Service consumption intelligent robot for preventing service failure and working method

Abstract

The present application relates to the technical field of intelligent robots, in particular to a service consumption intelligent robot for preventing service failure and a working method, which collects data in multiple dimensions by using a 3D vision sensor, a pressure sensor and an interactive interface log, and analyzes and predicts potential service failure risks. When detecting operation confusion risks or insufficient load risks, the mechanical arm is triggered in advance to swing with high frequency and small amplitude when confused, and to expand a horizontal platform and start hydraulic buffering when insufficiently loaded. Active intervention is made before service failure occurs, the service failure probability is greatly reduced, the limitation of relying on "after-the-fact remedy" in the prior art is broken through, and a complete prevention closed loop of "monitoring-prediction-intervention-optimization" is constructed.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot technology, specifically to a service consumption intelligent robot and its working method for preventing service failures. Background Technology

[0002] With the rapid development of artificial intelligence technology, intelligent robots are being used more and more widely in many fields, especially in the service industry, where they are gradually becoming an important service provider. However, in practical applications, the service quality of intelligent robots faces many challenges.

[0003] In intelligent robot service scenarios, technical limitations and complex and ever-changing service environments often lead to service failures. When service failures occur, existing technologies mostly rely on "post-event remediation" measures. For example, after a problem occurs, such as when a person is unable to operate the robot interface and is prompted that the operation has failed, the robot's robotic arm can be moved to restore control. This only provides a post-event remediation function.

[0004] In the robotic arm control process, as described in Chinese Patent Application No. 202411953555.8, published on March 18, 2025, and classified under B25J9 / 16, the robot includes a master control servo motor, m slave control robotic arms, and a camera device. The camera device is mounted on at least one of the slave control robotic arms. The method includes: acquiring environmental image data captured by the camera device; inputting the environmental image data into a target model for processing to obtain control commands for each of the m slave control robotic arms. The target model includes a first network and a second network. The first network is used to calculate joint motion data corresponding to each joint in the target robotic arm, where the target robotic arm is at least one of the m slave control robotic arms. The second network is used to perform mapping processing based on the joint motion data corresponding to each joint in the target robotic arm to obtain control commands for each of the m slave control robotic arms. This improves the accuracy and efficiency of robot control.

[0005] The robots described in the above literature primarily address issues that arise after the service has been provided. However, for consumer service robots, users' attention is easily focused on the user interface during the service process. If problems occur or waiting times are too long, anxiety arises. Existing intelligent robots lack effective ways to divert user attention and cannot proactively improve the user experience during the service process. For example, in a hotel check-in scenario, if the intelligent robot is slow in processing the check-in, users can only stare at the interface while waiting, without any measures to alleviate anxiety. This significantly reduces user satisfaction. The robotic arm, in contrast, could be controlled only when problems arise, thus further diminishing the user experience. Summary of the Invention

[0006] This invention provides a service consumption intelligent robot and its working method for preventing service failures. It can control the robot's robotic arm in advance based on the current reading status of the robot, thereby preventing failures and improving the user's service experience.

[0007] This invention provides a method for operating a service consumption intelligent robot to prevent service failures, comprising the following steps:

[0008] Step S1: Monitor user behavior characteristics in real time through the scanning area; user behavior characteristics include four categories: capturing user body movements, item carrying status and facial expression features, and the load-bearing status of the robotic arm; determine typical user behavior patterns through clustering algorithms, and update the weights of the four categories of behavior features in real time through learning methods to determine user behavior patterns.

[0009] Step S2: Analyze user needs based on user behavior patterns and predict potential service failure risk types;

[0010] Step S3: Trigger the corresponding robotic arm action mode based on the predicted type of potential service failure risk.

[0011] The above method scans user behavior characteristics, such as user body movements, the state of items carried, facial expressions, and the load-bearing status of the robotic arm. It then uses cluster analysis across four categories to determine the final behavior for each category. A learning method is used to update the weights of the four categories of behavioral characteristics in real time, thereby determining the user behavior pattern. For example, if a user's body movement is scratching their head and their facial expression is distressed, and the items are in good condition, the weight of the body movement and facial expression category is increased, thus determining the user behavior pattern as "not knowing how to operate." This user behavior pattern analysis reveals the user's need for guidance, and by predicting potential service timeliness risks, the robotic arm's action pattern is determined to be guided operation. This allows for early prediction of user failures and timely intervention, ensuring a good user experience and forming a complete preventative closed loop of "monitoring-prediction-intervention-optimization." This transforms the service process from "post-event remediation" to "pre-event prevention," establishing a new type of human-machine collaborative service relationship.

