Intra-hospital logistics robot distribution time prediction system based on artificial intelligence

Abstract

The invention discloses a hospital logistics robot distribution time prediction system based on artificial intelligence, relates to the technical field of artificial intelligence, and solves the problems that influence factors in a distribution scene are complex in composition, the personnel flow is large and changes dynamically in real time, congestion occurs in the robot distribution process, and the distribution time of a logistics robot in a hospital cannot be predicted. The prediction time of the robot distribution logistics is delayed, and the overall efficiency of time prediction is reduced. Comprising a data acquisition module for acquiring a distribution path; acquiring monitoring data corresponding to each distribution road section in the distribution path; the data analysis module is used for acquiring the global obstacle density and the plurality of local obstacle densities in the obstacle distribution diagram, and performing analysis based on the global obstacle density and the plurality of local obstacle densities to obtain a congestion coefficient; the time prediction module is used for performing analysis based on the congestion coefficient to obtain road section estimation time; performing simulation based on the congestion coefficient to obtain a prediction success rate; and the time prediction precision and the overall operation efficiency of robot distribution are improved.

Description

Technical Field

[0001] This application belongs to the field of artificial intelligence, involving artificial intelligence technology, and specifically relates to a distribution time prediction system for in-hospital logistics robots based on artificial intelligence. Background Art

[0002] The distribution time prediction system for logistics robots effectively reduces manual dependence through automated collection and analysis of road section time data, realizes intelligent management of logistics distribution time prediction, and significantly improves operation efficiency. In addition, the system uses robots to undertake part of the manual distribution tasks, which reduces labor costs and avoids time loss caused by human operation errors, providing an efficient and convenient solution for hospital logistics distribution.

[0003] In a conventional distribution scenario, the distribution environment is relatively simple, and there are fewer factors affecting the robot's execution of the distribution plan. Predicting the distribution time through a simple path has a relatively high accuracy; while in a hospital environment, the influencing factors in the distribution scenario are complex, the personnel flow is large and changes dynamically in real time, corresponding to congestion situations during the robot's distribution process, resulting in a delay in the predicted time of the robot's distribution logistics, and thus reducing the overall efficiency of time prediction. Summary of the Invention

[0004] This application aims to solve at least one of the technical problems existing in the prior art; for this purpose, this application proposes a distribution time prediction system for in-hospital logistics robots based on artificial intelligence, which is used to solve the technical problem that the influencing factors in the distribution scenario are complex, the personnel flow is large and changes dynamically in real time, corresponding to congestion situations during the robot's distribution process, resulting in a delay in the predicted time of the robot's distribution logistics, and thus reducing the overall efficiency of time prediction.

[0005] To achieve the above object, the first aspect of this application provides a distribution time prediction system for in-hospital logistics robots based on artificial intelligence, including: a data acquisition module, a data analysis module, a time prediction module, and a database.

[0006] Data acquisition module: Obtain the distribution path; obtain the monitoring data corresponding to each distribution section in the distribution path;

[0007] Data analysis module: Input the monitoring data into the distributed intelligent model to obtain the obstacle distribution map; obtain the global obstacle density and several local obstacle densities in the obstacle distribution map, and analyze based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient;

[0008] Time prediction module: Analyze based on the congestion coefficient to obtain the estimated time for each section of the road; simulate based on the congestion coefficient to obtain the prediction success rate; integrate the estimated time for each delivery section and the prediction success rate to obtain a prediction time report.

[0009] Among them, the delivery route is the distance that the robot travels from the warehouse to the receiving point through the corridor; the delivery route is divided into each delivery section;

[0010] Furthermore, the distributed intelligent model is obtained through training, including:

[0011] A number of monitoring data and corresponding obstacle distribution maps obtained from the database; the obstacle distribution map is obtained by experts analyzing a number of monitoring data; specifically, the obstacle positions are obtained from the images in the monitoring data and mapped to the floor plan of the delivery section; the position coordinates of each obstacle are marked using a scatter plot, and the position coordinates are integrated into an obstacle distribution map; the obstacle distribution map reflects the corresponding distribution of obstacles; a number of monitoring data and corresponding obstacle distribution maps are integrated into a number of training data and test data; the monitoring data includes images corresponding to corridors, halls, etc. in the delivery section; the obstacles include people, hospital beds, sundries, etc.

