An AI-based restaurant information control and management system and method

Abstract

The present invention discloses an AI-based restaurant information control and management system and method, which relates to the technical field of restaurant management, including a data acquisition module, a dining management module, a road recognition module, a path planning module and a comprehensive adjustment module; the data acquisition module is used to obtain the restaurant's historical dining data, the restaurant's real-time shooting data, the restaurant's order data and the food delivery robot's operation data: its technical points are: based on AI technology, the restaurant's historical data is analyzed and processed to understand the restaurant's dining situation under normal circumstances, and the dining is managed in combination with the restaurant's planning, the status of the diners is scientifically and accurately analyzed, the remaining dining time is reasonably analyzed, and then the number is further allocated, which can ensure that the diners in the restaurant are always in a relatively saturated state without affecting the restaurant's reputation, thereby ensuring that the restaurant can operate stably for a long time, with good use effect and good application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of restaurant management, and in particular to an AI-based restaurant information control and management system and method. Background Art

[0002] A restaurant is a facility or public dining house that provides food, drinks and other catering to the general public in a certain place.

[0003] In some large shopping malls, during peak meal times on weekends or holidays, a large number of employees flock to the restaurant, causing it to be overcrowded and with long queues. This will inevitably lead to a shortage of relevant service staff (for cost considerations, the number of service staff will be controlled within a certain range). At the same time, due to the long queues and insufficient service staff, the restaurant's reputation will gradually decline.

[0004] With the progress of society, restaurants are becoming more and more intelligent. In order to solve the above problems, people have invented various equipment to assist restaurant operations, including food delivery robots.

[0005] The existing patent application, CN108614449B, titled "Control Method for Smart Clean Energy Restaurants Based on Artificial Intelligence," describes the following: Step A: Using pressure sensors and infrared sensors to detect seat occupancy and obtain seat occupancy information; Step B: A processing device generates a current real-time seat status diagram and the restaurant's availability rate based on the seat occupancy information; Step C: Displaying the real-time seat status diagram and the restaurant's availability rate generated by the processing device to the user; The control method utilizes a dining management system for a smart clean energy restaurant, including: a clean energy power generation system, multiple seat sensing units, a wireless communication unit, a restaurant occupancy detection device, a processing device, a restaurant robot, and a restaurant drone. This system can optimize the dining experience, save time searching for seats, and improve the dining environment:

[0006] The existing application publication number is CN109146717B, and the invention patent named "A method for self-service ordering and delivery in an unmanned restaurant" records the following: the user obtains the identification information provided by the table through his or her personal mobile terminal; the user opens the login interface of the ordering system through his or her personal mobile terminal and logs in to the ordering system; the ordering system authenticates the received identification information, and the ordering system then provides the ordering interface to the user's personal mobile terminal, the user orders, and an order is formed in the ordering system; the ordering system automatically shares the information with the restaurant robot, the restaurant robot receives the order and identification information, and goes to the kitchen to pick up the food according to the order sequence; the restaurant robot locates the table according to the position information of the table in the identification information, and delivers the meal ordered by the user to the designated table; the present invention realizes the formation of intelligent and automated management between the user ordering and dining, reduces the number of restaurant waiters, and is conducive to reducing restaurant operating costs.

[0007] However, based on the above content and combined with the existing technology, the central solution recorded in the above patent is to display the number of empty seats to facilitate users to understand the number of remaining tables in the restaurant. On this basis, it can also cooperate with food delivery robots to deliver food, reduce manpower and reasonably control costs.

[0008] However, the above patent is only applicable to ordinary scenarios, where the user checks the remaining seats, then determines whether there are empty tables in the restaurant, and then decides whether to go to the restaurant;

[0009] But from the user's perspective, in actual application, restaurants and users are completely different;

[0010] From the restaurant's perspective, how can we provide standard services to more customers?

[0011] However, during holidays, when restaurants are packed and all tables are occupied, many restaurants still issue numbers as usual, resulting in long queues and long waiting times, which affects the mood of diners and the subsequent evaluation of the restaurant.

[0012] In addition, most restaurants have unlimited queues to increase their profits. When the queue is too long, they usually resort to adding tables temporarily. However, due to limited space inside the restaurant, adding tables will lead to an overcrowding of the restaurant and change the normal layout, resulting in a messy restaurant and affecting the normal restaurant robot delivery. At the same time, due to the excessive number of tables, the chefs are too busy and often cannot accurately control the quality of the dishes, which reduces the restaurant's reputation.

[0013] Therefore, the above solution has major defects in actual use and does not meet people's usage requirements. For this reason, we have developed an AI-based restaurant information control management system and method. Summary of the Invention

[0014] (1) Technical problems solved

[0015] In response to the shortcomings of the existing technology, the present invention provides an AI-based restaurant information control and management system and method, which analyzes and processes the restaurant's historical data based on AI technology, understands the restaurant's dining situation under normal circumstances, and manages dining in combination with the restaurant's planning, scientifically and accurately analyzes the status of diners, reasonably analyzes the remaining dining time, and then further allocates numbers, which can ensure that the diners in the restaurant are always in a relatively saturated state without affecting the restaurant's reputation, thereby ensuring that the restaurant can operate stably and for a long time, with good use effect and good application prospects.