[0012] Furthermore, the user behavior monitoring in step S1 specifically includes the following steps:

[0013] c1. Utilize 3D vision sensors to continuously capture user's body movements, the state of items being carried, and facial expressions, and use pressure sensors to detect the weight distribution data of the robotic arm in real time, and use the interactive interface log to record the user's operation trajectory in detail.

[0014] The system uses 3D vision sensors to record the user's hand clicks and swipes when operating the robot's interactive interface, as well as the size and shape of the items carried and the walking posture. Pressure sensors record the weight values ​​borne by various parts of the robotic arm at different time periods. The operation trajectory is recorded as the sequence of clicked buttons and page switching.

[0015] c2. Clean the collected raw data to remove erroneous data and outliers caused by sensor failure or other interference factors, and then standardize the different types of data.

[0016] 3D vision sensors capture limb movement data that does not conform to the normal logic of human movement. These data can be considered outliers and removed. Standardization involves unifying the range of weight data from pressure sensors and limb movement amplitude data detected by 3D vision sensors.

[0017] c3. Use clustering algorithms to analyze the preprocessed data, continuously adjust the parameters of the clustering algorithm, and determine the optimal clustering result through evaluation indicators;

[0018] c4. Determine the weights of the four clustering results and identify user behavior patterns through dynamic learning.

[0019] The above method groups a series of user behavior data with similar operation trajectories, similar body movements, and similar carried items into one category. It then uses cluster analysis to determine the optimal clustering result to ensure the rationality and accuracy of the cluster division. Furthermore, after obtaining the clustering results, it uses dynamic learning to adjust the weights of the clustering results, thereby determining the user behavior pattern.

[0020] Furthermore, step c4 also includes dynamic learning optimization, which involves establishing a user behavior baseline model, distinguishing typical behavior patterns through cluster analysis, updating the weights of behavioral features in real time using the FTRL online learning algorithm, and triggering an active calibration mechanism for abnormal data.

[0021] The above settings update the behavioral feature weights using the FTRL prior learning algorithm, thereby ensuring that the final user behavior pattern can more accurately determine user behavior.

[0022] Furthermore, in step S2, when predicting potential service failure risks, the following conditions are met: when the interface operation sequence deviates from the standard process, it is determined as an operation confusion risk; when the user's gaze wanders beyond a threshold, it is determined as an operation confusion risk; when the total volume of scanned items exceeds 80% of the maximum carrying capacity of the robotic arm, it is determined as an insufficient carrying capacity risk; when the variance of the item weight distribution exceeds a safety threshold, it is determined as an insufficient carrying capacity risk; and when items are detected being carried and the user's limbs are leaning towards one side of the items and their expression is one of pain, it is determined as a carrying inconvenience risk.

[0023] The above settings can analyze and confirm risks by detecting different situations, which facilitates the subsequent implementation of corresponding measures to address the risks.

[0024] Furthermore, step S3, which involves triggering the corresponding robotic arm action mode based on the risk type, includes:

[0025] b1. To address the risk of operational confusion, control the robotic arm to perform high-frequency, small-amplitude guiding swings, with a swing frequency of 5-8Hz.

[0026] b2. To address the risk of insufficient load-bearing capacity, control the robotic arm to deploy the horizontal platform and activate the hydraulic buffer.

[0027] b3. To mitigate the risk of inconvenience in carrying the item, control the robotic arm to move closer to the location of the item.

[0028] The above settings allow for better assistance to users by controlling the robotic arm to perform corresponding operations for different risks, thereby preventing poor user experience due to inadequate service.

[0029] Furthermore, step S3 also includes: shifting the user's attention focus through dynamic posture changes by using a visual saliency model to predict the user's gaze point; when the user is detected to be continuously staring at the wrong area, controlling the robotic arm to perform a sudden action and coordinating with lighting to guide the gaze shift.

[0030] The above settings allow the robotic arm to guide the user when they are continuously looking at the wrong area.

