An unmanned vehicle scheduling method and system based on multi-source data

Abstract

The application discloses a kind of based on multi-source data's unmanned vehicle scheduling method and system, it is related to transport capacity scheduling field, including: receiving module, for receiving unmanned vehicle operating state parameter, surrounding environment data, transport order information, to the received data executes standardization format conversion, obtains standardization three-dimensional data set;Dismantling module is used to obtain standardization three-dimensional data set, based on standardization three-dimensional data set Whole transport task is disassembled into subtask adapted to different unmanned vehicle transport capacity, the execution priority and constraint condition of each subtask are set simultaneously;The application improves data quality by multi-source data standardization processing and credibility weighting mechanism, accurately split and adapt transport capacity in combination with task complexity, redundancy design enhances scene adaptability, and efficient matching of task and transport capacity is realized through multidimensional adaptation evaluation.

Description

Technical Field

[0001] This invention relates to the field of transportation capacity scheduling technology, specifically to a method and system for scheduling unmanned vehicles based on multi-source data. Background Technology

[0002] Unmanned vehicles are core equipment in the logistics industry, based on Level 4 autonomous driving technology and integrating multiple sensors and precise algorithms to achieve environmental perception, path planning, and safe driving. Their applications cover scenarios such as last-mile delivery in cities, industrial park transfers, and transportation in remote areas, enabling 24 / 7 uninterrupted operation and significantly reducing labor costs and delivery errors.

[0003] The invention patent application with application number 202210339838.8 discloses an unmanned vehicle scheduling method, device and system. The application aims to solve the problem that "unmanned guided vehicles (AGVs) can be used in factories or workshops to realize intelligent handling. AGVs can realize navigation based on SLAM algorithm. When enterprises implement unmanned guided vehicle scheduling schemes that use laser SLAM unmanned vehicles, in application scenarios with small spaces and complex routes, they will encounter the problem of motion interference during vehicle driving, making it impossible to accurately maintain a safe distance between vehicles, thus leading to vehicle collisions."

[0004] However, the scheduling of unmanned vehicle transportation capacity is also a key focus in this type of scenario. A good scheduling configuration for unmanned vehicle transportation capacity can save transportation costs, reduce transportation risks, and make unmanned vehicle transportation scenarios more orderly.

[0005] To address this, we propose a method and system for scheduling unmanned vehicles based on multi-source data. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the existing technology, the present invention provides an unmanned vehicle scheduling method and system based on multi-source data, which can effectively solve the problems of the existing technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions; This invention discloses an unmanned vehicle scheduling system based on multi-source data, comprising: The system comprises the following modules: a receiving module (receiving module), a decomposition module, and a message module. The receiving module receives the autonomous vehicle's operational status parameters, surrounding environmental data, and transportation order information. It performs standardized format conversion on the received data to obtain a standardized 3D dataset. The decomposition module acquires the standardized 3D dataset and decomposes the overall transportation task into sub-tasks adapted to different autonomous vehicle capacities. It simultaneously sets execution priorities and constraints for each sub-task. The allocation module associates sub-tasks with autonomous vehicle resources, matching and allocating sub-tasks with autonomous vehicles based on the real-time status of the autonomous vehicles and the constraints of the sub-tasks. The control module plans the execution path for each vehicle based on the allocated sub-tasks and real-time surrounding environmental data. The feedback module collects the status data of the autonomous vehicles executing sub-tasks, estimates the risk of trip conflicts for each autonomous vehicle based on the status data, and feeds back the collected status data to the control module when the estimation result indicates a risk, triggering the control module to run again to iterate the trip parameter configuration. The message module captures system operation data and autonomous vehicle operation and scheduling iteration data in real time to generate autonomous vehicle operation and maintenance messages. When the control module first runs to plan the execution path of a single vehicle, it simultaneously configures the vehicle's speed. When the control module runs based on the feedback module's triggering, the iteration of the travel parameter configuration only includes the control of the speed. The receiving module is interconnected with a disassembly module and an allocation module via a wireless network. The disassembly module and the allocation module are interconnected with a control module via a wireless network. The control module is interconnected with a feedback module via a wireless network. The feedback module is interconnected with a message module via a wireless network.

[0008] Furthermore, when the receiving module performs standardized format conversion, it cleans the unmanned vehicle's operating status parameters, surrounding environment data, and transportation order information, and simultaneously constructs a three-dimensional data mapping model based on the dimensional attributes of multi-source data. Then, through normalization processing, it maps data of different magnitudes to a unified data range. ; In the formula: The data is standardized after normalization; This refers to the original data from the corresponding data source; Let be the credibility weight of the i-th type of data source; These are the historical maximum and minimum valid values ​​of the i-th type of data source; This is the scaling factor for the data range; This represents the offset of the data range. The three dimensions of the three-dimensional data mapping model correspond to the autonomous vehicle state dimension, environmental feature dimension, and order demand dimension, respectively. Sub-data items under each dimension are associated with corresponding fields of the standardized three-dimensional dataset through label mapping.

[0009] Furthermore, when the decomposition module decomposes the overall transportation task based on the standardized 3D dataset, it first calculates the total complexity value of the overall transportation task, and then, combined with the maximum capacity complexity of each unmanned vehicle, determines the number of sub-tasks to be divided and the complexity threshold of each sub-task. The number of sub-tasks to be divided is as follows: ; In the formula: This represents the total complexity value. Redundancy coefficients are allocated to tasks; Let j be the maximum transport capacity complexity of the unmanned vehicle. When determining the complexity threshold for each subtask, the following rule applies: the complexity threshold for each subtask does not exceed the complexity threshold of the corresponding assigned autonomous vehicle. And the sum of the complexities of all subtasks is not less than .