[0016] (2) Technical solution

[0017] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0018] An AI-based restaurant information control and management system, including a data acquisition module, a dining management module, a road recognition module, a route planning module, and a comprehensive adjustment module:

[0019] Data acquisition module: used to obtain restaurant historical dining data, restaurant real-time shooting data, restaurant order data and delivery robot operation data;

[0020] Dining Management Module: This module is used to input the restaurant's real-time camera data into the AI ​​model for analysis. Based on the analysis results and restaurant order data, it predicts the end time of each table in the restaurant. It then collects statistics on the end time of all meals, predicts the number of tables in the restaurant in the future based on the current state, and then allocates seats based on the predicted number of tables.

[0021] Road recognition module: This module continuously inputs real-time restaurant data into the AI ​​model for image analysis, identifies restaurant road data, compares multiple sets of collected restaurant road data with preset data, and marks locations with abnormalities in multiple comparisons as abnormal areas.

[0022] Path planning module: This module is used to conduct a comprehensive analysis based on abnormal area and restaurant order data, generate all the running paths of the food delivery robot, calculate the efficiency values ​​of all the running paths, and arrange the running paths according to the size of the delivery value;

[0023] Comprehensive adjustment module: used to compare the maximum delivery value with the preset safety threshold, and execute the corresponding delivery strategy based on the comparison results.

[0024] Furthermore, the restaurant's historical dining data includes the number of diners, the gender ratio of diners, the waiting time for dining, the age group of diners, and the number of dishes. The restaurant's real-time shooting data includes image data and restaurant road data taken in real time by a camera installed inside the restaurant. The restaurant's order data includes the categories and quantities of dishes. The food delivery robot's operation data includes the delivery robot's operation data and size data.

[0025] Furthermore, the steps for predicting the end time of each table in the restaurant are as follows:

[0026] S1. Use the YOLO real-time target detection algorithm to identify the real-time image data of the restaurant and determine the number of people and their distribution in the image;

[0027] S2. Segment the image according to the distribution results in the image, and then amplify and denoise the segmented images in turn;

[0028] S3, using the Gender and Age Detection algorithm to process the denoised image and identify the gender and age of the table occupants;

[0029] S4. Obtain the order data corresponding to the table staff, and calculate the total meal time based on the order data and the restaurant's historical meal data, and predict the meal end time based on the total meal time, current time and meal time.

[0030] Furthermore, the formula for calculating the total meal time is as follows:

[0031]

[0032] Where Tz is the total meal time, n is the number of dishes, is the additional meal time of the i-th dish, k is the number of factors affecting the meal time, is the influence ratio of the number of influencing factors of item j, wcxs is the error correction coefficient;

[0033] Meal end time = total meal time - meal time + current time.

[0034] Furthermore, the steps for predicting the number of tables in the restaurant at a future time based on the current state and then allocating numbers based on the predicted number of tables are as follows:

[0035] Get the number of empty tables in the current restaurant;

[0036] Compare the meal end time with the set maximum waiting time Td, mark the table numbers whose meal end time is less than the set maximum waiting time Td, and count the number of marked tables;

[0037] Calculate the number of numbers to be released, the number of numbers to be released = the number of empty tables in the current restaurant + the number of marked tables - the number of reserved tables.

[0038] Furthermore, the steps for continuously inputting the restaurant's real-time shooting data into the AI ​​model for image analysis and identifying restaurant road data are as follows:

[0039] Image data inside the restaurant is collected every T1 time interval;

[0040] The AI ​​model crops the image data of the restaurant interior to the same specifications and extracts the restaurant road image;

[0041] Grayscale threshold segmentation method is used to segment the restaurant road image and extract obstacle graphics on the road.

[0042] Furthermore, the steps of marking the locations where abnormalities are found in multiple comparisons as abnormal regions are as follows:

[0043] Construct a two-dimensional coordinate system with the bottom of the segmented image as the coordinate origin;

[0044] Enter the position data of all obstacle graphics into the two-dimensional coordinate system and compare whether the obstacles overlap;

[0045] If the obstacles do not overlap, they are not marked;

[0046] If obstacles overlap, the outlier value of the obstacle in the road is calculated;

[0047]

[0048] Where Out is the calculated outlier value of the obstacle in the road, β is a constant coefficient, β>1, Szmax is the number of pixels corresponding to the widest position of the obstacle in the image, Sd is the number of pixels corresponding to the road, and Sjmin is the number of pixels corresponding to the closest distance between the obstacle edge and the road edge in the image.

[0049] Compare the outliers with the preset anomaly threshold Compare, if Out is less than , it is marked as abnormal and an early warning message is sent to the restaurant service staff;

[0050] If Out is greater than or equal to , it is marked as a serious abnormality and an early warning message is sent to the restaurant service staff.