[0031] Furthermore, clustering algorithms include: Data points are divided into Clusters In the context of minimizing the sum of squared distances from each data point to the center of its cluster, let the data points be... Belongs to a cluster Clusters The center is Then the objective function is: ;

[0032] The K value of the clustering algorithm is continuously adjusted, and the optimal clustering result is determined by an evaluation metric, namely the silhouette coefficient. For a data point The contour coefficient is defined as ;

[0033] in yes The average distance to other data points in the same category yes The average distance to the nearest data point in other categories; the silhouette coefficient of the entire dataset is the average of the silhouette coefficients of all data points. ; Choose to make The largest K value is used as the optimal number of clusters to determine the objective function and obtain the clustering results.

[0034] The above settings, by continuously adjusting the K value of the clustering algorithm, determine the clustering result when the average value of the contour coefficients of the data points is maximized, thereby ensuring that the contour effect is good and more accurate.

[0035] Another aspect of the present invention provides a service consumption intelligent robot for preventing service failures, comprising: a robot body and a base disposed at its bottom, an interactive interface disposed on the exterior of the robot body, and robotic arms symmetrically disposed on both sides of the robot body, the robotic arms being used to perform preset posture actions to divert the user's attention according to the service status, a scanning area disposed on the exterior of the robot body for identifying user characteristics, and a control module disposed inside the robot body for controlling the switching of robotic arm action modes according to the input signal from the scanning area.

[0036] The above settings, by scanning user behavior characteristics such as body movements, the state of items carried, facial expressions, and the load-bearing status of the robotic arm, determine the final behavior for each of the four categories through cluster analysis. The weights of the four categories of behavioral features are updated in real time using a learning method to determine user behavior patterns. For example, if a user's body movement is scratching their head and their facial expression is distressed, and the items are in good condition, the weight of the body movement and facial expression categories is increased, thus determining the user behavior pattern as "not knowing how to operate." This allows analysis of user behavior patterns to identify the user's need for guidance. Furthermore, by predicting potential service timeliness risks, the robotic arm's action pattern is determined to be guided operation. This enables early prediction of user failures and early intervention, ensuring a good user experience and forming a complete preventative closed loop of "monitoring-prediction-intervention-optimization." It transforms the service process from "post-event remediation" to "pre-event prevention," establishing a new type of human-machine collaborative service relationship.

[0037] Furthermore, the robotic arm automatically extends and assumes a document delivery posture when it detects that the printing task is completed, and automatically adjusts to a horizontal carrying posture when it detects that the user is carrying multiple items in the scanning area.

[0038] The above settings allow for convenient control of the robotic arm through its various states. Attached Figure Description

[0039] Figure 1 This is a system flowchart of the present invention.

[0040] Figure 2 This is a schematic diagram of the structure of the present invention.

[0041] Figure 3 This is a diagram of the display interface of the present invention. Detailed Implementation

[0042] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. The embodiments of the present invention are given for illustrative and descriptive purposes only, and are not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described to better illustrate the principles and practical application of the invention, and to enable those skilled in the art to understand the invention and design various embodiments with various modifications suitable for a particular purpose.

[0043] Example:

[0044] Please see Figure 1 - Figure 3 This invention provides a technical solution: a service consumption intelligent robot for preventing service failures, comprising: a robot body 2 and a base 1 disposed at its bottom; an interactive interface is provided on the outside of the robot body 2; robotic arms are symmetrically arranged on both sides of the robot body 2; the robotic arms are used to perform preset posture actions according to the service status to divert the user's attention; a scanning area 3 is provided on the outside of the robot body 2; the scanning area 3 is used to identify user needs; and a control module is provided inside the robot body 2 for controlling the switching of robotic arm action modes according to the input signals from the scanning area.

[0045] When the printing task is detected as complete, the robotic arm automatically extends and assumes a document delivery posture. When the scanning area detects that the user is carrying multiple items, the robotic arm automatically adjusts to a horizontal carrying posture. When the robotic arm detects that the user does not know how to operate, it will adjust to a guiding state. When the user is carrying heavy items, the robotic arm will be controlled to move closer to the items. In this embodiment, the control module controls the robotic arm to move to the target position or perform the corresponding operation, which is the existing robotic arm control principle and is existing technology, so it will not be described in detail here.

[0046] like Figure 1 As shown, a method for a service consumption intelligent robot to prevent service failures includes the following steps:

[0047] Step S1: Monitor user behavior characteristics in real time through the scanning area; user behavior characteristics include four categories: capturing user body movements, item carrying status and facial expression features, and the load-bearing status of the robotic arm; determine typical user behavior patterns through clustering algorithms, and update the weights of the four categories of behavior features in real time through learning methods to determine user behavior patterns.

