A vehicle-human collaborative scheduling method, system, device and medium

Abstract

A vehicle-person cooperative scheduling method, system, device and medium, wherein the method comprises: obtaining personnel state data, vehicle state data, environmental parameters and a detection task list of a detection personnel; matching the personnel state data with each detection task to obtain a task matching degree and determine a detection personnel set for each detection task to divide a candidate vehicle-riding group; analyzing a first detection task location corresponding to a first detection personnel to obtain a first path dispersion coefficient and a first cost comparison index; adjusting the candidate vehicle-riding group based on the first path dispersion coefficient and the first cost comparison index to obtain a target vehicle-riding group; analyzing a second detection task location corresponding to each second detection personnel to obtain a second path dispersion coefficient and a second cost comparison index; determining a vehicle type of each target vehicle-riding group to obtain an allocation result; and generating a vehicle scheduling scheme. The present application can improve the efficiency of vehicle-person cooperative scheduling.

Description

Technical Field

[0001] This application relates to the field of vehicle-pedestrian collaborative scheduling technology, specifically to a vehicle-pedestrian collaborative scheduling method, system, device, and medium. Background Technology

[0002] In urban field inspection, quality supervision, and emergency response scenarios, there are often large-scale personnel deployments and transportation demands. To improve task execution efficiency and resource utilization, the dispatching system plays a crucial role, aiming to achieve the rational allocation and efficient scheduling of personnel and vehicle resources.

[0003] In existing technologies, personnel allocation and vehicle scheduling are typically handled as two relatively independent modules. For example, some systems prioritize matching personnel with the corresponding skills or qualifications based on task requirements, and then arrange transportation vehicles according to the task location and personnel location; other systems focus on optimizing vehicle routes and reducing travel costs, and then carry out personnel loading or allocation on this basis.

[0004] However, when faced with complex application scenarios such as wide task distribution areas, dispersed personnel starting points, and dynamic changes in traffic conditions, considering personnel matching or vehicle scheduling in isolation may lead to a lack of global coordination capabilities in the scheduling strategy. For example, although personnel may be highly matched with tasks, their routes may be scattered, resulting in high travel costs; or the vehicle routes may be optimal, but the personnel's capabilities may not match the task requirements, leading to resource waste and reduced efficiency in vehicle-person collaborative scheduling. Summary of the Invention

[0005] This application provides a vehicle-person cooperative scheduling method, system, device, and medium to improve the efficiency of vehicle-person cooperative scheduling.

[0006] The first aspect of this application provides a vehicle-pedestrian cooperative scheduling method applied in a server. The method includes: acquiring personnel status data, vehicle status data, environmental parameters, and a list of detection tasks; the personnel status data includes current location information, qualification level, and skill tag information; the environmental parameters include weather conditions and road traffic conditions; matching the personnel status data with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task, and determining the set of detection personnel for each detection task; dividing the set of detection personnel into multiple candidate passenger groups, and determining the first detection personnel corresponding to each first detection personnel in each candidate passenger group. The task location is determined, and the first path dispersion coefficient and first cost comparison index for each candidate passenger group are obtained. Based on the first path dispersion coefficient and first cost comparison index, the candidate passenger groups are adjusted to obtain the target passenger groups. Based on the preset ensemble model, the second detection task location corresponding to each second detection personnel in each target passenger group is analyzed to obtain the second path dispersion coefficient and second cost comparison index for each candidate passenger group. Based on the preset ensemble model and the second path dispersion coefficient and second cost comparison index, the vehicle type of each target passenger group is determined to obtain the allocation result. Based on the vehicle status data, allocation result, weather condition parameters, and road traffic status parameters, a vehicle scheduling plan is generated.

[0007] Optionally, the personnel status data is matched with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task. Specifically, this includes: constructing a task feature vector based on the skill requirements and qualification level requirements of each detection task in the detection task list; mapping the skill tag information and qualification level in the personnel status data to personnel feature vectors; calculating the qualification similarity between the personnel feature vector of each detection personnel and the task feature vector of each detection task to obtain a feature matching coefficient matrix for personnel-task pairs; calculating the spatial distance coefficient between the current location information of each detection personnel and the location information of each detection task to obtain a spatial distance coefficient matrix; and calculating the task matching degree based on the feature matching coefficient matrix and the spatial distance coefficient matrix.

[0008] Optionally, the first detection task location corresponding to each first detection person in each candidate ride-hailing group is determined, and the first path dispersion coefficient and first cost comparison index of each candidate ride-hailing group are obtained. Specifically, this includes: analyzing the location distribution characteristics of the first detection task locations of the first detection persons in each candidate ride-hailing group; determining the maximum path overlap between target paths based on environmental parameters, where the target path is the travel path of each first detection person from the current location to the first detection task location; calculating the first path dispersion coefficient of each candidate ride-hailing group based on the location distribution characteristics and the maximum path overlap; and calculating the connection cost of owned vehicles and ride-hailing vehicles based on the number of first detection persons in each candidate ride-hailing group and the first path dispersion coefficient, thereby obtaining the first cost comparison index.

[0009] Optionally, the maximum path overlap between target paths is determined based on environmental parameters, specifically including: determining the capacity adjustment coefficient for each traffic path based on weather condition parameters and road traffic status parameters; evaluating all candidate traffic paths for each first inspector based on the capacity adjustment coefficient, and selecting a set of target traffic paths that meet preset traffic conditions; calculating the length of overlapping road segments between the first target traffic path set and the second target traffic path set, and generating a path overlap weight between the first target traffic path set and the second target traffic path set by combining the length of overlapping road segments with the corresponding capacity adjustment coefficient of each overlapping road segment, wherein the first target traffic path set and the second target traffic path set are any set of target traffic paths from multiple target traffic path sets; and determining the maximum path overlap based on the path overlap weight.

[0010] Optionally, based on location distribution characteristics and maximum path overlap, a first path dispersion coefficient is calculated for each candidate passenger group. Specifically, this includes: constructing a path coordination adjustment factor based on the maximum path overlap; calculating the ratio of the average location spacing to the maximum path overlap in the location distribution characteristics to obtain the effective distribution dispersion; normalizing the maximum location span in the location distribution characteristics to obtain the target maximum location span; weighting the effective distribution dispersion with the target maximum location span to obtain the path dispersion value of the candidate passenger group; and normalizing the path dispersion value and mapping it to a preset standard score interval to obtain the first path dispersion coefficient.

[0011] Optionally, the candidate passenger groups are adjusted based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group. Specifically, this includes: screening the candidate passenger groups based on a preset path dispersion threshold and a cost comparison reference value to obtain the passenger group to be adjusted; calculating the path outlier contribution value of the target first detection task location to the first path dispersion coefficient of the adjusted passenger group based on the target first detection task location corresponding to each first detection personnel in the adjusted passenger group; identifying the first detection personnel whose path outlier contribution value is greater than the preset outlier rate threshold as the detection personnel to be adjusted; determining the migration plan for the detection personnel to be adjusted; and obtaining the target passenger group based on the migration plan.

[0012] Optionally, a migration plan for the personnel to be adjusted is determined, and a target passenger group is obtained based on the migration plan. Specifically, this includes: analyzing the path overlap between the target travel path of the personnel to be adjusted and each target travel path in the other candidate passenger groups and the passenger group to be adjusted; determining the migration passenger group based on the path overlap; simulating the migration of the personnel to be adjusted to the migration passenger group to obtain the target passenger group to be adjusted and the target migration passenger group after migration; calculating the first path dispersion coefficient of the target passenger group to be adjusted and the target migration passenger group respectively; when the first path dispersion coefficient of the target passenger group to be adjusted and the target migration passenger group are both less than a preset path dispersion threshold, a migration plan is obtained; and the passenger group to be adjusted is adjusted based on the migration plan to obtain the target passenger group.

