Resource acquisition method, system and unmanned vehicle

Abstract

This disclosure provides a resource acquisition method, system, and unmanned vehicle. The method includes: determining a target edge data center; acquiring target resources through a target resource acquisition path determined based on the target edge data center; the target edge data center is located at a service switching point and has pre-acquired the resources required by the vehicle; the service switching point is a geographical location area associated with changes in vehicle operation tasks.

Description

Technical Field

[0001] This disclosure relates to the fields of smart mining, autonomous driving, and vehicle control technology, and in particular to resource acquisition methods, systems, and unmanned vehicles. Background Technology

[0002] In the process of intelligent operation in the mining area, the mining fleet, as the core operation carrier, needs to frequently retrieve various key resources throughout the entire operation process. These resources include map tiles for vehicle positioning and route planning, AI models to support intelligent vehicle decision-making, blacklists and whitelists to ensure operational safety and access control, and task sheets to guide vehicle operation scheduling. The timely and efficient acquisition of these resources directly determines the operational efficiency and intelligent operation level of the mining fleet.

[0003] Currently, the key resources of the mining fleet are mainly acquired through a cloud distribution model, where the vehicle-mounted terminals directly establish a communication link with the cloud control platform to retrieve resources. However, this model has revealed many technical defects under the complex operating environment and communication conditions of the mining area, making it difficult to meet the operational data needs of the mining fleet. On the one hand, the mining area has many terrain obstacles such as deep pits and slopes, which can easily block wireless communication links, resulting in a significant decrease in the communication stability between the vehicle-mounted terminals and the cloud control platform, slow response speed of the first packet of data transmission, and even transmission interruption. On the other hand, the cellular communication network in the mining area is prone to congestion due to multiple vehicles communicating simultaneously, and the communication cost of cellular networks is high, further increasing the cost of data distribution and transmission latency. Moreover, under the cloud distribution model, when different vehicle-mounted terminals retrieve the same resources, a large number of duplicate requests to the source are easily generated, which not only consumes a lot of resources of the cloud control platform, but also exacerbates network congestion and reduces the overall data distribution efficiency. Summary of the Invention

[0004] This disclosure provides a resource acquisition method, system, and unmanned vehicle to address the problems existing in the prior art.

[0005] In view of the above problems, in a first aspect, this disclosure provides a resource acquisition method applied to a vehicle, comprising: Identify the target edge data center; The target resources are acquired by using the target resource acquisition path determined based on the target edge data center. The target edge data center is located at the service switching point and has pre-acquired the resources required by the vehicle; the service switching point is a geographical location area associated with changes in vehicle operation tasks.

[0006] In conjunction with the first aspect, in one possible implementation, determining the target edge data center includes: Assess the real-time status of edge data centers and determine the target edge data centers based on the assessment results; Preferably, the real-time status of the edge data center is evaluated, and the target edge data center is determined based on the evaluation results, including: The location of the vehicle is determined to be within each edge data center of its service radius; Edge data centers that meet the following conditions are identified as target edge data centers: the signal strength of the communication channel between the vehicle and the edge data center is higher than a first threshold, the network congestion of the communication channel is lower than a second threshold, and the communication channel meets the line-of-sight condition.

[0007] In conjunction with the first aspect, in one possible implementation, the target resource acquisition path is determined based on the target edge data center in the following manner: The target resource acquisition path is determined based on the weights of the resource acquisition paths; wherein the weights of the resource acquisition paths include: the weight corresponding to the target edge data center, the weight corresponding to the nearby vehicles, and the weight corresponding to the cloud control platform; Preferably, determining the target resource acquisition path based on the weight of the resource acquisition path includes: Determine the proportional relationship between the weights corresponding to the target edge data center, the weights corresponding to nearby vehicles, and the weights corresponding to the cloud control platform; Based on the aforementioned proportional relationship, multiple target resources are divided into subsets of a number related to the aforementioned proportional relationship, and the target resource acquisition path corresponding to each subset is determined.

[0008] In conjunction with the first aspect, in one possible implementation, if acquiring the target resource via the current target resource acquisition path fails or times out, the method further includes: According to the preset resource acquisition path priority, switch to the new target resource acquisition path in sequence, and use it as the new current target resource acquisition path; The target resource is obtained by switching to the new target resource acquisition path corresponding to the current path priority, and the acquisition status of the target resource is monitored until the download of the target resource is completed. Alternatively, if the acquisition of the target resource fails or times out, the system switches to the new target resource acquisition path corresponding to the next path priority. The resource acquisition paths, in descending order of priority, are: target edge data center, nearby vehicles, and cloud control platform.

[0009] Preferred options also include: If all attempts to acquire the target resource according to the preset resource acquisition path priorities fail, the resource requested and cached based on the resource request information is cancelled, and the original resources are maintained.

[0010] In conjunction with the first aspect, in one possible implementation, if the acquisition of the target resource fails or times out when the target edge data center is used as the target resource acquisition path, the method further includes: Continuously acquire target resources from the target edge data center; If the acquisition of target resources from the target edge data center fails or times out, the weight of the resource acquisition path of the target edge data center will be reduced according to the first rule. During the process of continuously acquiring target resources from the target edge data center based on the reduced weight, if there is a successful acquisition status when acquiring target resources from the target edge data center, the weight of the target edge data center is increased according to the second rule.

[0011] Secondly, a resource acquisition method is provided, applied to an edge data center, wherein the edge data center is located at a service switching point, and the service switching point is a geographical location area associated with changes in vehicle operation tasks. The method includes: Obtain resource pre-fetching tasks from the cloud control platform; Based on a preset multi-factor comprehensive priority model, the pre-fetch priority of each resource in the resource pre-fetching task is determined. Based on the pre-fetching priority of each resource, the system sends the corresponding target resource acquisition request to the cloud control platform in priority order; and receives the corresponding target resource and caches it locally so that the vehicle can acquire the target resource from the edge data center.

[0012] In conjunction with the second aspect, in one possible implementation, determining the prefetch priority of each resource in the resource prefetching task based on a preset multi-factor comprehensive priority model includes: For each type of resource, assign weights to each factor of that resource type; For each resource in the resource prefetching task, determine the priority score of that resource under each factor. Based on the weights of each factor of the resource, the priority scores under each factor are weighted and calculated to determine the prefetch priority of the resource.

[0013] In conjunction with the second aspect, in one possible implementation, the multi-factor comprehensive priority model includes at least one of the following factors: heat priority factor, time-sensitive factor, map association factor, and volume-friendly factor; The popularity priority factor is used to characterize the global request popularity of the resource on the cloud side and the historical request popularity of the corresponding content type of the resource in the edge data center. The time-sensitive factor is used to characterize the degree of correlation between the resource and the vehicle operation task, as well as the proximity of the current time to the relevant time node of the vehicle operation task. The map association factor is used to characterize the association strength between the resource and the map constraints; wherein the map constraints include at least one of the following: environmental awareness friendliness, path connectivity, and basic data freshness. The volume-friendly factor is used to characterize the fragment size of a resource.

[0014] In conjunction with the second aspect, in one possible implementation, when the resource type is map data, the association strength between the resource and map constraints is determined in the following manner: Determine the high-precision map static layer data and dynamic event layer data corresponding to the resource; The high-precision map static layer data is subjected to local rasterization processing to determine the line-of-sight probability and reachability assessment score; wherein, the line-of-sight probability is used to characterize environmental perception friendliness, and the reachability assessment score is used to characterize path connectivity. Based on the event impact area corresponding to the dynamic event layer data and the high-precision map static layer data, the map freshness corresponding to the resource is determined; the map freshness is used to characterize the freshness of the basic data. The direct-view probability, reachability assessment score, and map freshness are fused using a preset fusion rule to determine the correlation strength between the resource and map constraints.

[0015] In conjunction with the second aspect, in one possible implementation, the method further includes: Based on the target resource's lifetime and preset expiration date in the local cache, perform a resource eviction operation on the target resource; and / or, Based on the hit rate of the target resource in the local cache within the statistical period and the preset hit rate threshold, a resource eviction operation is performed on the target resource.

[0016] In conjunction with the second aspect, in one possible implementation, the type of the resource includes at least one of the following: map data, software packages, access control policies, and job tasks; and / or, The resource is split into equal-granularity fragments, and a directory is set for each fragment. The directory includes: version identifier, preset priority, and preset validity period.

[0017] Thirdly, an unmanned vehicle is provided that performs the steps of the resource acquisition method as described in the first aspect or any possible implementation thereof.

[0018] Fourthly, a resource acquisition system is provided, comprising: an edge data center and a vehicle; wherein, The edge data center is located at the service switching point, which is a geographical area associated with changes in vehicle operation tasks; The vehicle is used to perform the steps of the resource acquisition method as described in the first aspect or any possible implementation in conjunction with the first aspect.