[0012] Import a number of training data into the artificial intelligence model for training, and test the trained artificial intelligence model with the test data; specifically, input the monitoring data in the test data into the trained artificial intelligence model, and output the obstacle distribution map. Compare the obstacle distribution map with the obstacle distribution map recorded in the test data to determine whether it is within the acceptable range; if so, it means that this set of test data passes the test, and continue to test the next set of test data; if not, the relevant parameters of the artificial intelligence model need to be adjusted, and continue to use this set of test data for testing; until a set proportion of the test data passes the test; finally, an input of monitoring data and an obstacle distribution map, and an output of a distributed intelligent model are obtained; among them, the artificial intelligence model is a BEV model, etc.

[0013] Furthermore, obtaining the global obstacle density and a number of local obstacle densities in the obstacle distribution map includes:

[0014] Analyze the obstacle distribution map to obtain the global obstacle density;

[0015] Perform clustering analysis on the obstacle distribution map to obtain a number of local obstacle densities;

[0016] Furthermore, analyzing the obstacle distribution map to obtain the global obstacle density includes:

[0017] Obtain the total number of obstacles in the delivery section from the obstacle distribution map; obtain the maximum area that the obstacles in the delivery section can accommodate; record the ratio of the total number to the area of the delivery section as the global obstacle density;

[0018] Through the formula Calculate the global obstacle density QJY; where, ZAW represents the total number of obstacles in the delivery section; SP represents the area of the delivery section;

[0019] Furthermore, perform clustering analysis on the obstacle distribution map to obtain the several local obstacle densities, including:

[0020] Obtain the distribution coordinates of several obstacles from the obstacle distribution map; integrate the distribution coordinates into several feature vectors; perform clustering analysis on the several feature vectors to obtain several clustering clusters; obtain the clustering cluster density corresponding to each clustering cluster; mark the clustering cluster density as the local obstacle density.

[0021] Furthermore, analyze based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient, including:

[0022] Obtain the global obstacle density threshold and each local obstacle density threshold from the database; the global obstacle density threshold is the threshold of the density of obstacles in the delivery section; the local obstacle density threshold is the threshold of the density where the obstacles in the delivery section are concentrated;

[0023] When the global obstacle density is greater than the set global obstacle density threshold, then record the ratio of the global obstacle density to the adjustment factor affecting the congestion coefficient as the congestion coefficient;

[0024] Through the formula Calculate the congestion coefficient YD; where QJY represents the global obstacle density; AZ represents the unit density value of the obstacle; β represents the adjustment factor affecting the congestion coefficient; in this embodiment, it should be noted that through the formula, the greater the global obstacle density of the delivery section, the more obstacles there are for the robot in the delivery section, and the more congested the corresponding delivery section is; the greater the corresponding congestion coefficient; the adjustment factor affecting the congestion coefficient in this embodiment is the number of service windows in the delivery section, and the service window is a consulting room, etc.; when the number of service windows in the delivery section is large, its impact on reducing the number of people in the obstacles is greater, and the corresponding congestion coefficient will also decrease accordingly.

[0025] When the global obstacle density is less than the set global obstacle density threshold, then analyze based on several local obstacle densities to obtain the congestion coefficient.