[0048] Step S2: Analyze user needs based on user behavior patterns and predict potential service failure risk types;

[0049] Step S3: Trigger the corresponding robotic arm action mode based on the predicted type of potential service failure risk.

[0050] User behavior monitoring in step S1 specifically includes the following steps:

[0051] Step a1: Capture user's body movements, item carrying status, and facial expression features using a 3D vision sensor; In this embodiment, the 3D vision sensor captures user's body movements, item carrying status, and facial expression features based on existing visual recognition principles. Specifically, it determines the specific user's body movements, item carrying status, and facial expression features by comparing the visually scanned image with a preset standard image.

[0052] Step a2: Detect the weight distribution of the robotic arm using pressure sensors;

[0053] Step a3: Record the user's operation trajectory through the interactive interface log.

[0054] In one embodiment, the user behavior baseline model establishment step in step S1 is as follows:

[0055] c1. Utilize 3D vision sensors to continuously capture user's body movements, the state of items being carried, and facial expressions, and use pressure sensors to detect the weight distribution data of the robotic arm in real time, and use the interactive interface log to record the user's operation trajectory in detail.

[0056] The system uses 3D vision sensors to record the user's hand clicks and swipes when operating the robot's interactive interface, as well as the size and shape of the items carried and the walking posture. Pressure sensors record the weight values ​​borne by various parts of the robotic arm at different time periods. The operation trajectory is recorded as the sequence of clicked buttons and page switching.

[0057] c2. Clean the collected raw data to remove erroneous data and outliers caused by sensor failure or other interference factors, and then standardize the different types of data.

[0058] 3D vision sensors capture limb movement data that does not conform to the logic of normal human movement and can be considered outliers and discarded. Standardization involves unifying the range of weight data from pressure sensors and limb movement amplitude data detected by 3D vision sensors. For example, if the vision sensor captures the movement of other people or objects behind the user, it is judged as limb movement data that does not conform to the logic of normal human movement and is considered an outlier.

[0059] c3. Use clustering algorithms to analyze the preprocessed data, continuously adjust the parameters of the clustering algorithm, and determine the optimal clustering result through evaluation indicators;

[0060] A series of user behavior data with similar operation trajectories, similar body movements and similar items are grouped into one category. The optimal clustering result is determined by evaluating the silhouette coefficient to ensure the rationality and accuracy of the cluster division.

[0061] Taking the K-Means clustering algorithm as an example, its goal is to... Data points are divided into Clusters In this context, we want to find a way to minimize the sum of squared distances from each data point to the center of its cluster. Let the data points be... Belongs to a cluster Clusters The center is Then the objective function is: ;

[0062] The K value of the clustering algorithm is continuously adjusted, and the optimal clustering result is determined by evaluation metrics, with the silhouette coefficient being a commonly used metric. For a data point Its profile coefficient is defined as ;

[0063] in yes The average distance to other data points in the same category yes The average distance to the nearest data points in other categories. The silhouette coefficient of the entire dataset is the average of the silhouette coefficients of all data points. By trying different K values, the corresponding profile coefficients are calculated, and the one that makes the profile coefficients the most suitable is selected. The largest K value is used as the optimal number of clusters; then the cluster outline is obtained by substituting the K value into the objective function. In this embodiment, the K value can be adjusted by a preset standard value, such as reducing the K value when it is too large, or the K value can be automatically adjusted by a PID algorithm.

[0064] c4. Conduct in-depth analysis of each cluster, extract typical behavior models that can represent the core characteristics of the cluster, describe and define the determined typical behavior patterns in detail, and form a user behavior baseline model.

[0065] Within a certain cluster, it was found that most users followed a fixed sequence when operating the robot: first clicking a specific function button, then placing the item. Their physical movements were characterized by stable reaching and placing actions, which can be identified as a typical behavior pattern. The identified typical behavior pattern is described and defined in detail to form the basic framework of the user behavior baseline model. This model is used to distinguish between normal and abnormal behaviors and to provide a benchmark for dynamic learning and optimization.

[0066] For a cluster of limb movements, the mean vector of all limb movement vectors in the cluster can be calculated as a typical limb movement pattern. Let the cluster be... There is Individual limb movement vectors Typical body movement patterns .