[0019] The beneficial effects of the embodiments disclosed herein include: This disclosure provides a resource acquisition method, system, and unmanned vehicle, comprising: determining a target edge data center; acquiring target resources through a target resource acquisition path determined based on the target edge data center; the target edge data center is located at a service switching point and has pre-acquired the resources required by the vehicle; the service switching point is a geographical location area associated with changes in vehicle operation tasks. The resource acquisition method provided in this disclosure allows the vehicle to directly acquire resources from the edge node after the target edge data center has completed resource pre-caching, without needing to initiate a pull request from the origin cloud. This effectively avoids problems such as terrain obstructing cellular communication links in deep pits and slopes in mining areas, and network congestion caused by multiple concurrent vehicles leading to slow initial packet response and link interruptions in resource transmission. The target edge data center, using an excavator as a carrier, can move synchronously with the excavator's operating area, ensuring the edge node remains close to the vehicle's operating trajectory, further shortening the resource transmission link, reducing interference from complex terrain in mining areas on communication links, and significantly improving the stability of vehicle resource acquisition when combined with a pre-planned resource acquisition path. Mining area business switching points are high-frequency areas where convoys park and queue. Excavators, as core operating equipment in mining areas, operate at fixed locations or travel back and forth to these points for extended periods. By mounting the target edge data center on the excavator, edge nodes can be precisely deployed and dynamically covered to business switching points. During the window of waiting for operations, vehicles can acquire resources nearby, effectively making up for the technical deficiencies of existing technologies in adapting to highly constrained special scenarios such as parking and queuing at mining loading points. Attached Figure Description

[0020] Figure 1 One of the flowcharts for the resource acquisition method provided in the embodiments of this disclosure; Figure 2 This is a schematic diagram illustrating the real-time status assessment of an edge data center, provided as an embodiment of this disclosure. Figure 3 A schematic diagram illustrating the priority of resource acquisition paths provided in embodiments of this disclosure; Figure 4 A second flowchart of a resource acquisition method provided in this embodiment of the disclosure; Figure 5 A schematic diagram illustrating the determination of the prefetch priority of each resource according to an embodiment of this disclosure; Figure 6 This is a schematic diagram illustrating the determination of the correlation strength between resources and map constraints, provided in an embodiment of this disclosure. Detailed Implementation

[0021] This disclosure provides a resource acquisition method, system, and unmanned vehicle. Preferred embodiments of this disclosure are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustrative and explanatory purposes only and are not intended to limit the scope of this disclosure. Furthermore, the embodiments and features described herein can be combined with each other unless otherwise specified.

[0022] This disclosure provides a resource acquisition method applied to vehicles, such as... Figure 1 As shown, it includes: S101. Identify the target edge data center; S102. Obtain target resources by using the target resource acquisition path determined based on the target edge data center; the target edge data center is located at the service switching point and has pre-acquired the resources required by the vehicle; the service switching point is a geographical location area associated with the change of vehicle operation tasks.

[0023] In this embodiment of the disclosure, during the intelligent operation process in the mining area, the mining fleet typically serves as the core operational vehicle, and typically includes manned vehicles and unmanned vehicles. These vehicles need to frequently retrieve various key resources throughout the entire operation process, such as map tiles for vehicle positioning and route planning, AI models to support intelligent vehicle decision-making, blacklists and whitelists to ensure operational safety and access control, and task orders to guide vehicle operation scheduling. The timely and efficient acquisition of these resources directly determines the operational efficiency and intelligent operation level of the mining fleet. However, the current acquisition of these critical resources by mining fleets primarily relies on a cloud distribution model. This involves vehicle-mounted terminals directly establishing communication links with the cloud control platform to retrieve resources. However, this model suffers from significant technical flaws in the complex operating environment and communication conditions of mining areas, making it difficult to meet the actual resource acquisition needs of mining fleets. Firstly, the numerous deep pits, slopes, and other terrain obstacles within mining areas can easily obstruct wireless communication links, significantly reducing the communication stability between vehicle-mounted terminals and the cloud control platform, leading to delayed initial data transmission responses or even transmission interruptions. Secondly, the cellular communication network in the mining area is prone to congestion due to multiple vehicles simultaneously initiating communication requests. Furthermore, the high cost of cellular network communication further increases the overall cost of resource distribution and adds to data transmission latency. Thirdly, under the cloud distribution model, when different vehicle-mounted terminals retrieve the same critical resources, a large number of duplicate source requests are generated. This not only excessively consumes the computing and communication resources of the cloud control platform but also exacerbates network congestion in the mining area, significantly reducing overall resource distribution efficiency.

[0024] To address the aforementioned issues with cloud distribution models, existing technologies have proposed caching and collaborative streaming technologies based on V2X (vehicle-to-the-world communication) / RSU (roadside unit). These technologies utilize roadside caching nodes to achieve edge caching and localized distribution of resources, effectively improving vehicle data acquisition efficiency in highway scenarios. However, these technologies are designed for highway scenarios, where vehicles move at high speeds on open roads without fixed work areas. They fail to consider the specific constraints of mining operations and cannot directly adapt to the operational needs of mining fleets. Mining fleets exhibit significant characteristics of parking and queuing at loading points, creating a highly constrained operational scenario with dense parking and waiting. Existing V2X / RSU caching and collaborative streaming technologies are not optimized for this scenario. They fail to consider the high vehicle density and concentrated resource requests at loading points, and do not incorporate the operational scheduling patterns of mining fleets to achieve targeted caching and collaborative streaming at loading points. This results in low cache hit rates and poor resource distribution efficiency in mining loading point parking / queuing scenarios, failing to effectively address the technical shortcomings of cloud distribution models in mining areas.

[0025] In this embodiment, the target edge data center pre-caches resources, eliminating the need for vehicles to retrieve them from the origin cloud control platform. The resources cached in the target edge data center require periodic synchronization with the cloud control platform to ensure timeliness and consistency. The deployment location of the target edge data center is associated with service switching points. These service switching points are geographical areas corresponding to changes in vehicle operation tasks, including loading and unloading points in the mining area, boundaries of the work area, and fleet dispatch transfer points. These areas are key scenarios where mining fleets stop and queue, providing a time window for resource acquisition. The carrier of the target edge data center can be an excavator. Utilizing the fixed nature of excavators in the work area (e.g., long-term operation at loading points) or their mobility (e.g., cross-regional mining), the excavator carrier eliminates the need for additional fixed edge nodes, reusing existing mining equipment and significantly reducing edge deployment costs. Furthermore, pre-caching avoids repeated origin requests from multiple vehicles, reducing cloud control platform resource consumption and cellular network costs. The mobility of excavators also enables dynamic allocation of resource caches, improving overall resource distribution efficiency. In determining the target edge data center, the vehicle obtains its real-time location through its own positioning module (BeiDou, GPS, etc.). This location is then combined with a pre-stored map of mining area business switching points, excavator operation trajectory planning, and an edge data center deployment association table to match the current or upcoming business switching point and pinpoint the corresponding edge data center on the excavator. If multiple excavators with edge data centers exist at the same business switching point, link quality checks (e.g., line-of-sight verification, cache resource integrity verification), excavator operation status (e.g., whether it is in a stable operating state), and edge data center load assessment are used to select the optimal target edge data center. Furthermore, the vehicle and the target edge data center must complete two-way authentication to ensure data transmission security and prevent information leakage or tampering in the complex mining environment.

[0026] Furthermore, the target resource acquisition path is determined by prioritizing the target edge data center as the core resource acquisition path and using nearby vehicles and cloud control platforms as alternative resource acquisition paths. Following the principle of prioritizing core resources and providing backup options, and considering the characteristics of the target edge data center, coupled with the collaborative capabilities of nearby vehicles and the supplementary capabilities of the cloud control platform, efficient and stable resource retrieval is achieved. Based on the target edge data center, a direct line-of-sight (LoS) link between the vehicle and the target edge data center is prioritized. For example, leveraging the relative fixed position and height advantage of excavators during operation, the impact of terrain obstructions such as deep pits and slopes on communication is avoided. Furthermore, based on the communication coverage and signal strength distribution of the target edge data center, the shortest transmission path is optimized to reduce link loss and latency. Prioritizing short-range communication methods such as V2X reduces reliance on cellular networks, further controlling costs and transmission time. Alternative resource acquisition paths and their activation scenarios are clearly defined, forming a backup guarantee mechanism. Using nearby vehicles as alternative resource acquisition paths, vehicles use V2X communication to perceive nearby vehicles that have cached the target resource in real time. When the target edge data center is used as the core resource acquisition path, if link obstruction, congestion, or excessive load occurs at the target edge data center, the system automatically switches to the alternative resource acquisition path of nearby vehicles. If neither the target edge data center nor the alternative resource acquisition path of nearby vehicles can be used normally (e.g., when there are no nearby vehicles caching resources or the target edge data center is faulty), the system will use the optimal cloud link planned by the cloud control platform to pull resources. The cloud control platform can simultaneously optimize link parameters, avoid known congested areas, and minimize the drawbacks of the cloud distribution mode.