[0013] A second aspect of this application provides a vehicle-pedestrian cooperative scheduling system, comprising: an acquisition module for acquiring personnel status data, vehicle status data, environmental parameters, and a list of inspection tasks, wherein the personnel status data includes current location information, qualification level, and skill tag information, and the environmental parameters include weather condition parameters and road traffic status parameters; a matching module for matching the personnel status data with each inspection task in the inspection task list to obtain the task matching degree between the inspection personnel and each inspection task, and determining the set of inspection personnel for each inspection task; and a first determination module for dividing the set of inspection personnel into multiple candidate riding groups, determining the location of the first inspection task corresponding to each first inspection personnel in each candidate riding group, and obtaining... The system comprises the following modules: a first path dispersion coefficient and a first cost comparison index for each candidate passenger group; an adjustment module, used to adjust the candidate passenger groups based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group; an analysis module, used to analyze the second detection task location corresponding to each second detection personnel in each target passenger group based on a preset integrated model to obtain the second path dispersion coefficient and the second cost comparison index for each candidate passenger group; a second determination module, used to determine the vehicle type of each target passenger group based on the second path dispersion coefficient and the second cost comparison index according to the preset integrated model to obtain the allocation result; and a generation module, used to generate a vehicle scheduling plan based on vehicle status data, allocation results, weather condition parameters, and road traffic status parameters.

[0014] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described above.

[0015] In a fourth aspect, this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the foregoing descriptions.

[0016] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] 1. By acquiring multi-source heterogeneous information such as personnel status data, vehicle status data, environmental parameters, and detection task data, a unified perception and integration of human resources, transportation tools, task requirements, and the external environment can be achieved, constructing a scheduling information framework with dynamic adaptability. This provides comprehensive and accurate data support for human-vehicle collaborative scheduling, making subsequent scheduling decisions more rational and efficient. Utilizing task matching rules and clustering analysis algorithms, combined with a Stacking ensemble model, multi-dimensional evaluation and optimization of passenger group path distribution and cost indicators are performed, realizing intelligent segmentation and dynamic adjustment of passenger groups. Furthermore, through a scheduling type prediction module, vehicle status, meteorological factors, and road traffic conditions are integrated to generate a hybrid scheduling scheme covering both owned vehicles and ride-hailing vehicles. This scheme not only achieves optimal allocation of human and vehicle resources but also possesses good responsiveness and scalability. Ultimately, a task-scenario-oriented, real-time updated, and collaboratively efficient scheduling model is constructed, dynamically reflecting the spatial distribution of personnel and vehicles, task execution paths, and scheduling resource status, improving the efficiency of vehicle-human collaborative scheduling.

[0018] 2. Through the path aggregation evaluation module, combining the task location distribution characteristics and travel path overlap of the testers within candidate ride-hailing groups, the path concentration of different ride-hailing groups can be quantified, thereby constructing a path dispersion coefficient reflecting the potential for path optimization. This coefficient can effectively reveal the spatial correlation between tasks within a group, providing a quantitative basis for subsequent path integration and resource scheduling. Simultaneously, by combining the number of testers and the path dispersion coefficient, the cost difference between using owned vehicles and ride-hailing services is further evaluated, forming the primary cost comparison indicator and enabling intelligent selection of cost-sensitive scheduling strategies. This process fully integrates task geographic distribution, traffic environment parameters, and vehicle availability information, making the cost assessment results more objective, refined, and dynamic.

[0019] 3. By introducing a capacity adjustment coefficient and combining weather and road traffic parameters, the travel routes of each inspection personnel are dynamically evaluated and screened. This effectively eliminates route options with low traffic efficiency or risks, ensuring that route selection has high feasibility and safety. Furthermore, by calculating the overlapping road segments and their capacities among the target travel routes of different inspection personnel, a path overlap weight index is constructed. This index comprehensively reflects the degree of overlap between different passenger groups on their actual travel routes. This index not only considers the spatial overlap of routes but also introduces a capacity weight correction, making the path overlap assessment more realistic, accurate, and meaningful for actual traffic. The maximum path overlap determined by this path overlap weight provides a key input for path aggregation strategies, enabling quantitative analysis of path integration potential. Compared to traditional scheduling methods that only perform static or geometric analysis of path overlap, this improves the dynamism and practicality of path aggregation assessment. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the system architecture of an embodiment of a vehicle-pedestrian collaborative scheduling method or a vehicle-pedestrian collaborative scheduling system applied in this application.

[0021] Figure 2 This is a flowchart illustrating a vehicle-pedestrian collaborative scheduling method according to an embodiment of this application;

[0022] Figure 3 This is a schematic diagram of the structure of a vehicle-pedestrian collaborative scheduling system according to an embodiment of this application;

[0023] Figure 4 This is a schematic diagram of the structure of the electronic device in the embodiments of this application.

[0024] Explanation of reference numerals in the attached drawings: 301, Acquisition module; 302, Matching module; 303, First determination module; 304, Adjustment module; 305, Analysis module; 306, Second determination module; 307, Generation module; 401, Processor; 402, Communication bus; 403, User interface; 404, Network interface; 405, Memory. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0026] Figure 1 An exemplary system architecture 100 is shown, which can be applied to an embodiment of a vehicle-pedestrian cooperative scheduling method or a vehicle-pedestrian cooperative scheduling system of this application.

[0027] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables. Users can use terminal devices 101, 102, and 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications may be installed on terminal devices 101, 102, and 103, such as model training applications, video recognition applications, web browser applications, and social media platform software. Terminal devices 101, 102, and 103 can be hardware or software.

[0028] Server 105 can be a server providing various services, such as a backend server for processing data displayed on terminal devices 101, 102, and 103. The backend server can analyze and process the received data and feed back the processing results (such as recognition results) to the terminal devices. It should be noted that the server can be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules (such as multiple software programs or software modules used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0029] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included. In particular, if the target data does not need to be obtained remotely, the above system architecture may exclude the network and include only terminal devices or servers.

[0030] Figure 2 This is a flowchart illustrating a vehicle-pedestrian collaborative scheduling method in an embodiment of this application.

[0031] Please see Figure 2 This application provides a vehicle-pedestrian collaborative scheduling method, applied in a server, the method comprising:

[0032] S201. Obtain personnel status data, vehicle status data, environmental parameters, and a list of inspection tasks. The personnel status data includes current location information, qualification level, and skill tag information. The environmental parameters include weather conditions and road traffic conditions.

[0033] In this embodiment, in order to achieve efficient scheduling of vehicle-person collaboration, information collection operations are first performed on the server, specifically including obtaining personnel status data of inspection personnel, vehicle status data, environmental parameters, and inspection task list.

[0034] Personnel status data is collected in real-time via a connection to the mobile terminals (such as smartphones) of the testing personnel, obtaining their current location information through a positioning module. This current location information reflects the spatial distribution of personnel and forms the basis for subsequent cluster analysis and vehicle group segmentation. Simultaneously, the system retrieves the qualification level of each testing personnel from the personnel database. This level is set based on their training certificates, certifications, and other information to determine their ability to complete specific testing tasks. Furthermore, the system acquires skill tag information for each testing personnel. This information is constructed based on their past task experience, technical expertise, and other dimensions. The tags are stored in a structured format for easy matching with subsequent task requirements.

[0035] Vehicle status data is divided into two categories: primary vehicle status data for owned vehicles and secondary vehicle status data for ride-hailing vehicles. Primary vehicle status data is updated in real-time by the internal vehicle management system, including the vehicle's current location, availability, remaining mileage, and battery level, used to assess dispatch feasibility. Secondary vehicle status data is obtained through open interfaces with third-party ride-hailing platforms, acquiring information such as the current location, estimated arrival time, and service cost of vehicles to be dispatched. This allows for the dynamic introduction of social vehicle resources to supplement capacity when owned vehicle resources are insufficient or dispatch efficiency is low.

[0036] Simultaneously acquired environmental parameters include weather conditions and road traffic status parameters. Weather conditions, obtained through access to a meteorological service platform, include information such as rainfall and wind speed, used to assess the impact of severe weather on travel and vehicle scheduling. Road traffic status parameters, obtained through access to a traffic information system, include road congestion, closure information, and traffic restrictions, used to predict the efficiency of vehicle travel routes and potential delay risks. These environmental parameters, as dynamic variables input into the scheduling model, help generate more accurate and reliable scheduling schemes.

[0037] The inspection task list is pre-generated by the task assignment module and includes the task number, location, requirements, required skill tags, and completion deadline for all inspection tasks to be executed. This list serves as static reference data and, together with dynamically collected personnel status data, is input into the subsequent task matching module to build the foundation for vehicle-person task matching.

