A dynamic cooperative parking space sensing method based on open source honkong, device and medium

Abstract

The embodiment of the application discloses a kind of dynamic collaborative parking space sensing method, equipment and medium based on open source Hong Meng, belong to the technical field of Internet of Things, solve the existing fixed isolated parking space sensing device architecture, due to sensing blind area cause its reliability and sensing accuracy lower problem.It includes, by the distributed soft bus of open source Hong Meng, the multiple parking space sensing devices in parking lot are dynamically networked, and parking space sensing resource pool is constructed;Abnormal information in current parking lot is acquired, to determine target sensing range based on abnormal information;Abnormal information at least includes one of basic parking space sensing device abnormal information and parking behavior abnormal information;Based on target sensing range, filter multiple reference parking space sensing devices in parking space sensing resource pool to form collaborative sensing network, to carry out collaborative parking space sensing data acquisition to target sensing range;The data collected are fused and analyzed, and the collaborative parking space sensing information corresponding to target sensing range is generated.

Description

Technical Field

[0001] This application relates to the field of Internet of Things (IoT) technology, and in particular to a dynamic collaborative parking space sensing method, device, and medium based on the open-source HarmonyOS. Background Technology

[0002] With the deepening of smart city construction, smart parking has become a key to solving the parking problem. Traditional smart parking systems rely on fixed sensing devices deployed in parking lots to monitor the status of parking spaces. For example, fixed cameras are installed above each parking space, and image recognition algorithms are used to determine the status of the parking space.

[0003] However, in complex real-world application scenarios, these systems face severe challenges: First, for non-standard parking behaviors, such as parking over the lines, angled parking, or the front / rear of the vehicle exceeding the permitted limits, fixed sensing devices suffer from limited sensing range, resulting in poor reliability and a high false alarm rate. Second, if key node devices such as the aggregation gateway or specific cameras in a certain area malfunction, it will lead to inaccurate parking space status across a large area, creating sensing blind spots, thus resulting in low reliability and sensing accuracy of the smart parking system. Summary of the Invention

[0004] This application provides a dynamic collaborative parking space sensing method, device, and medium based on open-source HarmonyOS to solve the following technical problem: Existing smart parking systems adopt a fixed and isolated parking space sensing device architecture. When the sensing device malfunctions or parking behavior is detected, the reliability and accuracy of the sensing are low due to the sensing blind spot.

[0005] The embodiments of this application adopt the following technical solutions: This application provides a dynamic collaborative parking space perception method based on the open-source HarmonyOS. It includes: dynamically networking multiple parking space perception devices in a parking lot using the distributed soft bus of HarmonyOS to construct a parking space perception resource pool; wherein the parking space perception devices include at least one of the following: a camera, a geomagnetic sensor, and a Bluetooth beacon; acquiring abnormal information in the current parking lot to determine the target perception range based on the abnormal information; wherein the abnormal information includes at least one of the following: abnormal information from basic parking space perception devices and abnormal parking behavior information; based on the target perception range, selecting multiple reference parking space perception devices from the parking space perception resource pool to form a collaborative perception network to collect collaborative parking space perception data within the target perception range; fusing and analyzing the collected collaborative parking space perception data, and dynamically adjusting the weight of each parking space perception data in the decision-making process based on the real-time status of each data source to generate collaborative parking space perception information corresponding to the target perception range.

[0006] In one implementation of this application, multiple parking space sensing devices in a parking lot are dynamically networked using the open-source HarmonyOS distributed soft bus to construct a parking space sensing resource pool. Specifically, this includes: automatically acquiring multiple parking space sensing devices in the parking lot that support the open-source HarmonyOS protocol via the open-source HarmonyOS distributed soft bus, and dynamically networking the discovered parking space sensing devices to construct a device network; encapsulating the heterogeneous sensing capabilities of each parking space sensing device in the device network into sensing service instances with a unified calling format; wherein the sensing service instances are decoupled from the native interfaces of the parking space sensing devices; registering each encapsulated sensing service instance and its corresponding device metadata to a metadata library maintained by a distributed data management service; wherein the device metadata includes at least one of the following: device location, device capability type, and device real-time status; and aggregating the registered sensing service instances to construct the parking space sensing resource pool, and providing a unified service calling interface decoupled from the parking space sensing devices for upper-layer sensing applications.

[0007] In one implementation of this application, abnormal information within the current parking lot is obtained to determine the target perception range based on the abnormal information. Specifically, this includes: if the abnormal information is abnormal information of the basic parking space sensing device, determining the target perception range based on the location information of the basic parking space sensing device and the corresponding perception range of the basic parking space sensing device; if the abnormal information is abnormal parking behavior information, generating a vehicle outline rectangle based on the real-time location coordinates of the target vehicle, calculating the spatial geometric relationship between the vehicle outline rectangle and the boundary coordinates of the vacant parking spaces in the parking lot digital map, and determining the target perception range corresponding to the target vehicle based on the calculation result.