[0027] In this embodiment, the target edge data center pre-caches resources, eliminating the need for vehicles to retrieve them from the cloud. This avoids issues such as slow initial packets and transmission interruptions caused by deep pits, slope obstructions, and cellular congestion. Furthermore, the excavator can move with the work area, keeping the target edge data center close to the vehicle's work trajectory, further shortening the transmission link and reducing the impact of terrain on communication. Combined with pre-planned paths, this significantly improves the stability of resource acquisition. Service switching points are high-frequency areas where mining convoys park and queue. Excavators, as core operational equipment in mining areas, often stay at or travel to these locations. By mounting the target edge data center on the excavator, placing it at the service switching point, precise and dynamic coverage of the service switching point can be achieved. Vehicles can obtain resources nearby while waiting for work, overcoming the shortcomings of existing technologies in adapting to the special scenarios of mining areas.

[0028] In another embodiment of this disclosure, step S101 above, determining the target edge data center, includes: Step 1: Assess the real-time status of the edge data center and determine the target edge data center based on the assessment results; Preferably, the real-time status of the edge data center is evaluated, and the target edge data center is determined based on the evaluation results, including: Step 2: Determine the location of the vehicle within each edge data center of its service radius; Step 3: Identify the edge data centers that meet the following conditions as target edge data centers: the signal strength of the communication channel between the vehicle and the edge data center is higher than the first threshold, the network congestion of the communication channel is lower than the second threshold, and the communication channel meets the line-of-sight condition.

[0029] In this embodiment, an edge data center whose service radius covers the current location of the vehicle is determined to narrow down the candidate range. Then, verification is performed using signal strength, network congestion, and line-of-sight conditions to ultimately determine the target edge data center that meets the needs of the complex communication environment in the mining area, ensuring the stability and efficiency of the subsequent resource acquisition link. Regarding step 1 above, the real-time status of the edge data center is evaluated. This evaluation needs to cover multiple key dimensions, adapting to the mining operation scenario and the characteristics of the excavator. The evaluation content includes, but is not limited to: communication link status (signal strength, transmission rate, link stability), network load status (number of currently connected vehicles, data transmission throughput, network congestion level), resource caching status (whether the resources required by the vehicle are cached, resource version integrity, remaining cache space), and carrier operation status (if it is an excavator, it needs to be evaluated whether the excavator is in a stable operating state and whether it will leave the current area in the short term). During the evaluation process, the vehicle-mounted terminal collects the status data of each edge data center in real time through V2X communication, cloud control platform synchronization, etc., and performs comprehensive scoring or threshold judgment on each edge data center. Ultimately, the edge data center with the best status and the strongest adaptability is selected as the target edge data center. If there are multiple nodes that meet the status, a second screening can be conducted by combining factors such as distance and load balancing to ensure that the evaluation results meet the real-time operation needs of the vehicle.

[0030] Regarding step 2 above, each edge data center (especially the mobile node mounted on the excavator) needs to be pre-configured with a service radius parameter. This parameter can be dynamically adjusted based on communication frequency band, antenna power, and mining terrain characteristics (e.g., expanding the service radius in open areas and reducing it in deep pits or densely hilly areas), and synchronized in real time to the vehicle terminal and cloud control platform. The vehicle obtains its precise real-time location through its own positioning module (BeiDou, GPS, etc.), and combines this with the pre-stored location information and service radius parameters of each edge data center. Through spatial distance calculation, all edge data centers whose service radii can cover the vehicle's current location are selected. This allows for the rapid elimination of nodes that are outside the coverage area and cannot achieve effective communication, avoiding the need for complex indicator evaluations of invalid edge data centers, significantly reducing the computational burden on the vehicle terminal, improving the efficiency of determining target edge data centers, and laying the foundation for subsequent accurate verification.

[0031] Regarding step 3 above, the vehicle-mounted terminal monitors the signal strength (e.g., RSRP value) of the communication channel between the vehicle and the candidate edge data centers in real time, retaining only edge data centers with signal strength higher than the first threshold. The first threshold can be dynamically configured according to the mining area scenario, adapting to the impact of terrain obstruction (e.g., appropriately lowering the threshold in sloping areas and increasing the threshold in open loading points) to ensure sufficient signal strength to support stable resource transmission and avoid transmission interruptions or excessively high error rates due to weak signals. By collecting network congestion data (e.g., link utilization, data transmission latency, queue length) of the candidate edge data centers in real time, it is determined whether their values ​​are below a preset second threshold. The second threshold can be dynamically adjusted in conjunction with scenarios such as peak mining operations and concurrent requests from multiple vehicles to ensure that the selected edge data centers have low network load, can quickly respond to vehicle resource requests, avoid increased transmission latency due to congestion, and meet the resource acquisition needs of concentrated mining fleet operations. It is verified whether the communication channel between the vehicle and the candidate edge data centers meets the line-of-sight (LoS) condition, i.e., there are no obstacles obstructing the geographical location of the vehicle and the edge data center. For example, a straight spatial link is drawn from the communication antenna of the unmanned vehicle to the communication base station antenna of the edge data center. In the three-dimensional coordinate system of the high-precision map, there are no obstacles obstructing the straight link. For edge data centers mounted on excavators, the height advantage of the excavators can be used to increase the probability of forming a direct line of sight link and avoid the obstruction effect of complex terrain in the mining area. Edge data centers that meet the above conditions are identified as target edge data centers.

[0032] like Figure 2As shown, unmanned vehicles A203 and C205 are located within the service radius of edge data center A202. Unmanned vehicle A203 evaluates the real-time status of edge data center A202; if the above conditions are met, edge data center A202 is identified as the target edge data center. Since there is an obstruction 207 between unmanned vehicle C205 and edge data center A202, the above conditions are not met, and unmanned vehicle C205 cannot obtain the target resource through edge data center A202. Furthermore, there are no neighboring vehicles around unmanned vehicle C205; therefore, unmanned vehicle C205 obtains the target resource through cloud control platform 201. Unmanned vehicle B204 is not located within the service radius of edge data center A202, but it has a neighboring vehicle A203; therefore, unmanned vehicle B204 can obtain the target resource from neighboring vehicle A203. Unmanned vehicle D206 is not located within the service radius of edge data center A202, and there are no neighboring vehicles around unmanned vehicle D206; therefore, unmanned vehicle D206 obtains the target resource through cloud control platform 201.

[0033] The edge data center onboard the excavator moves with the operation, and real-time status assessment and service radius selection can dynamically adapt to changes in the edge data center's location, ensuring that the vehicle always locks onto the optimal edge data center within its current coverage area. This compensates for the lack of adaptability of fixed edge nodes to dynamic mining operations. Verification conditions for signal strength, low congestion, and line-of-sight ensure low link loss, low transmission latency, and strong anti-interference capabilities between the target edge data center and the vehicle, providing reliable support for subsequent acquisition of core path resources and further optimizing the initial packet response speed.

[0034] In another embodiment of this disclosure, the target resource acquisition path is determined based on the target edge data center in the following manner: Step 1: Determine the target resource acquisition path based on the weights of the resource acquisition paths; the weights of the resource acquisition paths include: the weight corresponding to the target edge data center, the weight corresponding to nearby vehicles, and the weight corresponding to the cloud control platform. Preferably, the target resource acquisition path is determined based on the weight of the resource acquisition path, including: Step 2: Determine the proportional relationship between the weights corresponding to the target edge data center, the weights corresponding to nearby vehicles, and the weights corresponding to the cloud control platform; Step 3: Based on the proportional relationship, divide the multiple target resources into subsets of quantities related to the proportional relationship, and determine the target resource acquisition path corresponding to each subset.

[0035] In this embodiment, the target resource is divided into a corresponding number of subsets based on the proportional relationship of the weights of the target edge data center, nearby vehicles, and the cloud control platform. Each subset is matched with a corresponding target resource acquisition path, achieving multi-path collaborative resource distribution and solving the problems of insufficient adaptability and low transmission efficiency of a single path. Nearby vehicles refer to mining operation vehicles located in the same geographical area as the vehicle currently performing resource acquisition operations (such as loading points, unloading points, or other business switching points in the mining area), and within a preset communication range (usually the V2X communication coverage area). These vehicles have cached some or all of the target resources, or have the ability to establish stable communication links with the target edge data center and cloud control platform. They can achieve resource forwarding or sharing through V2X, serving as an important carrier for nearby supplementary transmission in mining scenarios. A cloud control platform can refer to a mining area global operation control and resource management platform deployed in the cloud. It is the core storage and distribution source of mining operation resources, and can store all key resources such as map tiles, AI models, blacklists and whitelists, and task orders. It also has functions such as global communication scheduling, network status monitoring, and operation task management. It can issue scheduling instructions and resource update packages to mining vehicles and edge data centers, and is the ultimate guarantee for resource acquisition.