[0038] S202. Match the personnel status data with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task, and determine the set of detection personnel for each detection task;

[0039] In step S202, the personnel status data is matched one by one with each detection task in the detection task list, and the task matching degree between each detection personnel and each detection task is calculated. The matching degree, as a comprehensive indicator measuring the suitability between personnel and tasks, integrates the matching relationship between personnel capabilities and task requirements, as well as the impact of spatial location information on scheduling efficiency. The matching degree result serves as the basis for subsequently determining the detection personnel set. By comparing it with preset matching rules, the detection personnel set for each detection task is determined, thus providing a personnel foundation for subsequent passenger group division and scheduling path optimization. Optionally, matching the personnel status data with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task may include: constructing a task feature vector based on the skill requirements and qualification level requirements of each detection task in the detection task list; mapping the skill tag information and qualification level in the personnel status data to personnel feature vectors; calculating the qualification similarity between the personnel feature vector of each detection personnel and the task feature vector of each detection task to obtain a feature matching coefficient matrix for personnel-task pairs; calculating the spatial distance coefficient between the current location information of each detection personnel and the location information of each detection task to obtain a spatial distance coefficient matrix; and calculating the task matching degree based on the feature matching coefficient matrix and the spatial distance coefficient matrix.

[0040] In the specific implementation process, the first step is to construct the task feature vector for each detection task. The server extracts the core competency requirements for each task in the detection task list based on its skill and qualification level requirements, and converts these requirements into a structured, multi-dimensional numerical vector, forming the task feature vector. For example, if a detection task requires the skill of "high-voltage electrical detection" and the qualification level of "advanced operator," the system combines the corresponding encoding values ​​of the task's skill tags in the skill tag library with the level weights to construct the task feature vector. During vector construction, a standardization process is used to unify the feature values ​​of different dimensions to the same order of magnitude to ensure the accuracy and stability of subsequent similarity calculations. Through vectorization, the originally discrete and unstructured task requirement information can be transformed into a mathematical model form that is easy for computers to process.

[0041] Subsequently, the server vectorizes the skill tag information and qualification level contained in the personnel status data to generate personnel feature vectors. This process is consistent with the task feature vector construction method, ensuring that the two vector systems have the same dimensions and feature dimension order. The skill tag information comes from the individual skill profile and contains multiple skill items, each of which is assigned a corresponding skill intensity value or experience value. The qualification level is converted from textual information to numerical level through a level mapping table. In this way, the system realizes quantitative modeling of personnel capabilities, which facilitates subsequent similarity calculation between vectors. Taking a certain person as an example, whose skill tags include "high voltage detection" and "instrument calibration", and whose level is "intermediate operator", their personnel feature vector may be (0.8, 0.6, 2), where the first two items represent skill intensity and the last item represents the level value.

[0042] Next, the server calculates the similarity between the personnel feature vector of each inspector and the task feature vector of each inspection task, obtaining a feature matching coefficient. The similarity calculation uses a cosine similarity algorithm: this algorithm evaluates similarity based on the cosine of the angle between two vectors; the smaller the angle, the higher the similarity. The calculation results form a personnel-task pair feature matching coefficient matrix, where each element reflects the degree of matching between a person and a task in terms of ability and qualifications. This matching coefficient serves as a quantitative basis, reflecting whether a person possesses the ability to complete the task. For example, if the feature matching coefficient between person A and task X is 0.95, it indicates that person A's abilities are highly matched with the requirements of task X.

[0043] Based on matching personnel and task capabilities, the server further considers task execution efficiency and scheduling costs, incorporating spatial factors into the matching evaluation. Specifically, the server calculates the spatial distance between personnel based on their current location information in the personnel status data and the task location information in the task list. Spatial distance is calculated using geographic coordinate methods, such as the Haversine formula for route distance estimation. The calculated distance value is converted into a spatial distance coefficient, representing the ease with which a person can reach the task location. All spatial distance coefficients between personnel and tasks constitute a spatial distance coefficient matrix. The spatial distance coefficient reflects the accessibility and efficiency of the scheduling path and is a crucial factor affecting scheduling time costs. For example, if personnel B is only 1.5 kilometers from task Y, their spatial distance coefficient will be higher than that of personnel C, who is 10 kilometers from task Y.

[0044] Finally, the server performs a weighted fusion of the feature matching coefficient matrix and the spatial distance coefficient matrix to calculate the overall task matching degree. The matching degree calculation uses a weighted linear combination model, where the weight of the feature matching coefficient reflects the importance of task completion capability, and the weight of the spatial distance coefficient reflects the priority of scheduling efficiency. The weight coefficients can be dynamically adjusted based on factors such as task urgency and personnel scarcity. For example, for highly specialized tasks, the weight of the feature matching coefficient can be increased; for time-sensitive tasks, the weight of the spatial distance coefficient can be increased. The fused task matching degree serves as the final adaptation indicator between personnel and tasks, providing a quantitative reference for subsequent personnel set selection. Based on the task matching degree ranking results for each task, the system, combined with preset matching rules (such as matching degree thresholds, maximum candidate number limits, etc.), selects a set of testing personnel and binds them to tasks to complete the initial task allocation.

[0045] Taking a real-world example, Task Z requires "infrared detection" skills and an "advanced operator" level, and is located in Area A. Personnel A possesses these skills and levels and is currently located near Area A, with a matching score of 0.93. Personnel B also possesses the same skills but is located further away, with a matching score of 0.78. The system prioritizes assigning Personnel A to Task Z based on their matching score, achieving a dual optimization of capability and efficiency.

[0046] After the task matching degree is calculated, the set of personnel for each detection task is determined. Specifically, the server first sorts the matching degree of all personnel corresponding to each detection task, ranking them from highest to lowest to form a list of matching personnel for each task. Then, the system uses preset matching rules to filter this list. These preset matching rules are set by the scheduling strategy module and typically include, but are not limited to: minimum matching degree threshold, maximum candidate limit, skill category balance requirements, and task priority restrictions. For example, the system can set a minimum matching degree threshold, requiring matching personnel for a given task to have a matching degree of at least 0.75; personnel with a lower value, even if ranked high in the sorting, will not be included in the set of detection personnel. The maximum candidate limit controls the upper limit of task resource allocation, preventing excessive concentration of personnel resources on a small number of tasks. During the filtering process, the system can also dynamically adjust based on the urgency of the task. For example, for tasks with high urgency, the system can appropriately relax the matching degree threshold to increase personnel coverage; while for tasks with high technical barriers, the weight of skill matching degree is increased, prioritizing personnel with higher qualifications. In addition, to prevent a certain skill category from being too concentrated in the personnel set, the system can also enable a skill tag distribution balancing mechanism to constrain the skill types in the personnel set and ensure that multi-skill collaborative tasks are actually feasible.

[0047] Finally, the server selects personnel who meet the matching requirements and comply with the matching rules as the set of personnel for the detection task, and binds this set to the task to form a structured task-person relationship table.

[0048] S203. Divide the set of testing personnel into multiple candidate bus groups, determine the first testing task location corresponding to each first testing personnel in each candidate bus group, and obtain the first path dispersion coefficient and the first cost comparison index for each candidate bus group.

[0049] After completing the screening of the detection personnel set for the detection task, the server performs cluster analysis on the current location information of the target detection personnel in all detection personnel sets, and divides the target detection personnel who are relatively close in location into multiple candidate bus groups based on geographic spatial distribution.

[0050] In the specific implementation process, the server first extracts the current location information of all personnel being detected. This information comes from real-time location data in the personnel status data, typically expressed in latitude and longitude coordinates. To ensure the geographic accuracy of the clustering analysis, the server performs format normalization and spatial projection transformation on the coordinate data before processing it, ensuring that all coordinate points are comparable in a unified plane coordinate system. Subsequently, the server inputs the processed location information into the clustering analysis model, which uses the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm. This algorithm can automatically identify multiple relatively dense personnel gathering areas based on the spatial density relationship between points without needing to preset the number of clusters.

[0051] The core principle of the DBSCAN algorithm lies in setting two parameters: neighborhood radius and minimum number of neighboring points. For any coordinate point of a target person, if there are other person points within a specified neighborhood radius that are no less than the minimum number of neighboring points, then this point is considered a core point, and all points densely connected to it are grouped into the same cluster. This algorithm can effectively identify naturally formed areas of personnel clustering and has strong noise resistance; that is, isolated individual points that do not form a clustering effect can be automatically identified as noise points, thus avoiding their interference with the passenger group division. This clustering method does not require manual specification of the number of groups and can automatically adapt to the actual distribution of target personnel, exhibiting strong flexibility and practicality.

