Dynamic scheduling method and system for automobile service tasks

Abstract

The invention discloses a dynamic scheduling method and system for an automobile service task, and the method is executed by a centralized scheduling system, and comprises the steps: responding to service requests from a plurality of automatic driving vehicles, and obtaining multi-dimensional dynamic data including a vehicle state, a service facility state, a traffic environment and user preference; inputting the service request and the multi-dimensional dynamic data into a pre-trained scheduling decision model, and generating a scheduling instruction including a service station, a service time window and a vehicle traveling path; and issuing the instruction to a corresponding vehicle to guide the vehicle to complete a service task. According to the invention, the problem of low competition and cooperation efficiency of service resources in a multi-vehicle multi-task scene is solved, and global optimization scheduling and efficient utilization of city-level service resources are realized.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and in particular to a dynamic scheduling method and system for vehicle service tasks. Background Technology

[0002] With the maturation of autonomous driving technology, a large number of highly autonomous intelligent vehicles will appear in future cities. These vehicles will not only need to drive autonomously from point A to point B, but also autonomously complete a series of service tasks such as charging, washing, maintenance, and repair. Currently, existing solutions mostly focus on reservations and navigation for single service types, such as electric vehicle charging station reservation systems. However, the inventors recognized that these systems have significant limitations: First, they are isolated; a single charging reservation system cannot coordinate car wash services, which may require vehicles to travel to different locations, resulting in inefficiency. Second, they lack a global perspective; when multiple vehicles simultaneously request the same service station, it can easily lead to resource competition and queuing congestion, making optimal coordination at the city level impossible.

[0003] Furthermore, traditional scheduling algorithms (such as first-come, first-served and shortest-job-first) struggle to handle such complex dynamic environments. This is because scheduling decisions require comprehensive consideration of dynamic information across multiple dimensions, including real-time vehicle status (e.g., battery level), real-time service supply and demand, real-time city traffic conditions, and personalized user preferences—a typical NP-hard problem. Summary of the Invention

[0004] This application provides a dynamic scheduling method and system for vehicle service tasks, aiming to solve the problems in the prior art where scheduling systems for different service types are isolated from each other, lack global coordination and optimization capabilities, resulting in low vehicle service efficiency, severe resource competition and queuing congestion, and the difficulty of traditional scheduling algorithms in coping with multi-dimensional dynamic decision-making environments.

[0005] Firstly, a dynamic scheduling method for vehicle service tasks is provided, the method being executed by a centralized scheduling system, comprising:

[0006] In response to service requests from multiple autonomous vehicles, multidimensional dynamic data related to the service requests is acquired, including vehicle status data, service facility status data, traffic environment data, and user preference data.

[0007] The service request and the multidimensional dynamic data are input into a pre-trained scheduling decision model to generate a scheduling instruction corresponding to the service request, wherein the scheduling instruction specifies at least the service station, the service time window, and the vehicle travel path.

[0008] The dispatch instructions are then sent to the corresponding autonomous vehicles.

[0009] Optionally, in the above scheme, the method further includes:

[0010] Monitor the execution status of scheduling commands and environmental changes;

[0011] When a vehicle fails to arrive as planned, a service facility malfunctions, or a higher-priority emergency request is received, the scheduling decision model is triggered to recalculate, and an updated scheduling instruction is generated and issued based on the recalculation result.

[0012] In the above scheme, optionally, the scheduling decision model is constructed and trained in the following manner:

[0013] Define all pending service requests and the current service resource status in the system as the state space;

[0014] The set of actions that assign each service request to a specific service site is defined as the action space;

[0015] The system evaluates and optimizes its decisions based on a pre-defined reward function, which is designed to positively incentivize the overall service throughput, average latency, and resource utilization of the system.

[0016] In the above scheme, optionally, the reward function also includes a fairness constraint term, which is used to reward and compensate vehicle requests whose waiting time exceeds a preset threshold, so as to avoid scheduling discrimination.