[0026] Furthermore, the analysis based on several local obstacle densities to obtain the congestion coefficient includes:

[0027] Obtain the density of each local obstacle in the current delivery section, and compare the density of each local obstacle with the corresponding local obstacle density threshold in turn; when the local obstacle density exceeds the local obstacle density threshold, mark the status of the local obstacle density as congested; when the local obstacle density does not exceed the local obstacle density threshold, mark the status of the local obstacle density as normal;

[0028] Obtain several local obstacle densities with the status of congestion; analyze them to obtain the corresponding local congestion coefficients; record the sum of each local congestion coefficient as the congestion coefficient;

[0029] The congestion coefficient is the ratio of the sum of the densities of each local obstacle to the adjustment factor affecting the congestion coefficient;

[0030] Through the formula Calculate the congestion coefficient YD; where, JBMm represents the density of each local obstacle; AZ represents the unit density value of the obstacle; m represents the number corresponding to the density of each local obstacle; M is the total number of the densities of each local obstacle; α represents other influencing factors affecting the congestion coefficient; such as the number of service windows, etc.;

[0031] This application analyzes the congestion coefficient through the global obstacle density and several local obstacle densities, so as to more accurately predict the time for the subsequent robot on the delivery path and improve the accuracy of the overall delivery.

[0032] Furthermore, analyzing based on the congestion coefficient to obtain the estimated time of the section includes:

[0033] Obtain the normal running time TC of the delivery section from the database; the sum of the normal running time and the congestion adjustment time is the estimated time of the section; the product of the running time corresponding to the unit congestion coefficient and the congestion coefficient is recorded as the congestion adjustment time;

[0034] The formula for obtaining the estimated time of the section:

[0035] TD = TC + YD × JK

[0036] where, TD represents the estimated time of the section, and JK represents the running time corresponding to the unit congestion coefficient;

[0037] Furthermore, simulating based on the congestion coefficient to obtain the prediction success rate includes:

[0038] Obtain each delivery section between the current position of the robot and the receiving point; obtain each congestion coefficient corresponding to each time period of the delivery section; fit the congestion coefficients of the same delivery section in the order of time periods to obtain the corresponding congestion change curve; then, obtain the congestion change curves corresponding to each delivery section in turn; and generate each congestion slope through each change curve; sum up each slope and perform normalization processing to obtain the prediction success rate;

[0039] Further, fitting each congestion coefficient to obtain the corresponding congestion change curve includes:

[0040] Obtain the congestion change curve by fitting the congestion coefficients of the delivery section in the order of their corresponding time periods through a set fitting method; the fitting method includes interpolation method, etc.;

[0041] Compared with the prior art, the beneficial effects of the present application are:

[0042] 1. The present application obtains the delivery path; obtains the monitoring data corresponding to each delivery section in the delivery path; inputs the monitoring data into the distributed intelligent model to obtain the obstacle distribution map; obtains the global obstacle density and several local obstacle densities in the obstacle distribution map, analyzes based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient; analyzes based on the congestion coefficient to obtain the estimated time of the section; simulates based on the congestion coefficient to obtain the prediction success rate; integrates the estimated time of each delivery section and the prediction success rate corresponding to each delivery section to obtain the prediction time report; the simulation of the estimated time of the section and the prediction success rate based on the congestion coefficient effectively avoids the delivery delay caused by the congestion of the path during the peak period of hospital traffic, dynamically predicts the time of robot delivery in real time, and improves the overall efficiency of prediction time;

[0043] 2. The present application analyzes the congestion coefficient through the global obstacle density and several local obstacle densities, which can more accurately predict the time for the robot on the delivery path and improve the accuracy of the overall delivery. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0045] Figure 1 It is a schematic diagram of the principle of the present application;

[0046] Figure 2 It is a system flowchart of the present application. Specific Embodiments

[0047] The technical solutions of the present application will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0048] Please refer to Figure 1 , an embodiment of the first aspect of the present application provides a time prediction system for in-hospital logistics robots based on artificial intelligence, including:

[0049] Data acquisition module: Obtain the delivery route; obtain the monitoring data corresponding to each delivery section in the delivery route;

[0050] Data analysis module: Input the monitoring data into the distributed intelligence model to obtain the obstacle distribution map; obtain the global obstacle density and several local obstacle densities in the obstacle distribution map, and analyze based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient;

[0051] Time prediction module: Analyze based on the congestion coefficient to obtain the estimated time for each section; simulate based on the congestion coefficient to obtain the prediction success rate; integrate the estimated time for each delivery section and the prediction success rate to obtain the prediction time report.