[0008] In one implementation of this application, a collaborative sensing network is formed by selecting multiple reference parking space sensing devices from a parking space sensing resource pool based on the target sensing range. Specifically, this includes: determining the overlap between the sensing coverage of each parking space sensing device and the target sensing range based on the target sensing range, as well as the location information and functional attribute information of each parking space sensing device; acquiring the operating status information corresponding to each parking space sensing device; wherein the operating status information is related to the load status and operational health status of the parking space sensing device; weighting the operating status information corresponding to each parking space sensing device to determine the optimal score for each parking space sensing device; sorting the parking space sensing devices based on the optimal score, and selecting reference parking space sensing devices sequentially based on the sorting order to construct a reference device set; determining the sensing coverage blind zone of the reference device set for the target sensing range based on the overlap between each reference parking space sensing device and the target sensing range; and constructing a collaborative sensing network based on the reference device set when the area of ​​the sensing coverage blind zone is less than a preset blind zone area threshold.

[0009] In one implementation of this application, the collected parking space sensing data is fused and analyzed, and the weights of each parking space sensing data in the decision-making process are dynamically adjusted based on the real-time status of each data source to generate collaborative parking space sensing information corresponding to the target sensing range. Specifically, this includes: spatiotemporal alignment of multi-source parking space sensing data; wherein the multi-source parking space sensing data includes at least one of image data, geomagnetic data, and Bluetooth signals; dynamically adjusting the weights of the multi-source parking space sensing data based on the real-time status and data quality of each data source; wherein the real-time status refers to the operating status of each parking space sensing device acquiring the parking space sensing data, and the data quality refers to the accuracy of the parking space sensing data; and based on the synthesis rules of evidence theory, conflict resolution and decision synthesis are performed on the dynamically weighted multi-source parking space sensing data to generate collaborative parking space sensing information corresponding to the target sensing range.

[0010] In one implementation of this application, based on the synthesis rules of evidence theory, conflict resolution and decision synthesis are performed on multi-source parking space perception data after dynamic weight adjustment to generate collaborative parking space perception information corresponding to the target perception range. Specifically, this includes: converting multi-source parking space perception data from different parking space perception devices into a basic probability allocation function in evidence theory to construct parking space status evidence bodies corresponding to each data source; determining the conflict degree factor between evidence bodies based on the distance and directional similarity of each parking space status evidence body in the vector space; when a conflict is detected between evidence bodies, fusing the parking space status evidence bodies based on the conflict degree factor and a preset evidence conflict synthesis strategy to obtain a comprehensive support score for multiple reference parking space statuses; determining the collaborative parking space perception information corresponding to the target perception range based on the comprehensive support score, and determining the confidence level corresponding to the status information based on the distribution characteristics of the comprehensive support score, so that the collaborative parking space perception information corresponding to the target perception range is determined when the confidence level is greater than a preset threshold; wherein, the ratio between the distribution characteristics and the comprehensive support level is related to the information entropy corresponding to the comprehensive support level.

[0011] In one implementation of this application, after selecting multiple reference parking space sensing devices from the parking space sensing resource pool to form a collaborative sensing network for collaborative parking space sensing data collection of the target sensing range, the method further includes: real-time monitoring of the reference parking space sensing devices; if a fault is detected in a reference parking space sensing device, transforming the optimization objective corresponding to the selection of the reference parking space sensing device, so as to select a replacement parking space sensing device corresponding to the faulty reference parking space sensing device from the parking space sensing resource pool through the transformed optimization objective; and combining and adjusting the replacement parking space sensing devices so that the sensing range of the combined replacement parking space sensing devices covers the sensing range of the faulty reference device.

[0012] In one implementation of this application, alternative parking space sensing devices corresponding to a faulty reference parking space sensing device are selected from a parking space sensing resource pool. Specifically, this includes: selecting a first group of parking space sensing devices with the same sensing capability as the faulty reference parking space sensing device from the parking space sensing resource pool; selecting a second group of parking space sensing devices from the first group whose sensing range overlaps with that of the faulty reference parking space sensing device; determining the substitution value score of each sensing device in the second group based on its real-time load indicators and communication costs in the network topology; selecting devices sequentially from the second group according to their substitution value scores until the comprehensive sensing range of the selected device covers the sensing range of the faulty reference parking space sensing device, generating a preferred device set; and performing collaborative readiness verification on the preferred device set, using the verified devices as alternative parking space sensing devices.

[0013] This application provides a dynamic collaborative parking space sensing device based on the open-source HarmonyOS, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to: dynamically network multiple parking space sensing devices in a parking lot via the distributed soft bus of the open-source HarmonyOS to construct a parking space sensing resource pool; wherein the parking space sensing devices include at least one of a camera device, a geomagnetic sensor, and a Bluetooth beacon in the parking lot; acquire abnormal information in the current parking lot to determine a target sensing range based on the abnormal information; wherein the abnormal information includes at least one of abnormal information from basic parking space sensing devices and abnormal information from parking behavior; based on the target sensing range, select multiple reference parking space sensing devices from the parking space sensing resource pool to form a collaborative sensing network for collaborative parking space sensing data collection within the target sensing range; perform fusion analysis on the collected collaborative parking space sensing data, and dynamically adjust the weight of each parking space sensing data in the decision-making process based on the real-time status of each data source to generate collaborative parking space sensing information corresponding to the target sensing range.