[0036] For step one above, the weights of the target edge data center, nearby vehicles, and cloud control platform need to be quantitatively set based on the actual communication environment, resource status, and transmission requirements of the mining area. The weight values ​​are positively correlated with the transmission advantages of the path. The weight of the target edge data center is mainly set based on its link quality with the vehicle (signal strength, line-of-sight, congestion level), resource cache integrity, and load status. The better the link quality, the more complete the cached resources, and the lower the load, the higher the weight value. The weight of nearby vehicles is mainly set based on their communication distance with the current vehicle, link stability, the proportion of cached target resources, and vehicle operation status. The closer the distance, the more stable the link, and the more cached resources, the higher the weight value. The weight of the cloud control platform is mainly set based on the overall congestion level of the mining area's cellular network, the load status of the cloud control platform, and the cost of resource transmission. The lower the network congestion, the lower the platform load, and the lower the cost, the higher the weight value. Furthermore, in typical mining scenarios, the weight of the target edge data center > the weight of nearby vehicles > the weight of the cloud control platform. Determining the target resource acquisition path based on the weight of the resource acquisition path can include: selecting the path corresponding to the entity with the highest weight value as the target resource acquisition path, suitable for scenarios with small target resource data volumes and extremely high requirements for transmission timeliness; and also including: allocating resource transmission ratios according to the weight values ​​to form a multi-path collaborative target resource acquisition path system, suitable for typical mining scenarios with large target resource data volumes and concentrated requests from multiple vehicles.

[0037] For step two above, the proportional relationships between the weights corresponding to the target edge data center, nearby vehicles, and the cloud control platform are determined. For example, the weight values ​​are normalized and converted into a unitless proportional relationship. If the weight of the target edge data center is 8, the weight of nearby vehicles is 1, and the weight of the cloud control platform is 1, then the proportional relationship among the three is 8:1:1. The proportional relationship is updated synchronously with the real-time adjustment of the weight values. When the communication environment in the mining area changes (such as the target edge data center link being blocked, causing the weight value to decrease), the weights are recalculated and the proportions are converted to ensure that the proportional relationship always matches the actual transmission requirements.

[0038] Regarding step three above, based on the proportional relationship, the overall target resource is divided into a corresponding number of subsets, with the number and weight ratio consistent. For example, if the ratio is 8:1:1, the target resource is divided into subsets with proportions of 80%, 10%, and 10%, respectively. Based on the proportional relationship and the subset division results, a corresponding target resource acquisition path is determined for each resource subset. For example, 80% of the subset matches the target edge data center path, 10% of the subset matches nearby vehicles, and 10% of the subset matches the cloud control platform path, achieving a precise match between path advantages and resource transmission needs.

[0039] The target resource acquisition path is determined based on weights. Factors such as the link characteristics, resource storage status, and communication costs of the target edge data center, nearby vehicles, and cloud control platform are quantified into weights. This avoids subjective path selection and makes the target resource acquisition path more suitable for the dynamic communication environment of the mining area, such as deep pits / slopes and network congestion, thus improving the adaptability of the target resource acquisition path. Resources are divided into subsets according to weight ratios and matched with corresponding target resource acquisition paths. Paths with a high weight ratio are assigned to carry more resource transmission tasks, while paths with a low weight ratio are assigned to supplementary transmission. This fully leverages the transmission advantages of each path (such as low latency of the edge data center and comprehensive resources of the cloud control platform), reduces invalid transmissions, and lowers cellular network costs and cloud control platform resource consumption.

[0040] In another embodiment of this disclosure, if acquiring the target resource through the current target resource acquisition path fails or times out, the method further includes: Step (1): Switch to a new target resource acquisition path according to the preset resource acquisition path priority, and use it as the new current target resource acquisition path; Step (2): Obtain the target resource by switching to the new target resource acquisition path corresponding to the current path priority, and monitor the target resource acquisition status of the new target resource acquisition path corresponding to the current path priority until the target resource is downloaded, or, if the target resource acquisition fails or times out, switch to the new target resource acquisition path corresponding to the next path priority. The resource acquisition paths, from highest to lowest priority, are: target edge data center, nearby vehicles, and cloud control platform.

[0041] Preferred options also include: Step (3): If all attempts to obtain the target resource according to the priority of each preset resource acquisition path fail, cancel the resource requested and cached based on the resource request information, and maintain the original resource.

[0042] In this embodiment, if the current target resource acquisition path fails to acquire resources or times out, the path is switched and retried in order of resource acquisition path priority (target edge data center > nearby vehicles > cloud control platform), while the target resource acquisition status of the new current target resource acquisition path is monitored until successful acquisition. To avoid resource chaos and invalid occupation, an abnormal rollback step is preferably added. If the acquisition of target resources according to each preset resource acquisition path priority fails, the requested cached resources are canceled and the original resources are maintained to ensure the stability of vehicle operation resources and completely solve the problem of resource acquisition interruption caused by single path failure.

[0043] For step (1) above, the resource acquisition path priority from high to low is "target edge data center > nearby vehicles > cloud control platform". This priority is set based on the characteristics of the mining area scenario. The target edge data center caches resources nearby, has the lowest transmission latency and the best stability, and is used as the core path; nearby vehicles coordinate closely through V2X without relying on cellular networks, and are used as the first backup path; the cloud control platform is the source of resources and has the most comprehensive resources, and is used as the final fallback path. If the current target resource acquisition path "fails" to acquire resources (such as link interruption, incomplete resource transmission, or identity authentication failure) or "times out" (transmission time exceeds a preset threshold), the switching process is initiated. Switching is performed step by step according to priority order, without skipping levels, such as... Figure 3 As shown, if the current target resource acquisition path is 301 in the target edge data, after failure / timeout, it will first switch to the nearby vehicle 302; if the nearby vehicle 302 is used as the new current target resource acquisition path, and the acquisition of the target resource also fails / timeouts, it will then switch to the cloud control platform 201, ensuring that a better backup path is selected in each round of switching to avoid resource waste. Furthermore, the target resource acquisition path identifier of the vehicle terminal is updated synchronously during switching, and the switching reason (such as obstruction, nearby vehicle offline) is recorded, providing data support for subsequent target resource acquisition path optimization.

[0044] Regarding step (2) above, after switching to the new current target resource acquisition path, the vehicle terminal sends a resource request to the corresponding entity (nearby vehicle 302 / cloud control platform 201) and starts a real-time monitoring mechanism. The monitoring dimensions include link signal strength, transmission rate, resource packet integrity (such as CRC checksum matching degree), transmission time, etc., and tracks the resource acquisition progress throughout the process. If the target resource transmission is monitored to be normal and all target resources are downloaded and integrity verified within the preset time, the switching process is terminated and the target resource download is completed. If the new current target resource acquisition path is monitored to fail to acquire the target resource (such as a nearby vehicle suddenly leaving the communication range, or the cloud control platform link congestion causing transmission interruption) or timeout (transmission time exceeds the preset threshold of the corresponding path), the next round of switching is triggered, and the system jumps to the next priority path, repeating the acquisition and monitoring operations of this step.

[0045] Regarding step (3) above, if the attempts to obtain the target resource through the target edge data center 301, the nearby vehicle 302, and the cloud control platform 201 all fail / time out (i.e., after switching and retrying all priority paths, the target resource is still not successfully obtained), the vehicle terminal performs resource rollback, automatically cleans up the cached data generated by this resource request, including incomplete resource packages, unverified resource fragments, temporary cache files, etc., and releases the corresponding cache space to avoid invalid resources occupying terminal storage and computing resources; retains the historical valid resources already stored in the terminal (such as old map tiles, previous task orders) to ensure that the vehicle can still perform basic operations based on the original resources (such as returning along the original route, temporary parking) after the resource acquisition fails, thus avoiding operation interruption.

[0046] The resource acquisition path priority, from highest to lowest, is "target edge data center > nearby vehicles > cloud control platform." Prioritizing shortest, lowest-cost, and lowest-latency paths is always the first priority. Backup paths are only activated when the core path fails, minimizing cellular network costs and cloud-to-origin latency, thus balancing resource acquisition stability and cost-effectiveness. With the target edge data center as the core path and nearby vehicles and the cloud control platform as backup paths, a safety net system is formed. This effectively addresses unforeseen issues such as edge link interruptions caused by deep pits / slopes in the mining area and collaborative link failures due to the movement of nearby vehicles, preventing operational stoppages caused by single-path failures and ensuring continuous resource acquisition.