[0052] Among them, the delivery route is the route that the robot passes through corridors, lobbies, rest areas, etc. from the warehouse to the receiving point; the delivery route is divided into each delivery section; the obstacle distribution map is the map of the obstacle distribution positions in the delivery section; the global obstacle density is the density of obstacles in the delivery section; the local obstacle density is the density of the area where obstacles are concentrated in the delivery section; the congestion coefficient is a quantitative index used to measure the congestion degree of the delivery section; the estimated time for each section is the predicted time for the delivery section; the prediction success rate is the success probability of predicting that the robot will reach the receiving point within the predicted time; the prediction time report is the result of the predicted time for the delivery route.

[0053] Specifically, the distributed intelligence model is obtained through training, including:

[0054] A number of monitoring data obtained from the database and the corresponding obstacle distribution map; the obstacle distribution map is obtained by experts analyzing a number of monitoring data; specifically, the obstacle positions are mapped from the images in the monitoring data to the floor plan of the distribution section; the position coordinates of each obstacle are marked using a scatter plot, and the position coordinates are integrated into an obstacle distribution map; the obstacle distribution map reflects the corresponding distribution of obstacles; a number of monitoring data and the corresponding obstacle distribution maps are integrated into a number of training data and test data; the monitoring data includes images corresponding to corridors, halls, etc. in the distribution section; the obstacles include people, hospital beds, sundries, etc.

[0055] Import a number of training data into the artificial intelligence model for training, and test the trained artificial intelligence model with the test data; specifically, input the monitoring data in the test data into the trained artificial intelligence model, and output the obstacle distribution map. Compare the obstacle distribution map with the obstacle distribution map recorded in the test data to determine whether it is within the acceptable range; if so, it means that this group of test data passes the test, and continue to test the next group of test data; if not, the relevant parameters of the artificial intelligence model need to be adjusted, and continue to use this group of test data for testing; until a set proportion of the test data passes the test; finally, an input is the monitoring data and the obstacle distribution map, and the output is the distribution intelligent model; where the artificial intelligence model is a BEV model, etc.

[0056] Specifically, obtain the global obstacle density and a number of local obstacle densities in the obstacle distribution map, including:

[0057] Obtain the global obstacle density through analysis of the obstacle distribution map;

[0058] Obtain a number of local obstacle densities through cluster analysis of the obstacle distribution map;

[0059] Specifically, obtaining the global obstacle density through analysis of the obstacle distribution map includes:

[0060] Obtain the total number of obstacles in the distribution section through the obstacle distribution map; obtain the maximum area occupied by the obstacles in the distribution section; record the ratio of the total number to the area of the distribution section as the global obstacle density;

[0061] Through the formula Calculate the global obstacle density QJY; where, ZAW represents the total number of obstacles in the distribution section; SP represents the area of the distribution section;

[0062] Specifically, obtaining the number of local obstacle densities through cluster analysis of the obstacle distribution map includes:

[0063] Obtain the distribution coordinates of several obstacles from the obstacle distribution map; integrate the distribution coordinates into several feature vectors; perform clustering analysis on the several feature vectors to obtain several clustering clusters; obtain the clustering cluster density corresponding to each clustering cluster; label the clustering cluster density as the local obstacle density.

