorting mechanical arm (1) is provided with a high-definition scanner, and the high-definition scanner acquires the spatial position information of the articles to be sorted on the current conveyor belt device (11) and outputs the accurate coordinates of the position of the gravity center point of the articles;

the optimal gripping action includes a touch location action and an migration action.

2. The warehouse item sort robot with AI identification of claim 1, wherein: the intelligent clamping mechanism (2) comprises: the upper machine disc (21) is vertically provided with four connecting rods (23) below, a positioning disc (22) is fixed below the connecting rods (23), a machine seat (24) is vertically arranged between the upper machine disc (21) and the positioning disc (22) in a sliding manner, and the machine seat (24) is in sliding connection with each connecting rod (23);

A plurality of connecting arms (25) are symmetrically distributed on the positioning disc (22), one ends of the connecting arms (25) are respectively connected with the positioning disc (22) in a rotating way, a plurality of adjusting rods (26) corresponding to the connecting arms (25) are distributed on the base (24), and one ends of the adjusting rods (26) are connected with the connecting arms (25) in a rotating way;

a hydraulic rod (27) is fixed on the upper machine disc (21), and one end of the hydraulic rod (27) is connected with the machine base (24);

An L-shaped driving arm (28) is further fixed on each connecting arm (25), and the L-shaped driving arms (28) are driven by a hydraulic cylinder to carry out folding deformation;

the end part of the L-shaped driving arm (28) is fixed with a clamping component (3).

3. The warehouse item sort robot with AI identification of claim 2, wherein: the intelligent clamping mechanism is characterized in that a positioner is arranged in the middle of an upper machine disc on the intelligent clamping mechanism (2), the positioner collects and outputs accurate coordinates of the upper machine disc (21) in a three-dimensional space in real time, and a sorting module on the sorting mechanical arm (1) plans an optimal motion track of the sorting mechanical arm based on the coordinate difference between the space coordinates of the obtained object gravity center point and the space coordinates of the upper machine disc (21), so that the intelligent clamping mechanism (2) is suspended above the object gravity center point before sorting the objects.

4. A warehouse item sort robot with AI identification as claimed in claim 3, wherein: the adjusting rods (26) are two-section telescopic adjusting mechanisms, the adjusting rods (26) are independently controlled, and the L-shaped driving arms (28) are independently controlled;

The contact positioning action of the intelligent clamping mechanism (2) is analyzed as follows: the adjusting rods (26) on the connecting arms (25) conduct telescopic action, so that the clamping assemblies (3) are contacted with the surface of an article, the hydraulic cylinders on the L-shaped driving arms (28) conduct telescopic action adjustment, at the moment, the clamping assemblies (3) are contacted with comprehensive optimal clamping points positioned on the surface of the article, and the hydraulic rods (27) control the driving base (24) to conduct sliding adjustment and apply contact pressure to the article through the clamping assemblies (3).

5. The warehouse item sort robot with AI identification of claim 2, wherein: the middle part of the lower end face of the positioning disc (22) is provided with a pressure plate (29) through an electric telescopic rod, the front scanning module (12) compares image data and label data of adjacent articles on the conveyor belt device (11), and when the image data and the label data of the adjacent articles are identical, the intelligent clamping mechanism (2) adopts an optimal clamping action to carry out rear stacking on the articles close to one side of the transferring area and then carries out a plurality of simultaneous transferring.

6. The warehouse item sort robot with AI identification of claim 2, wherein: the gripping assembly (3) comprises: the outer clamping frame (31) is rotatably connected with a main shaft (32) through a bearing, one end of the main shaft (32) is sleeved with a transmission tooth, the outer clamping frame (31) is provided with a miniature motor, and the output end of the miniature motor is meshed with the transmission tooth for transmission; the main shaft (32) is coaxially fixed with a guide disc (4), a plurality of supporting brackets (5) are transversely distributed on the guide disc (4) at equal intervals, and an anti-slip lining (33) is covered on the guide disc (4).

7. The warehouse item sort robot with AI identification of claim 6, wherein: an annular pressure cavity (41) is arranged in the guide disc (4), hydraulic oil is stored in the pressure cavity (41), a feeler lever (51) is connected in a sealing sliding manner in the pressing support (5), a connecting sleeve (52) is fixed at the end part of the feeler lever (51), a sealing joint (53) is connected to one side of the pressing support (5), and the sealing joint (53) is embedded and fixed on the guide disc (4) and communicated with the pressure cavity (41);

The guide disc (4) is internally provided with a guide cavity (42), the guide cavity (42) is communicated with the pressure cavity (41) through an inner channel (43), a piston is slidably connected in the guide cavity (42) through a fine adjustment rod (45), and one side of the piston is connected with a supporting spring (44).

8. The warehouse item sort robot with AI identification of claim 7, wherein: the end part of each feeler lever (51) is also provided with a displacement sensor, the guide disc (4) is provided with a signal receiver, and the signal receiver acquires displacement information of the feeler lever in the working process;

The intelligent clamping mechanism (2) is provided with a travel monitoring module, the signal receiver converts displacement information into an electric signal and sends the signal to the travel monitoring module through a built-in data interface, and the travel monitoring module gathers first displacement characteristic data of each feeler lever (51) on the guide disc (4) in the current working state, obtains a mapping point corresponding to each feeler lever (51) and builds a first contact surface three-dimensional model;

the stroke monitoring module comprises an early warning model, and the first risk level is comprehensively analyzed and output on the stress points of the three-dimensional model of the first contact surface in each clamping assembly (3) through the preset early warning model.

9. The warehouse item sort robot with AI identification of claim 8, wherein: when the first risk level is higher than a threshold value, the micro motor drives the guide disc (4) to deflect, the travel monitoring module gathers real-time displacement characteristic data of each feeler lever (51) on the guide disc (4), builds a real-time contact surface three-dimensional model, comprehensively analyzes stress points of the real-time contact surface three-dimensional model in each clamping assembly (3) through a preset early warning model, and outputs the real-time risk level.