[0017] Optionally, in the above scheme, the service time window is a time period; the method further includes:

[0018] Based on the service time window, speed guidance suggestions or safe waiting suggestions are sent to the autonomous vehicle via vehicle-road cooperative communication so that the vehicle's arrival time falls within the time period.

[0019] Optionally, in the above scheme, the vehicle status data includes one or more of the following: vehicle location, remaining energy, vehicle health, and service urgency.

[0020] The service facility status data includes one or more of the following: service capacity of the target service station, real-time queuing information, estimated service duration, and service price.

[0021] The traffic environment data includes real-time road condition information, weather information, and estimated travel time from the vehicle's current location to each candidate service station calculated based on the aforementioned information.

[0022] The user preference data includes one or more of the following: user preference for service brands, sensitivity to service prices, and urgency regarding service completion time.

[0023] Secondly, a dynamic scheduling system for automotive service tasks is provided, comprising:

[0024] The request receiving module is used to receive multiple service requests from multiple autonomous vehicles.

[0025] The data perception module is used to acquire multi-dimensional dynamic data related to the service request. The multi-dimensional dynamic data includes vehicle status data, service facility status data, traffic environment data, and user preference data.

[0026] The intelligent decision-making module has a built-in reinforcement learning scheduling model, which is used to receive the service request and the multi-dimensional dynamic data, and generate a scheduling instruction corresponding to the service request. The scheduling instruction specifies at least the service station, the service time window and the vehicle travel path.

[0027] The scheme distribution module is used to distribute the scheduling instructions to the corresponding autonomous vehicles.

[0028] Optionally, the above scheme also includes a dynamic rescheduling module, which is used to monitor the execution status and environmental changes of the scheduling instruction after the scheduling instruction is issued; when it is detected that a vehicle fails to arrive as planned, a service facility malfunctions, or an emergency request with higher priority is received, the scheduling decision model is triggered to recalculate, and an updated scheduling instruction is generated and issued based on the recalculation result.

[0029] Optionally, in the above scheme, the data sensing module includes:

[0030] The vehicle status data acquisition unit is used to acquire vehicle location, remaining energy, vehicle health, and service urgency.

[0031] The service facility status data acquisition unit is used to acquire the service capacity, current queuing status, estimated service duration and service price of each service station.

[0032] The traffic environment data acquisition unit is used to acquire real-time traffic information, weather information, and calculate the estimated travel time from the vehicle's current location to each candidate service station.

[0033] The user preference data acquisition unit is used to acquire information on users' preference for service brands, their sensitivity to service prices, and their urgency regarding service completion time.

[0034] Thirdly, an electronic device is provided, comprising:

[0035] Memory, used to store computer programs;

[0036] A processor for executing the computer program to implement the method described above.

[0037] Compared with the prior art, this application has at least the following beneficial effects:

[0038] Based on further analysis and research into the problems of existing technologies, this application recognizes that existing scheduling systems for different service types are isolated from each other, lacking global coordination and optimization capabilities. This leads to low vehicle service efficiency, severe resource competition and queuing congestion, and the inability of traditional scheduling algorithms to cope with multi-dimensional dynamic decision-making environments. This application constructs a unified centralized scheduling system that integrates multi-dimensional dynamic data such as vehicle status, service facility status, real-time traffic environment, and user preferences. It utilizes a pre-trained scheduling decision model for real-time calculation and decision-making, achieving global collaborative scheduling and dynamic path planning for multiple vehicles and multiple service tasks. This breaks down scheduling barriers between different service types, significantly improves service resource utilization efficiency, effectively alleviates resource competition and queuing congestion, and endows the system with high adaptability and intelligent decision-making capabilities in complex dynamic environments.

[0039] This application also has at least the following beneficial effects:

[0040] 1. From isolation to collaboration: It breaks down the barriers between different types of car services, realizes collaborative scheduling of "one-click multi-service", and reduces vehicle empty driving and waiting time.

[0041] 2. From static to dynamic: Reinforcement learning-based scheduling models have online learning and adaptive capabilities, which can effectively cope with dynamic changes such as traffic congestion and sudden failures of service facilities, making them more robust and intelligent than static rules or traditional algorithms.