[0014] This application provides a non-volatile computer storage medium storing computer-executable instructions. These instructions are configured to: dynamically network multiple parking space sensing devices within a parking lot using the distributed soft bus of the open-source HarmonyOS to construct a parking space sensing resource pool; wherein the parking space sensing devices include at least one of the following: a camera, a geomagnetic sensor, and a Bluetooth beacon; acquire abnormal information within the current parking lot to determine the target sensing range based on the abnormal information; wherein the abnormal information includes at least one of the following: abnormal information from basic parking space sensing devices and abnormal parking behavior information; based on the target sensing range, select multiple reference parking space sensing devices from the parking space sensing resource pool to form a collaborative sensing network for collaborative parking space sensing data collection within the target sensing range; perform fusion analysis on the collected collaborative parking space sensing data, and dynamically adjust the weight of each parking space sensing data in the decision-making process based on the real-time status of each data source to generate collaborative parking space sensing information corresponding to the target sensing range.

[0015] The above-mentioned technical solutions adopted in this application embodiment can achieve the following beneficial effects: This application embodiment realizes dynamic networking of sensing devices in parking lots through the open-source HarmonyOS distributed soft bus, and realizes the screening and management of sensing devices in parking lots by constructing a parking space sensing resource pool. Secondly, for the received abnormal information in the parking lot, this application embodiment can screen multiple parking space sensing devices to build a collaborative sensing network, which not only improves resource utilization, but also enhances the robustness and coverage integrity of the sensing system through the complementarity of multi-source parking space sensing data. Finally, through the spatiotemporal alignment and intelligent fusion of multi-source parking space sensing data, the decision weight can be dynamically adjusted according to the real-time quality and reliability of each data source, solving the problem of misjudgment caused by the failure of a single sensing device or environmental interference. The final output of collaborative parking space sensing information has a higher accuracy, enabling the parking space sensing system to maintain a high-precision and high-reliability operating state in complex parking lot environments. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart of a dynamic collaborative parking space perception method based on open-source HarmonyOS provided in this application embodiment; Figure 2 A dynamic resource pool construction diagram provided in this application embodiment; Figure 3This is a schematic diagram of the structure of a dynamic collaborative parking space sensing device based on the open-source HarmonyOS, provided as an embodiment of this application.

[0017] Figure label: 200: Dynamic collaborative parking space sensing device based on open source HarmonyOS; 201: Processor; 202: Memory. Detailed Implementation

[0018] This application provides a dynamic collaborative parking space sensing method, device, and medium based on the open-source HarmonyOS.

[0019] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0020] Figure 1 A flowchart of a dynamic collaborative parking space perception method based on open-source HarmonyOS provided in this application embodiment is shown below. Figure 1 As shown, the dynamic collaborative parking space perception method based on open-source HarmonyOS includes the following steps: S101. Through the open-source HarmonyOS distributed soft bus, multiple parking space sensing devices in the parking lot are dynamically networked to build a parking space sensing resource pool.

[0021] In one implementation of this application, multiple parking space sensing devices supporting the open-source HarmonyOS protocol within the parking lot are automatically acquired via the open-source HarmonyOS distributed soft bus. These devices are then dynamically networked to construct a device network. The heterogeneous sensing capabilities of each parking space sensing device in the device network are encapsulated into sensing service instances with a unified calling format; these sensing service instances are decoupled from the native interfaces of the parking space sensing devices. Each encapsulated sensing service instance, along with its corresponding device metadata, is registered in a metadata repository maintained by a distributed data management service; the device metadata includes at least one of the following: device location, device capability type, and device real-time status. The registered sensing service instances are aggregated to form a parking space sensing resource pool, providing a unified service calling interface decoupled from the parking space sensing devices for upper-layer sensing applications.

[0022] Specifically, through the distributed soft bus of the open-source HarmonyOS, all parking space sensing devices supporting the open-source HarmonyOS protocol within the parking lot are automatically acquired and managed. These devices include not only dedicated geomagnetic sensors and fixed parking space cameras, but also panoramic monitoring cameras and rotating cameras belonging to the security system, as well as Bluetooth beacons for positioning. Distributed device virtualization technology abstracts the sensing capabilities of these physical devices, such as video streaming, magnetic field sensing, and distance detection, into a unified interface service, forming a parking space sensing resource pool within the system. The metadata (location, capabilities, and status) of all devices is synchronized in the distributed data management service.

[0023] Furthermore, through the open-source HarmonyOS distributed soft bus, multiple parking space sensing devices supporting the open-source HarmonyOS protocol within the parking lot are automatically acquired. Communication with each physical device is established via the soft bus connection. The native capability list reported by the device is read, and then, according to the predefined parking space sensing service abstraction layer specification, heterogeneous hardware instructions are encapsulated into a unified software service interface, achieving decoupling from the native interface. After encapsulation, each device's capabilities are presented as one or more software service instances with unique service IDs. After encapsulation, the resource pool management service generates complete metadata for each sensing service instance. By calling the interface provided by the open-source HarmonyOS distributed data management service, this metadata is registered as a record in its maintained distributed metadata repository. The dynamic attributes in the metadata repository are continuously updated through the soft bus's heartbeat mechanism or event notification mechanism. Any change in the state of any node is synchronized in real time to all system components that have subscribed to this information through the distributed data management service, ensuring that the entire system's view of the resource pool remains consistent and up-to-date. Once all service instances have completed registration, the parking space sensing resource pool is fully constructed.