[0067] For a cluster of pressure sensor data, the mean vector and standard deviation vector of the load weight distribution vector within the cluster can be calculated. This represents the average load distribution of this cluster, expressed as a standard deviation vector. It reflects the degree of dispersion of the data.

[0068] For clusters of interactive interface operation trajectories, the most frequently occurring sequence of operation events can be identified as typical operation trajectory patterns. If within a cluster... In the sequence of operation events Number of times At most, then This can serve as a typical operational trajectory pattern for this type of cluster.

[0069] For a certain type of cluster, it can be described as: when the user presents... Characterized by limb movements, the robotic arm bears a weight distribution that is close to... And the operation trajectory follows This pattern represents typical user behavior of this type.

[0070] c4. Determine the weights of the clustering results and identify user behavior patterns through dynamic learning. In this embodiment, the FTRL online learning algorithm is used to update the weights of behavioral features in real time; an active calibration mechanism is triggered for abnormal data.

[0071] For example, if the feature weights K1, K2, and K3 are preset for each clustering result at the beginning, when a clustering result is detected, the feature weight value corresponding to the clustering result that differs significantly from the preset evaluation criteria is increased. The increased value is the preset value, so that the clustering result that differs significantly from the preset evaluation criteria is better reflected. This clustering result can be used as the main influencing factor to determine the user behavior pattern. For example, for the user, the current situation is an operation failure in terms of operation clustering, but for the physical action clustering, such as carrying multiple items or the items being too heavy, if the item clustering result exceeds the preset clustering result by 60%, the operation error interface clustering result exceeds the preset clustering result by 10%, and the physical action clustering result exceeds the preset clustering result by 0%, then the feature weight of the item clustering result is increased by 10, so that the entire user behavior is judged as insufficient carrying risk.

[0072] In step S2, when predicting potential service failure risks, the following conditions are met: when the interface operation sequence deviates from the standard process, it is determined as an operation confusion risk; when the user's gaze wanders beyond a threshold, it is determined as an operation confusion risk; when the total volume of scanned items exceeds 80% of the maximum carrying capacity of the robotic arm, it is determined as an insufficient carrying capacity risk; when the variance of the item weight distribution exceeds a safety threshold, it is determined as an insufficient carrying capacity risk; and when items are detected being carried and the user's limbs are leaning towards one side of the items and their expression is one of pain, it is determined as a carrying inconvenience risk.

[0073] The step S3, which involves triggering the corresponding robotic arm action mode based on the risk type, includes:

[0074] b1. To mitigate the risk of operational confusion, control the robotic arm to perform high-frequency, small-amplitude guiding swings, with a swing frequency of 5-8Hz.

[0075] b2. To address the risk of insufficient load-bearing capacity, control the robotic arm to deploy the horizontal platform and activate the hydraulic buffer.

[0076] b3. To mitigate the risk of inconvenience in carrying the item, control the robotic arm to move closer to the location of the item.

[0077] The working principle of this invention is as follows: By scanning user behavior characteristics, such as user body movements, the state of items carried, facial expressions, and the load-bearing state of the robotic arm, the final behavior of each category is determined through cluster analysis of four categories. The weights of the four categories of behavioral characteristics are updated in real time through a learning method to determine the user behavior pattern. For example, if the user's body movement is scratching their head and their facial expression is distressed, and the items are in good condition, the weight of the category of user body movement and facial expression is increased, thus determining the user behavior pattern as "not knowing how to operate." This allows analysis of the user behavior pattern to identify the need for guidance. Furthermore, by predicting potential service timeliness risks, the robotic arm's action pattern is determined to be guided operation. This enables early prediction of user failures and early intervention, ensuring a good user experience and forming a complete preventative closed loop of "monitoring-prediction-intervention-optimization." This transforms the service process from "post-event remediation" to "pre-event prevention," establishing a new type of human-machine collaborative service relationship.

Claims

1. A method for operating a service consumption intelligent robot to prevent service failures, characterized in that, include: Step S1: Monitor user behavior characteristics in real time through the scanning area; User behavior features include four categories: capturing user body movements, item carrying status and facial expressions, and the load-bearing status of the robotic arm; typical user behavior patterns are determined through clustering algorithms, and the weights of the four categories of behavior features are updated in real time through learning methods to determine user behavior patterns. Step S2: Analyze user needs based on user behavior patterns and predict potential service failure risk types; Step S3: Trigger the corresponding robotic arm action mode based on the predicted type of potential service failure risk; In step S2, when predicting potential service failure risks, the following conditions are met: when the interface operation sequence deviates from the standard process, it is determined as an operation confusion risk; when the user's gaze wanders beyond a threshold, it is determined as an operation confusion risk; when the total volume of scanned items exceeds 80% of the maximum carrying capacity of the robotic arm, it is determined as an insufficient carrying capacity risk; when the variance of the item weight distribution exceeds a safety threshold, it is determined as an insufficient carrying capacity risk; and when items are detected being carried and the user's limbs are leaning towards one side of the items and their expression is one of pain, it is determined as a carrying inconvenience risk.