[0042] 3. From local to global: With the goal of optimizing the overall efficiency of the urban system, the system smooths out the load peaks of each service station through intelligent allocation and time window management, and significantly improves the utilization rate of the entire service network resources.

[0043] 4. User experience optimization: By incorporating user preference factors, personalized services are provided while pursuing system efficiency, meeting the differentiated needs of different users. Attached Figure Description

[0044] Figure 1 A flowchart illustrating a dynamic scheduling method for vehicle service tasks provided in one embodiment of this application;

[0045] Figure 2 A schematic diagram of the overall process of a dynamic scheduling method for vehicle service tasks provided in one embodiment of this application;

[0046] Figure 3 A schematic diagram illustrating the structure of a multidimensional decision factor system provided in one embodiment of this application;

[0047] Figure 4 A schematic diagram illustrating the training and decision-making principles of a reinforcement learning scheduling model provided in one embodiment of this application;

[0048] Figure 5 This diagram illustrates the comparison of average system waiting time between the scheduling method described in this application and the first-come, first-served method. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0050] In the description of this application, unless otherwise stated, the terms "including", "comprising", "having", etc., also mean "not limited to" (certain units, components, materials, steps, etc.).

[0051] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of the relevant data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, and necessary confidentiality measures have been taken. They do not violate public order and good morals, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0052] In some embodiments, reference Figures 1-5 A dynamic scheduling method for vehicle service tasks is provided, the method being executed by a centralized scheduling system, comprising:

[0053] In response to service requests from multiple autonomous vehicles, multidimensional dynamic data related to the service requests is acquired, including vehicle status data, service facility status data, traffic environment data, and user preference data.

[0054] The service request and the multidimensional dynamic data are input into a pre-trained scheduling decision model to generate a scheduling instruction corresponding to the service request, wherein the scheduling instruction specifies at least the service station, the service time window, and the vehicle travel path.

[0055] The dispatch instructions are then sent to the corresponding autonomous vehicles.

[0056] This method aims to solve the resource conflict and efficiency optimization problems in multi-vehicle, multi-service scenarios. Its execution process can be summarized into the following core steps:

[0057] 1. Receiving service requests and acquiring multi-dimensional dynamic data. The centralized scheduling system continuously receives service requests from multiple autonomous vehicles, which may include various service types such as charging, car washing, maintenance, and repair. To make globally optimal scheduling decisions, the system synchronously acquires multi-dimensional dynamic data related to the requests, specifically including:

[0058] Vehicle status data: such as current vehicle location, remaining battery / fuel level, health status, service urgency level, etc.;

[0059] Service facility status data: such as the current service capacity of each service station, queuing status, estimated service time, and rates;

[0060] Traffic environment data: such as real-time road conditions, weather conditions, estimated travel time, etc.;

[0061] User preference data: such as users' preference levels for service brands, prices, and time urgency.

[0062] 2. Generate scheduling instructions based on the scheduling decision model. The system inputs the service request along with the aforementioned multi-dimensional dynamic data into the pre-trained scheduling decision model. This model is typically built using artificial intelligence methods such as reinforcement learning, and can learn and optimize scheduling strategies in a constantly changing system environment to maximize overall system efficiency and fairness. The scheduling instructions output by the model are a set of structured decision results, including at least:

[0063] Designated service stations: Assign appropriate service stations to each vehicle;

[0064] Recommended service time window: Provides a time period rather than a single moment to smooth site load;

[0065] Plan vehicle travel routes: Provide the optimal or optimized route from the current location to the service station by combining real-time traffic data.

[0066] 3. Issue instructions and guide the vehicles to execute them.

[0067] The generated dispatch instructions are sent to the corresponding autonomous vehicles in real time via vehicle-to-everything (V2X) or other communication methods. The vehicles proceed to the designated service stations according to the instructions, and can adjust their speed according to the system's suggested parameters or wait en route to ensure arrival within the recommended time window, thereby achieving balanced utilization of service resources and overall coordination of traffic flow.