[0051] Furthermore, the steps to generate all the running paths of the food delivery robot are as follows:

[0052] Input restaurant road data, mark the starting and ending locations of the delivery robots, and then generate all the delivery robot paths;

[0053] Compare all generated delivery robot paths with the severely abnormal areas one by one, delete the delivery robot paths that pass through the severely abnormal areas, and record the remaining running paths as all running paths;

[0054] The formula for calculating the efficiency value of all running paths is as follows:

[0055]

[0056] Where η is the efficiency value of the operation path, A is the number of roads of different widths in the operation path, is the length of the i2th road, e is a constant data, C is the maximum width of the restaurant road, is the width of the i2th section of road, Zxc is the number of turns in the running path, Zxs is the influence coefficient of turns in the running path on efficiency, Rs is the number of diners at the tables on both sides of the running path, and Yxs is the influence coefficient of diners on efficiency.

[0057] Furthermore, the preset safety threshold is the efficiency lower limit η x and the repeat check value η c , the maximum delivery value η max Compare with the preset safety threshold and execute the corresponding delivery strategy according to the comparison result as follows;

[0058] If the calculated η max <η x , then the service staff will be prompted to use manual delivery instead;

[0059] If the calculated η x ≤η max ≤η c , then collect the restaurant's real-time shooting data within the set time period T1 and recalculate the efficiency value. When the recalculated efficiency value ≤η c When the food is delivered manually, the service staff will be prompted to switch to manual delivery.

[0060] If the calculated η c ≤η max , then the maximum delivery value η max The corresponding running path is set as the delivery route of the delivery robot.

[0061] Furthermore, an AI-based restaurant information control and management method includes the following steps:

[0062] Obtain restaurant historical dining data, restaurant real-time shooting data, restaurant order data, and delivery robot operation data;

[0063] The AI ​​model analyzes the restaurant's real-time camera data and predicts the end time of each table's meal based on the analysis results and restaurant order data. The model then calculates the end time of all meals and predicts the number of tables in the restaurant based on the current state in the future. The model then allocates seats based on the predicted number of tables.

[0064] Continuously input the restaurant's real-time shooting data into the AI ​​model for image analysis, identify restaurant road data, and compare multiple sets of collected restaurant road data with preset data. Locations with abnormalities in multiple comparisons are marked as abnormal areas;

[0065] Based on a comprehensive analysis of abnormal areas and restaurant order data, all the running paths of the food delivery robot are generated, the efficiency values ​​of all the running paths are calculated, and the running paths are arranged according to the size of the delivery value;

[0066] The maximum delivery value is compared with the preset safety threshold, and the corresponding delivery strategy is executed according to the comparison result.

[0067] It also includes a statistical analysis module, which comprehensively calculates η when the fixed number of meal deliveries o is reached. max , calculate η max Average value , , where is the maximum delivery value of the i3th delivery route;

[0068] If the average Less than the preset standard value η BZ , then analyze and investigate the reputation. If the reputation decreases, adjust the restaurant's internal space layout. If the reputation is stable or rising, change the standard value η BZ ;

[0069] If the average Greater than or equal to the preset standard value η BZ , no adjustment is required.

[0070] (3) Beneficial effects

[0071] The present invention provides an AI-based restaurant information control and management system and method, which has the following beneficial effects:

[0072] 1. The present invention provides an AI-based restaurant information control and management system and method, which analyzes and processes restaurant historical data based on AI technology, understands the restaurant's dining situation under normal circumstances, and manages dining in combination with the restaurant's planning, scientifically and accurately analyzes the status of diners, reasonably analyzes the remaining dining time, and then further allocates numbers, which can ensure that the diners in the restaurant are always in a relatively saturated state without affecting the restaurant's reputation, thereby ensuring that the restaurant can operate stably for a long time, with good use effect and good application prospects.

[0073] 2. The present invention provides an AI-based restaurant information control and management system and method, which actually collects the internal road environment of the restaurant in real time, and performs real-time analysis and processing on the collected data to determine whether there is any abnormal state in the restaurant, and helps relevant technical personnel to deal with the abnormality in a timely manner, thereby ensuring that the relevant food delivery robots can deliver food stably. It further analyzes the factors affecting food delivery in the restaurant, conducts a comprehensive analysis of all factors, and calculates the efficiency value of the operation path. Compared with the existing food delivery robot that delivers food along the shortest distance, it can effectively reduce external interference and improve the efficiency of food delivery. The overall use effect is good and it has a good prospect for use. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 This is a flowchart of an AI-based restaurant information control and management system of the present invention;

[0075] Figure 2 A path planning diagram for use of an AI-based restaurant information control and management system of the present invention;

[0076] Figure 3 This is a diagram of abnormal areas marked when using the AI-based restaurant information control and management system of the present invention. DETAILED DESCRIPTION

[0077] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0078] Existing restaurants all adopt a number-issuing system, with no limit on the number of diners and no restrictions on queuing. However, when there are too many people queuing, they adopt methods such as adding tables, which affect the restaurant's reputation, resulting in a decrease in repeat customers and affecting the restaurant's survival.

[0079] Therefore, during the research and development process, the initial idea was how to reasonably plan the crowd and appropriately allocate numbers. Therefore, the initial idea was to identify empty seats and allocate a fixed number of numbers that exceeded the number of tables.