[0010] Furthermore, when the allocation module matches and allocates sub-tasks to unmanned vehicles, it simultaneously constructs a multi-dimensional adaptability evaluation model. The evaluation dimensions of the model include the unmanned vehicle's current remaining capacity, distance from the starting point of the sub-task, historical task completion accuracy, and degree of satisfaction with constraints. Each dimension is assigned a preset weight and then the comprehensive adaptability value is calculated. The allocation module sorts the subtasks from highest to lowest execution priority. For each subtask, it selects the top 3 autonomous vehicles with the highest comprehensive suitability as candidate matching objects. Then, it combines the total complexity of the subtasks already assigned to the candidate autonomous vehicles and selects the candidate objects whose total complexity does not exceed their maximum capacity complexity to complete the allocation. If none of the candidate objects meet the requirements, the subtask is marked as pending allocation, and the decomposition module is triggered to re-evaluate the complexity threshold of the subtask, perform a second decomposition, and then execute the allocation process. Before being used to calculate the overall fit value, the unmanned vehicle's current remaining capacity, distance from the sub-task starting point, historical task completion accuracy, and degree of satisfaction with constraints are all normalized, so that the values ​​of the unmanned vehicle's current remaining capacity, distance from the sub-task starting point, historical task completion accuracy, and degree of satisfaction with constraints are limited to the range of (0,1).

[0011] Furthermore, when planning the execution path for a single vehicle, the control module generates all candidate feasible paths that satisfy the start and end points and constraints of the sub-tasks through regional road network topology modeling based on electronic map data and real-time surrounding environment data. Then, it constructs a path planning model with the goal of minimizing the overall path cost. ; In the formula: , , The weights are path length weight, environmental resistance weight, and traffic efficiency weight. This represents the total physical length of the path. Number of path segments; Let i be the environmental resistance coefficient of the i-th path segment; Let be the length of the i-th path segment; This represents the historical average traffic speed. This is the real-time traffic efficiency coefficient.

[0012] Furthermore, when estimating the conflict risk of each autonomous vehicle's journey, the feedback module uses any two autonomous vehicles performing sub-tasks as analysis units to calculate the conflict risk value: ; In the formula: Let be the real-time straight-line distance between the i-th autonomous vehicle and the j-th autonomous vehicle; The relative speed between the two vehicles; This represents the path conflict coefficient. This refers to the path overlap weighting coefficient. The length of the overlapping section of the two vehicle paths; Let be the total length of the sub-task paths of the i-th and j-th autonomous vehicles; when When the risk exceeds the preset risk threshold, it is determined that there is a risk of travel conflict, and the feedback module triggers the control module to run.

[0013] Furthermore, when the control module is triggered by the feedback module, it first determines the dominant party in the conflict between the two autonomous vehicles, and then performs speed control on the dominant party: Based on the real-time speed and angle of travel of the two vehicles Calculate the relative velocity contribution value ; If the contribution value is positive, then the i-th driverless car is the dominant party in the conflict; If the contribution value is negative, then the j-th driverless vehicle is the dominant party in the conflict; Speed ​​control is applied to the dominant party in the conflict, while the non-dominant parties maintain their original speeds. The target speed for the dominant party is... ; In the formula: The current movement speed of the party in control of the conflict; Risk response coefficient; The collision risk value between the two vehicles; To preset risk thresholds; The environmental adaptability coefficient of the current driving segment of the party in the conflict; The control module simultaneously sends the target speed to the autonomous vehicle that is in control of the conflict.

[0014] Furthermore, the operation and maintenance messages generated by the message module include operation messages and maintenance messages: The operation messages are generated at preset time intervals and include the real-time location, speed, sub-task execution progress, current road segment environmental data, path adjustment records, and conflict risk iteration data of each unmanned vehicle. Maintenance messages are generated according to the subtask completion nodes, including the state parameter change curves of the unmanned vehicle during the execution of the subtask, the number and magnitude of speed adjustment, and the impact assessment data of scheduling iteration on task completion efficiency. Among them, the impact of scheduling iteration on task completion efficiency is evaluated by quantitatively representing the ratio of actual transportation completion time to preset transportation completion time limit.

[0015] On the other hand, an unmanned vehicle scheduling method based on multi-source data includes: The system receives unmanned vehicle (UAV) operating status parameters, surrounding environment data, and transportation order information. Through data cleaning, normalization, and 3D data mapping modeling, a standardized 3D dataset is obtained. Based on this standardized 3D dataset, the overall transportation task complexity is calculated. Combined with the maximum capacity complexity of the UAV, the number of subtasks and complexity thresholds are determined, and subtask execution priorities and constraints are set simultaneously. A multi-dimensional adaptability evaluation model is constructed, and candidate UAVs are selected and assigned according to subtask priority. If the assignment conditions are not met, a secondary subtask split is triggered. Candidate feasible paths are generated based on electronic maps and real-time environmental data. A single-vehicle execution path is planned with the goal of minimizing the overall path cost, and an initial speed is configured. UAV operating status data is collected, and the risk value of conflicts between two vehicles is calculated. When the risk value exceeds a preset threshold, the dominant conflict target is identified, and its speed is adjusted. System and UAV-related data are captured in real time, generating operation and maintenance messages.