[0068] Overall, the method described in this embodiment achieves real-time matching, dynamic sorting, and global optimization of multi-vehicle, multi-service tasks through a centralized intelligent scheduling architecture, significantly improving the overall efficiency, resource utilization, and user experience of city-level automotive service networks.

[0069] In some embodiments, the method further includes:

[0070] Monitor the execution status of scheduling commands and environmental changes;

[0071] When a vehicle fails to arrive as planned, a service facility malfunctions, or a higher-priority emergency request is received, the scheduling decision model is triggered to recalculate, and an updated scheduling instruction is generated and issued based on the recalculation result.

[0072] After the initial dispatch instruction is issued, the system does not end the task, but enters the execution monitoring phase. The system continuously tracks two key pieces of information: (1) the execution status of the dispatch instruction: such as whether the vehicle travels along the planned route and whether there is a deviation in the expected arrival time; (2) changes in the relevant environment: such as sudden road congestion, changes in the status of service facilities (such as charging pile failure), or whether new service requests are generated.

[0073] The system has predefined rules to determine when a recalculation needs to be initiated. Typical triggering events include:

[0074] Plan deviation: The vehicle failed to arrive at the service station within the planned time window due to unforeseen circumstances (such as traffic congestion or vehicle breakdown);

[0075] Resource anomaly: The designated service site experiences a sudden failure or insufficient resources (such as power outage or equipment maintenance).

[0076] Priority update: A higher priority emergency service request has been received (such as a vehicle battery about to run out or an emergency task involving public safety).

[0077] Once any of the above conditions are triggered, the system will automatically recall the pre-trained scheduling decision model. The model will be recalculated and optimized based on the latest, global system state (including all incomplete requests, real-time vehicle locations, the latest facility status, updated road conditions, etc.).

[0078] The model outputs updated scheduling instructions, which may involve: reallocating service stations to affected vehicles; adjusting original service time windows or travel routes; and coordinating resources for emergency requests (such as advising other vehicles to wait). These new instructions are sent to the relevant vehicles in real time to ensure the system can respond quickly to changes and maintain overall operational efficiency and task completion rates.

[0079] By introducing this dynamic rescheduling mechanism, the method described in this embodiment is upgraded from "one-time scheduling" to an intelligent closed-loop system of "continuous optimization and adjustment", which significantly enhances the robustness, flexibility and reliability of the scheduling scheme in real complex environments.

[0080] In some embodiments, the scheduling decision model is constructed and trained in the following manner:

[0081] Define all pending service requests and the current service resource status in the system as the state space;

[0082] The set of actions that assign each service request to a specific service site is defined as the action space;

[0083] The system evaluates and optimizes its decisions based on a pre-defined reward function, which is designed to positively incentivize the overall service throughput, average latency, and resource utilization of the system.

[0084] The scheduling decision model employs a reinforcement learning framework, autonomously learning the optimal scheduling strategy through simulated continuous interaction with the environment. Its core elements are constructed as follows:

[0085] The model's state space is defined as a snapshot of all pending service requests and the current state of service resources in the system. It is not focused on a single vehicle, but rather aims to capture the global status of the dispatching system at a given moment, which may include: details of all pending requests (service type, vehicle location, user preferences, etc.); the real-time status of all available service stations (queue length, service capacity, estimated idle time, etc.); and relevant dynamic environmental information (regional traffic flow, weather impact, etc.).

[0086] The model's action space is defined as the set of actions that assign any pending service request to a specific service site. At each decision point, the model selects an action from this action space (i.e., makes an "assignment" decision). Through successive action selections, the model progressively completes the scheduling of all requests.

[0087] The model's training revolves around a carefully designed reward function. This function acts as a "compass" for the model's learning, designed to positively incentivize key indicators of overall system efficiency, including: overall system service throughput: encouraging more vehicles to be served per unit time; average waiting time: encouraging a reduction in the total waiting time from request to receiving service (including driving wait and service queuing); and resource utilization: encouraging balanced and efficient use of the service capacity of each service station, avoiding resource idleness or excessive congestion.