[0080] However, in actual use, it was found that using this method, there would still be a situation where some people would have to wait too long. For example, the people inside the restaurant had basically just finished their meal, while the people in line had to wait for a full cycle.

[0081] Therefore, in the middle of the research and development, we changed our thinking, adjusted the plan, studied the factors affecting the meal time, and conducted a comprehensive and integrated analysis to predict the remaining meal time for each table. Based on the remaining meal time, we made further plans and then allocated numbers, which was very effective.

[0082] However, in actual use, it was found that due to the busy meal period, food delivery would be affected by the environment and people. According to the fixed delivery method, food often could not be delivered in the first time, resulting in food accumulation and untimely delivery.

[0083] Therefore, in the later stage of research and development, we fully combined the previous plans and developed an AI-based restaurant information control management system and method. It fully combines historical data, analyzes historical data and indoor environment, and makes global real-time planning, so that the planned route is better and the delivery efficiency is higher.

[0084] Study plan

[0085] like Figure 1 As shown, an AI-based restaurant information control and management system includes a data acquisition module, a dining management module, a road recognition module, a path planning module, and a comprehensive adjustment module:

[0086] This system is based on the existing database of existing restaurants and the camera devices in existing restaurants.

[0087] During use, in order to improve accuracy, the number of camera devices can be increased to improve the accuracy of image recognition, thereby improving the effect of subsequent analysis.

[0088] The hardware it requires is a server.

[0089] Data collection

[0090] When the system is in use, it must be combined with historical data and comprehensively analyzed in order to make subsequent accurate predictions. The process of comprehensive data collection is based on the data acquisition module.

[0091] The data acquisition module obtains restaurant historical dining data, restaurant real-time shooting data, restaurant order data and delivery robot operation data;

[0092] The restaurant's historical dining data includes the number of diners, the gender ratio of diners, the waiting time for dining, the age group of diners, and the number of dishes. The restaurant's real-time shooting data includes image data taken in real time by cameras installed inside the restaurant and restaurant road data. The restaurant's order data includes the categories and quantities of dishes. The food delivery robot's operation data includes the delivery robot's operation data and size data.

[0093] In addition to the restaurant's real-time shooting data, the restaurant's historical dining data, restaurant order data, and delivery robot operation data are directly obtained from the existing restaurant's management system.

[0094] The results of the AI ​​model calculation can be compared with the restaurant order data to determine the AI ​​model's recognition effect. (The restaurant order data will record the number of people at each table, which can be compared with the results of the AI ​​model calculation)

[0095] Headcount Management

[0096] After understanding the restaurant's historical data, it is necessary to analyze and apply it in combination with the existing restaurant situation in order to predict the number of remaining tables in the future and predict future situations, so that the number can be allocated scientifically and the long-term operation of the restaurant can be guaranteed. This process is based on the dining management module.

[0097] The dining management module is used to input the restaurant's real-time camera data into the AI ​​model for analysis. Based on the analysis results and restaurant order data, it predicts the end time of each table in the restaurant. It then collects statistics on the end time of all meals and predicts the number of tables in the restaurant in the future based on the current state. It then allocates tables based on the predicted number of tables.

[0098] The steps to predict the end time of each table in the restaurant are as follows:

[0099] S1. Use the YOLO real-time target detection algorithm to identify the real-time image data of the restaurant and determine the number of people and their distribution in the image;

[0100] The YOLO real-time target detection algorithm can identify the location of the target when there are multiple people in the image, making it easier to process the image later.

[0101] S2. Segment the image according to the distribution results in the image, and then amplify and denoise the segmented images in turn;

[0102] Image segmentation is to separate the target objects and perform separate calculations on them. This method can effectively improve the accuracy of recognition.

[0103] If the YOLO real-time target detection algorithm is used directly for recognition, the recognition accuracy will be greatly reduced in remote multi-person scenarios. Therefore, a separate algorithm is needed.

[0104] After enlarging the image, the feature recognition can be increased, and combined with denoising processing, some interference factors can be removed, thereby improving the accuracy of subsequent recognition.

[0105] S3, using the Gender and Age Detection algorithm to process the denoised image and identify the gender and age of the table occupants;

[0106] The Gender and Age Detection algorithm can determine the gender and age of a person. Based on the previous processing steps and this algorithm, the accuracy of gender and age recognition can reach 90%, which can meet the needs of restaurants and has good performance.

[0107] S4. Obtain the order data corresponding to the table staff, and calculate the total meal time based on the order data and the restaurant's historical meal data, and predict the meal end time based on the total meal time, current time and meal time.

[0108] The formula for calculating total meal time is as follows:

[0109]

[0110] Where Tz is the total meal time, n is the number of dishes, is the additional meal time of the i-th dish, k is the number of factors affecting the meal time, is the influence ratio of the number of influencing factors of item j, wcxs is the error correction coefficient;

[0111] The increased meal time for a dish is the time assigned by the AI ​​model based on historical data analysis, not the actual meal time. This data corresponds one-to-one to the dish and is a fixed value.