[0024] Figure 2 A dynamic resource pool construction diagram is provided for embodiments of this application, such as... Figure 2 As shown, this dynamic resource pool adopts a three-layer design. The top layer is the physical device layer, which includes various heterogeneous parking space sensing devices such as geomagnetic sensors, fixed parking space cameras, panoramic monitoring cameras, rotating cameras, and Bluetooth beacons. The middle layer uses distributed soft bus technology to achieve automatic networking and unified management of devices, possessing device virtualization capabilities, interface standardization, and distributed data management functions, enabling the synchronization of metadata and status maintenance for all connected parking space sensing devices. The bottom layer constructs a parking space sensing resource pool, which abstracts the capabilities of various devices into standardized service modules such as video streaming services, magnetic field sensing services, and location detection services through a resource registration mechanism, while maintaining a unified device metadata database. Data communication between each layer is achieved through status synchronization and resource registration mechanisms, providing standardized sensing service interfaces for upper-layer applications.

[0025] S102. Obtain abnormal information in the current parking lot, and determine the target perception range based on the abnormal information.

[0026] This application prioritizes low-cost parking space sensing devices, such as geomagnetic sensors or fixed parking space cameras. The aim is to determine the status of most common parking spaces with minimal computing power and network overhead. The system receives magnetic field change signals from the geomagnetic sensor or single-view image streams from the fixed parking space camera. If the geomagnetic signal is stable in an untriggered state, it is directly determined to be idle; if it is stable in a triggered state, it is initially determined to be occupied. A lightweight edge-side image recognition model quickly analyzes the image. If no vehicle is detected, it is determined to be idle; if a vehicle is clearly identified as parked properly within the parking space lines, it is determined to be occupied. When the two basic sensing results are consistent, the AI ​​analysis and decision engine outputs a parking space status result with a standard confidence level and completes the sensing task. When the basic sensing results are ambiguous or lack sufficient confidence, a heterogeneous collaborative judgment process is initiated.

[0027] First, construct an abnormal behavior recognition model. This includes: 1. Data collection and annotation: During daily operation, the system automatically collects raw data that triggers the following conditions: Low confidence samples: The confidence level of the AI ​​analysis and decision engine in judging the parking space status is lower than the preset threshold; Conflicting samples: Samples whose judgment results are inconsistent among different sensory sources; Manually reviewed samples: Samples used by administrators when confirming or correcting system alerts in the background.

[0028] These samples, along with their corresponding real-state labels, timestamps, device IDs, and environmental metadata, are saved together, and these labeled image pairs constitute a scene-specific incremental dataset.

[0029] 2. Model fine-tuning training, the training process is carried out in a simultaneous training and validation manner: Base model: Load the deployed general vehicle detection model as pre-trained weights; Training strategy: Employ the fine-tuning strategy from transfer learning. Typically, freeze the feature extraction network at the bottom layer of the model and train only the top classification head or bounding box regression head. This effectively utilizes general features while learning scene-specific features with relatively low computational cost.

[0030] Loss function and optimization: The standard object detection loss function is used. The model is trained on a dataset that mixes incremental data and a portion of historical data. The parameters are updated through backpropagation and an optimizer to prevent the model from forgetting the original knowledge.

[0031] 3. Offline evaluation and verification: After training, the new model is rigorously evaluated offline on a subset of historical data that was never used in the training. Key metrics: Primarily assess accuracy and robustness; Qualification criteria: Only when the new model shows a significant and stable improvement over the current online model in these metrics will it proceed to the next stage.

[0032] 4. Model Deployment and Closed Loop: The validated new model will be hot-deployed as a new version on edge servers or in the cloud, gradually replacing the old model in a seamless and smooth manner to ensure uninterrupted service. After the new model goes live, it will generate new runtime data, thus starting the next cycle of data collection, training, evaluation, and deployment, enabling the model to continuously evolve.

[0033] Furthermore, this application embodiment pre-defines an iterative abnormal behavior model library, whose main abnormal types and initial judgment criteria include: (1) Parking over the line / crossing the space: The intersection area of ​​the vehicle outline and the parking space line exceeds the preset threshold; (2) Angled / lateral parking: The angle between the centerline of the vehicle and the centerline of the parking space exceeds the allowable range; (3) Vehicle front / rear exceeds: The vehicle's bounding box exceeds the parking space line boundary; (4) Parking space occupied by non-vehicle objects: Fixed obstacles are detected in the parking space; (5) Conflict in parking space perception data: For example, the geomagnetic field continuously reports that the parking space is occupied, but the camera continuously reports that it is idle.

[0034] Specifically, when basic perception results conflict, or when the confidence level calculated by the AI ​​analysis and decision engine falls below a preset threshold, the dynamic resource filter is triggered. Based on device metadata from the distributed data management service, the filter intelligently selects nearby rotatable security camera viewing angles, or retrieves video streams from other panoramic cameras covering the side of the parking space, to obtain multi-angle, multi-dimensional perception information. Furthermore, as needed, it retrieves the vehicle's Bluetooth movement trajectory corresponding to the target perception range, using changes in the strength of Bluetooth beacon signals as an auxiliary basis for determining the vehicle's presence.