[0047] In another embodiment of this disclosure, if the acquisition of the target resource fails or times out when the target edge data center is used as the target resource acquisition path, the method further includes: Step (1): Continuously acquire target resources from the target edge data center; Step (II): If the acquisition of target resources from the target edge data center fails or times out, the weight of the resource acquisition path of the target edge data center will be reduced according to the first rule. Step (3): In the process of continuously acquiring target resources from the target edge data center based on the reduced weight, if there is a successful acquisition status when acquiring target resources from the target edge data center, the weight of the target edge data center will be increased according to the second rule.

[0048] In this embodiment, if resource acquisition through the target edge data center fails or times out, the system switches to a backup path and continuously retryes acquiring the target resource from the target edge data center to cope with temporary link fluctuations in the mining area. If the retry still fails or times out, its weight is reduced according to the first rule, weakening the priority of the path. If subsequent retry based on the reduced weight is successful, the weight is increased according to the second rule, restoring its priority. This approach maximizes the usability of the core path while flexibly adapting to changes in link status, balancing resource acquisition stability with the scientific nature of path selection. For step (i) above, a "periodic retry + number / duration constraint" mechanism is adopted to continuously acquire the target resource from the target edge data center, taking into account both fault tolerance and efficiency. The retry period can be dynamically adjusted (e.g., an initial period of 1 second, extended to 2 seconds after every 3 retries to avoid frequent retries consuming terminal resources); and a maximum number of retries (e.g., 5 times) or a maximum retry duration (e.g., 10 seconds) is set to prevent infinite retries from causing a complete halt in resource acquisition. To address the specific characteristics of the mining area, the link signal strength and line-of-sight conditions are monitored simultaneously during the retry process. If the signal gradually recovers and the line-of-sight link is re-established (e.g., the excavator adjusts its working position to avoid slope obstruction), the retry cycle can be shortened, accelerating resource acquisition. If the signal remains at 0 and the line-of-sight condition is completely lost, the retry can be terminated early, proceeding to the next weight adjustment stage. Regarding step (II) above, if the maximum number of retries or the longest duration are reached without successfully acquiring the target resource (including multiple failures and repeated timeouts), it is determined that there is a persistent fault in the target edge data center link (e.g., the excavator has been away for an extended period, or the link is permanently blocked), and the weight reduction process is initiated. The first rule adopts a tiered weight reduction method to avoid reducing the weight to 0 all at once, ensuring that the path remains an alternative. For example, with a base weight of 10, the weight is reduced by 2 for every 3 consecutive failed retries, down to a minimum of 5 (higher than the cloud control platform weight of 3, lower than the nearby vehicle weight of 6). The reduced weight can be significantly lower than the weight under normal conditions, for example, lower than the nearby vehicle weight. After the weight is reduced, the vehicle terminal can determine the target resource acquisition path based on the reduced weight in subsequent target resource requests. It also periodically retryes the target edge data center path to monitor whether its link status has recovered. Regarding step (iii) above, after the weight corresponding to the resource acquisition path of the target edge data center is reduced, if the vehicle terminal successfully acquires the target resource from the target edge data center through periodic retries or subsequent resource requests (resource integrity verification must be completed to ensure valid acquisition), it is determined that the target edge data center link has recovered and stabilized (e.g., the excavator returns to the work point, obstructions are removed), and the weight increase process is initiated. The second rule uses a step-by-step weight increase method to avoid a one-time revert to the baseline weight, preventing frequent weight increases and decreases when the link is unstable; and the weight increase magnitude can be linked to the continuity of successful acquisition, ensuring that the weight adjustment aligns with the actual stability of the link.If resources are successfully acquired at fixed service switching points such as loading and unloading points, the weighting can be increased appropriately, as these areas have higher link stability. If resources are successfully acquired during mobile operations, the weighting can be reduced, and the link stability can be continuously observed before gradually returning to the baseline weight. For example, if the acquisition status is maintained continuously for 8 seconds (no failures / timeouts), the weighting is increased significantly (by 3 points each time) if the signal strength is higher than the baseline threshold upon success; if the signal strength is lower than the baseline threshold but higher than the minimum threshold, the weighting is increased less (by 1 point each time), and the weighting stops after reaching the baseline weight. The edge data center on the excavator may experience temporary link interruptions due to operational movement. Continuous retries can wait for the excavator to return to a stable operating position and the link to reconnect. Dynamic weight adjustment can avoid over-reliance on the path during excavator movement and allow for a rapid return to the core path after stabilization, adapting to the characteristics of mobile edge nodes. By adjusting the weight in real time to reflect the link quality of the target edge data center, the weight is reduced when there is a failure and increased when there is a success, so that path selection no longer depends on fixed settings but adapts to dynamic link changes, avoiding the use of the core path with poor status as the first choice, and also allowing the core path to be reused in a timely manner after the link is restored.

[0049] This disclosure provides a resource acquisition method applied to an edge data center. The edge data center is located at a service switching point, which is a geographical location associated with changes in vehicle operation tasks. Figure 4 As shown, the method includes: S401, Obtain resource pre-fetching tasks from the cloud control platform; S402. Based on the preset multi-factor comprehensive priority model, determine the pre-fetch priority of each resource in the resource pre-fetching task; S403. Based on the prefetching priority of each resource, send the corresponding target resource acquisition request to the cloud control platform in order of priority; and receive the corresponding target resource cache locally so that the vehicle can acquire the target resource from the edge data center.

[0050] In this embodiment, the edge data center is deployed at a service switching point, which is a geographical area associated with changes in vehicle operation tasks. The edge data center is built on an excavator, which employs an industrial gateway (including multi-core CPU, memory, and NVMe) and is equipped with near-field wireless (e.g., Wi-Fi / PC5 / 802.11p) and wide-field origin (e.g., 5G) modules. It also has built-in prefetcher, cache management, near-field service, statistics, and health check functions. The vehicle's onboard terminal has a scheduler, directory management, V2X capabilities, and near-field connectivity. It can automatically switch resource acquisition strategies based on the service radius R and map signal strength. The cloud control platform is responsible for resource directory and popularity publishing, policy distribution, token and permission management, and log and visualization. The edge data center establishes a secure communication link with the cloud control platform through its built-in wide-field origin module (e.g., 5G). It completes identity authentication and permission verification based on a token issued by the cloud control platform (the token is short-term valid, bound to vehicle and mining area identifiers, and only allows access to near-field whitelisted resources). After successful verification, it receives resource prefetching tasks from the cloud control platform. Resource pre-fetching tasks can include information such as resource identifiers, estimated demand levels, and the scope of associated business switchover points. These are generated by the cloud control platform in conjunction with the entire mine's resource catalog, vehicle operation scheduling plans, and historical access data, ensuring that pre-fetched resources align with the task change requirements of vehicles at the corresponding business switchover points. The edge data center has a built-in health check function that synchronously verifies the stability of the cloud control platform's connection and the integrity of the tasks. If a connection anomaly is detected, a reconnection mechanism is triggered to ensure the reliability of the pre-fetching task acquisition. A pre-defined multi-factor comprehensive priority model determines the pre-fetching priority of each resource in the resource pre-fetching task. For example, the multi-factor comprehensive priority model incorporates core factors such as resource popularity, relevance to business switchover tasks, resource size, version iteration urgency, and validity period. Each factor is weighted according to the mine's operational needs (e.g., resources associated with high-frequency tasks at business switchover points have the highest weight). The edge data center calls the built-in pre-fetcher to parse the resource information in the pre-fetching task, combines it with the resource catalog issued by the cloud control platform (which contains the version number, priority, and validity period of each resource), extracts the corresponding factor data, and substitutes it into the model to calculate the priority score. Furthermore, by combining the service radius R with the signal distribution characteristics of the service switching point map, the priority is fine-tuned. For example, the priority of resources required in densely parked areas is increased, and the priority of lightweight resources adapted to weak signal areas is improved. Finally, the pre-fetching priority ranking result of each resource is determined, providing a basis for subsequent hierarchical pre-fetching.

[0051] The edge data center sends target resource acquisition requests to the cloud control platform sequentially from high to low priority via a wide-area backhaul link. Each request carries a token for secondary permission verification, and the entire link employs encrypted transmission to ensure data security. On the cloud control platform, resources are divided into granular shards. For identical shards arriving from multiple sources, the edge data center performs deduplication and integrity checks using a hash consistency algorithm, retaining only one valid shard stored on the local NVMe storage device, centrally managed by the cache management module. Version number, priority, and expiration information are synchronously associated during resource caching. When a new version of a resource is detected by the vehicle's onboard terminal, it does not immediately switch to it. Instead, it continuously monitors the coverage of "newly arrived shards." When the coverage reaches a preset threshold, such as 70%, it updates to the new version. Older versions of resources are retained until their expiration date to prevent operational jitter due to incomplete resources during vehicle cold starts. The edge data center has a built-in near-source service module. After caching, it updates the local resource catalog, synchronizes it to the vehicle-side catalog management module, and broadcasts resource availability notifications to parked vehicles at the service switching point via a near-domain wireless module (such as Wi-Fi / PC5 / 802.11p). This allows vehicles to obtain resources through near-source connections or V2X collaboration. Furthermore, the statistics and health check module records resource prefetch hit rates, first packet times, failure rates, and version rollback events, generating audit logs that are uploaded to the cloud control platform for subsequent priority model optimization and job verification. If an abnormal source request is detected, its priority is automatically reduced to strengthen security control.

