CN110689380A

CN110689380A - Intelligent vending system

Info

Publication number: CN110689380A
Application number: CN201910949361.3A
Authority: CN
Inventors: 王忠斌; 曾果君; 陈海燕
Original assignee: Shenzhen Touches Technology Co Ltd
Current assignee: Shenzhen Touches Technology Co Ltd
Priority date: 2019-10-08
Filing date: 2019-10-08
Publication date: 2020-01-14

Abstract

The invention discloses an intelligent vending system which comprises a display, a sound pick-up, a loudspeaker, a saleable commodity list, a voice recognition device, a point touch sensing device, a human body recognition device, a shipment control device and a main control unit. The method is characterized in that: when the system is idle, the display plays video media, when the system detects that a person appears in front of the display, or recognizes that the person calls a preset awakening language, or receives a point touch induction signal, the system displays a humanoid robot on the display, and voice interaction is carried out with the person. The customer selects the goods by voice or click. When the goods selection instruction of the customer can indicate the goods, putting the goods into a goods frame to be paid; when the goods selection instruction of the customer can not indicate the goods, the goods meeting the conditions are highlighted in the goods selection frame to prompt the customer to further select. Meanwhile, the robot has various expressions and corresponds to different vending states with different expressions.

Description

Intelligent vending system

Technical Field

The present invention relates to the field of vending machines.

Background

Vending machine that uses in the existing market, intelligent degree is low, lack man-machine interaction function, and user's shopping experience is poor. The invention aims to provide an intelligent vending machine capable of human-computer natural language interaction, which provides help for shopping of customers through a robot and improves shopping experience of the customers.

Disclosure of Invention

The intelligent vending system is organically combined with the robot, the automatic vending and the media playing technologies, so that the intelligent vending system is realized, the shopping experience of customers is improved, and the commercial value of equipment is improved.

The intelligent vending device comprises a display, a sound pick-up, a loudspeaker, a saleable commodity list, a voice recognition device, a human body induction device, a shipment control device, a main control unit and the like.

It is characterized in that: when the system is in an idle state, the display circularly plays video media, dynamic pictures and the like; when the human body sensing device detects that a person appears in front of the display, or the voice recognition device recognizes that the person calls a preset awakening language, or the point touch device senses a point touch action, the system enters a vending state, a human-shaped robot is displayed on the display, natural voice interaction is carried out with a customer, the customer is welcomed to arrive, and the customer is guided to shop, pay and pick up goods.

Further: the humanoid robot has multiple expressions and is respectively associated with different states of the vending process. When the equipment enters a vending mode, the robot meets the customer in a smiling expression; when the customer finishes payment, the robot expresses the thank you in a thank you expression; when the customer does not select any desirable goods, welcoming the next time the customer is in unfortunate expression; when the device fails, a apology is expressed to the customer in the refractory expression.

Further: in the vending mode, the display displays a vendible commodity selection frame, a to-be-paid commodity frame and a payment information frame besides the robot, and is characterized in that: if the commodity name of the voice goods selection instruction of the client can indicate a certain commodity, putting the commodity into a commodity frame to be paid; if the commodity name of the voice goods selection instruction of the customer cannot accurately indicate the commodity, displaying all commodities meeting the conditions in a special mode in a commodity selection frame, and prompting the customer to further select.

Further: the saleable goods list in the system is characterized in that: the commodity information includes fields for the name, category, brand, alias, function, and feature of the commodity. The commodity designation of the customer's voice pick instruction may be the name, category, brand, alias, etc. of the commodity; meanwhile, the customer can also put forward the own purchasing demand, and the system can match commodities with certain functions or characteristics according to the customer demand.

Further: the human body sensing device in the system can be an image recognition device with client identity recognition, such as face recognition, iris recognition and the like. If the system can identify the identity of the customer, after the customer finishes shopping, the business of the customer is counted besides the overall sales volume.

Further: when the system switches from the media playing mode to the vending mode, if the system already identifies the identity of the customer, placing the commodities frequently purchased by the customer in the prominent position of the commodity selection box; if the system is unable to identify the customer, the item with the large overall sales is placed in a prominent position in the item selection box.

Drawings

FIG. 1 is a block diagram of an embodiment of the present invention.

Fig. 2 is an operation mode transition diagram of the embodiment.

Fig. 3 is a screen layout in the robot vending mode.

FIG. 4 is a table of vendable items and a set of designated items.

FIG. 5 is a semantic framework of shopping instructions.

Detailed Description

In the typical embodiment of the invention, the display is an LCD large screen (101), and the point touch sensing device is an infrared correlation touch frame (102) attached to the screen; a microphone (103) and a camera (104) are arranged on the outer frame of the screen; a cargo storage box (105) with a cargo channel is arranged behind the screen, and a cargo outlet control device (106) is arranged below the cargo channel; the main control unit (107) comprises a computer board and a network communication module (108) and runs system control software.