[0035] In one implementation of this application, the abnormal information includes at least one of basic parking space sensing device abnormality information and parking behavior abnormality information. If the abnormal information is basic parking space sensing device abnormality information, the target sensing range is determined based on the location information and sensing range of the basic parking space sensing device. If the abnormal information is parking behavior abnormality information, a vehicle outline rectangle is generated based on the real-time location coordinates of the target vehicle. The spatial geometric relationship between the vehicle outline rectangle and the boundary coordinates of vacant parking spaces in the parking lot digital map is calculated, and the target sensing range corresponding to the target vehicle is determined based on the calculation result.

[0036] Specifically, when the abnormal information indicates a problem with the basic parking space sensing device, such as a faulty geomagnetic sensor or camera in a fixed parking space, the system first retrieves the installation location coordinates and preset sensing range parameters of the faulty device from the device management database. These parameters include, for example, the horizontal and vertical field of view, installation height, and pitch angle of the camera, or the detection radius of the geomagnetic sensor. Based on these parameters, the system uses a geometric calculation model to draw the target sensing range that the device was originally responsible for monitoring on the digital map layer of the parking lot. When the abnormal information indicates abnormal parking behavior, such as the inability to detect whether a vehicle is crossing a line due to obstruction, the system first obtains the real-time location coordinates of the target vehicle. Using the vehicle's coordinates as a reference point, the system constructs the overall outline of the target vehicle based on this appearance data. The overall outline of the vehicle is then drawn in the image, and this pixel-level rectangle is converted into a vehicle outline rectangle in the world coordinate system using camera calibration parameters, thus completing the geometric modeling of the space occupied by the vehicle.

[0037] The coordinates of the generated vehicle outline rectangle are compared with the precise boundary coordinates of all currently vacant parking spaces obtained from the parking lot digital map. First, it is calculated whether the centroid of the vehicle rectangle falls inside the boundary polygon of a vacant parking space; if so, the parking space is directly identified as a candidate. If the centroid cannot match (common in irregular situations such as parking over the lines), the intersection-union ratio (IUR) of the vehicle rectangle with each vacant parking space polygon is calculated. Finally, parking spaces whose centroids fall inside, or those with the highest IUR exceeding a preset threshold, are identified as the target perception range corresponding to the target vehicle.

[0038] S103. Based on the target perception range, select multiple reference parking space perception devices from the parking space perception resource pool to form a collaborative perception network, so as to collect collaborative parking space perception data within the target perception range.

[0039] In one implementation of this application, the overlap between the sensing coverage of each parking space sensing device and the target sensing range is determined based on the location information of the target sensing range, as well as the location information and functional attribute information of each parking space sensing device. The operating status information corresponding to each parking space sensing device is obtained; wherein, the operating status information is related to the load status and operational health status of the parking space sensing device. The operating status information corresponding to each parking space sensing device is weighted to determine the optimal score for each parking space sensing device. Based on the optimal score, the parking space sensing devices are sorted, and reference parking space sensing devices are selected sequentially according to the sorting order to construct a reference device set. Based on the overlap between each reference parking space sensing device and the target sensing range, the sensing coverage blind zone of the reference device set for the target sensing range is determined. If the area of ​​the sensing coverage blind zone is less than a preset blind zone area threshold, a collaborative sensing network is constructed based on the reference device set.

[0040] Specifically, firstly, based on the location information of the target perception range, as well as the location information and functional attribute information of each parking space sensing device, the overlap between the perception coverage of each parking space sensing device and the target perception range is determined. This includes: establishing a spatial point-based perception probability model for each parking space sensing device based on the spatial changes in device functional attribute information; establishing a semantic value distribution map for the target perception range based on the perception task type; whereby the semantic value distribution map reflects the distribution of importance levels of different areas within the target perception range for the current perception task; for each parking space sensing device, determining the geometric intersection between spatial areas with perception probabilities greater than a preset threshold and spatial areas in the semantic value distribution map with importance greater than a preset threshold; determining the value integral of the semantic value distribution map within the geometric intersection, and using the value integral as the effective perception value of the parking space sensing device for the target perception range; and using the ratio between the effective perception value and the total semantic value of the target perception range as the overlap between the perception coverage of the parking space sensing device and the target perception range.

[0041] Specifically, based on the functional attributes of each parking space sensing device, such as the effective detection distance, resolution, and environmental attenuation characteristics of the cameras, a spatial perception probability model for the parking space sensing devices is constructed. This model sets the physical location of the parking space sensing device as the origin of the coordinate system. Through radial basis function or Gaussian process regression, it calculates the probability value of the device successfully sensing the target state at any point in space, forming a perception capability probability field covering the three-dimensional space surrounding the device with continuously changing probability values. This probability field reflects the non-uniform attenuation characteristics of the device's perception capability with distance, angle, and obstructions. Furthermore, based on the location information of the current target perception range, the importance level of different areas of the parking space to completing the task is analyzed and defined. For example, the central area of ​​the parking space is crucial for determining whether the vehicle is parked stably, while the boundary area is relatively important for determining whether the wheels are crossing the lines. Based on this definition, a value distribution model is constructed for the target perception range, where each spatial coordinate point is assigned a quantified semantic value score, reflecting the point's contribution to the current perception task, forming a structured semantic value distribution map.