[0052] In another embodiment of this disclosure, step S402 above, determining the prefetch priority of each resource in the resource prefetching task based on a preset multi-factor comprehensive priority model, includes: Step 1) For each type of resource, assign weights to each factor of that type of resource; Step 2) For each resource in the resource prefetching task, determine the priority score of that resource under each factor; Step 3) Based on the weights of each factor of the resource, perform a weighted calculation of the priority score under each factor to determine the prefetch priority of the resource.

[0053] In this embodiment, differentiated weights are configured for each factor according to resource type to adapt to the pre-fetching requirements of different types of resources; then, for each resource in the pre-fetching task, its priority score under each factor is calculated one by one; finally, the scores are weighted and summed in combination with the corresponding factor weights to obtain the comprehensive priority score of each resource, and the final pre-fetching priority is determined according to the score, so as to ensure that the pre-fetching resources are in line with the changes in mining area vehicle operation tasks and edge caching characteristics.

[0054] Regarding step 1), resources are categorized by function type, including map tile resources, AI model resources, task list resources, blacklist / whitelist resources, etc., to adapt to the core resource types required for mining vehicle operations. Secondly, for each resource category, based on its role in mining operations, edge prefetching requirements, and transmission and storage characteristics, corresponding weights are assigned to each factor in the multi-factor comprehensive priority model. The total weight is 1 (normalized), and the weight is positively correlated with the importance of the factor to that resource category. For example, for map tile resources, the map association factor weight is set to 0.4, the popularity priority factor weight to 0.3, the volume-friendly factor weight to 0.2, and the time-sensitive factor weight to 0.1. Regarding step 2), the edge data center calls the built-in prefetcher to parse each resource in the resource prefetching task one by one. Combining the resource catalog (including version number, priority, validity period, etc.) issued by the cloud control platform, historical access data, and geographical information of business switching points, the priority score of the resource under each factor in the model is calculated. Each factor score uses a standardized scoring system of 0-10 points. A higher score indicates a higher priority in that dimension. The scoring rules are set individually for each factor, tailored to the characteristics of the mining area scenario. For step 3) above, for each resource, first match the weights of each factor corresponding to its resource type, then multiply each factor score by its corresponding weight to obtain a weighted score for each factor. Finally, sum the weighted scores of all factors to calculate the comprehensive priority score for the resource (range 0-10). The higher the comprehensive priority score, the higher the resource prefetching priority. The edge data center sorts all resources in the prefetching task from highest to lowest comprehensive score, forming the final resource prefetching priority sequence. For example, a map tile resource (belonging to the map tile category, with weighted configurations: map association factor 0.4, popularity priority factor 0.3, volumetric friendliness factor 0.2, and time sensitivity factor 0.1) has scores of 10, 7, 6, and 4 for each factor, respectively. Its comprehensive priority score is calculated as follows: 10 × 0.4 + 7 × 0.3 + 6 × 0.2 + 4 × 0.1 = 4 + 2.1 + 1.2 + 0.4 = 7.7. If another task resource has a comprehensive score of 8.5, then the task resource has a higher prefetch priority than this map tile resource. By converting core requirements such as popularity, time sensitivity, and map association into quantifiable scores and weights, and calculating the comprehensive priority through weighted averages, subjective judgment is replaced, making the ranking results more aligned with the prefetching capabilities of edge data centers, vehicle parking needs, and the communication environment of mining areas, thus improving the scientific nature of prefetching.

[0055] In another embodiment of this disclosure, the multi-factor comprehensive priority model includes at least one of the following factors: heat priority factor, time-sensitive factor, map association factor, and volume-friendly factor; The popularity priority factor is used to characterize the global request popularity of resources on the cloud side and the historical request popularity of the corresponding content type of resources in the edge data center; The time-sensitive factor is used to characterize the correlation between resources and vehicle operation tasks, as well as the proximity of the current time to the relevant time nodes of the vehicle operation tasks; Map association factors are used to characterize the strength of the association between resources and map constraints; among which map constraints include at least one of the following: environmental awareness friendliness, path connectivity, and basic data freshness; The volume-friendly factor is used to characterize the fragment size of a resource.

[0056] In this embodiment, the resource prefetching priority is quantified by factors such as heat priority, time sensitivity, map association, and volume friendliness to generate a prefetching scoring and sorting queue; then the prefetcher distributes resources to hierarchical cache structures such as memory hotspot cache and NVMe main cache; finally, the cache is dynamically cleaned up by the LRU (Least Recently Used) + TTL (Time to Live) eviction policy to ensure that high-value resources are stored first and accessed efficiently.

[0057] like Figure 5 As shown, the popularity priority factor integrates the global request popularity (501) of resources on the cloud side (the frequency of access to the resource by vehicles throughout the entire mining area) with the historical request popularity (502) of the corresponding content type of the resource in the edge data center, such as the historical request popularity of map tile type resources in the edge data center. The higher the popularity, the higher the factor score. For example, map tiles that are frequently used at the loading point in the mining area have a significantly higher popularity priority factor score than resources in other areas.

[0058] Time-sensitive factors can be used to assess the correlation between resources and vehicle operation tasks (such as the binding relationship between task orders, temporary blacklists and whitelists and current operation tasks), as well as the proximity between the current time and the task time node (such as the resource score of task shift 503 is higher 1 hour before execution than the resource score of 24 hours in advance), to ensure that resources associated with urgent tasks are prioritized for pre-retrieval.

[0059] Map association factors can measure the strength of the association between resources and map constraints, including environmental perception friendliness (e.g., whether the resource contains perceptual features of terrain such as deep pits and slopes), path connectivity (e.g., whether the resource covers the connecting nodes of the vehicle operation path), and basic data freshness (e.g., the update time of map tiles). The higher the association, the higher the score, which is more suitable for the operation needs of complex terrain in mining areas.

[0060] The volume-friendly factor can be evaluated based on the fragment size (507) of the resource. The smaller the fragment size (507), the lower the transmission and storage costs, and the higher the score. For example, blacklist / whitelist fragments (<10MB) score higher than AI model fragments (>1GB). Prioritizing the prefetching of small-volume resources can reduce the transmission pressure on edge nodes. After calculating the multi-factor scores for each resource, a weighted sum is performed based on the weights corresponding to the resource type to obtain the comprehensive priority score for each resource. The resources are then sorted from highest to lowest score to generate a prefetching scoring and sorting queue (508).

[0061] The prefetcher pulls resources from the cloud control platform and distributes them to the corresponding tiered cache structure based on the prefetch sorting queue, achieving efficient tiered management of cache resources: The memory hotspot cache (509) stores the highest-priority, frequently accessed resources in the prefetch queue (such as real-time task orders at the current loading point and frequently accessed map tiles). Leveraging the high-speed read / write capabilities of memory, it achieves millisecond-level response times for vehicle resource requests, meeting the rapid resource acquisition needs when vehicles dock in the mining area. The NVMe main cache (510) stores resources with the next highest priority and moderate size (such as AI model shards and map tiles covering multiple work areas). Utilizing the high IO performance and large capacity of NVMe, it balances storage capacity and access speed to meet the storage requirements of concurrent requests from multiple vehicles. To ensure the effectiveness and storage efficiency of cached resources, a hybrid eviction strategy of LRU + TTL (511) is used for dynamic cache management. For LRU (Least Recently Used), for both the memory hotspot cache and the NVMe main cache, the least recently accessed resources are automatically cleaned up, ensuring that frequently accessed resources are always retained in the high-speed cache, adapting to the dynamic access characteristics of mining area vehicle operations. For TTL (Time to Live), combined with the resource's validity period (such as the execution end time of a task order and the update cycle of a map tile), expired resources are automatically cleaned up to avoid invalid resources occupying cache space and ensure the freshness of cached resources. The statistics reporting system 512 is used to count the number of hits, failure rates, and rollback events of vehicle resource requests, and reports them to the cloud control platform. This allows the cloud control platform, edge data center, and vehicle strategies to iterate synchronously, adapting to dynamic changes in mining operations (such as excavator movement and changes in vehicle parking density), and improving the overall system's resilience and adaptability.