[0088] During training, the model learns to select scheduling actions that maximize long-term cumulative rewards in complex, multi-dimensional system states by continuously trying (selecting actions), observing results (receiving rewards), and learning and adjusting. This achieves the approximation of the preset optimization goal.

[0089] In some embodiments, the reward function further includes a fairness constraint term for rewarding vehicle requests whose waiting time exceeds a preset threshold, in order to avoid scheduling discrimination.

[0090] This embodiment significantly improves the optimization objective of the scheduling decision model by introducing a mechanism to ensure individual fairness while pursuing overall system efficiency (such as throughput and average waiting time). This constraint aims to address the "starvation" or "discrimination" problems that may arise from purely efficiency-oriented scheduling, where individual vehicles (e.g., due to remote starting locations, special request types, or multiple rescheduling) may experience abnormally long waiting times due to prolonged lack of service. By introducing this constraint, the system is explicitly guided to focus on and improve the situation of the worst-performing users, ensuring that scheduling decisions achieve a balance between overall optimization and individual fairness.

[0091] The system sets a reasonable preset threshold for vehicle waiting time. This threshold can be a fixed value or dynamically adjusted based on historical data or service level agreements. In the reward function, when the cumulative waiting time of a service request exceeds this threshold, the system allocates an additional positive reward to that request. This compensation mechanism makes the model tend to prioritize vehicle requests with excessively long waiting times and unfavorable conditions when making decisions, thereby proactively avoiding "scheduling discrimination."

[0092] In some embodiments, the service time window is a time period; the method further includes:

[0093] Based on the service time window, speed guidance suggestions or safe waiting suggestions are sent to the autonomous vehicle via vehicle-road cooperative communication so that the vehicle's arrival time falls within the time period.

[0094] In some embodiments, the vehicle status data includes one or more of the following: vehicle location, remaining energy, vehicle health, and service urgency.

[0095] The service facility status data includes one or more of the following: service capacity of the target service station, real-time queuing information, estimated service duration, and service price.

[0096] The traffic environment data includes real-time road condition information, weather information, and estimated travel time from the vehicle's current location to each candidate service station calculated based on the aforementioned information.

[0097] The user preference data includes one or more of the following: user preference for service brands, sensitivity to service prices, and urgency regarding service completion time.

[0098] In some embodiments, a dynamic scheduling system for vehicle service tasks is also provided, comprising:

[0099] The request receiving module is used to receive multiple service requests from multiple autonomous vehicles.

[0100] The data perception module is used to acquire multi-dimensional dynamic data related to the service request. The multi-dimensional dynamic data includes vehicle status data, service facility status data, traffic environment data, and user preference data.

[0101] The intelligent decision-making module has a built-in reinforcement learning scheduling model, which is used to receive the service request and the multi-dimensional dynamic data, and generate a scheduling instruction corresponding to the service request. The scheduling instruction specifies at least the service station, the service time window and the vehicle travel path.

[0102] The scheme distribution module is used to distribute the scheduling instructions to the corresponding autonomous vehicles.

[0103] In some embodiments, a dynamic rescheduling module is also included, which is used to monitor the execution status and environmental changes of the scheduling instruction after the scheduling instruction is issued; when it is detected that a vehicle fails to arrive as planned, a service facility malfunctions, or an emergency request with higher priority is received, the scheduling decision model is triggered to recalculate, and an updated scheduling instruction is generated and issued based on the recalculation result.

[0104] The system's recommended service time window is a time interval with start and end times, rather than a single, precise moment. This design gives the dispatch system inherent flexibility, allowing vehicles to arrive within a certain time frame as acceptable. This flexibility forms the basis for addressing urban traffic uncertainties (such as momentary congestion and intersection delays) and also provides adjustment space for global system optimization.