[0064] Specifically, analyzing based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient includes:

[0065] Obtain the global obstacle density threshold and each local obstacle density threshold from the database; the global obstacle density threshold is the threshold of the density of obstacles in the delivery section; the local obstacle density threshold is the threshold of the density where obstacles are concentrated in the delivery section;

[0066] When the global obstacle density is greater than the global obstacle density threshold, then the ratio of the global obstacle density to the adjustment factor affecting the congestion coefficient is recorded as the congestion coefficient;

[0067] The congestion coefficient YD is calculated through the formula ; where QJY represents the global obstacle density; AZ represents the unit density value of the obstacle; β represents the adjustment factor affecting the congestion coefficient; in this embodiment, it should be noted that through the formula, the greater the global obstacle density of the delivery section, the more obstacles there are for the robot in the delivery section, and the more congested the corresponding delivery section is; the greater the corresponding congestion coefficient; the adjustment factor affecting the congestion coefficient in this embodiment is the number of service windows in the delivery section, and the service window is a consulting room, etc.; when the number of service windows in the delivery section is large, its impact on reducing the number of people in the obstacles is greater, and the corresponding congestion coefficient will also decrease accordingly.

[0068] When the global obstacle density is less than the global obstacle density threshold, then analyze based on several local obstacle densities to obtain the congestion coefficient.

[0069] Specifically, the analysis based on several local obstacle densities to obtain the congestion coefficient includes:

[0070] Obtain the local obstacle density of each current delivery section, and compare each local obstacle density with the corresponding local obstacle density threshold in turn; when the local obstacle density exceeds the local obstacle density threshold, then mark the status of the local obstacle density as congested; when the local obstacle density does not exceed the local obstacle density threshold, then mark the status of the local obstacle density as normal;

[0071] Obtain several local obstacle densities with the status of congested; and analyze them to obtain the corresponding local congestion coefficients; record the sum of each local congestion coefficient as the congestion coefficient.

[0072] The status of the local obstacle density is marked as congested, which is used to indicate that the area corresponding to the local obstacle density in the distribution section is in a congested state, and congestion occurs during the robot distribution process; the status of the local obstacle density is marked as normal, which is used to indicate that the area corresponding to the local obstacle density in the distribution section is in a normal state, and the robot operates normally during the distribution process.

[0073] The congestion coefficient is the ratio of the sum of each local obstacle density to the adjustment factor affecting the congestion coefficient;

[0074] Each marked local obstacle density is expressed as JBMm; each marked local obstacle density has a corresponding adjustment factor αm for the local congestion coefficient; m = 1, 2,..., M;

[0075] Through the formula The congestion coefficient YD is calculated; where, AZ represents the unit density value of the obstacle, and M represents the total number of marked ones;

[0076] In this embodiment, the greater the marked local obstacle density in the distribution section, the more congested the corridor corresponding to the marked local obstacle density of the robot, and the greater the corresponding congestion coefficient; the adjustment factor for the local congestion coefficient in this embodiment is the number of service windows in the distribution section, and the service window is a consulting room, etc.; when the number of service windows in the corridor corresponding to the marked local obstacle density is large, its impact on the reduction of the number of people in the obstacle is greater, and the corresponding congestion coefficient will also be reduced accordingly.

[0077] This application analyzes the congestion coefficient through the global obstacle density and several local obstacle densities, so as to more accurately predict the time for the robot on the distribution path in the future and improve the accuracy of the overall distribution.

[0078] Specifically, analyzing based on the congestion coefficient to obtain the estimated time of the section includes:

[0079] Obtain the normal running time TC of the distribution section from the database; the sum of the normal running time and the congestion adjustment time is the estimated time of the section; the product of the running time corresponding to the unit congestion coefficient and the congestion coefficient is recorded as the congestion adjustment time;

[0080] The formula for obtaining the estimated time of the section:

[0081] TD = TC + YD × JK

[0082] Among them, TD represents the estimated time of the road section, and JK represents the running time corresponding to the unit congestion coefficient. In this embodiment, it should be noted that the normal running time is the time when the robot runs without congestion. Through the formula, it shows that the larger the congestion coefficient, the more serious the congestion of the corresponding distribution road section, the longer the running time of the robot, and the longer the estimated time of the corresponding road section.