[0112] The number of factors that affect the length of dining time includes the number of diners, the gender ratio of diners, the waiting time for dining and the corresponding age groups of diners. Other relevant influencing factors can also be added.

[0113] The weight coefficient is determined using the coefficient of variation method, which is a method of assigning weights to each indicator based on the degree of variation between the current value of each evaluation indicator and the target value. If the numerical difference of an indicator is large, it can clearly distinguish the evaluated objects, indicating that the indicator has rich discrimination information, and thus the indicator should be given a larger weight. On the contrary, if the numerical difference of each evaluated object on a certain indicator is small, then the ability of this indicator to distinguish the evaluation objects is weak, and thus the indicator should be given a smaller weight. This method directly uses the information contained in each indicator to obtain the weight of the indicator through calculation, and therefore is objective.

[0114] The other weight coefficients are discussed by experts through the above method, and most companies in this field have corresponding weight coefficients.

[0115] By subdividing the influencing factors and adjusting the error correction coefficient, the prediction of the total dining time of most restaurants is made more accurate.

[0116] Meal end time = total meal time - meal time + current time.

[0117] The meal time can be checked directly from the order data, which is more convenient.

[0118] The steps to predict the number of tables in the restaurant at a future time based on the current state and then allocate seats based on the predicted number of tables are as follows:

[0119] Get the number of empty tables in the current restaurant by directly comparing the total number of tables with the number of orders placed in the order book;

[0120] Compare the meal end time with the set maximum waiting time Td, mark the table numbers whose meal end time is less than the set maximum waiting time Td, and count the number of marked tables;

[0121] This step can ensure that the queueing people can eat within the set maximum waiting time Td, and can effectively reduce the impact of too many people on the restaurant service, with good use effect.

[0122] Calculate the number of numbers to be released, the number of numbers to be released = the number of empty tables in the current restaurant + the number of marked tables - the number of reserved tables.

[0123] The number of reserved tables includes the number of scheduled tables and the number of empty tables reserved in advance to avoid subsequent abnormalities, thereby preventing the restaurant from falling into over-operation. It has good use effect and good prospects for use.

[0124] The present invention provides an AI-based restaurant information control and management system, which analyzes and processes restaurant historical data based on AI technology, understands the restaurant's dining situation under normal circumstances, and manages dining in combination with the restaurant's planning, scientifically and accurately analyzes the status of diners, reasonably analyzes the remaining dining time, and then further allocates numbers, which can ensure that the diners in the restaurant are always in a relatively saturated state without affecting the restaurant's reputation, thereby ensuring that the restaurant can operate stably for a long time, with good use effect and good application prospects.

[0125] Sports route planning

[0126] When the number of people in a restaurant is limited, the layout of the restaurant will not change during normal use. However, some changes may occur occasionally, such as users moving tables, chairs, and trash cans. These changes may cause other items to be placed in the road, requiring relevant personnel to clean up.

[0127] In order to effectively avoid the above situation, it is necessary to detect the road environment inside the restaurant before use to determine whether there is interference. This process is based on the road recognition module.

[0128] The road recognition module continuously inputs the restaurant's real-time shooting data into the AI ​​model for image analysis, identifies the restaurant's road data, compares the collected multiple sets of restaurant road data with the preset data, and marks the locations with abnormalities in multiple comparisons as abnormal areas;

[0129] The steps for continuously inputting restaurant real-time shooting data into the AI ​​model for image analysis and identifying restaurant road data are as follows:

[0130] Image data from the restaurant interior is collected every T1 interval to reduce the interference of moving objects, that is, diners, and determine whether they are abnormal;

[0131] The AI ​​model performs isotropic cropping on the restaurant interior image data and extracts restaurant road images. The cameras are fixed, so their shooting angles and directions are the same. Therefore, isotropic cropping is sufficient.

[0132] Grayscale threshold segmentation method is used to segment the restaurant road image and extract obstacle graphics on the road.

[0133] After extracting the restaurant road image, the grayscale threshold of the restaurant floor is close. By setting the threshold interval, objects in other threshold intervals can be extracted, thereby achieving the effect of extracting obstacles on the road.

[0134] like Figure 2-3 As shown in the figure, the steps for marking the locations with abnormalities in multiple alignments as abnormal regions are as follows:

[0135] Construct a two-dimensional coordinate system with the bottom of the segmented image as the coordinate origin;

[0136] Enter the position data of all obstacle graphics into the two-dimensional coordinate system and compare whether the obstacles overlap;

[0137] If the obstacles do not overlap, they are not marked;

[0138] Non-overlapping means that the obstacle can be moved, and most of them are caused by users passing by. When this situation exists in all the pictures taken, it means that people or objects have been occupying the road for a long time, and this situation is considered abnormal.

[0139] If obstacles overlap, the outlier value of the obstacle in the road is calculated;

[0140]

[0141] Where Out is the calculated outlier value of the obstacle in the road, β is a constant coefficient, β>1, Szmax is the number of pixels corresponding to the widest position of the obstacle in the image, Sd is the number of pixels corresponding to the road, and Sjmin is the number of pixels corresponding to the closest distance between the obstacle edge and the road edge in the image.