[0042] Furthermore, for each parking space sensing device, in its corresponding spatial sensing probability model, a set of all spatial points with a sensing probability greater than a preset confidence threshold is extracted, constituting the area that the device can reliably sense. Simultaneously, in the semantic value distribution map of the target sensing range, a set of all spatial points with a semantic value score greater than a preset importance threshold is extracted, constituting the area that is relevant to the current task. Subsequently, an intersection operation is performed on these two sets of spatial points to calculate the common spatial range that the device can reliably sense and simultaneously cover the key task area of ​​the parking space, resulting in a geometric intersection area with both high confidence and high value. Furthermore, for the obtained geometric intersection region, the value scores defined in the semantic value distribution map are integrated within its continuous spatial range. Specifically, the geometric intersection region is discretized into tiny spatial units, and the product of the volume of each unit and the semantic value score of the unit's center point is calculated. The sum of all unit products is then obtained, and the result is the effective perception value generated by the parking space sensing device for the target perception range. The integration process in this embodiment quantifies the contribution value that the device can provide to completing the core perception task while ensuring perception reliability. The total semantic value of the target perception range in the entire semantic value distribution map is calculated, that is, the value score of the entire spatial range of the parking space is globally integrated. The effective perception value of a single device is then compared with the total semantic value of the parking space. The result of this ratio is the overlap between the perception coverage of the parking space sensing device and the target perception range. The larger the value, the more comprehensive and reliable the device's coverage contribution to the current task.

[0043] Secondly, the device monitoring agent of the distributed soft bus collects the real-time operating status information of each parking space sensing device. This operating status information includes at least: real-time computing load, network load, and device physical health. These indicators are normalized using a pre-defined health assessment model to generate a comprehensive operating health score for each device. The calculated overlap rate and the resulting comprehensive operating health score are used as inputs to calculate the optimal score for each parking space sensing device using a pre-defined weighted scoring function. The weighted scoring function is a weighted sum of the overlap rate and the comprehensive operating health score, with the weighting coefficients dynamically configured based on the reliability or real-time requirements of the sensing task.

[0044] Next, all parking space sensing devices are sorted in descending order based on the calculated optimal scores. Starting with the device with the highest score, devices are sequentially added to the candidate reference device set until the cumulative coverage area of ​​the candidate set reaches a preset initial coverage target (e.g., 90% of the parking space area), thus completing the construction of the initial reference device set. Based on the constructed initial reference device set, and combined with the calculated precise sensing range model of each device, the overall sensing coverage area of ​​the set for the target sensing range is calculated using geometric Boolean operations, and the uncovered sensing coverage blind spots are identified. If the total blind spot area is less than a preset blind spot area threshold, the initial reference device set is confirmed as the finally selected collaborative sensing network; if the blind spot area exceeds the threshold, the next device is selected from the sorted sequence to be added to the set, iteratively optimizing until the blind spots meet the requirements, thus finally constructing the collaborative sensing network.

[0045] S104. The collected collaborative parking space perception data is fused and analyzed, and the weight of each parking space perception data in the decision-making is dynamically adjusted based on the real-time status of each data source, so as to generate collaborative parking space perception information corresponding to the target perception range.

[0046] In one implementation of this application, multi-source parking space sensing data is spatiotemporally aligned; wherein the multi-source parking space sensing data includes at least one of image data, geomagnetic data, and Bluetooth signals. Based on the real-time status and data quality of each data source, the data weights are dynamically adjusted, where the real-time status refers to the operating status of each parking space sensing device acquiring the parking space sensing data, and the data quality refers to the accuracy of the parking space sensing data. Based on the synthesis rules of evidence theory, conflict resolution and decision synthesis are performed on the dynamically weighted multi-source parking space sensing data to generate collaborative parking space sensing information corresponding to the target sensing range.

[0047] First, the multi-source parking space perception data is spatiotemporally aligned to ensure that all data correspond to the same target at the same time and in the same space. Then, an AI engine performs parallel feature extraction on different types of data. For image data, high-level semantic features are extracted using convolutional neural networks, such as vehicle bounding boxes, contours, attitude angles, and relative positions to parking lines. For geomagnetic signals, time-domain / frequency-domain features such as signal strength, stability, and waveform variations are extracted. For Bluetooth signals, features such as signal strength, connection stability, and MAC address identifiers are extracted.

[0048] Secondly, multi-source parking space sensing data from different parking space sensing devices are transformed into basic probability allocation functions in evidence theory to construct parking space status evidence bodies corresponding to each data source. Based on the distance and directional similarity of each parking space status evidence body in vector space, a conflict degree factor between evidence is determined. When a conflict is detected between evidence, the parking space status evidence bodies are fused based on the conflict degree factor and a preset evidence conflict synthesis strategy to obtain a comprehensive support score for multiple reference parking space statuses. According to the comprehensive support score, the collaborative parking space sensing information corresponding to the target sensing range is determined, and based on the distribution characteristics of the comprehensive support score, the confidence level corresponding to the status information is determined. When the confidence level is greater than a preset threshold, the collaborative parking space sensing information corresponding to the target sensing range is sent to the parking user; wherein, the ratio between the distribution characteristics and the comprehensive support level is related to the information entropy corresponding to the comprehensive support level.