[0083] Specifically, the prediction success rate is obtained through simulation based on the congestion coefficient, including:

[0084] Obtain each distribution road section between the current position of the robot and the receiving point; obtain each congestion coefficient corresponding to each time period of the distribution road section; fit the congestion coefficients of the same distribution road section in the order of time periods into the corresponding congestion change curve; then, obtain the congestion change curves corresponding to each distribution road section in turn; and generate each congestion slope through each change curve; sum up each slope and then perform normalization processing to obtain the prediction success rate.

[0085] In this embodiment, specifically, the corresponding slope is obtained by obtaining each congestion change curve. When the slope is positive, it indicates that the degree of congestion during distribution is gradually increasing, and at this time, the probability that the robot arrives within the estimated time is relatively low; when the slope is negative, it indicates that the degree of congestion during distribution is gradually decreasing; at this time, the probability that the robot arrives within the estimated time is relatively high; through the formula The prediction success value YC is calculated; where ki represents each slope value; then, the prediction success value is normalized to obtain the prediction success rate; the closer the prediction success rate is to the negative value, the greater the probability that the robot arrives within the estimated time; the closer the prediction success rate is to the positive value, the smaller the probability that the robot arrives within the estimated time; the congestion slope is the slope of the congestion coefficient predicted for the next time period through the congestion change curve; the normalization is Z-score standardization.

[0086] Specifically, fitting each congestion coefficient into the corresponding congestion change curve includes:

[0087] The congestion coefficients of the distribution road section are obtained in the order of their corresponding time periods through a set fitting method to obtain the congestion change curve; the fitting method includes interpolation method, etc.; the congestion change curve is a curve used to reflect the congestion change situation.

[0088] Some data in the above formula are calculated by removing the dimension and taking their numerical values. The formula is obtained by software simulation of a large amount of collected data to obtain a formula closest to the actual situation; the preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0089] Working principle of this application: By obtaining the delivery route; obtaining the monitoring data corresponding to each delivery section in the delivery route; inputting the monitoring data into the distributed intelligent model to obtain the obstacle distribution map; obtaining the global obstacle density and several local obstacle densities in the obstacle distribution map, and analyzing based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient; analyzing based on the congestion coefficient to obtain the estimated time of the section; simulating based on the congestion coefficient to obtain the prediction success rate; integrating the estimated time of each delivery section and the prediction success rate to obtain the prediction time report; based on the simulation of the estimated time of the section and the prediction success rate of the congestion coefficient, effectively avoiding the delivery delay caused by the path congestion during the peak hospital flow period, making real-time dynamic prediction of the time of robot delivery, and improving the overall efficiency of the prediction time.

[0090] Please refer to Figure 2 , the second aspect embodiment of this application provides a flowchart of a delivery time prediction system for in-hospital logistics robots based on artificial intelligence;

[0091] The above embodiments are only used to illustrate the technical method of this application and not to limit it. Although this application has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of this application can be modified or equivalently replaced without departing from the spirit and scope of the technical method of this application.

Claims

1. An in-hospital logistics robot delivery time prediction system based on artificial intelligence, characterized in that Data acquisition module: Obtain the delivery route; obtain the monitoring data corresponding to each delivery section in the delivery route; Data analysis module: Input the monitoring data into the distributed intelligence model to obtain the obstacle distribution map; obtain the global obstacle density and several local obstacle densities in the obstacle distribution map, and analyze based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient; Time prediction module: Analyze based on the congestion coefficient to obtain the estimated time for each section; simulate based on the congestion coefficient to obtain the prediction success rate; integrate the estimated time for each delivery section and the prediction success rate to obtain the prediction time report.

2. The time prediction system for in-hospital logistics robot distribution based on artificial intelligence according to claim 1, wherein, The distributed intelligence model is obtained through training, including: Several monitoring data and corresponding obstacle distribution maps obtained from the database; the obstacle distribution map is obtained by experts analyzing several monitoring data; integrate several monitoring data and corresponding obstacle distribution maps into several training data and test data; Import several training data into the artificial intelligence model for training, and test the trained artificial intelligence model with the test data; finally obtain the distributed intelligence model with the monitoring data and obstacle distribution map as the input and the output.