[0016] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects: This invention improves data quality through multi-source data standardization and a credibility weighting mechanism, accurately decomposes tasks based on complexity and adapts them to available capacity, enhances scenario adaptability through redundant design, achieves efficient matching of tasks and available capacity through multi-dimensional adaptation evaluation, reduces pending allocation, optimizes planning schemes based on comprehensive path costs to reduce travel consumption and time costs, monitors trip conflict risks in real time and precisely controls movement speed to avoid collisions while ensuring execution efficiency, and comprehensively records operation and maintenance data to support scheduling strategy optimization and equipment maintenance. Overall, it effectively improves the intelligence level and operational stability of unmanned vehicle scheduling and increases transportation efficiency. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0018] Figure 1 This is a schematic diagram of the structure of an unmanned vehicle scheduling system based on multi-source data; Figure 2 This is a flowchart illustrating an unmanned vehicle scheduling method based on multi-source data. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] The present invention will be further described below with reference to embodiments.

[0021] Example 1: This embodiment presents an unmanned vehicle scheduling system based on multi-source data, such as... Figure 1 As shown, it includes: The receiving module is used to receive unmanned vehicle operating status parameters, surrounding environment data, and transportation order information, and to perform standardized format conversion on the received data to obtain a standardized three-dimensional dataset. When the receiving module performs standardized format conversion, it cleans the unmanned vehicle's operating status parameters, surrounding environment data, and transportation order information. Specifically, it removes outliers based on a preset anomaly threshold, and simultaneously constructs a three-dimensional data mapping model based on the dimensional attributes of the multi-source data. Then, it maps data of different magnitudes to a unified data range through normalization processing. ; In the formula: The data is standardized after normalization; This refers to the original data from the corresponding data source; Let be the credibility weight of the i-th type of data source; These are the historical maximum and minimum valid values ​​of the i-th type of data source; This is the scaling factor for the data range; This represents the offset of the data range. The above formula takes into account the reliability differences of different data sources. First, it defines the data range by the historical maximum and minimum valid values ​​of various data sources. Then, it makes targeted corrections to the original data by using the confidence weight. Finally, it maps the corrected data to a unified range by using the scaling factor and offset. This not only solves the problem of difficult collaborative analysis of data of different scales, but also incorporates data quality considerations by using the confidence weight, making the standardized results more in line with the actual data value. The three dimensions of the three-dimensional data mapping model correspond to the autonomous vehicle state dimension, environmental feature dimension, and order demand dimension, respectively. The sub-data items under each dimension are associated with the corresponding fields of the standardized three-dimensional dataset through label mapping. in, , Pre-set based on the target mapping range; ; In the formula: The factory accuracy level score for the data acquisition device of type i is determined based on the accuracy parameter table provided by the device manufacturer, and the value range is [0.5, 1]. This represents the deviation between the actual calibration cycle and the standard calibration cycle of the data acquisition device of type i. This is the standard calibration cycle for the equipment; The cumulative usage time of the data acquisition device for the i-th type of data source; The design lifespan of the data acquisition device for the i-th type of data source; The environmental adaptability rating score for the data acquisition device of type i is determined based on the working stability test results of the device in the current use environment, and the value range is [0.6,1]. The stronger the environmental adaptability, the closer the score is to 1. This represents the total number of data source categories participating in data collection within the system. It should be noted that after the credibility weights of various data sources are calculated, normalization is performed to make the sum of the weights of all data sources equal to 1. The above formula incorporates key influencing factors such as equipment factory precision, calibration cycle deviation, cumulative usage time, design service life, and environmental adaptability. Through the collaborative calculation of multi-dimensional parameters, it achieves dynamic evaluation of data source reliability, enabling weight allocation to accurately reflect the comprehensive impact of equipment performance and actual usage status on data quality. The decomposition module is used to acquire a standardized 3D dataset. Based on the standardized 3D dataset, the overall transportation task is decomposed into sub-tasks that are adapted to different unmanned vehicle capacity. At the same time, execution priorities and constraints are set for each sub-task. When the decomposition module decomposes the overall transportation task based on the standardized 3D dataset, it first calculates the total complexity of the overall transportation task, and then, combined with the maximum capacity complexity of each autonomous vehicle, determines the number of subtasks and the complexity threshold of each subtask. The number of subtasks is as follows: ; In the formula: This represents the total complexity value. Redundancy coefficients are allocated to tasks; Let j be the maximum transport capacity complexity of the unmanned vehicle. To ensure that the subtask splitting is accurately matched with the autonomous vehicle's capacity, the above formula first reserves buffer space by combining the total task complexity with the redundancy coefficient, and then determines the number of splits by rounding up based on the minimum and maximum capacity complexity of the autonomous vehicles in the system. This not only avoids the subtasks from exceeding the autonomous vehicle's carrying capacity, but also improves the adaptability of the splitting scheme to complex scenarios by dynamically adjusting the redundancy coefficient according to the fleet's vehicle condition and transportation environment, ensuring the stable progress of the scheduling process. When determining the complexity threshold for each subtask, the following rule applies: the complexity threshold for each subtask does not exceed the complexity threshold of the corresponding assigned autonomous vehicle. And the sum of the complexities of all subtasks is not less than ; in, ; In the formula: a, b, c, and d are the weight coefficients