[0105] To ensure vehicles can effectively utilize this flexibility, the system provides dynamic on-route guidance via a vehicle-to-everything (V2X) communication network. Based on the vehicle's real-time location, speed, road conditions, distance to service stations, and time window requirements, the system calculates and issues two types of optimization suggestions in real time:

[0106] Speed ​​guidance suggestion: It is recommended that vehicles increase or decrease their speed appropriately to "fine-tune" their estimated arrival time.

[0107] Safe waiting suggestions en route: If a vehicle is expected to arrive too early, the system can guide it to wait briefly in a safe area that does not affect traffic (such as a designated parking area or a slow-moving section of road).

[0108] Through this dynamic guidance, the system can proactively distribute the arrival times of different vehicles evenly within their respective service time windows, thereby avoiding instantaneous congestion and resource competition caused by a large number of vehicles arriving at the service station at the same time. This mechanism achieves "peak shaving and valley filling" of the service station load, smoothing out the reception pressure and significantly improving the station's operational efficiency and service experience. At the same time, it also reduces the situation where vehicles arrive too early and waste time idling or queuing outside the station.

[0109] In some embodiments, the data sensing module includes:

[0110] The vehicle status data acquisition unit is used to acquire vehicle location, remaining energy, vehicle health, and service urgency.

[0111] The service facility status data acquisition unit is used to acquire the service capacity, current queuing status, estimated service duration and service price of each service station.

[0112] The traffic environment data acquisition unit is used to acquire real-time traffic information, weather information, and calculate the estimated travel time from the vehicle's current location to each candidate service station.

[0113] The user preference data acquisition unit is used to acquire information on users' preference for service brands, their sensitivity to service prices, and their urgency regarding service completion time.

[0114] By constructing the dynamic data system covering the four dimensions of "vehicles, facilities, roads, and people," the method described in this embodiment establishes a panoramic real-time perception network. This enables scheduling decisions to move beyond simple matching based on local or static information. Instead, it allows for global, personalized, and real-time-adaptive optimal calculations based on a complete, dynamic, and fine-grained information graph. This is the fundamental guarantee for achieving its superior scheduling performance.

[0115] Example 1

[0116] In a smart city demonstration zone, three autonomous vehicles simultaneously send service requests to the cloud platform: Vehicle A (15% battery remaining, needs charging, user prefers fast charging), Vehicle B (sufficient battery, requests automatic car wash), and Vehicle C (20% battery remaining, needs charging, user is price-sensitive).

[0117] refer to Figure 2 The platform's multi-dimensional decision factor construction module quickly acquired the status and preferences of these vehicles, as well as real-time information on two charging stations and one car wash in the area (including queuing status, rates, charging power, etc.). The dynamic scheduling model then calculated and output the following decision:

[0118] Dispatch vehicle A to charging station 1, which has high charging power but slightly higher prices and faster charging speeds, and plan the fastest route for it.

[0119] Dispatch vehicle B to the car wash.

[0120] Vehicle C was relocated to another charging station 2, which offered promotional activities but had a slightly slower charging speed, because this decision satisfied its price-sensitive preferences without significantly impacting the overall efficiency of the system.

[0121] The platform issued this plan to all vehicles, and the three vehicles followed the instructions and completed the task efficiently, avoiding congestion caused by vehicles A and C rushing to the same charging station at the same time.

[0122] Example 2

[0123] Following Example 1, while vehicle A is en route to charging station 1, one of the charging piles at charging station 1 experiences a sudden malfunction. Upon detecting this event, the platform immediately triggers a dynamic rescheduling process. After recalculation, the model decides to reassign vehicle A to charging station 2. However, because vehicle A has a low battery level and high priority, the model, via V2X communication, suggests that a vehicle at charging station 2 that is about to finish charging should wait to make way for vehicle A, and sends this updated route and new service time window to vehicle A. The entire process is completed automatically by the system, ensuring vehicle A's urgent needs are met.

[0124] In summary, this application provides an efficient, reliable, and intelligent global scheduling solution for the service needs of massive numbers of autonomous vehicles in future smart cities.

[0125] In some embodiments, an electronic device is also provided, comprising:

[0126] Memory, used to store computer programs;

[0127] A processor is configured to execute the computer program to implement the steps of the method provided in the above embodiments.