[0142] For example, when the object is in the middle of the road, the smaller Sjmin is, the larger Out is, the more abnormal it is, and the more necessary it is to process.

[0143] When more objects occupy the road, Szmax becomes larger. In this case, Out becomes larger, which is more abnormal and needs to be processed.

[0144] Compare the outliers with the preset anomaly threshold Compare, if Out is less than , it is marked as abnormal and an early warning message is sent to the restaurant service staff;

[0145] If Out is greater than or equal to , it is marked as a serious abnormality and an early warning message is sent to the restaurant service staff.

[0146] Abnormalities also vary depending on the situation. If it only occupies a small amount, no action can be taken and the restaurant staff can be notified. If it occupies too much or is located in the middle of the road, the calculated abnormal value is large and it is marked as an abnormality. At this time, relevant personnel need to handle it.

[0147] Path Generation

[0148] When the abnormal situation of the restaurant is known, the effect of analyzing the path by combining the abnormal situation and the location of the destination is better. The path analysis is based on the path planning module.

[0149] Path planning module: This module is used to conduct a comprehensive analysis based on abnormal area and restaurant order data, generate all the running paths of the food delivery robot, calculate the efficiency values ​​of all the running paths, and arrange the running paths according to the size of the delivery value;

[0150] like Figure 2 As shown in the figure, the steps to generate all the running paths of the food delivery robot are as follows:

[0151] Enter restaurant road data, mark the starting and ending locations of the delivery robot, and then generate all the delivery robot paths. This is a basic function of existing delivery robot-related software, namely the navigation function, which will not be described in detail here;

[0152] Compare all generated delivery robot paths with the severely abnormal areas one by one, delete the delivery robot paths that pass through the severely abnormal areas, and record the remaining running paths as all running paths;

[0153] This step is to eliminate abnormal situations, prevent the food delivery robot from being stuck in abnormal positions for a long time, and ensure that the food delivery robot can deliver food in a timely and fast manner.

[0154] Existing food delivery robots generally wait for human intervention when encountering an anomaly, requiring the cooperation of restaurant staff and making their use more cumbersome.

[0155] In addition, some intelligent food delivery robots adopt the methods of waiting or detouring, but both methods have a certain response time, so the efficiency of food delivery is not high.

[0156] The formula for calculating the efficiency value of all running paths is as follows:

[0157]

[0158] Where ROUNDUP is a function that rounds up to one decimal place, η is the efficiency value of the operation path, A is the number of roads of different widths in the operation path, is the length of the i2th road, e is a constant data, C is the maximum width of the restaurant road, is the width of the i2th section of road, Zxc is the number of turns in the running path, Zxs is the influence coefficient of turns in the running path on efficiency, Rs is the number of diners at the tables on both sides of the running path, and Yxs is the influence coefficient of diners on efficiency.

[0159] For example, Can clearly reflect the impact of the road, The larger the value, the smaller the efficiency value. The larger the maximum width C of the restaurant road, the larger the efficiency value. The more turns there are in the operation path, the slower the delivery, and the larger the corresponding Zxc×Zxs. At this time, the smaller the efficiency value. The more people are dining, the higher the possibility of its impact. The larger the corresponding Rs×Yxs, the smaller the efficiency value. The above formula can clearly understand the efficiency value of each operation path, so that we can determine which operation path is more suitable for delivery.

[0160] The efficiency value of the running path can clearly and clearly judge the quality of the running path. The larger the efficiency value of the running path, the better the running path is and the more suitable it is for the food delivery robot to deliver food.

[0161] Comprehensive adjustment module: used to compare the maximum delivery value with the preset safety threshold, and execute the corresponding delivery strategy based on the comparison results.

[0162] The preset safety threshold is the lower efficiency limit η x and the repeat check value η c , the maximum delivery value η max Compare with the preset safety threshold and execute the corresponding delivery strategy according to the comparison result as follows;

[0163] If the calculated η max <η x , then the service staff will be prompted to use manual delivery instead;

[0164] If the calculated η x ≤η max ≤η c , then collect the restaurant's real-time shooting data within the set time period T1 and recalculate the efficiency value. When the recalculated efficiency value ≤η c When the food is delivered manually, the service staff will be prompted to switch to manual delivery.

[0165] In this case, the main purpose is to avoid calculation errors. In order to reduce the workload of related manual labor, the calculation can be performed again quickly. When repeated calculations fall into this situation, in order to avoid affecting the efficiency of food delivery, manual delivery is used at this time, which is more effective.

[0166] If the calculated η c ≤η max , then the maximum delivery value η max The corresponding running path is set as the delivery route of the delivery robot. Compared with the existing method of selecting the nearest route for delivery, this delivery method can reduce the interference of the external environment and has a good use effect.