[0062] In another embodiment of this disclosure, when the resource type is map data, the association strength between the resource and map constraints is determined in the following manner, including: Step 1) Determine the static layer data and dynamic event layer data of the high-precision map corresponding to the resource; Step 2) Perform local rasterization processing on the static layer data of the high-precision map to determine the line-of-sight probability and the reachability assessment score; whereby the line-of-sight probability is used to characterize the environmental perception friendliness, and the reachability assessment score is used to characterize the path connectivity. Step 3) Determine the map freshness corresponding to the resource based on the event impact area corresponding to the dynamic event layer data and the high-precision map static layer data; map freshness is used to characterize the freshness of basic data; Step 4) Use preset fusion rules to fuse the direct-view probability, reachability assessment score and map freshness to determine the correlation strength between resources and map constraints.

[0063] In this embodiment of the disclosure, the static layer data of the high-precision map is rasterized to obtain the direct line-of-sight probability and the reachability assessment score. The map freshness is calculated by combining the dynamic event layer data. Finally, the direct line-of-sight probability, the reachability assessment score and the map freshness are fused by a preset fusion rule to output the final association strength value, which provides accurate map association factor scores for the multi-factor priority model.

[0064] like Figure 6As shown, regarding step one above), the high-precision map static layer data 601 can refer to long-term stable geographic information data such as mine terrain, roads, and buildings, which is the foundation for vehicle environmental perception and path planning. The dynamic event layer data 602 can refer to real-time changing event information such as temporary construction in the mine, road blockage, and adjustment of work areas, used to update the dynamic attributes of the map data. The high-precision map static layer data 601 corresponding to the current business switching point (such as the loading point) is extracted from the map resources cached in the edge data center, including information such as terrain elevation, road topology, and obstacle distribution. The dynamic event layer data 602 of this area is obtained from the cloud control platform, including the latest dynamic information such as temporary construction areas, road blockage events, and adjustment of work areas. This determines the high-precision map static layer data 601 and dynamic event layer data 602 corresponding to the resource. Regarding step two above), rasterization processing 603 involves dividing the continuous map space into regular raster units (such as 1m×1m). By calculating the attributes of each raster, raster data is obtained, enabling a refined evaluation of the map data. The line-of-sight probability (LOS) of 604 represents the probability that there is no terrain obstruction on the communication link between a vehicle and the edge data center within the raster data, reflecting the friendliness of environmental perception in that area. The reachability assessment score of 605 evaluates the path connectivity score from a given raster to the edge data center; a higher score indicates better path accessibility. The high-precision map static layer data 601 is rasterized 603 into multiple raster units of various scales, each raster associated with terrain elevation, road attributes, and other information. For each raster unit, the LOS probability 604 of the communication link is calculated based on the vehicle antenna height, the edge data center antenna height, and the raster terrain elevation. For example, if there is slope obstruction within the raster, the LOS probability 604 represents the proportion of the raster with no obstruction; a higher probability indicates stronger environmental perception friendliness. Based on the road connectivity and obstacle distribution of the raster, the path reachability from the raster to the edge data center is evaluated, resulting in the reachability assessment score 605. For example, a standardized score of 0-10 is used, with a higher score indicating better path connectivity. Regarding step three above, map freshness score 606 measures the degree of matching between map data and the current actual scene. It is calculated by fusing the event update time and impact range of dynamic event layer data 602 with static layer data. The impact area 607, occurrence time, and duration of each event in dynamic event layer data 602 are extracted and spatially matched with the raster cells of high-precision map static layer data 601. If a raster is within the impact range of a dynamic event, and the more recent the event occurrence time, the higher the map freshness score 606. For example, if a temporary construction area is added to the loading point, the map freshness score 606 of the raster in that area will increase as the construction event is updated.Regarding step four above, the preset fusion rule 608 is a rule for weighted or algorithmically fusion of the direct-view probability 604, reachability assessment score 605, and map freshness 606. The weights of each indicator are typically configured according to the operational needs of the mining area (e.g., when the weight of mining area environmental perception is higher, the weight of direct-view probability 604 is larger). The direct-view probability 604 (0-100%), reachability assessment score 605 (0-10 points), and map freshness 606 (0-10 points) are normalized to a unified score of 0-10. The preset fusion rule 608 is configured according to the mining area operational scenario, for example, setting weights such as direct-view probability 604 (0.4), reachability assessment score 605 (0.3), and map freshness 606 (0.3). The weighted summation yields the association strength value (0-10 points). A higher value indicates a stronger association between the map resource and map constraints 609, thus providing accurate map association factor scores for the multi-factor priority model.

[0065] Through rasterization and direct-view probability calculation, obstructed areas such as deep pits and slopes in mining areas can be accurately identified, improving the reliability of vehicle environmental perception. The reachability assessment score directly supports connectivity verification of mining operation paths, preventing vehicles from planning invalid routes. Dynamic event layer data combined with static layer calculations of map freshness can reflect real-time dynamic changes such as temporary construction and road blockages in the mining area, ensuring the freshness of map data and adapting to the dynamic nature of mining operation scenarios. Map constraints are transformed into quantifiable probabilities, scores, and freshness values, avoiding subjective judgment bias and making the scores of map-related factors more accurate, thereby improving the reliability of the multi-factor priority model. Map data with higher correlation strength is pre-fetched and cached first, ensuring that vehicles can obtain highly adaptable map resources at business switching points (such as loading points), improving operational efficiency and safety.

[0066] In another embodiment of this disclosure, the method further includes: Step 01: Based on the target resource's lifetime and preset expiration date in the local cache, perform a resource eviction operation on the target resource; and / or, Step 02: Based on the hit rate of the target resource in the local cache within the statistical period and the preset hit rate threshold, perform resource eviction operations on the target resource.

[0067] In this embodiment, the Time-to-Live (TTL) and preset validity period rules automatically clean up expired and invalid target resources; low-value and low-access target resources are automatically cleaned up through a popularity rule based on the hit rate of a statistical period and a preset hit rate threshold. Regarding step 01 above, TTL can refer to the length of time a target resource is cached in the edge data center, starting from the moment the target resource is cached. The preset validity period can be the validity period set by the cloud control platform according to the resource type (e.g., 2 hours for a task order, 7 days for a map tile), with different validity periods configured for different types of resources. The cache management module of the edge data center periodically scans all target resources in the local cache and compares their TTL with the preset validity period one by one. When the TTL of a target resource is greater than or equal to the preset validity period, a resource eviction operation is triggered, directly deleting the cache file of the target resource and releasing the corresponding storage space. Regarding step 02 above, the statistical period can be a fixed time window (e.g., 1 hour, 12 hours) used to count the access frequency of target resources. The hit rate can refer to the proportion of the number of times a vehicle successfully retrieves the target resource directly from the local cache within the statistical period, relative to the total number of requests. The preset hit rate threshold can be set by the cloud control platform as the minimum acceptable hit rate (e.g., 10%). Target resources with a hit rate lower than this preset threshold are judged as low-value resources. The statistical reporting module of the edge data center calculates the hit rate of all cached target resources at the end of each statistical period. The cache management module compares the hit rate of each resource with the preset hit rate threshold. If the hit rate is less than the preset hit rate threshold, a resource eviction operation is triggered, and the cache of that target resource is deleted. The time-driven eviction mechanism can automatically clean up expired task orders, temporary blacklists and whitelists, and other time-sensitive resources to avoid operational errors caused by vehicles obtaining invalid information, adapting to the needs of dynamic adjustment of mining tasks. The popularity-driven eviction mechanism cleans up low-hit-rate resources, and the freed-up space can be used to store high-frequency hot resources (such as loading point map tiles, AI models), reducing invalid cache occupation and improving the utilization efficiency of excavator storage resources. High-value resources are retained in the cache, reducing the frequency of vehicles returning to the source cloud control platform, reducing cellular network costs and transmission latency, and avoiding retrieval time caused by cache redundancy, further improving resource acquisition speed.

[0068] In another embodiment of this disclosure, the type of resource includes at least one of the following: map data, software packages, access control policies, and job tasks; and / or, Resources are split into equally granular shards, and a directory is set for each shard. The directory includes: version identifier, preset priority, and preset validity period.

[0069] In this embodiment, resource type refers to the functional classification of resources required for mining operations, including map data (such as map tiles), software packages (such as AI models and system upgrade packages), access control policies (such as blacklists / whitelists and permission rules), and work tasks (such as loading instructions and task orders). Map data, such as map tiles, including high-precision map static layer and dynamic event layer data, is the core foundational resource for vehicle environmental perception and path planning. Software packages include AI model packages and vehicle system upgrade packages, supporting vehicle intelligent decision-making and system iteration. Access control policies include blacklists / whitelists and permission verification rules, ensuring the safety and compliance of mining operations. Work tasks include loading instructions and task orders, which are the direct basis for vehicles to execute operations. Equal-granularity fragmentation refers to splitting a complete resource into equal-volume fragments of a fixed size (such as 64MB / fragment), enabling each fragment to have independent transmission, caching, and verification capabilities. The cloud control platform can split resources into equal-granularity fragments. The fragment directory is a set of metadata bound to each fragment, including: version identifier, preset priority, and preset validity period. Version identifiers are used to identify the resource version to which a fragment belongs, distinguishing between old and new versions and avoiding confusion.