[0052] After clustering analysis, the server groups individuals belonging to the same cluster into candidate ride-sharing groups based on the clustering results. Members in each candidate ride-sharing group are spatially adjacent, making it feasible for them to share ride-sharing resources. Individuals not identified as core or boundary points by the clustering algorithm, i.e., noise points, are not immediately included in candidate ride-sharing groups by the server. Instead, they are awaited for subsequent ride-sharing group adjustments or individual scheduling to ensure optimal allocation of scheduling resources.

[0053] To achieve intelligent evaluation of the path aggregation effect and scheduling cost of candidate passenger groups, the server calls a preset ensemble model deployed in the scheduling system to analyze the first detection task location corresponding to each first detection person in each candidate passenger group. This ultimately yields the first path dispersion coefficient and first cost comparison index for each candidate passenger group. The preset ensemble model can be a Stacking model. As a multi-level ensemble learning structure, the Stacking model integrates multiple basic models and a meta-learner, enabling complementary predictive capabilities among different models, thereby significantly improving the accuracy and robustness of the results. The Stacking model is preset as the core of the scheduling decision, internally including a path aggregation evaluation module and a scheduling type prediction module. The path aggregation evaluation module is specifically used to analyze the spatial distribution and path overlap of task locations, and calculates the scheduling cost index in conjunction with environmental parameters, providing basic data for subsequent vehicle type allocation and passenger group adjustment. Step S203, determining the first detection task location corresponding to each first detection person in each candidate passenger group and obtaining the first path dispersion coefficient and first cost comparison index for each candidate passenger group, may include steps S2031-S2034:

[0054] S2031. Analyze the location distribution characteristics of the first detection task location for the first detection personnel in each candidate passenger group;

[0055] The server first extracts the coordinate information of the first detection task locations corresponding to all first detection personnel in each candidate passenger group. This coordinate information is structured latitude and longitude data, derived from the task address locations recorded in the detection task list. The geocoding module accurately converts the address information into standardized coordinates. These coordinate points are then uniformly mapped to a two-dimensional geospatial coordinate system to construct the task location set for that passenger group. Based on this, the path aggregation evaluation module performs spatial statistical analysis on this set. First, the geometric center point of all task points is calculated as the central reference for the task distribution. The geometric center point is obtained by the arithmetic mean of all coordinate points on the horizontal and vertical axes, used to measure the overall centroid of the task point distribution. Subsequently, the server calculates the Euclidean distance of each task point relative to the geometric center point and determines the standard deviation of all distances; a larger standard deviation indicates a higher degree of dispersion between task points. Furthermore, the spatial expansion degree can be further evaluated by calculating the minimum circumscribed circle radius or the convex hull area of ​​the task point set. All these indicators together constitute a measurement system for location distribution characteristics.

[0056] To facilitate subsequent model processing, the server normalizes the aforementioned spatial statistical indicators, converting them into standardized spatial concentration coefficients. The numerical range is typically limited to (0, 1). A value closer to 0 indicates a more concentrated distribution of task points and stronger path aggregation potential; a value closer to 1 indicates a more dispersed distribution of task points and higher difficulty in unified path scheduling. This spatial concentration coefficient serves as a crucial input in calculating the path dispersion coefficient, working in conjunction with subsequent path overlap information to assess the consistency level of scheduled paths. For example, a candidate passenger group may contain three first-stage inspectors whose task locations are situated in different factories within the same industrial park. The maximum distance between their coordinates is no more than 1.2 kilometers, and the standard deviation of all task points relative to the center point is only 0.35 kilometers. The final normalized spatial concentration coefficient is 0.12, indicating a high concentration of task locations for this passenger group and a good foundation for path aggregation.

[0057] S2032. Determine the maximum path overlap between target paths based on environmental parameters. The target path is the travel path of each first inspection personnel from the current location to the first inspection task location.

[0058] After analyzing the location distribution characteristics of the first detection task locations in the candidate passenger groups, the maximum path overlap between target paths is determined based on environmental parameters. Path overlap, as a core indicator measuring the degree of overlap between multiple travel paths, directly affects the path sharing rate and connection efficiency during unified scheduling. Since environmental conditions such as weather and road traffic conditions significantly affect path feasibility and traffic capacity, a traffic capacity adjustment mechanism needs to be introduced during the path overlap assessment process to dynamically correct the path assessment and more accurately reflect the feasibility and overlap value of the actual paths. Optionally, determining the maximum path overlap between target paths based on environmental parameters may include the following steps: determining the capacity adjustment coefficient for each traffic path based on weather condition parameters and road traffic status parameters; evaluating all candidate traffic paths for each first inspector based on the capacity adjustment coefficient, and selecting a set of target traffic paths that meet preset traffic conditions; calculating the length of overlapping road segments between the first target traffic path set and the second target traffic path set, and generating a path overlap weight between the first target traffic path set and the second target traffic path set by combining the length of overlapping road segments with the corresponding capacity adjustment coefficient of each overlapping road segment, wherein the first target traffic path set and the second target traffic path set are any set of target traffic paths from multiple target traffic path sets; and determining the maximum path overlap based on the path overlap weight.

[0059] In practice, the server first determines the capacity adjustment coefficient for each travel path based on weather and road traffic status parameters included in the environmental parameters. Weather parameters include indicators such as rainfall, snowfall, visibility, and temperature, reflecting the impact of current weather conditions on travel safety and speed. Road traffic status parameters include congestion levels, traffic incidents, and traffic restriction information, reflecting road accessibility and efficiency. The server maps these parameters to a path's capacity adjustment coefficient using a rule engine. This coefficient is a normalized value between 0 and 1, with lower values ​​indicating lower path efficiency. For example, if a road's average speed decreases by 30% due to congestion and continuous rainfall, its capacity adjustment coefficient can be set to 0.7 to reduce the path's weight in subsequent evaluations.

[0060] After obtaining the traffic capacity adjustment coefficient, the server evaluates all alternative routes for each first inspection personnel from their current location to the first inspection task location based on this coefficient, and filters out a set of target routes that meet preset traffic conditions. The route generation is based on a map path engine; the system generates several possible routes for each personnel, including the shortest path, fastest path, and traffic avoidance paths, and filters out paths with traffic capacity below a set threshold using the traffic capacity adjustment coefficient, retaining only the set of target routes that are practically feasible under the current environmental conditions. This filtering process ensures that subsequent path overlap calculations are based on the feasible path set, avoiding the impact of abnormal weather or road conditions causing inaccessible routes on the evaluation results.

[0061] After obtaining the target travel path sets of all first-time inspectors, the server further analyzes the path overlap between any two target travel path sets and calculates their path overlap weight. To this end, the system first performs path segment encoding on each path, converting each path into a sequence of segments composed of continuous road segment identifiers. Then, it compares the paths in any two target travel path sets pairwise, identifies overlapping road segments, and calculates the total length of these overlapping segments. Based on this, the length of each overlapping road segment is weighted by its corresponding capacity adjustment coefficient to generate a path overlap weight. This weight not only reflects the degree of overlap in path structure but also incorporates the actual travel value under environmental influences, thus providing a more comprehensive evaluation of the sharing potential between paths.

[0062] After calculating the path overlap weights among all path pairs, the server selects the path pair with the highest weight value as the maximum path overlap for the candidate passenger group. Maximum path overlap, as a key indicator for path aggregation evaluation, reflects the overlap capability of the optimal path-sharing pair within the passenger group. A higher value indicates a greater degree of path overlap during unified passenger scheduling, stronger path aggregation of the passenger group, lower scheduling costs, and higher scheduling efficiency. This indicator will serve as one of the core variables in the subsequent calculation of the path dispersion coefficient and indirectly influence the selection of vehicle scheduling methods.

[0063] For example, if the optimal travel routes of passengers A and B in a candidate passenger group overlap by 3.5 kilometers, and the capacity adjustment coefficient for this segment is 0.9, then their path overlap weight is 3.15 kilometers. Passengers A and C have a 2.8-kilometer overlap, but their capacity coefficient is 0.95, so their overlap weight is 2.66 kilometers. The system uses 3.15 kilometers as the overlap weight corresponding to the maximum path overlap, serving as a representative indicator of the passenger group's path aggregation capability, providing a quantitative basis for subsequent path dispersion calculations and cost comparisons.