of the influencing factors; Q, D, T, and E are the total amount of goods in the order, the transportation distance, the maximum transportation time required by the order, and the comprehensive complexity of the surrounding environment. Let be the capacity adaptation coefficient of the j-th unmanned vehicle; , , The weights for load adaptation of the j-th autonomous vehicle, the combined weights for power and range of the j-th autonomous vehicle, and the weights for the influence of the current position of the j-th autonomous vehicle are: Let J be the rated maximum load of the j-th unmanned vehicle; Let be the rated maximum speed of the j-th unmanned vehicle; This represents the remaining driving time of the j-th driverless vehicle. Let be the straight-line distance between the current position of the j-th unmanned vehicle and the starting point of the transportation task; The above formula integrates four core factors: total cargo volume, transportation distance, time requirements, and environmental complexity. The weight allocation highlights the degree of influence of each factor. The unmanned vehicle capacity combines performance parameters such as load, power range, and current location, while also introducing a capacity adaptation coefficient to consider the vehicle's historical performance. This ensures that the calculation of both covers key objective indicators while taking into account individual differences in actual use, providing effective support for task decomposition and allocation. in, The preset value range is [0.05, 0.3]. The value is larger when the average service life of the unmanned vehicles in the scheduling system is long, the power performance is significantly degraded, or the transportation area environment is complex. The value is smaller when the unmanned vehicle fleet is in good overall condition, the transportation capacity is stable, and the transportation environment is relatively stable. a, b, c, and d are all positive numbers, and their sum is 1; , , All are positive numbers, and their sum is 1; The value range is preset to [0.6, 1.2]. The value is larger when the j-th unmanned vehicle has a high on-time completion rate of historical tasks, low component wear during execution, and no travel conflict records. The value is smaller when the historical task delay rate is high, the load fluctuates greatly, or the failure frequency exceeds the preset value. The overall complexity E of the surrounding environment ranges from [0,1]. The larger the value is when the obstacle density is high, the coverage area is wide, the road condition is poor, and the weather has a significant impact, the smaller the value is when the obstacle density is low, the coverage area is small, the road condition is good, and the weather has no adverse impact. The allocation module is used to associate subtasks with autonomous vehicle resources, and to match and allocate subtasks with autonomous vehicles by combining the real-time status of autonomous vehicles and the constraints of subtasks. When the allocation module matches and allocates sub-tasks to unmanned vehicles, it simultaneously constructs a multi-dimensional suitability evaluation model. The evaluation dimensions of the model include the unmanned vehicle's current remaining capacity, distance from the starting point of the sub-task, historical task completion accuracy, and degree of satisfaction of constraints. Each dimension is assigned a preset weight and then the comprehensive suitability value is calculated. The allocation module sorts subtasks from highest to lowest execution priority. For each subtask, it selects the top 3 autonomous vehicles with the highest overall suitability as candidate matching objects. Then, it combines the total complexity of the subtasks already assigned to the candidate autonomous vehicles and selects the candidate objects whose total complexity does not exceed their maximum capacity complexity to complete the allocation. If none of the candidate objects meet the requirements, the subtask is marked as pending allocation, and the decomposition module is triggered to re-evaluate the complexity threshold of the subtask, perform a second decomposition, and then execute the allocation process. When, after the second decomposition, the sum of the total capacity complexity of all available autonomous vehicles in the system is still less than the total complexity of the subtasks to be allocated, the system automatically generates a capacity shortage warning message, prompting manual intervention to adjust the allocation or extend the task time limit. Before being used to calculate the overall fit value, the unmanned vehicle's current remaining capacity, distance from the sub-task starting point, historical task completion accuracy, and degree of satisfaction with constraints are all normalized, so that the values ​​of the unmanned vehicle's current remaining capacity, distance from the sub-task starting point, historical task completion accuracy, and degree of satisfaction with constraints are limited to the range of (0,1). The degree of satisfaction of constraints is based on various constraint indicators of the sub-task, including time constraints, load constraints, and path restriction constraints. The fitting ratio between the actual capability parameters of the unmanned vehicle and the constraint threshold under each constraint indicator is calculated. After assigning a preset weight to each fitting ratio, the values ​​are summed, and the sum is the quantitative value of the degree of satisfaction of constraints. The control module is used to plan the execution path of a single vehicle based on the assigned sub-tasks and real-time surrounding environmental data; When the control module plans the execution route for a single vehicle, it generates all candidate feasible routes that meet the start and end points and constraints of the subtasks through regional road network topology modeling based on electronic map data and real-time surrounding environment data. Then, it constructs a path planning model with the objective of minimizing the overall path cost. ; In the formula: , , The weights are path length weight, environmental resistance weight, and traffic efficiency weight. This represents the total physical length of the path. Number of path segments; Let i be the environmental resistance coefficient of the i-th path segment; Let be the length of the i-th path segment; This represents the historical average traffic speed. This is the real-time traffic efficiency coefficient; The above formula constructs a cost model from three core dimensions: path length, environmental resistance, and traffic efficiency. It balances the importance of each dimension under different scheduling scenarios through dynamically adapted weight coefficients. The environmental resistance coefficient accurately quantifies the actual impact of road terrain, wind speed, obstacle distribution, etc., while the traffic efficiency is adjusted by combining historical average speed and real-time road conditions, so that the planned path is more in line with actual driving needs. in, , , Based on dynamic adaptation to scheduling scenarios, and , , All are positive numbers, and , , The sum is 1; The preset value range is (0,2]. The larger the terrain slope, the higher the wind