[0070] Preset priority refers to the priority set by the cloud control platform for resources. Based on this preset priority, the edge data center can determine the prefetch priority of each resource in the resource prefetching task based on a preset multi-factor comprehensive priority model, thus determining the order of prefetching and caching. For example, if the preset priority of resource A and resource B is level one, and the preset priority of resource C is level two, the edge data center will then determine the prefetch priority of resources A and B within level one. If the prefetch priority of resource B is greater than that of resource A, the prefetching order will be: resource B, resource A, resource C. The preset validity period is the effective storage time of the shard in the edge data center. After expiration, automatic eviction will be triggered to ensure cache freshness.

[0071] Based on the same disclosed concept, this disclosure also provides an unmanned vehicle. Since the principle by which these unmanned vehicles solve the problem is similar to the aforementioned resource acquisition method, the implementation of this unmanned vehicle can refer to the implementation of the aforementioned method, and the repeated parts will not be described again.

[0072] This disclosure provides an unmanned vehicle that performs the steps of the resource acquisition method as described in any of the above embodiments.

[0073] Based on the same disclosed concept, this disclosure also provides a resource acquisition system. Since the principle of solving the problem by these systems is similar to that of the aforementioned resource acquisition methods, the implementation of this system can refer to the implementation of the aforementioned methods, and the repeated parts will not be described again.

[0074] This disclosure provides a resource acquisition system, including: an edge data center and a vehicle; wherein, The edge data center is located at the service switching point, which is a geographical area associated with changes in vehicle operation tasks; The vehicle is used to perform the steps of the resource acquisition method as described in any of the above embodiments.

[0075] Through the above description of the embodiments, those skilled in the art can clearly understand that the embodiments of this disclosure can be implemented in hardware or by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions of the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this disclosure.

[0076] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes in the drawings are not necessarily essential for implementing this disclosure.

[0077] Those skilled in the art will understand that the modules in the apparatus of the embodiments can be distributed in the apparatus of the embodiments as described in the embodiments, or they can be located in one or more devices different from this embodiment with corresponding changes. The modules of the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0078] The sequence numbers of the embodiments disclosed above are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0079] Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from its spirit and scope. Therefore, if such modifications and variations fall within the scope of the claims of this disclosure and their equivalents, this disclosure is also intended to include such modifications and variations.

Claims

1. A resource acquisition method, applied to vehicles, characterized in that, include: Identify the target edge data center; The target resources are acquired by using the target resource acquisition path determined based on the target edge data center. The target edge data center is located at the service switching point and has pre-acquired the resources required by the vehicle; the service switching point is a geographical area associated with changes in vehicle operation tasks.

2. The method as described in claim 1, characterized in that, The determination of the target edge data center includes: Assess the real-time status of edge data centers and determine the target edge data centers based on the assessment results; Preferably, the real-time status of the edge data center is evaluated, and the target edge data center is determined based on the evaluation results, including: The location of the vehicle is determined to be within each edge data center of its service radius; Edge data centers that meet the following conditions are identified as target edge data centers: the signal strength of the communication channel between the vehicle and the edge data center is higher than a first threshold, the network congestion of the communication channel is lower than a second threshold, and the communication channel meets the line-of-sight condition.

3. The method as described in claim 1, characterized in that, The target resource acquisition path is determined based on the target edge data center using the following method: The target resource acquisition path is determined based on the weights of the resource acquisition paths; wherein the weights of the resource acquisition paths include: the weight corresponding to the target edge data center, the weight corresponding to the nearby vehicles, and the weight corresponding to the cloud control platform; Preferably, determining the target resource acquisition path based on the weight of the resource acquisition path includes: Determine the proportional relationship between the weights corresponding to the target edge data center, the weights corresponding to nearby vehicles, and the weights corresponding to the cloud control platform; Based on the aforementioned proportional relationship, multiple target resources are divided into subsets of a number related to the aforementioned proportional relationship, and the target resource acquisition path corresponding to each subset is determined.

4. The method as described in claim 1, characterized in that, In cases where acquiring the target resource via the current acquisition path fails or times out, the following also applies: According to the preset resource acquisition path priority, switch to the new target resource acquisition path in sequence, and use it as the new current target resource acquisition path; The target resource is obtained by switching to the new target resource acquisition path corresponding to the current path priority, and the acquisition status of the target resource is monitored until the download of the target resource is completed. Alternatively, if the acquisition of the target resource fails or times out, the system switches to the new target resource acquisition path corresponding to the next path priority. The resource acquisition paths, in descending order of priority, are: target edge data center, nearby vehicles, and cloud control platform. Preferred options also include: If all attempts to acquire the target resource according to the preset resource acquisition path priorities fail, the resource requested and cached based on the resource request information is cancelled, and the original resource is maintained.

5. The method as described in claim 1, characterized in that, In the event that the acquisition of the target resource fails or times out when the target edge data center is used as the target resource acquisition path, the following additional steps are also included: Continuously acquire target resources from the target edge data center; If the acquisition of target resources from the target edge data center fails or times out, the weight of the resource acquisition path of the target edge data center will be reduced according to the first rule. During the process of continuously acquiring target resources from the target edge data center based on the reduced weight, if there is a successful acquisition status when acquiring target resources from the target edge data center, the weight of the target edge data center is increased according to the second rule.

6. A resource acquisition method applied to an edge data center, characterized in that, The edge data center is located at a service switching point, which is a geographical area associated with changes in vehicle operation tasks. The method includes: Obtain resource pre-fetching tasks from the cloud control platform; Based on a preset multi-factor comprehensive priority model, the pre-fetch priority of each resource in the resource pre-fetching task is determined. Based on the pre-fetching priority of each resource, the system sends the corresponding target resource acquisition request to the cloud control platform in priority order; and receives the corresponding target resource and caches it locally so that the vehicle can acquire the target resource from the edge data center.

7. The method as described in claim 6, characterized in that, The method for determining the prefetch priority of each resource in the resource prefetching task based on the preset multi-factor comprehensive priority model includes: For each type of resource, assign weights to each factor of that resource type; For each resource in the resource prefetching task, determine the priority score of that resource under each factor. Based on the weights of each factor of the resource, the priority scores under each factor are weighted and calculated to determine the prefetch priority of the resource.

8. The method as described in claim 6 or 7, characterized in that, The multi-factor comprehensive priority model includes at least one of the following factors: heat priority factor, time-sensitive factor, map association factor, and volume-friendly factor; The popularity priority factor is used to characterize the global request popularity of the resource on the cloud side and the historical request popularity of the corresponding content type of the resource in the edge data center. The time-sensitive factor is used to characterize the degree of correlation between the resource and the vehicle operation task, as well as the proximity of the current time to the relevant time node of the vehicle operation task. The map association factor is used to characterize the association strength between the resource and the map constraints; wherein the map constraints include at least one of the following: environmental awareness friendliness, path connectivity, and basic data freshness. The volume-friendly factor is used to characterize the fragment size of a resource.

9. The method as described in claim 8, characterized in that, When the resource type is map data, the association strength between the resource and map constraints is determined using the following methods: Determine the high-precision map static layer data and dynamic event layer data corresponding to the resource; The high-precision map static layer data is subjected to local rasterization processing to determine the line-of-sight probability and reachability assessment score; wherein, the line-of-sight probability is used to characterize environmental perception friendliness, and the reachability assessment score is used to characterize path connectivity. Based on the event impact area corresponding to the dynamic event layer data and the high-precision map static layer data, the map freshness corresponding to the resource is determined; the map freshness is used to characterize the freshness of the basic data. The direct-view probability, reachability assessment score, and map freshness are fused using a preset fusion rule to determine the correlation strength between the resource and map constraints.

10. The method as described in claim 6, characterized in that, The method further includes: Based on the target resource's lifetime and preset expiration date in the local cache, perform a resource eviction operation on the target resource; and / or, Based on the hit rate of the target resource in the local cache within the statistical period and the preset hit rate threshold, a resource eviction operation is performed on the target resource.

11. The method as described in claim 6, characterized in that, The types of resources include at least one of the following: map data, software packages, access control policies, and job tasks; and / or, The resource is split into equal-granularity fragments, and a directory is set for each fragment. The directory includes: version identifier, preset priority, and preset validity period.

12. An unmanned vehicle, characterized in that, Perform the steps of any of the resource acquisition methods as claimed in claims 1 to 5.

13. A resource acquisition system, characterized in that, include: Edge data centers and vehicles; among them, The edge data center is located at the service switching point, which is a geographical area associated with changes in vehicle operation tasks; The vehicle is used to perform the steps of any of the resource acquisition methods as claimed in claims 1 to 5.

Images