[0064] S2033. Based on location distribution characteristics and maximum path overlap, calculate the first path dispersion coefficient for each candidate passenger group;

[0065] After analyzing the location distribution characteristics of candidate passenger groups and calculating the maximum path overlap, to further quantify the scheduling difficulty of each passenger group in terms of path coordination and spatial distribution, the server calculates the first path dispersion coefficient for each candidate passenger group based on the location distribution characteristics and the maximum path overlap. The first path dispersion coefficient, as a key indicator measuring whether the task execution paths within a passenger group are highly dispersed, directly affects the path integration efficiency and travel cost during vehicle scheduling. This coefficient needs to comprehensively consider the spatial dispersion of task points and the structural aggregation of travel paths to construct a standard quantitative parameter that can be used for scheduling optimization. This may include the following steps: constructing a path coordination adjustment factor based on the maximum path overlap; calculating the ratio of the average location spacing to the maximum path overlap in the location distribution characteristics to obtain the effective distribution dispersion; normalizing the maximum location span in the location distribution characteristics to obtain the target maximum location span; weighting the effective distribution dispersion and the target maximum location span to obtain the path dispersion value of the candidate passenger group; and normalizing the path dispersion value and mapping it to a preset standard score interval to obtain the first path dispersion coefficient.

[0066] In the specific implementation process, to calculate the path dispersion coefficient, the server first constructs a path coordination adjustment factor based on the maximum path overlap. The path coordination adjustment factor reflects the overlap advantage between task paths in the passenger group. A higher overlap indicates a longer shareable path segment among the detection personnel, stronger path coordination, and greater scheduling integration potential. The server normalizes the maximum path overlap, typically retaining the original ratio, and uses it as a positive factor in adjusting dispersion in subsequent calculations to enhance path aggregation capability. Based on the constructed path coordination adjustment factor, the server further combines the average location spacing in the location distribution characteristics with the maximum path overlap to calculate the effective distribution dispersion. The average location spacing refers to the average straight-line distance between all first detection task locations in the passenger group, reflecting the uniformity of task points in space. By ratioing this average spacing to the maximum path overlap, the effective distribution dispersion is obtained, representing the actual spatial dispersion between task points considering path merging. A higher index indicates that although paths partially overlap, the task points themselves are still relatively dispersed, increasing scheduling difficulty.

[0067] After obtaining the effective distribution dispersion, the server normalizes the maximum location span in the location distribution characteristics to obtain the target maximum location span. The maximum location span refers to the distance between the two farthest task points within a passenger group, representing the limit of the overall task coverage. To avoid evaluation bias caused by differences in city or regional scales, the server normalizes this value according to the regional scope set by the city management unit, transforming it into a standardized indicator. This indicator is used to assess whether the scheduling path has the risk of cross-regional travel; the larger the span, the higher the difficulty of path integration. Subsequently, the server performs a weighted calculation of the effective distribution dispersion and the target maximum location span to obtain the path dispersion value. The weighted model adopts a linear combination form, where the effective distribution dispersion has a higher weight than the target maximum location span because the former integrates path structure and distribution characteristics, and is a more direct indicator affecting scheduling efficiency. The weight ratio of the weighted model can be set to 0.6 and 0.4 without special strategy settings. Simultaneously, the system supports flexible adjustment of the weight distribution based on task urgency or time window, realizing dynamic adaptive adjustment of the scheduling strategy.

[0068] Finally, the server maps the obtained path dispersion values ​​to a preset standard score range, generating the first path dispersion coefficient. This mapping process uses a normalization function, such as the Min-Max normalization method, to uniformly compress all path dispersion values ​​into a continuous numerical range between 0 and 1, facilitating subsequent horizontal comparisons with the path dispersion coefficients of other passenger groups. The closer the first path dispersion coefficient is to 1, the higher the overall path dispersion of the passenger group, making it unsuitable for shared scheduling; the closer it is to 0, the higher the path aggregation, making it suitable for unified passenger connections.

[0069] S2034. Based on the number of first testing personnel and the first path dispersion coefficient of each candidate passenger group, calculate the connection cost of owned vehicles and ride-hailing vehicles to obtain the first cost comparison index.

[0070] After obtaining the first path dispersion coefficient for each candidate passenger group, the server further calculates the connection cost required to complete the dispatch task for that passenger group using its own vehicles and using ride-hailing vehicles, based on the number of first detection personnel and the first path dispersion coefficient for each candidate passenger group, thus obtaining the first cost comparison index. In specific implementation, the server first determines the number of first detection personnel for each candidate passenger group, which is directly obtained from the set of detection personnel generated in the previous task matching step. The number of personnel, as a fundamental parameter affecting vehicle capacity demand and the number of dispatched vehicles, determines the minimum number of vehicles required for dispatch. Based on this, the server combines the first path dispersion coefficient to evaluate the dispersion of the passenger group's task path. The more dispersed the path, the more complex the vehicle route, and the higher the dispatch cost. Therefore, the number of personnel and the path dispersion coefficient are simultaneously used as cost weighting factors in the dispatch cost model for modeling calculations.

[0071] The server uses both the proprietary vehicle cost model and the ride-hailing cost model to estimate dispatch costs. The proprietary vehicle cost model calculates costs based on parameters such as fixed dispatch cost per vehicle, fuel consumption cost, vehicle mileage, and path complexity factor. The path complexity factor is modeled using a first path dispersion coefficient; a higher coefficient indicates a more dispersed vehicle path, requiring a proportional increase in cost per unit mileage. The ride-hailing cost model predicts the total cost required to dispatch the ride group based on the current ride-hailing platform's pricing rules, combined with factors such as the base starting price, vehicle type, total travel distance, waiting time, and peak-hour surcharges. To ensure the timeliness of the evaluation results, the server obtains current ride-hailing price parameters in real-time via the platform API and performs dynamic calculations based on distance and travel time information from the path planning results. After completing the two cost models, the server compares the obtained proprietary vehicle connection costs with the ride-hailing connection costs, calculating the cost difference, proportional difference, and cost-efficiency ratio after path dispersion adjustment, ultimately generating the first cost comparison index. This indicator not only reflects the difference in scheduling costs between the two types of vehicles in the current scheduling scenario, but also takes into account the complexity of the scheduling path. It uses the path dispersion coefficient to correct the predicted cost for risk, ensuring that the decision-making basis is realistically feasible.

[0072] S204. Adjust the candidate passenger groups based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group;

[0073] In this embodiment, to further optimize the aggregation effect of vehicle scheduling paths and improve the path consistency and resource utilization efficiency of passenger groups, after obtaining the first path dispersion coefficient and the first cost comparison index of candidate passenger groups, the server filters and adjusts the structure of candidate passenger groups based on these two indicators to obtain the final target passenger groups that can be used for scheduling execution. The core of this step is to identify individuals with significant abnormal contributions in the path structure and reduce path dispersion and increase path overlap through reasonable personnel migration strategies, thereby improving overall scheduling efficiency and economy. Optionally, adjusting candidate passenger groups based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group may include the following steps: filtering candidate passenger groups based on a preset path dispersion threshold and a cost comparison reference value to obtain passenger groups to be adjusted; calculating the path outlier contribution value of the target first detection task location to the first path dispersion coefficient of the adjusted passenger group based on the target first detection task location corresponding to each first detection personnel in the adjusted passenger group, and identifying the first detection personnel whose path outlier contribution value is greater than the preset outlier rate threshold as detection personnel to be adjusted; determining the migration plan for the detection personnel to be adjusted, and obtaining the target passenger group based on the migration plan.

[0074] In the implementation process, the server first filters candidate passenger groups based on a preset path dispersion threshold and a cost comparison reference value. Passenger groups with high path dispersion or high scheduling costs under the current structure are identified as those to be adjusted. The path dispersion threshold is the upper limit of the acceptable scheduling range set by the system. If the first path dispersion coefficient of a passenger group exceeds this threshold, it means that the differences between its task paths are large, and it lacks good path aggregation. Similarly, if the first cost comparison index for a passenger group shows that the cost of the current scheduling method is significantly higher than another method, it can also be identified as an optimization target. By combining these two indicators for screening, it is possible to effectively focus on passenger groups that have significant room for improvement in path structure or cost control.