[0049] Specifically, multi-source parking space sensing data from different parking space sensing devices are transformed into basic probability allocation functions in evidence theory based on their respective data characteristics and historical performance, in order to construct parking space status evidence bodies corresponding to each data source. That is, for the recognition results output by image sensors, the confidence level is proportionally allocated to state hypotheses such as occupied, vacant, and abnormal, based on its classification confidence and historical accuracy, and the remaining confidence level is allocated to an uncertain set containing all hypotheses to represent the uncertainty of the evidence. For geomagnetic sensors, the membership degree of each state is calculated based on the relationship between its signal change amplitude and a preset threshold, and the basic probability allocation is determined by combining the sensor's own historical error rate. For Bluetooth signals, basic probabilities are assigned to each state hypothesis based on the stability of its signal strength and the connection duration. This step generates an independent evidence body for each data source, providing input for subsequent fusion.

[0050] Based on the distance and directional similarity of each parking space status evidence body in vector space, a conflict degree factor between evidence is determined. First, each evidence body is represented as a multi-dimensional vector, with its dimensions corresponding to the basic probability assignment values ​​of different state hypotheses. Then, the Jousselme distance between any two evidence vectors is calculated; this distance quantifies the degree of difference between the evidence, with a larger distance indicating a more significant conflict. Simultaneously, the cosine similarity of the angle between evidence vectors is calculated to assess the similarities and differences of the evidence in the direction supporting the state. Finally, by weighting and fusing the distance and similarity, a normalized conflict degree factor is calculated. This conflict degree factor comprehensively and quantitatively reflects the consistency and contradictions among the evidence.

[0051] Furthermore, each parking space status evidence is a point in a multi-dimensional vector space. For example, the evidence vector from a camera is [Idle: 0.8, Occupied: 0.15, Abnormal: 0.05], and the evidence vector from geomagnetism is [Idle: 0.1, Occupied: 0.85, Abnormal: 0.05]). The distance between each pair of these evidence points is calculated. For the set of evidence points, the average distance between all evidence pairs is calculated, and the calculated scalar value is the conflict factor. The larger the conflict factor value, the more dispersed the distribution of evidence points in the vector space, and the higher the degree of contradiction.

[0052] This application embodiment includes an evidence conflict synthesis strategy library, which defines fusion algorithms corresponding to different conflict intervals. When a conflict is detected between pieces of evidence, the fusion strategy is dynamically selected or adjusted based on the calculated conflict severity factor. For example, when the conflict severity factor value is less than a low threshold, it indicates that the evidence is basically consistent, and the computationally efficient Dempster combination rule can be used for fusion. When the conflict severity factor is at a medium level, a weighted average method can be used, that is, the basic probability distribution of all evidence is averaged first to smooth the differences, and then Dempster combination is performed. When the conflict severity factor exceeds a high threshold, it indicates that the evidence is highly conflicting, and a weighted correction method based on evidence similarity is used: first, the reliability weight of each piece of evidence is calculated, then the evidence is discounted and corrected using the weights, and finally the corrected evidence is synthesized using the Dempster rule. This application embodiment, through an adaptive strategy, finally obtains the comprehensive support score of each candidate parking space status (vacant, occupied, abnormal), forming a normalized support distribution vector.

[0053] Furthermore, based on the integrated support score distribution obtained after fusion, the state with the highest support is determined as the final state information within the target perception range. Subsequently, the confidence level of this decision is quantified based on the statistical characteristics of this support distribution, including calculating the sharpness ratio and information entropy. Specifically, the sharpness ratio is calculated, which is the ratio of the highest support to the second highest support; this value reflects the clarity of the decision. Secondly, the information entropy is calculated; the entropy value measures the uncertainty of the entire distribution, and the lower the entropy, the more concentrated the probability quality. Finally, the confidence level is calculated using the formula: Confidence Level = (Highest Support) * (Normalized Sharpness Ratio) * (1 - Normalized Entropy). The resulting confidence level assessment integrates the absolute support level, relative advantage, and overall uncertainty of the decision, thus deriving a credibility index. If the obtained confidence level is higher than a preset value, state feedback can be provided based on the obtained state information within the target perception range; if the confidence level is not higher than the preset value, the user can be prompted to perform manual testing to avoid misjudgment.

[0054] In one implementation of this application, the embodiment further monitors the reference parking space sensing device in real time. If a fault is detected in the reference parking space sensing device, the optimization target used when selecting the reference parking space sensing device is transformed. Based on the transformed optimization target, a replacement parking space sensing device corresponding to the faulty reference parking space sensing device is selected from the parking space sensing resource pool. The replacement parking space sensing devices are then combined and adjusted so that the sensing range of the combined parking space sensing device covers the sensing range of the faulty reference device.

[0055] Specifically, this application embodiment continuously monitors the health status of all devices in the resource pool. When a device failure or network interruption is detected, the dynamic resource filter immediately updates the resource pool status and dynamically and seamlessly switches the sensing tasks handled by that device to other heterogeneous collaborative devices that can cover the same area. For example, if a fixed camera fails, the system can temporarily adjust the patrol paths of its two adjacent rotating cameras to cover its blind spots.

[0056] Specifically, the health monitoring module monitors the device's heart rate and data quality in real time. Once a device malfunction is detected, it immediately alerts the dynamic resource filter and marks all incomplete sensing tasks of the malfunctioning device as pending takeover. In this scenario, the filter's optimization objective shifts from optimal to fastest feasible.