The control software of the system comprises a voice recognition module which can be used for recognizing the voice signal collected by the microphone. The system divides the identified guest instructions into two groups: wake up instructions and shopping instructions. In order to make the system more humanized, the robot can be named as small beauty, big girl and the like. The wake-up words may be set to: "little beauty in the horse" and "big girl in the horse"

The control software of the system also comprises a face recognition module for recognizing the face appearing in front of the lens. The lens of the camera is adjusted to a position where a human face can only be shot when the human face appears in front of the screen. The face recognition comprises two processes, namely recognizing the face appearing in the visual field of the lens and informing the system. Then extracting an image of the Face part, sending the image to a server to retrieve from a Face library, setting the retrieved Face ID as the current client ID if the retrieval is successful, and extracting the shopping statistics of the client; if the retrieval fails, the current client ID is nulled.

The system controls the software to be divided into two working modes, namely a media playing mode and a robot selling mode. When the system works in a media playing mode, the display plays video media, such as advertisements, notifications, enterprise culture and the like. At the same time, a "shopping" button is displayed at an appropriate location on the screen (e.g., lower left corner).

Fig. 2 is a transition diagram of the operation mode. When the system is in a playing mode, the voice recognition module intercepts the awakening instruction, and the face recognition module retrieves the face in the image captured by the camera. When the voice recognition module detects that a person calls a 'wakeup word'; or the face recognition module detects the face in the image; or the touch screen senses the point touch signal, and the system is switched to the vending mode from the current playing mode.

When the system is in the robot vending mode, the system divides the display into four areas, a robot (301), a vendable item selection box (302), a to-be-paid item selection box (303), and a payment information box (304). In order to improve the affinity of the device, the robot occupies a major portion of the screen, giving the customer the feeling of talking to the real person.

The robot is realized by dynamic continuous playing of a plurality of static pictures expressing different expressions and mouth shapes. The pictures are divided into different emoticons, and when the system is in different vending states, different emoticons are displayed. The expression group of the robot comprises: smile, thank you, regret, and too hard, etc., each emoticon picture contains different mouth shapes and eye shapes, such as open, closed, etc. When the robot sounds, the mouth shape is synchronous with the sound; the eye shape is switched at regular time, so that the blinking action can be formed, and the agility of the robot is increased.

When the system enters the vending mode, the picture of the smile group is displayed, and meanwhile, welcome words, such as welcome, are played. When payment information is received from the customer, the picture of the thank you group is used. When the customer does not pick the merchandise beyond a specified time, or has picked the merchandise but has not completed payment for a long time, the pictures of the "regret" group are used to inform the customer that the time has expired. When a failure occurs at the time of shipment of the equipment, the customer is given an apology using the "refractory" group of pictures.

According to the spoken grammar of the modern Chinese language, we describe the grammar of the system's shopping instructions as follows:

< shopping instruction > is defined as:

< order of picking >

< Instructions for abandonment >

< requirement instruction >

< Exit instruction >

The < pick order > is defined as:

{ < pick > } { < quantity > } { < Unit > } < Commodity designation >

< pick > is defined as:

i want | give me | take |

< quantity > is defined as:

< unit > is defined as:

one | box | can | tin |

The < commercial designation > is defined as:

< name > < alias >

< Brand > { < Commodity designation > }

< feature > { of } { < Commodity designation > }

The < reject order > is defined as:

< discard > { < quantity > < Unit > } < Commodity designation >

Missing | Do this | this I don't | { this } is not what I want

< reject > was defined as:

get rid of | < take away |)

The < demand instruction > is defined as:

{ me } thirst, | { me } starves | { me } sleeps

The < exit instruction > is defined as:

i do not buy I quit

The list of items available for sale (401) in the system includes the following fields: codes, names, categories, brands, functions, features, and the like. The functions of the commodity are as follows: quenching thirst, relieving hunger, refreshing mind, etc. The characteristics of the commodity include: sweet, salty, spicy, sour, bitter, etc.

When the system is initialized, classifying the commodities according to names, categories, brands, functions and characteristics in a saleable commodity table to form a designated commodity set (402) of each word; classifying according to the functions of the commodities to form a functional commodity set (403) of each function; the articles are classified by their characteristics to form a characteristic article set (404) for each characteristic.

In the vending mode, the system monitors shopping instructions of customers all the time, and carries out voice recognition, grammar analysis and semantic analysis on the shopping instructions. The semantics of the instructions adopt a framework structure, see fig. 5. Wherein the < semanticism > comprises: < add >, < discard >, < end >. The semantic framework includes slots such as < quantity >, < unit >, < feature >, < designation-1 >, < designation-2 >, etc. The system defines a default value for each slot. Each slot corresponds to a grammar rule, and the corresponding relation is as follows:

frame groove	Grammar rules
		<Number of>	<Number of>
<Unit of>	<Unit of>
		<Feature(s)>	<Feature(s)>、<Function(s)>
<Scale-1>	<Brand>
		<Finger scale-2>	<Name (R)>、<Alias name>

The grammar analysis takes the vocabulary recognized by the voice as input, and carries out the protocol analysis from bottom to top according to the grammar rule to form a complete semantic frame.