speed, and the more concentrated the obstacle distribution density, the larger the value. The smaller the value is when the terrain is flatter, the wind speed is lower, and the obstacle distribution is sparser. The feedback module is used to collect the status data of the autonomous vehicles performing sub-tasks, estimate the risk of conflict in the journey of each autonomous vehicle based on the status data, and when the estimation result indicates that there is a risk, the collected status data is fed back to the control module, triggering the control module to run again to iterate the configuration of the journey parameters. When estimating the risk of conflict in each autonomous vehicle's journey, the feedback module uses any two autonomous vehicles performing sub-tasks as the analysis unit to calculate the conflict risk value: ; In the formula: Let be the real-time straight-line distance between the i-th autonomous vehicle and the j-th autonomous vehicle; The relative speed between the two vehicles; This represents the path conflict coefficient. This refers to the path overlap weighting coefficient. The length of the overlapping section of the two vehicle paths; Let be the total length of the sub-task paths of the i-th and j-th autonomous vehicles; Among them, a minimum distance threshold d is preset. min ,like Less than the minimum distance threshold d min In the calculation of conflict risk value, the minimum distance threshold is used instead. Perform calculations; The above formula comprehensively considers three key factors: spatial distance between the two vehicles, relative speed, and path overlap. The closer the distance and the greater the relative speed, the higher the base risk value. When paths overlap, the risk assessment results are further amplified by the conflict coefficient and overlap weight. At the same time, the weight is dynamically adjusted according to the traffic priority and road condition complexity of the overlapping road sections, so that the risk value calculation can fully reflect the comprehensive impact of vehicle motion state and path characteristics, so as to achieve early warning of potential conflicts. when When the risk exceeds the preset risk threshold, it is determined that there is a risk of travel conflict, and the feedback module triggers the control module to run; When the routes of the two vehicles overlap, The value is 1, otherwise The value is 0.1; The value range is within the preset interval (0.5, 2]. Its value is positively correlated with the traffic priority and road condition complexity of the overlapping road sections of the two vehicle paths. That is, the higher the traffic priority of the overlapping road sections and the greater the road condition complexity, the larger the value; the lower the traffic priority of the overlapping road sections and the smaller the road condition complexity, the smaller the value. When the control module is triggered by the feedback module, it first determines the dominant party in the conflict between the two autonomous vehicles, and then performs speed control on the dominant party: Based on the real-time speed and angle of travel of the two vehicles Calculate the relative velocity contribution value ; If the contribution value is positive, then the i-th driverless car is the dominant party in the conflict; If the contribution value is negative, then the j-th driverless vehicle is the dominant party in the conflict; Speed ​​control is applied to the dominant party in the conflict, while the non-dominant parties maintain their original speeds. The target speed for the dominant party is... ; In the formula: The current movement speed of the party in control of the conflict; Risk response coefficient; The collision risk value between the two vehicles; To preset risk thresholds; The environmental adaptability coefficient of the current driving segment of the party in the conflict; The above formula determines the dominant party in the conflict by the angle between the real-time speed and the direction of travel of the two vehicles. Only the speed of the dominant party is adjusted, while the non-dominant party maintains its original speed. The target speed is dynamically calculated by combining the current speed, the degree of risk, the vehicle's braking and power adjustment capabilities, and the road environment adaptability coefficient. This avoids the efficiency loss caused by blind adjustment and ensures that the speed adjustment is safe and reasonable through multi-parameter coordination, thereby quickly reducing the conflict risk to a safe range. The control module simultaneously sends the target speed to the autonomous vehicle that is in control of the conflict, and records the parameters before and after the speed adjustment, the change data of the conflict risk, and the real-time status data of the non-controlling party, to ensure that the relative speed between the two vehicles is reduced to a safe range after the adjustment. in, The preset value range is [0.3, 1.2]. Its value increases with the stronger the braking performance of the unmanned vehicle and the higher the power adjustment accuracy, and decreases with the weaker the braking performance and the lower the power adjustment accuracy. The preset value range is [0.6-1]. Its value increases with the higher the road condition level and the greater the road surface friction coefficient, and decreases with the lower the road condition level and the smaller the road surface friction coefficient. The message module is used to capture system operation data and unmanned vehicle operation and scheduling iteration data in real time to generate unmanned vehicle operation and maintenance messages; The operation and maintenance messages generated by the message module include operation messages and maintenance messages: The operation message is generated at preset time intervals and includes the real-time location, speed, sub-task execution progress, current road segment environmental data, path adjustment records, and conflict risk iteration data of each unmanned vehicle. Maintenance messages are generated according to the subtask completion nodes, including the state parameter change curves of the unmanned vehicle during the execution of the subtask, the number and magnitude of speed adjustment, and the impact assessment data of scheduling iteration on task completion efficiency. Among them, the impact of scheduling iteration on task completion efficiency is evaluated by quantitatively representing the ratio of actual transportation completion time to preset transportation completion time limit; When the control module first runs to plan the execution path of a single vehicle, it simultaneously configures the vehicle's speed. When the control module runs based on the feedback module's triggering, the iteration of the travel parameter configuration only includes the control of the speed. The receiving module is interconnected with the disassembly module and the distribution module via a wireless network. The disassembly module and the distribution module are interconnected with the control module via a wireless network. The control module is interconnected with the feedback module via a wireless network. The feedback module is interconnected with the message module via a wireless network.