[0167] The present invention provides an AI-based restaurant information control and management system, which actually collects the internal road environment of the restaurant in real time, and performs real-time analysis and processing on the collected data to determine whether there is any abnormal state in the restaurant, and helps relevant technical personnel to deal with the abnormality in a timely manner, thereby ensuring that the relevant food delivery robots can deliver food stably. It further analyzes the factors affecting food delivery in the restaurant, conducts a comprehensive analysis of all factors, and calculates the efficiency value of the operation path. Compared with the existing food delivery robot that delivers food along the shortest distance, it can effectively reduce external interference and improve the efficiency of food delivery. The overall use effect is good and it has a good prospect for use.

[0168] Real-time feedback

[0169] In order for a restaurant to operate for a long time, it must meet the requirements of users. However, user requirements will change over time, and they will gradually enjoy a larger space. Therefore, the layout inside the restaurant needs to be adjusted in real time, and this process is based on the statistical analysis module.

[0170] When the statistical analysis module reaches a fixed number of food deliveries o, the comprehensive statistics η max , calculate η max Average value , , where is the maximum delivery value of the i3th delivery route;

[0171] If the average Less than the preset standard value η BZ , then analyze and investigate the reputation. If the reputation decreases, adjust the restaurant's internal space layout. If the reputation is stable or rising, change the standard value η BZ ;

[0172] If the average Greater than or equal to the preset standard value η BZ , no adjustment is required.

[0173] In daily operations, when analyzing and investigating word of mouth, word of mouth is declining, but η max Greater than or equal to the preset standard value η BZ , the standard value η can be adjusted upward BZ .

[0174] To analyze and investigate the reputation, you can use questionnaires or directly check the corresponding software scores. You can directly understand the reasons for the decline in reputation and make corresponding adjustments, so as to better maintain the restaurant's reputation and enable the restaurant to operate stably and effectively in the long term.

[0175] An AI-based restaurant information control and management method includes the following steps:

[0176] Obtain restaurant historical dining data, restaurant real-time shooting data, restaurant order data, and delivery robot operation data;

[0177] The AI ​​model analyzes the restaurant's real-time camera data and predicts the end time of each table's meal based on the analysis results and restaurant order data. The model then calculates the end time of all meals and predicts the number of tables in the restaurant based on the current state in the future. The model then allocates seats based on the predicted number of tables.

[0178] Continuously input the restaurant's real-time shooting data into the AI ​​model for image analysis, identify restaurant road data, and compare multiple sets of collected restaurant road data with preset data. Locations with abnormalities in multiple comparisons are marked as abnormal areas;

[0179] Based on a comprehensive analysis of abnormal areas and restaurant order data, all the running paths of the food delivery robot are generated, the efficiency values ​​of all the running paths are calculated, and the running paths are arranged according to the size of the delivery value;

[0180] The maximum delivery value is compared with the preset safety threshold, and the corresponding delivery strategy is executed according to the comparison result.

[0181] The above formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the latest real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0182] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0183] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0184] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. An AI-based restaurant information control and management system, characterized by: include: Data acquisition module: used to obtain restaurant historical dining data, restaurant real-time shooting data, restaurant order data and delivery robot operation data; Dining Management Module: This module is used to input the restaurant's real-time camera data into the AI ​​model for analysis. Based on the analysis results and restaurant order data, it predicts the end time of each table in the restaurant. It then collects statistics on the end time of all meals, predicts the number of tables in the restaurant in the future based on the current state, and then allocates seats based on the predicted number of tables. The steps to predict the end time of each table in the restaurant are as follows: S1. Use the YOLO real-time target detection algorithm to identify the real-time image data of the restaurant and determine the number of people and their distribution in the image; S2. Segment the image according to the distribution results in the image, and then amplify and denoise the segmented images in turn; S3, using the Gender and Age Detection algorithm to process the denoised image and identify the gender and age of the table occupants; S4. Obtain the order data corresponding to the table staff, calculate the total meal time based on the order data and the restaurant's historical meal data, and predict the meal end time based on the total meal time, current time, and meal time; The formula for calculating total meal time is as follows: Where Tz is the total meal time, n is the number of dishes, is the additional meal time of the i-th dish, k is the number of factors affecting the meal time, is the influence ratio of the number of influencing factors of item j, wcxs is the error correction coefficient; Meal end time = total meal time - meal time + current time; The steps to predict the number of tables in the restaurant at a future time based on the current state and then allocate seats based on the predicted number of tables are as follows: Get the number of empty tables in the current restaurant; Compare the meal end time with the set maximum waiting time Td, mark the table numbers whose meal end time is less than the set maximum waiting time Td, and count the number of marked tables; Calculate the number of numbers to be released, which is the number of empty tables in the restaurant + the number of marked tables - the number of reserved tables. Road recognition module: This module continuously inputs real-time restaurant data into the AI ​​model for image analysis, identifies restaurant road data, compares multiple sets of collected restaurant road data with preset data, and marks locations with abnormalities in multiple comparisons as abnormal areas. Path planning module: This module is used to conduct a comprehensive analysis based on abnormal area and restaurant order data, generate all the running paths of the food delivery robot, calculate the efficiency values ​​of all the running paths, and arrange the running paths according to the size of the delivery value; Comprehensive adjustment module: used to compare the maximum delivery value with the preset safety threshold, and execute the corresponding delivery strategy based on the comparison results.