3. The time prediction system for the in-hospital logistics robot based on artificial intelligence according to claim 1, wherein Obtaining the global obstacle density and several local obstacle densities in the obstacle distribution map includes: Analyze through the obstacle distribution map to obtain the global obstacle density; Perform clustering analysis through the obstacle distribution map to obtain several local obstacle densities.

4. The time prediction system for the in-hospital logistics robot based on artificial intelligence according to claim 3, wherein, Analyzing through the obstacle distribution map to obtain the global obstacle density includes: Obtain the total number of obstacles in the delivery section through the obstacle distribution map; obtain the maximum area that the obstacles in the delivery section can accommodate; record the ratio of the total number to the area of the delivery section as the global obstacle density.

5. The time prediction system for in-hospital logistics robot distribution based on artificial intelligence according to claim 3, wherein, Performing clustering analysis through the obstacle distribution map to obtain the several local obstacle densities includes: Obtain the distribution coordinates of several obstacles from the obstacle distribution map; integrate the distribution coordinates into several feature vectors; perform clustering analysis on several feature vectors to obtain several clustering clusters; obtain the clustering cluster density corresponding to each clustering cluster; mark the clustering cluster density as the local obstacle density.

6. The time prediction system for in-hospital logistics robot distribution based on artificial intelligence according to claim 1, wherein, Analyzing based on the global obstacle density and several local obstacle densities to obtain the congestion coefficient includes: Obtain the global obstacle density threshold and each local obstacle density threshold from the database; When the global obstacle density is greater than the set global obstacle density threshold, calculate the congestion coefficient through the ratio of the global obstacle density to the adjustment factor affecting the congestion coefficient; When the global obstacle density is less than the set global obstacle density threshold, analyze based on several local obstacle densities to obtain the congestion coefficient.

7. The time prediction system for in-hospital logistics robot distribution based on artificial intelligence according to claim 6, wherein, The analysis based on several local obstacle densities to obtain the congestion coefficient includes: Obtain the density of each local obstacle in the current delivery section, and compare each local obstacle density with the corresponding local obstacle density threshold in turn; when the local obstacle density exceeds the local obstacle density threshold, mark the status of the local obstacle density as congested; when the local obstacle density does not exceed the local obstacle density threshold, mark the status of the local obstacle density as normal; Obtain several local obstacle densities with the status of congested; and analyze them to obtain the corresponding local congestion coefficients; record the sum of each local congestion coefficient as the congestion coefficient; The congestion coefficient is the ratio of the sum of each local obstacle density to the adjustment factor affecting the congestion coefficient.

8. The time prediction system for the in-hospital logistics robot based on artificial intelligence according to claim 1, characterized in that, Analyze based on the congestion coefficient to obtain the estimated time of the section, including: Obtain the normal running time of the delivery section from the database; the sum of the normal running time and the congestion adjustment time is the estimated time of the section; the congestion adjustment time is the product of the running time corresponding to the unit congestion coefficient and the congestion coefficient.

9. The time prediction system for in-hospital logistics robot distribution based on artificial intelligence according to claim 1, wherein, Simulate based on the congestion coefficient to obtain the prediction success rate, including: Obtain each delivery section between the current position of the robot and the receiving point; obtain each congestion coefficient corresponding to each time period of the delivery section; fit the congestion coefficients of the same delivery section in the order of time periods into the corresponding congestion change curve; then, obtain each congestion change curve corresponding to each delivery section in turn; and generate each congestion slope through each change curve; sum and normalize each slope to obtain the prediction success rate.

10. The time prediction system for in-hospital logistics robot distribution based on artificial intelligence according to claim 9, characterized in that, Fit each congestion coefficient into the corresponding congestion change curve, including: Obtain the congestion change curve by fitting the congestion coefficients of the delivery section in the order of their corresponding time periods through a set fitting method.

Images