[0022] In this embodiment, the receiving module receives unmanned vehicle (UAV) operating status parameters, surrounding environment data, and transportation order information. It performs standardized format conversion on the received data to obtain a standardized 3D dataset. The decomposition module then obtains the standardized 3D dataset and decomposes the overall transportation task into sub-tasks adapted to different UAV capacities based on the standardized 3D dataset. Simultaneously, it sets execution priorities and constraints for each sub-task. The allocation module then associates the sub-tasks with UAV resources, and performs matching and allocation of sub-tasks and UAVs based on the real-time status of the UAVs and the constraints of the sub-tasks. The control module plans the execution path of a single vehicle based on the allocated sub-tasks and real-time surrounding environment data. The feedback module further collects the status data of the UAVs executing sub-tasks and estimates the risk of trip conflicts for each UAV based on the status data. When the estimation result indicates the existence of a risk, the collected status data is fed back to the control module, triggering the control module to run again to iterate the trip parameter configuration. Finally, the message module captures system operation data and UAV operation and scheduling iteration data in real time to generate UAV operation and maintenance messages.

[0023] In the above embodiments, the system is based on multi-source data integration and accurate calculation. The transportation task can be adapted to the reasonable division of unmanned vehicle capacity, the route planning is more in line with the actual scenario, the trip conflict can be avoided in time, and the transportation efficiency and safety can be improved. At the same time, the operation and maintenance related data are recorded in real time, which improves the overall intelligence level and operational reliability of unmanned vehicle scheduling.

[0024] Application example: Within the XX logistics park, three unmanned vehicles (numbered 1, 2, and 3) are tasked with transporting a batch of electronic products. The park is using the system described in the above example. The system's receiving module first collects the operational status parameters of the three unmanned vehicles (including remaining load, current speed, and remaining driving time), environmental data of the surrounding area (obstacle distribution, road condition level, real-time wind speed, etc.), and transportation order information (total cargo volume, transportation origin and destination, and maximum transportation time limit, etc.). After data cleaning and outlier removal, normalization is performed to map data of different magnitudes to a unified range. The credibility weights of various data sources are calculated to be 0.35, 0.42, and 0.23, respectively, ultimately constructing a standardized three-dimensional dataset containing three dimensions: unmanned vehicle status, environmental characteristics, and order requirements.

[0025] The task breakdown module calculated the overall transportation task complexity to be 8.6 based on the dataset. Since the autonomous vehicles in the park are relatively new and the transportation environment is stable, the task breakdown redundancy coefficient α was set to 0.1. Combined with the maximum transport complexity of the three autonomous vehicles (3.2, 3.5, and 3.0 respectively), the number of subtasks was determined to be three. Subsequently, complexity thresholds for each subtask were set to 3.0, 2.8, and 2.8, respectively, all of which did not exceed the maximum transport complexity of the corresponding adapted autonomous vehicle, and the sum of the three was not less than the total complexity. At the same time, execution priorities were set for each subtask, and time and load constraints were clearly defined.

[0026] The allocation module constructs a multi-dimensional suitability evaluation model. After normalizing the current remaining capacity of the autonomous vehicle, its distance from the sub-task starting point, historical task completion accuracy, and constraint satisfaction, it assigns preset weights to calculate the overall suitability. Sub-tasks are sorted from high to low priority. The first sub-task selects the top 3 autonomous vehicles (1, 2, and 3) based on their overall suitability. After verification, the total complexity of the tasks assigned to these three vehicles does not exceed their maximum capacity. Based on their suitability, these three sub-tasks are finally assigned to autonomous vehicle 2. The second sub-task is assigned to autonomous vehicle 1, and the third sub-task is assigned to autonomous vehicle 3.

[0027] When the control module runs for the first time, it generates multiple candidate paths that meet the start and end points and constraints of the sub-tasks based on the park's electronic map data and real-time environmental data. With the goal of minimizing the overall cost of the paths, the overall cost of the optimal path is calculated to be 4.2. At the same time, the initial moving speeds of unmanned vehicles 1, 2, and 3 are configured to be 15 km / h, 14 km / h, and 16 km / h, respectively.

[0028] During the operation of the autonomous vehicles, the feedback module collects status data in real time, focusing on analyzing the risk of travel conflict between autonomous vehicles 1 and 2. It calculates relevant parameters such as the real-time straight-line distance between the two vehicles, their relative speed, and path overlap. The final conflict risk value is 0.85, exceeding the preset risk threshold of 0.6, thus determining that a travel conflict risk exists. The status data is immediately fed back to the control module.

[0029] Upon receiving feedback, the control module triggers a secondary operation. By calculating the relative speed contribution corresponding to the real-time speeds and directional angles of the two vehicles, it determines that driverless vehicle 2 is the dominant party in the conflict. Combining parameters such as the risk response coefficient and the current road environment adaptation coefficient, the target speed for driverless vehicle 2 is calculated to be 10 km / h, while driverless vehicle 1 maintains its original speed. The control module then sends the target speed to driverless vehicle 2, and after adjustment, the relative speed between the two vehicles drops to a safe range, thus eliminating the conflict risk.

[0030] The message module generates operation messages at preset time intervals, recording data such as the real-time position, speed, task execution progress, and path adjustment records of each unmanned vehicle. After all sub-tasks are completed, a maintenance message is generated, which includes task completion efficiency evaluation data such as the unmanned vehicle status parameter change curve, the speed adjustment of unmanned vehicle 2 (amplitude 4km / h), and the ratio of actual transportation time to preset time limit of 0.9, providing a reference for subsequent scheduling optimization and equipment maintenance.