[0075] For the selected passenger groups requiring adjustment, the server further calculates the specific impact of the target first detection task location corresponding to each first detection personnel on the current passenger group's path dispersion, i.e., the path outlier contribution value. This value is obtained through sensitivity analysis of the path dispersion coefficient. Specifically, while keeping other personnel unchanged, a certain detection personnel is temporarily removed, and the path dispersion coefficient is recalculated. If the dispersion coefficient decreases significantly after removal, it indicates that the personnel's task path has a strong pulling effect on the overall path dispersion of the passenger group. The server records the path outlier contribution value of each detection personnel and compares it with the outlier rate threshold set by the system. If the contribution value exceeds the threshold, the detection personnel is marked as a detection personnel requiring adjustment for subsequent structural optimization operations.

[0076] After identifying the personnel to be adjusted for testing, a relocation plan for these personnel is determined, and the target passenger group is obtained based on the relocation plan. Optionally, determining the relocation plan for the personnel to be adjusted and obtaining the target passenger group based on the relocation plan may include the following steps: analyzing the path overlap between the target travel path of the personnel to be adjusted and other candidate passenger groups and each target travel path in the passenger group to be adjusted, and determining the relocation passenger group based on the path overlap.

[0077] The personnel to be adjusted are simulated to be moved to the migration group to obtain the target group to be adjusted and the target migration group. The first path dispersion coefficient of the target group to be adjusted and the target migration group are calculated respectively. When the first path dispersion coefficient of the target group to be adjusted and the target migration group are both less than the preset path dispersion threshold, the migration plan is obtained. The group to be adjusted is adjusted based on the migration plan to obtain the target group.

[0078] During the migration plan development process, the server first analyzes the path overlap between the target travel path of the inspector to be adjusted and all target travel paths in other candidate passenger groups. The travel path is the recommended route for the inspector from their current location to their first inspection task location. Path overlap measures the proportion of shareable road segments between different travel paths. The server converts all paths into a structured sequence of road segment identifiers using path encoding and identifies overlapping segments with paths in each candidate passenger group using set intersection operations. It then calculates an overlap score based on the length of these overlapping segments. A higher path overlap score indicates greater compatibility between the inspector's path structure and the target passenger group, resulting in less disruption to path consistency after migration, and potentially even improving the path aggregation level of the target passenger group. Based on the path overlap score, the server initially selects a group of passenger groups suitable for migration.

[0079] Once potential migration targets are identified, the server simulates the migration of the personnel to be adjusted to each migration bus group, forming several post-migration structural combinations. After each simulated migration, the server calculates the first path dispersion coefficient between the original bus group (i.e., the target bus group to be adjusted) and the newly admitted target migration bus group. This calculation process is consistent with the aforementioned path dispersion assessment method, still based on weighted modeling of task location distribution characteristics and path overlap, ultimately outputting a standardized path dispersion coefficient. By comparing the path dispersion coefficients before and after the migration, the server can determine whether the migration operation effectively alleviated the path dispersion problem of the original bus group and assess whether the target migration bus group experienced path structure deterioration as a result.

[0080] When the simulated migration results show that the first path dispersion coefficients of both the target passenger group to be adjusted and the target migrating passenger group are less than the preset path dispersion threshold set by the system, it indicates that the migration operation will not deteriorate the path structure of any passenger group and is feasible. The server can then determine the simulated migration path as the official migration plan. Subsequently, the server updates the personnel composition of the original passenger group and the target migrating passenger group based on this migration plan, forming a structurally optimized set of target passenger groups. Finally, passenger groups with better path structures and reduced scheduling costs will be retained for subsequent vehicle scheduling type prediction and scheduling path generation.

[0081] S205. Based on the path aggregation evaluation module, analyze the location of the second detection task corresponding to each second detection person in each target passenger group to obtain the second path dispersion coefficient and the second cost comparison index for each candidate passenger group.

[0082] To ensure that the target passenger groups ultimately used for scheduling tasks exhibit good path aggregation and scheduling economy during actual operation, the server re-analyzes the second detection task locations corresponding to each second detection personnel in each target passenger group based on the path aggregation evaluation module, obtaining the second path dispersion coefficient and second cost comparison index for each target passenger group. Although the first path dispersion coefficient and first cost comparison index have undergone preliminary screening and adjustment, due to changes in the target passenger group structure and the potential dynamic updating of detection task locations, it is necessary to re-evaluate and confirm the path aggregation efficiency and scheduling cost. In specific execution, the server uses the same path aggregation evaluation method as in the first stage to model the location distribution characteristics of the second detection task locations and the maximum path overlap of the travel paths, and recalculates the second path dispersion coefficient by combining the number of second detection personnel and the task path structure in the target passenger group; based on this, and considering the current environmental parameters and vehicle status, the connection costs of owned vehicles and ride-hailing vehicles are estimated respectively, generating the second cost comparison index.

[0083] S206. Based on the preset integrated model, determine the vehicle type for each target passenger group according to the second path dispersion coefficient and the second cost comparison index, and obtain the allocation result;

[0084] In the specific implementation process, the server first extracts structured input features for each target passenger group. These features include the second path dispersion coefficient, the second cost comparison index, the number of detection personnel, the average path distance, the maximum path overlap, weather condition parameters, and road traffic status parameters. The second path dispersion coefficient reflects the spatial divergence of the passenger group's task path, indirectly affecting the complexity and integration difficulty of the scheduling path; the second cost comparison index quantifies the difference in scheduling costs required to use different vehicle resources, and is the most direct economic indicator in decision-making; the number of personnel affects vehicle capacity requirements; and the path overlap and average distance jointly determine the path integration efficiency. These features are normalized before input to ensure that the model has stable discriminative ability under different numerical dimensions.

[0085] After feature preparation, the server invokes the scheduling type prediction module to classify the vehicle type. This module is built on a Stacking ensemble learning architecture, integrating multiple basic models, including logistic regression, random forest, and support vector machine models. Each model predicts the scheduling type from the perspectives of linear discrimination, nonlinear structure recognition, and feature interaction. These basic models output the probability value for "owned vehicle" or "ride-hailing" for each target passenger group. The server uses these probability outputs as new feature combinations and inputs them into the upper-level meta-learner. The meta-learner is trained using gradient boosting trees, possessing strong capabilities in nonlinear feature combination and decision boundary fitting. It ultimately outputs a classification label: "0" indicates a recommendation for owned vehicle, and "1" indicates a recommendation for ride-hailing. Before system deployment, this model underwent supervised training using historical scheduling data and cross-validation in typical scenarios to ensure its stability and accuracy in actual operation.

[0086] The server generates the final vehicle allocation result based on the classification labels output by the scheduling type prediction module and the current vehicle status data. If the prediction result for a target passenger group is "0", the server searches for available self-owned vehicle resources in the first vehicle status data and generates a first vehicle scheduling plan for that passenger group. If the prediction result is "1", the server calls the second vehicle status data, initiates a pick-up task request to the ride-hailing platform, and generates a second vehicle scheduling plan. This allocation result not only determines the subsequent vehicle scheduling execution method but also serves as the input basis for the route scheduling optimization module, directly affecting scheduling route planning, timeliness control, and task allocation efficiency. For example, a target passenger group includes 3 second-level inspectors, with a path dispersion coefficient of 0.28, a maximum path overlap of 0.83, and a second cost comparison indicator showing that using self-owned vehicles costs 88 yuan while using ride-hailing services costs 132 yuan. The server uses this information as input features and sends it to the scheduling type prediction module. After the underlying model prediction and meta-learner fusion judgment, the output classification label is "0", indicating that self-owned vehicles are recommended.

[0087] S207. Based on the vehicle status data, the allocation result, the weather condition parameters, and the road traffic status parameters, generate a vehicle dispatching plan.

[0088] In step S207, the vehicle dispatching scheme not only needs to meet the path integration and time requirements of the current task, but also needs to dynamically adapt to environmental changes and sudden dispatching needs. Therefore, in the process of generating the dispatching scheme, the server needs to comprehensively consider multiple factors such as path feasibility, vehicle accessibility, resource availability and dispatching flexibility, and generate the first vehicle dispatching scheme and the second vehicle dispatching scheme for its own vehicles and ride-hailing vehicles respectively, and have a rapid response mechanism during the task execution process.