2. The AI-based restaurant information control and management system according to claim 1, characterized in that: The restaurant's historical dining data includes the number of diners, the gender ratio of diners, the waiting time for dining, the age range of diners, and the number of dishes. The restaurant's real-time shooting data includes image data taken in real time by a camera installed inside the restaurant and restaurant road data. The restaurant's order data includes the categories and quantities of dishes. The food delivery robot's operation data includes the delivery robot's operation data and size data.

3. The AI-based restaurant information control and management system according to claim 2, characterized in that: The steps for continuously inputting real-time restaurant data into the AI ​​model for image analysis and identifying restaurant road data are as follows: Image data inside the restaurant is collected every T1 time interval; The AI ​​model crops the image data of the restaurant interior to the same specifications and extracts the restaurant road image; Grayscale threshold segmentation method is used to segment the restaurant road image and extract obstacle graphics on the road.

4. The AI-based restaurant information control and management system according to claim 3, characterized in that: The steps to mark locations with abnormalities in multiple alignments as abnormal regions are as follows: Construct a two-dimensional coordinate system with the bottom of the segmented image as the coordinate origin; Enter the position data of all obstacle graphics into the two-dimensional coordinate system and compare whether the obstacles overlap; If the obstacles do not overlap, they are not marked; If obstacles overlap, the outlier value of the obstacle in the road is calculated; Where Out is the calculated outlier value of the obstacle in the road, β is a constant coefficient, β>1, Szmax is the number of pixels corresponding to the widest position of the obstacle in the image, Sd is the number of pixels corresponding to the road, and Sjmin is the number of pixels corresponding to the closest distance between the obstacle edge and the road edge in the image. Compare the outliers with the preset anomaly threshold Compare, if Out is less than , it is marked as abnormal and an early warning message is sent to the restaurant service staff; If Out is greater than or equal to , it is marked as a serious abnormality and an early warning message is sent to the restaurant service staff.

5. The AI-based restaurant information control and management system according to claim 4, characterized in that: The steps to generate all the running paths of the food delivery robot are as follows: Input restaurant road data, mark the starting and ending locations of the delivery robots, and then generate all the delivery robot paths; Compare all generated delivery robot paths with the severely abnormal areas one by one, delete the delivery robot paths that pass through the severely abnormal areas, and record the remaining running paths as all running paths; The formula for calculating the efficiency value of all running paths is as follows: Where η is the efficiency value of the operation path, A is the number of roads of different widths in the operation path, is the length of the i2th road, e is a constant data, C is the maximum width of the restaurant road, is the width of the i2th road segment, Zxc is the number of turns in the running path, Zxs is the influence coefficient of the turns in the running path on the efficiency, Rs is the number of diners at the tables on both sides of the running path, and Yxs is the influence coefficient of the diners on the efficiency; The preset safety threshold is the lower efficiency limit η x and the repeat check value η c , the maximum delivery value η max Compare with the preset safety threshold and execute the corresponding delivery strategy according to the comparison result as follows; If the calculated η max <η x , then the service staff will be prompted to use manual delivery instead; If the calculated η x ≤η max ≤η c , then collect the restaurant's real-time shooting data within the set time period T1 and recalculate the efficiency value. When the recalculated efficiency value ≤η c When the food is delivered manually, the service staff will be prompted to switch to manual delivery. If the calculated η c ≤η max , then the maximum delivery value η max The corresponding running path is set as the delivery route of the delivery robot.

6. The AI-based restaurant information control and management system according to claim 5, characterized in that: It also includes a statistical analysis module, which comprehensively calculates η when the fixed number of meal deliveries o is reached. max , calculate η max Average value , , where is the maximum delivery value of the i3th delivery route; If the average Less than the preset standard value η BZ , then analyze and investigate the reputation. If the reputation decreases, adjust the restaurant's internal space layout. If the reputation is stable or rising, change the standard value η BZ ; If the average Greater than or equal to the preset standard value η BZ , no adjustment is required.

7. An AI-based restaurant information control and management method, using the system according to any one of claims 1 to 6, characterized in that: The steps include: Obtain restaurant historical dining data, restaurant real-time shooting data, restaurant order data, and delivery robot operation data; The AI ​​model analyzes the restaurant's real-time camera data and predicts the end time of each table's meal based on the analysis results and restaurant order data. The model then calculates the end time of all meals and predicts the number of tables in the restaurant based on the current state in the future. The model then allocates seats based on the predicted number of tables. Continuously input the restaurant's real-time shooting data into the AI ​​model for image analysis, identify restaurant road data, and compare multiple sets of collected restaurant road data with preset data. Locations with abnormalities in multiple comparisons are marked as abnormal areas; Based on a comprehensive analysis of abnormal areas and restaurant order data, all the running paths of the food delivery robot are generated, the efficiency values ​​of all the running paths are calculated, and the running paths are arranged according to the size of the delivery value; The maximum delivery value is compared with the preset safety threshold, and the corresponding delivery strategy is executed according to the comparison result.

Images