[0031] Example 2: At the implementation level, based on Example 1, this example refers to... Figure 2 A more detailed description of the unmanned vehicle scheduling system based on multi-source data in Example 1 is provided below: A method for scheduling unmanned vehicles based on multi-source data includes: The system receives unmanned vehicle operating status parameters, surrounding environment data, and transportation order information. Through data cleaning, normalization processing, and 3D data mapping modeling, a standardized 3D dataset is obtained. The overall complexity of the transportation task is calculated based on a standardized 3D dataset. The number of subtasks and their complexity thresholds are determined by combining the maximum capacity complexity of the unmanned vehicle. The execution priority and constraints of the subtasks are set simultaneously. A multi-dimensional adaptability evaluation model is constructed to sort candidate unmanned vehicles according to the priority of sub-tasks and complete the allocation. If the allocation conditions are not met, the sub-tasks are split into secondary parts. Candidate feasible paths are generated based on electronic maps and real-time environmental data. The execution path of a single vehicle is planned with the goal of minimizing the overall path cost, and the initial speed is configured. Collect autonomous vehicle operation status data, calculate the risk value of the conflict between the two vehicles, and when the risk value exceeds the preset threshold, identify the dominant conflict target and adjust the speed of the dominant conflict target. Real-time capture of system and unmanned vehicle related data, generating operation and maintenance messages.

[0032] In summary, the systems and methods described above improve data quality through multi-source data standardization and a credibility weighting mechanism, accurately decompose and adapt to transportation capacity based on task complexity, enhance scenario adaptability through redundant design, achieve efficient matching of tasks and transportation capacity through multi-dimensional adaptation evaluation, reduce pending allocation, optimize planning schemes based on comprehensive path costs to reduce travel consumption and time costs, monitor travel conflict risks in real time and accurately adjust movement speed to avoid collisions while ensuring execution efficiency, and comprehensively record operation and maintenance data to support scheduling strategy optimization and equipment maintenance. Overall, these methods effectively improve the intelligence level and operational stability of unmanned vehicle scheduling and increase transportation efficiency.

[0033] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An unmanned vehicle scheduling system based on multi-source data, characterized in that, include: The receiving module is used to receive unmanned vehicle operating status parameters, surrounding environment data, and transportation order information, and to perform standardized format conversion on the received data to obtain a standardized three-dimensional dataset. The decomposition module is used to acquire a standardized 3D dataset. Based on the standardized 3D dataset, the overall transportation task is decomposed into sub-tasks that are adapted to different unmanned vehicle capacity. At the same time, execution priorities and constraints are set for each sub-task. The allocation module is used to associate subtasks with autonomous vehicle resources, and to match and allocate subtasks with autonomous vehicles by combining the real-time status of autonomous vehicles and the constraints of subtasks. The control module is used to plan the execution path of a single vehicle based on the assigned sub-tasks and real-time surrounding environmental data; The feedback module is used to collect the status data of the autonomous vehicles performing sub-tasks, estimate the risk of conflict in the journey of each autonomous vehicle based on the status data, and when the estimation result indicates that there is a risk, the collected status data is fed back to the control module, triggering the control module to run again to iterate the configuration of the journey parameters. The message module is used to capture system operation data and unmanned vehicle operation and scheduling iteration data in real time to generate unmanned vehicle operation and maintenance messages; When the control module first runs and plans the execution path of a single vehicle, it simultaneously configures the vehicle's speed. When the control module runs based on the feedback module's triggering, the iteration of the travel parameter configuration only includes the control of the speed.

2. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, When the receiving module performs standardized format conversion, it cleans the unmanned vehicle's operating status parameters, surrounding environment data, and transportation order information. Simultaneously, it constructs a three-dimensional data mapping model based on the dimensional attributes of multi-source data and maps data of different magnitudes to a unified data range through normalization processing. ； In the formula: The data is standardized after normalization; This refers to the original data from the corresponding data source; Let be the credibility weight of the i-th type of data source; These are the historical maximum and minimum valid values ​​of the i-th type of data source; This is the scaling factor for the data range; This represents the offset of the data range. The three dimensions of the three-dimensional data mapping model correspond to the autonomous vehicle state dimension, environmental feature dimension, and order demand dimension, respectively. Sub-data items under each dimension are associated with corresponding fields of the standardized three-dimensional dataset through label mapping.

3. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, When the decomposition module decomposes the overall transportation task based on the standardized 3D dataset, it first calculates the total complexity value of the overall transportation task, and then, combined with the maximum capacity complexity of each unmanned vehicle, determines the number of sub-tasks and the complexity threshold of each sub-task. The number of sub-tasks is as follows: ； In the formula: This represents the total complexity value. Redundancy coefficients are allocated to tasks; Let j be the maximum transport capacity complexity of the unmanned vehicle. When determining the complexity threshold for each subtask, the following rule applies: the complexity threshold for each subtask does not exceed the complexity threshold of the corresponding assigned autonomous vehicle. And the sum of the complexities of all subtasks is not less than 1. .

4. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, When the allocation module matches and allocates sub-tasks to unmanned vehicles, it simultaneously constructs a multi-dimensional adaptability evaluation model. The evaluation dimensions of the model include the unmanned vehicle's current remaining capacity, distance from the starting point of the sub-task, historical task completion accuracy, and degree of satisfaction of constraints. Each dimension is assigned a preset weight and then the comprehensive adaptability value is calculated. The allocation module sorts the subtasks from highest to lowest execution priority. For each subtask, it selects the top 3 autonomous vehicles with the highest comprehensive suitability as candidate matching objects. Then, it combines the total complexity of the subtasks already assigned to the candidate autonomous vehicles and selects the candidate objects whose total complexity does not exceed their maximum capacity complexity to complete the allocation. If none of the candidate objects meet the requirements, the subtask is marked as pending allocation, and the decomposition module is triggered to re-evaluate the complexity threshold of the subtask, perform a second decomposition, and then execute the allocation process. Before being used to calculate the overall fit value, the unmanned vehicle's current remaining capacity, distance from the sub-task starting point, historical task completion accuracy, and degree of satisfaction with constraints are all normalized, so that the values ​​of the unmanned vehicle's current remaining capacity, distance from the sub-task starting point, historical task completion accuracy, and degree of satisfaction with constraints are limited to the range of (0,1).

5. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, When planning the execution path for a single vehicle, the control module generates all candidate feasible paths that meet the start and end points and constraints of the subtasks based on electronic map data and real-time surrounding environment data through regional road network topology modeling. Then, it constructs a path planning model with the objective of minimizing the overall path cost. ; In the formula: , , The weights are path length weight, environmental resistance weight, and traffic efficiency weight. This represents the total physical length of the path. Number of path segments; Let i be the environmental resistance coefficient of the i-th path segment; Let i be the length of the i-th path segment; This represents the historical average traffic speed. This is the real-time traffic efficiency coefficient.

6. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, When estimating the conflict risk of each autonomous vehicle's journey, the feedback module uses any two autonomous vehicles performing sub-tasks as analysis units to calculate the conflict risk value: ； In the formula: Let be the real-time straight-line distance between the i-th autonomous vehicle and the j-th autonomous vehicle; The relative speed between the two vehicles; This is the path conflict coefficient; This refers to the path overlap weighting coefficient. The length of the overlapping section of the two vehicle paths; Let be the total length of the sub-task paths of the i-th and j-th autonomous vehicles; when When the risk exceeds the preset risk threshold, it is determined that there is a risk of travel conflict, and the feedback module triggers the control module to run.

7. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, When the control module is triggered by the feedback module, it first determines the dominant party in the conflict between the two autonomous vehicles, and then performs speed control on the dominant party: Based on the real-time speed and angle of travel of the two vehicles Calculate the relative velocity contribution value ; If the contribution value is positive, then the i-th driverless car is the dominant party in the conflict; If the contribution value is negative, then the j-th driverless vehicle is the dominant party in the conflict; Speed ​​control is applied to the dominant party in the conflict, while the non-dominant parties maintain their original speeds. The target speed for the dominant party is... ; In the formula: The current movement speed of the party in control of the conflict; Risk response coefficient; The collision risk value between the two vehicles; To preset risk thresholds; The environmental adaptability coefficient of the current driving segment of the party in the conflict; The control module simultaneously sends the target speed to the autonomous vehicle that is in control of the conflict.

8. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, The operation and maintenance messages generated by the message module include operation messages and maintenance messages: The operation messages are generated at preset time intervals and include the real-time location, speed, sub-task execution progress, current road segment environmental data, path adjustment records, and conflict risk iteration data of each unmanned vehicle. Maintenance messages are generated according to the subtask completion nodes, including the state parameter change curves of the unmanned vehicle during the execution of the subtask, the number and magnitude of speed adjustment, and the impact assessment data of scheduling iteration on task completion efficiency. Among them, the impact of scheduling iteration on task completion efficiency is evaluated by quantitatively representing the ratio of actual transportation completion time to preset transportation completion time limit.

9. The unmanned vehicle scheduling system based on multi-source data according to claim 1, characterized in that, The receiving module is interconnected with a disassembly module and an allocation module via a wireless network. The disassembly module and the allocation module are interconnected with a control module via a wireless network. The control module is interconnected with a feedback module via a wireless network. The feedback module is interconnected with a message module via a wireless network.

10. A method for scheduling unmanned vehicles based on multi-source data, wherein the method is an implementation method of an unmanned vehicle scheduling system based on multi-source data as described in any one of claims 1-9, characterized in that, include: The system receives unmanned vehicle operating status parameters, surrounding environment data, and transportation order information. Through data cleaning, normalization processing, and 3D data mapping modeling, a standardized 3D dataset is obtained. The overall complexity of the transportation task is calculated based on a standardized 3D dataset. The number of subtasks and their complexity thresholds are determined by combining the maximum capacity complexity of the unmanned vehicle. The execution priority and constraints of the subtasks are set simultaneously. A multi-dimensional adaptability evaluation model is constructed to sort candidate unmanned vehicles according to the priority of sub-tasks and complete the allocation. If the allocation conditions are not met, the sub-tasks are split into secondary parts. Candidate feasible paths are generated based on electronic maps and real-time environmental data. The execution path of a single vehicle is planned with the goal of minimizing the overall path cost, and the initial speed is configured. Collect autonomous vehicle operation status data, calculate the risk value of the conflict between the two vehicles, and when the risk value exceeds the preset threshold, identify the dominant conflict target and adjust the speed of the dominant conflict target. Real-time capture of system and unmanned vehicle related data, generating operation and maintenance messages.

Images