[0089] In practice, the server first groups all target passenger groups according to vehicle type based on the allocation results determined in the preceding steps. For passenger groups assigned to use their own vehicles, the server extracts currently available vehicles from the first vehicle status data, including their current location, remaining available time, passenger capacity, fuel level, or battery status. This information is then combined with the number of inspectors and the route structure within the passenger group to match vehicles with passenger groups. Using a scheduling route planning engine and supported by map data, the server generates a complete route for its own vehicles to depart from their current location, pick up all inspectors in the optimal route order, and proceed to their respective inspection task locations. Weather and road traffic status parameters are incorporated into this route generation process to adjust route capacity in real time, avoiding congested, closed, or high-risk areas, ensuring the scheduling route is practically executable. The final output of the first vehicle scheduling plan includes key scheduling information such as vehicle number, pick-up order, estimated travel time, route segments, and estimated completion time.

[0090] For passenger groups assigned to ride-hailing services, the server generates a second vehicle dispatch plan based on the second vehicle status data and the available dispatchable resources on the current ride-hailing platform. During implementation, the server calls the ride-hailing service through the platform interface, selecting suitable vehicle types and service types based on the passenger group's origin location, task time window, and route dispersion. It also evaluates the route suitability of multiple ride-hailing drivers, prioritizing vehicles with high overlap between their current route and the passenger group's task route, and short service response times. Simultaneously, the system adjusts the estimated pick-up time and price parameters based on weather and road conditions to ensure the dispatch plan aligns with current external execution capabilities. The second vehicle dispatch plan output includes ride-hailing vehicle identification, driver information, order acceptance time, estimated arrival time, and execution route, ensuring the dispatch process is controllable, transparent, and traceable.

[0091] To further enhance the robustness of the scheduling scheme, an emergency task response mechanism is introduced into the scheduling scheme. When a high-priority task is added to the task list, or when the time or location of an existing task is adjusted due to an unforeseen event (such as a sudden outbreak of an epidemic, equipment failure, or personnel changes), the server immediately triggers the scheduling reconfiguration logic. The scheduling system first identifies the affected passenger groups and vehicle resources, reassesses their task paths and reachability, and calculates the path overlap and scheduling time cost between the new task and the current scheduling path. If the currently assigned vehicles still have service capabilities, the server inserts the emergency task into the original scheduling path and adjusts the connection order and time nodes; if the original vehicle cannot carry the load, the server will search for a backup vehicle based on the status data of the first and second vehicles and generate an independent scheduling path. The reconfigured scheduling scheme will replace the original plan and be pushed to vehicle terminals and passenger mobile terminals in real time to ensure the timeliness and accuracy of the scheduling response.

[0092] For example, a passenger group was originally dispatched by its own vehicle A, and the dispatch route had been planned and pushed to the vehicle system. During execution, the system received an emergency inspection task that required inserting a new inspector B into the current route. The server calculated that there was a 1.5-kilometer overlap between inspector B's task point and the original route. After insertion, the total travel time would only increase by 8 minutes, still within the dispatch time window. Based on this, the server updated the first vehicle dispatch plan, inserting inspector B's pick-up and task point into the existing route, and updating vehicle navigation and task reminders on the passenger's end in real time. If the inserted route exceeded the dispatch window, the server would call the ride-hailing platform to generate an independent second vehicle dispatch plan, arranging a ride-hailing vehicle to pick up and drop off inspector B, ensuring an immediate response to the emergency task.

[0093] Please see Figure 3 This is a schematic diagram of the structure of a vehicle-pedestrian collaborative scheduling system provided in an embodiment of this application. The vehicle-pedestrian collaborative scheduling system 300 specifically includes: an acquisition module 301, used to acquire personnel status data, vehicle status data, environmental parameters and a list of detection tasks. The personnel status data includes current location information, qualification level and skill tag information. The environmental parameters include weather condition parameters and road traffic status parameters.

[0094] The matching module 302 is used to match the personnel status data with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task, and to determine the set of detection personnel for each detection task.

[0095] The first determining module 303 is used to divide the set of testing personnel into multiple candidate bus groups, determine the first testing task location corresponding to each first testing personnel in each candidate bus group, and obtain the first path dispersion coefficient and the first cost comparison index for each candidate bus group.

[0096] The adjustment module 304 is used to adjust the candidate passenger group based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group;

[0097] Analysis module 305 is used to analyze the second detection task location corresponding to each second detection person in each target passenger group based on the preset integrated model, and obtain the second path dispersion coefficient and second cost comparison index for each candidate passenger group;

[0098] The second determining module 306 is used to determine the vehicle type of each target passenger group based on the preset integrated model according to the second path dispersion coefficient and the second cost comparison index, and obtain the allocation result;

[0099] The generation module 307 is used to generate a vehicle dispatching plan based on the vehicle status data, the allocation result, the weather condition parameters, and the road traffic status parameters.

[0100] Optionally, the matching module 302 is specifically used for: constructing a task feature vector based on the skill requirements and qualification level requirements of each detection task in the detection task list; mapping the skill tag information and qualification level in the personnel status data to personnel feature vectors; calculating the qualification similarity between the personnel feature vector of each detection personnel and the task feature vector of each detection task to obtain a feature matching coefficient matrix for personnel-task pairs; calculating the spatial distance coefficient between the current location information of each detection personnel and the location information of each detection task to obtain a spatial distance coefficient matrix; and calculating the task matching degree based on the feature matching coefficient matrix and the spatial distance coefficient matrix.

[0101] Optionally, the first determining module 303 is specifically used for: analyzing the location distribution characteristics of the first detection task location of the first detection personnel in each candidate ride-hailing group; determining the maximum path overlap between target paths based on environmental parameters, wherein the target path is the travel path of each first detection personnel from its current location to the first detection task location; calculating the first path dispersion coefficient of each candidate ride-hailing group based on the location distribution characteristics and the maximum path overlap; and calculating the connection cost of owned vehicles and ride-hailing vehicles according to the number of first detection personnel in each candidate ride-hailing group and the first path dispersion coefficient, thereby obtaining a first cost comparison index.

[0102] Optionally, the first determining module 303 is further specifically used for: determining the capacity adjustment coefficient of each traffic path based on weather condition parameters and road traffic status parameters; evaluating all candidate traffic paths for each first inspector based on the capacity adjustment coefficient, and selecting a set of target traffic paths that meet preset traffic conditions; calculating the length of overlapping road segments between the first target traffic path set and the second target traffic path set, and generating a path overlap weight between the first target traffic path set and the second target traffic path set by combining the length of overlapping road segments with the corresponding capacity adjustment coefficient of each overlapping road segment, wherein the first target traffic path set and the second target traffic path set are any set of target traffic paths among multiple target traffic path sets; and determining the maximum path overlap degree based on the path overlap weight.

[0103] Optionally, the first determining module 303 is further specifically used for: constructing a path coordination adjustment factor based on the maximum path overlap; calculating the ratio of the average location spacing to the maximum path overlap in the location distribution characteristics to obtain the effective distribution dispersion; normalizing the maximum location span in the location distribution characteristics to obtain the target maximum location span; weighting the effective distribution dispersion and the target maximum location span to obtain the path dispersion value of the candidate passenger group; and normalizing the path dispersion value and mapping it to a preset standard score interval to obtain the first path dispersion coefficient.

[0104] Optionally, the adjustment module 304 is specifically used for: screening candidate passenger groups based on a preset path dispersion threshold and a cost comparison reference value to obtain passenger groups to be adjusted; calculating the path outlier contribution value of the target first detection task location to the first path dispersion coefficient of the adjusted passenger group based on the target first detection task location corresponding to each first detection person in the adjusted passenger group, and identifying the first detection person whose path outlier contribution value is greater than the preset outlier rate threshold as the detection person to be adjusted; determining the migration plan for the detection person to be adjusted, and obtaining the target passenger group based on the migration plan.