2. The working method of a service consumption intelligent robot for preventing service failures according to claim 1, characterized in that: User behavior monitoring in step S1 specifically includes the following steps: c1. Utilize 3D vision sensors to continuously capture user's body movements, the state of items being carried, and facial expressions, and use pressure sensors to detect the weight distribution data of the robotic arm in real time, and use the interactive interface log to record the user's operation trajectory in detail. The system uses 3D vision sensors to record the user's hand clicks and swipes when operating the robot's interactive interface, as well as the size and shape of the items carried and the walking posture. Pressure sensors record the weight values ​​borne by various parts of the robotic arm at different time periods. The operation trajectory is recorded as the sequence of clicked buttons and page switching. c2. Clean the collected raw data to remove erroneous data and outliers caused by sensor failure or other interference factors, and then standardize the different types of data. 3D vision sensors capture limb movement data that does not conform to the normal human movement logic. These can be regarded as outliers and removed. Standardization processing is to unify the range of weight data from pressure sensors and limb movement amplitude data detected by 3D vision sensors. c3. Use clustering algorithms to analyze the preprocessed data, continuously adjust the parameters of the clustering algorithm, and determine the optimal clustering result through evaluation indicators; c4. Determine the weights of the four clustering results and identify user behavior patterns through dynamic learning.

3. The working method of the intelligent service consumption robot for preventing service failures according to claim 2, characterized in that: Step c4 further includes dynamic learning optimization, which involves establishing a user behavior baseline model, distinguishing typical behavior patterns through cluster analysis, updating the weights of behavioral features in real time using the FTRL online learning algorithm, and triggering an active calibration mechanism for abnormal data.

4. The working method of a service consumption intelligent robot for preventing service failures according to claim 1, characterized in that: The step S3, which involves triggering the corresponding robotic arm action mode based on the risk type, includes: b1. To address the risk of operational confusion, control the robotic arm to perform high-frequency, small-amplitude guiding swings, with a swing frequency of 5-8Hz. b2. To address the risk of insufficient load-bearing capacity, control the robotic arm to deploy the horizontal platform and activate the hydraulic buffer. b3. To mitigate the risk of inconvenience in carrying the item, control the robotic arm to move closer to the location of the item.

5. The working method of a service consumption intelligent robot for preventing service failures according to claim 4, characterized in that: Step S3 further includes: shifting the user's attention focus through dynamic posture changes by using a visual saliency model to predict the user's gaze point; when the user is detected to be continuously staring at the wrong area, controlling the robotic arm to perform a sudden action and coordinating with lighting to guide the gaze shift.

6. The working method of a service consumption intelligent robot for preventing service failures according to claim 1, characterized in that: Clustering algorithms include: Data points are divided into Clusters In the context of minimizing the sum of squared distances from each data point to the center of its cluster, let the data points be... Belongs to a cluster Clusters The center is Then the objective function is: ; The K value of the clustering algorithm is continuously adjusted, and the optimal clustering result is determined by an evaluation metric, namely the silhouette coefficient. For a data point The contour coefficient is defined as ; in yes The average distance to other data points in the same category yes The average distance to the nearest data point in other categories; the silhouette coefficient of the entire dataset is the average of the silhouette coefficients of all data points. ; Choose to make The largest K value is used as the optimal number of clusters to determine the objective function and obtain the clustering results.

7. The working method of a service consumption intelligent robot for preventing service failures according to claim 1, comprising: The robot body and its base are provided at the bottom. The robot body has an interactive interface on its exterior. The robot body has symmetrical robotic arms on both sides. The robotic arms are used to perform preset posture actions to divert the user's attention according to the service status. The robot body has a scanning area on its exterior for identifying user needs. The robot body has a control module inside for controlling the switching of robotic arm action modes according to the input signals from the scanning area.

Images