[0128] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A dynamic scheduling method for vehicle service tasks, characterized in that, The method is executed by a centralized scheduling system and includes: In response to service requests from multiple autonomous vehicles, multidimensional dynamic data related to the service requests is acquired, including vehicle status data, service facility status data, traffic environment data, and user preference data. The service request and the multidimensional dynamic data are input into a pre-trained scheduling decision model to generate a scheduling instruction corresponding to the service request, wherein the scheduling instruction specifies at least the service station, the service time window, and the vehicle travel path. The dispatch instructions are then sent to the corresponding autonomous vehicles.

2. The method according to claim 1, characterized in that, The method further includes: Monitor the execution status of scheduling commands and environmental changes; When a vehicle fails to arrive as planned, a service facility malfunctions, or a higher-priority emergency request is received, the scheduling decision model is triggered to recalculate, and an updated scheduling instruction is generated and issued based on the recalculation result.

3. The method according to claim 1, characterized in that, The scheduling decision model is constructed and trained in the following ways: Define all pending service requests and the current service resource status in the system as the state space; The set of actions that assign each service request to a specific service site is defined as the action space; The system evaluates and optimizes its decisions based on a pre-defined reward function, which is designed to positively incentivize the overall service throughput, average latency, and resource utilization of the system.

4. The method according to claim 3, characterized in that, The reward function also includes a fairness constraint term, which is used to reward and compensate vehicle requests whose waiting time exceeds a preset threshold, so as to avoid scheduling discrimination.

5. The method according to claim 1, characterized in that, The service time window is a time period; the method further includes: Based on the service time window, speed guidance suggestions or safe waiting suggestions are sent to the autonomous vehicle via vehicle-road cooperative communication so that the vehicle's arrival time falls within the time period.

6. The method according to claim 1, characterized in that, The vehicle status data includes one or more of the following: vehicle location, remaining energy, vehicle health, and service urgency. The service facility status data includes one or more of the following: service capacity of the target service station, real-time queuing information, estimated service duration, and service price. The traffic environment data includes real-time road condition information, weather information, and estimated travel time from the vehicle's current location to each candidate service station calculated based on the aforementioned information. The user preference data includes one or more of the following: user preference for service brands, sensitivity to service prices, and urgency regarding service completion time.

7. A dynamic scheduling system for automobile service tasks, characterized in that, include: The request receiving module is used to receive multiple service requests from multiple autonomous vehicles. The data perception module is used to acquire multi-dimensional dynamic data related to the service request. The multi-dimensional dynamic data includes vehicle status data, service facility status data, traffic environment data, and user preference data. The intelligent decision-making module has a built-in reinforcement learning scheduling model, which is used to receive the service request and the multi-dimensional dynamic data, and generate a scheduling instruction corresponding to the service request. The scheduling instruction specifies at least the service station, the service time window and the vehicle travel path. The scheme distribution module is used to distribute the scheduling instructions to the corresponding autonomous vehicles.

8. The system according to claim 7, characterized in that, It also includes a dynamic rescheduling module, which monitors the execution status and environmental changes of the scheduling instructions after they are issued; when a vehicle fails to arrive as planned, a service facility malfunctions, or an emergency request with a higher priority is received, the scheduling decision model is triggered to recalculate, and an updated scheduling instruction is generated and issued based on the recalculation result.

9. The system according to claim 7, characterized in that, The data sensing module includes: The vehicle status data acquisition unit is used to acquire vehicle location, remaining energy, vehicle health, and service urgency. The service facility status data acquisition unit is used to acquire the service capacity, current queuing status, estimated service duration and service price of each service station. The traffic environment data acquisition unit is used to acquire real-time traffic information, weather information, and calculate the estimated travel time from the vehicle's current location to each candidate service station. The user preference data acquisition unit is used to acquire information on users' preference for service brands, their sensitivity to service prices, and their urgency regarding service completion time.

10. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the method as described in any one of claims 1 to 6.

Images