[0105] Optionally, the adjustment module 304 is further specifically used for: analyzing the path overlap between the target travel path of the personnel to be adjusted and each target travel path in the other candidate bus groups and the bus group to be adjusted; determining the bus group to be migrated based on the path overlap; simulating the migration of the personnel to be adjusted to the bus group to be migrated, obtaining the target bus group to be adjusted and the target bus group to be migrated after migration; calculating the first path dispersion coefficient of the target bus group to be adjusted and the target bus group to be migrated respectively; when the first path dispersion coefficient of the target bus group to be adjusted and the target bus group to be migrated are both less than the preset path dispersion threshold, obtaining the migration plan; and adjusting the bus group to be adjusted based on the migration plan to obtain the target bus group.

[0106] This embodiment also discloses an electronic device, as shown in the reference. Figure 4The electronic device may include: at least one processor 401, at least one communication bus 402, a user interface 403, a network interface 404, and at least one memory 405. The communication bus 402 is used to enable communication between these components. The user interface 403 may include a display screen or a camera; optionally, the user interface 403 may also include a standard wired interface or a wireless interface. The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 405, and by accessing data stored in the memory 405. It is understood that the aforementioned modem may also not be integrated into the processor 401, but may be implemented as a separate chip.

[0107] The memory 405 may include random access memory (RAM) or read-only memory. For example... Figure 4 As shown, the memory 405, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a vehicle-person cooperative scheduling method. Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 401 can be used to call an application program stored in the memory 405 for a vehicle-person cooperative scheduling method. When executed by one or more processors 401, the electronic device executes one or more methods as described in the above embodiments.

Claims

1. A vehicle-pedestrian collaborative scheduling method, characterized in that, When applied to a server, the method includes: The system acquires personnel status data, vehicle status data, environmental parameters, and a list of inspection tasks. The personnel status data includes current location information, qualification level, and skill tag information. The environmental parameters include weather conditions and road traffic conditions. The personnel status data is matched with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task, and the set of detection personnel for each detection task is determined. The set of testing personnel is divided into multiple candidate bus groups. The first testing task location corresponding to each first testing personnel in each candidate bus group is determined, and the first path dispersion coefficient and the first cost comparison index of each candidate bus group are obtained. The candidate passenger groups are adjusted based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group; Based on the analysis of the second detection task location corresponding to each second detection person in each target passenger group, the second path dispersion coefficient and the second cost comparison index of each candidate passenger group are obtained. Based on the preset integrated model, the vehicle type of each target passenger group is determined according to the second path dispersion coefficient and the second cost comparison index, and the allocation result is obtained; A vehicle dispatching plan is generated based on the vehicle status data, the allocation results, the weather condition parameters, and the road traffic status parameters.

2. The method according to claim 1, characterized in that, The step of matching the personnel status data with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task specifically includes: Based on the skill requirements and qualification level requirements of each detection task in the detection task list, a task feature vector is constructed; Map the skill tag information and qualification level in the personnel status data into a personnel feature vector; Calculate the eligibility similarity between the personnel feature vector of each of the detection personnel and the task feature vector of each of the detection tasks to obtain the feature matching coefficient matrix of personnel-task pairs; Calculate the spatial distance coefficient between the current location information of each of the detection personnel and the location information of each detection task to obtain a spatial distance coefficient matrix; The task matching degree is calculated based on the feature matching coefficient matrix and the spatial distance coefficient matrix.

3. The method according to claim 1, characterized in that, The step of determining the first detection task location corresponding to each first detection personnel in each candidate passenger group, and obtaining the first path dispersion coefficient and first cost comparison index for each candidate passenger group, specifically includes: Analyze the location distribution characteristics of the first detection task location for the first detection personnel in each of the candidate passenger groups; The maximum path overlap between target paths is determined based on the environmental parameters, wherein the target path is the travel path of each first inspection personnel from the current location to the first inspection task location; Based on the location distribution characteristics and the maximum path overlap, calculate the first path dispersion coefficient for each candidate passenger group; Based on the number of the first detection personnel in each candidate passenger group and the first path dispersion coefficient, the connection cost of owned vehicles and ride-hailing vehicles is calculated to obtain the first cost comparison index.

4. The method according to claim 3, characterized in that, The determination of the maximum path overlap between target paths based on the environmental parameters specifically includes: Based on the weather condition parameters and road traffic status parameters, determine the traffic capacity adjustment coefficient for each of the traffic routes; Based on the traffic capacity adjustment coefficient, all the candidate traffic paths for each of the first inspection personnel are evaluated, and a set of target traffic paths that meet the preset traffic conditions is selected. Calculate the length of overlapping road segments between the first target travel path set and the second target travel path set, and combine the length of overlapping road segments with the corresponding capacity adjustment coefficient of each overlapping road segment to generate the path overlap weight between the first target travel path set and the second target travel path set. The first target travel path set and the second target travel path set are any of the multiple target travel path sets. The maximum path overlap is determined based on the path overlap weight.

5. The method according to claim 4, characterized in that, The calculation of the first path dispersion coefficient for each candidate passenger group based on the location distribution characteristics and the maximum path overlap specifically includes: Based on the maximum path overlap, a path coordination adjustment factor is constructed. The effective distribution dispersion is obtained by calculating the ratio of the average location spacing to the maximum path overlap in the location distribution characteristics. The maximum location span in the location distribution features is normalized to obtain the target maximum location span; The effective distribution dispersion is weighted by the target maximum location span to obtain the path dispersion value of the candidate passenger group; The path dispersion value is normalized and mapped to a preset standard score range to obtain the first path dispersion coefficient.

6. The method according to claim 1, characterized in that, The step of adjusting the candidate passenger group based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group specifically includes: The candidate passenger groups are screened based on a preset path dispersion threshold and a cost comparison reference value to obtain the passenger groups to be adjusted. Based on the target first detection task location corresponding to each first detection person in the adjusted passenger group, calculate the path outlier contribution value of the target first detection task location to the first path dispersion coefficient of the adjusted passenger group, and determine the first detection person whose path outlier contribution value is greater than a preset outlier rate threshold as the detection person to be adjusted. Determine the relocation plan for the personnel to be adjusted for testing, and obtain the target passenger group based on the relocation plan.

7. The method according to claim 6, characterized in that, The process of determining the relocation plan for the personnel to be adjusted for testing, and obtaining the target passenger group based on the relocation plan, specifically includes: Analyze the path overlap between the target travel path of the personnel to be adjusted and the target travel path of each candidate travel group and the travel group to be adjusted, and determine the travel group to be moved based on the path overlap. The personnel to be adjusted are simulated to be relocated to the relocated vehicle group to obtain the target vehicle group to be adjusted and the target relocated vehicle group. The first path dispersion coefficient of the target vehicle group to be adjusted and the target relocated vehicle group are calculated respectively. When the first path dispersion coefficient of the target vehicle group to be adjusted and the target relocated vehicle group are both less than the preset path dispersion threshold, the relocation plan is obtained. The vehicle group to be adjusted is adjusted based on the relocation plan to obtain the target vehicle group.

8. A vehicle-pedestrian collaborative scheduling system, characterized in that, include: The acquisition module is used to acquire personnel status data, vehicle status data, environmental parameters, and a list of inspection tasks. The personnel status data includes current location information, qualification level, and skill tag information. The environmental parameters include weather conditions and road traffic conditions. The matching module is used to match the personnel status data with each detection task in the detection task list to obtain the task matching degree between the detection personnel and each detection task, and to determine the set of detection personnel for each detection task. The first determining module is used to divide the set of testing personnel into multiple candidate bus groups, determine the first testing task location corresponding to each first testing personnel in each candidate bus group, and obtain the first path dispersion coefficient and the first cost comparison index for each candidate bus group. The adjustment module is used to adjust the candidate passenger group based on the first path dispersion coefficient and the first cost comparison index to obtain the target passenger group; The analysis module is used to analyze the second detection task location corresponding to each second detection person in each target passenger group based on a preset integrated model, and to obtain the second path dispersion coefficient and the second cost comparison index for each candidate passenger group. The second determining module is used to determine the vehicle type of each target passenger group based on the preset integrated model, according to the second path dispersion coefficient and the second cost comparison index, and obtain the allocation result; The generation module is used to generate a vehicle dispatching plan based on the vehicle status data, the allocation results, the weather condition parameters, and the road traffic status parameters.

9. An electronic device, characterized in that, include: One or more processors and memory; The memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium comprising instructions, characterized in that, When the instructions are executed on an electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 1-7.

Images