If the final specification is < goods selection instruction >, the user sends a goods selection instruction, and the semantic symbol is < add >; if the specification result is < abandon instruction >, the user sends an instruction of abandoning the goods, and the semantic symbol is < abandon >; if the analysis result is < demand instruction >, the user puts forward a purchase demand, and the semanteme symbol is < add >; if the analysis result is < exit instruction >, the operation instruction is < end >.

If the result of the syntactic analysis is < pick instruction > or < abandon instruction >, the frame information of < quantity >, < unit >, < feature >, < designation > and the like of the goods is recorded in the analysis process. If the command is the < requirement command >, the commodity function set corresponding to the command is directly filled into the < feature > slot of the semantic framework.

After the parsing is completed, a semantic framework is formed. Semantic analysis will detect the integrity and legitimacy of the semantic frame content and compute the final commodity index set.

And performing intersection operation on the non-empty commodity sets corresponding to the < feature >, < designation-1 > and < designation-2 > according to the analyzed semantic framework.

If the result set contains only one item member, the designation of the pick instruction identifies the item and places it in the pick box for the item to be paid. If the result set contains a plurality of commodity members, the commodities are marked out by adopting a highlighted frame in the commodity selection box which can be sold, and meanwhile, the customer is informed of further pointing out the commodities by voice. If the result set is empty, carrying out parallel operation on the commodity set corresponding to the member of the named expression, and if the result set is also the empty set, informing the customer that the commodity selected by the customer does not exist; if the union is not empty, the commodities are highlighted in the commodity selection box and are prompted by voice to ask the customer to indicate the commodities.

When the commodity frame to be paid of the customer is not empty, the system displays a payment two-dimensional code in the payment information bar for the customer to scan the code for payment through the mobile phone. At this time, the system control software will continuously inquire the payment status of the customer to the server through the network. If the payment is successful, the goods output control device is driven to output goods, and the customer is informed of taking the goods by voice, and the selling mode is exited, and the previous media playing is continued. If the payment information is not received for a long time, the mobile terminal quits overtime and returns to the play mode.

Claims

1. The invention relates to an intelligent vending system, which comprises a display, a sound pick-up, a loudspeaker, a saleable commodity list, a voice recognition device, a point touch induction device, a human body recognition device, a shipment control device and a main control unit, and is characterized in that: when the system is in an idle state, the display plays video media or displays pictures; when the system detects that a person appears in front of the display through the human body recognition device, or recognizes that the person calls a preset awakening language through the voice recognition device, or receives a sensing signal of a point touch sensing device, the system is switched to a vending mode, a human-shaped robot is displayed on the display, and the human-shaped robot performs voice interaction with the person to guide a customer to shop, pay and pick up goods; the customer selects goods through voice or touch; after the customer completes payment, the system drives the delivery device to deliver goods and carries out sales statistics.

2. The display of claim 1, which displays the selectable box of the vendable commodity, the box of the commodity to be paid and the payment information box in addition to the display of the robot after the system enters the vending mode, and is characterized in that: when the commodity of the voice goods selection instruction of the client can indicate a certain commodity, adding the icon of the commodity into a commodity frame to be paid, and displaying payment information in a payment information frame; if the commodity name of the voice goods selection instruction of the customer can not indicate the commodity, displaying the commodity meeting the condition in a special mode in a commodity available for sale selection frame, and prompting the customer to further indicate the commodity.

3. The robot of claim 1 having a plurality of expressions, including smiling, thank you, regretting, sad, etc., and associating different vending states with these expressions: when the system enters a vending mode, the system meets the customer with smiling expression; when the customer finishes payment, expressing the thank you in a thank you expression; when the client does not find the needed commodity, the client is welcomed to come next time in an unfortunate expression; when the device fails, a apology is expressed to the customer in the refractory expression.

4. The list of vendible items of claim 1, further comprising: the commodity information comprises fields such as the name, the category, the brand, the alias, the function and the characteristic of the commodity, and the commodity name of the voice goods selection instruction of the client can be the name, the category, the brand, the alias and the like of the commodity; meanwhile, the customer can also put forward the own purchasing demand, and the system can match commodities with certain functions or characteristics according to the customer demand.

5. The human body sensing device of claim 1, which is an image recognition device capable of identifying the identity of a customer, and when the system is capable of identifying the identity of the customer, the system will also count the goods purchased by the customer.

6. The vendible item selection box of claim 2, further comprising: when the system enters a vending mode, if the system identifies the identity of a customer, displaying commodities with a large number of purchases at the significant position of a commodity selection box according to the shopping statistics of the customer; if the identity of the customer cannot be confirmed, the goods with large sales in the sales statistical table are displayed in a prominent position.