[0057] Furthermore, within the parking space sensing resource pool, a first group of parking space sensing devices with the same sensing capabilities as the faulty reference parking space sensing device is selected. Within this first group, a second group of parking space sensing devices is selected whose sensing range overlaps with that of the faulty reference parking space sensing device. Within the second group, based on the real-time load indicators of each sensing device and its communication cost within the network topology, the substitution value score of each sensing device is determined. Based on these substitution value scores, devices are selected sequentially within the second group until the comprehensive sensing range of the selected device covers the sensing range of the faulty reference parking space sensing device, generating a preferred device set. The preferred device set undergoes collaborative readiness verification, and devices that pass the verification are designated as replacement parking space sensing devices.

[0058] Specifically, the system quickly filters out all healthy devices from the resource pool that have the same or similar sensing capabilities as the faulty device and whose spatial coverage overlaps with the faulty area. Among these devices, the device with the lightest current load is selected as the priority takeover device. If multiple devices exist, the device with the closest network topology distance is selected. If a single device cannot completely cover the faulty area, the filter calculates a minimum set of devices that can jointly cover the original area.

[0059] Furthermore, by using a pre-defined substitution value assessment model and pre-defined assessment dimensions, the substitution value score corresponding to each candidate parking space sensing device in the second parking space sensing device group is determined. The pre-defined assessment dimensions include at least one of the following: the matching degree between the performance feature vector of the candidate parking space sensing device and the task requirement vector of the faulty reference parking space sensing device; the load change trend of the candidate parking space sensing device; and the network path between the candidate parking space sensing device and the faulty reference parking space sensing device. Based on the substitution value scores, the candidate parking space sensing devices are ranked, establishing a substitution value spectrum. According to the substitution value spectrum and the urgency parameter of the sensing task, a spectrum truncation threshold is determined, and candidate devices with scores greater than the spectrum truncation threshold are selected to generate a preferred device set. Verification commands are sent to each device in the preferred device set to perform collaborative readiness verification. Based on the collaborative readiness verification results, the device with the best response quality and the shortest stabilization time is selected from the preferred device set as the substitution parking space sensing device.

[0060] Specifically, a multi-dimensional quantitative evaluation of each candidate parking space sensing device in the second parking space sensing device group is conducted using a pre-defined alternative value assessment model. Specifically, the performance feature vector matching degree between the candidate device and the faulty reference device is calculated. This matching degree is obtained by comparing the cosine similarity of the two devices in features such as sensor type, accuracy, resolution, field of view, and algorithm support capabilities. Simultaneously, based on the historical load data sequence of the candidate devices, their load change trends are calculated using first-order differencing to predict their future load trends. Furthermore, the historical data transmission records of the network path between the candidate device and the faulty device are analyzed to assess path stability and uncertainty. The quantitative results of the above dimensions are then integrated using a pre-defined weighted aggregation function to finally calculate the alternative value score corresponding to each candidate device.

[0061] Based on the calculated substitution value scores of each candidate parking space sensing device, all candidate devices are sorted in descending order. The sorting results form a sequence, which is the substitution value spectrum, intuitively reflecting the priority of each candidate device's ability to replace a faulty device. According to this sequence, devices are selected sequentially in the second parking space sensing device group until the comprehensive sensing range corresponding to the selected device covers the sensing range corresponding to the faulty reference parking space sensing device, generating a preferred device set.

[0062] A standardized verification command is sent to each device in the preferred set of devices. This command contains a lightweight simulated task, similar in function and resource consumption to the original task of the faulty device but on a scaled-down version. The system synchronously monitors and records the response process of each device to the verification command, collecting multiple metrics including response latency, task execution success rate, resource consumption compliance, and preliminary quality of the returned data. Using a pre-defined verification evaluation algorithm, these metrics are combined to generate a collaborative readiness verification result for each device. This result reflects the device's actual ability and status to perform the replacement task in a real-time environment.

[0063] Based on the collaborative readiness verification results, the final replacement parking space sensing device is selected from the preferred device set. First, the response quality of each device is evaluated based on indicators such as response data quality and task execution success rate from the verification results. Second, the stabilization time is evaluated based on the time it takes for the device to output a stable and reliable result after receiving a command. Finally, a comprehensive selection function is constructed, which assigns a positive weight to response quality and a negative weight to stabilization time. The comprehensive score of each device in the preferred set under this function is calculated. If the score meets the preset screening criteria, the verification passes; if the score does not meet the preset screening criteria, device selection continues from the second parking space sensing device group until the selected device can cover the sensing range corresponding to the faulty sensing device, and the device's response time and stability meet the requirements.

[0064] Figure 3 This is a schematic diagram of the structure of a dynamic collaborative parking space sensing device based on the open-source HarmonyOS, provided as an embodiment of this application. Figure 3 As shown, a dynamic collaborative parking space sensing device 200 based on the open-source HarmonyOS includes: at least one processor 201; and a memory 202 communicatively connected to the at least one processor 201. The memory 202 stores instructions executable by the at least one processor 201, which, when executed, enable the at least one processor 201 to: dynamically network multiple parking space sensing devices within a parking lot via the distributed soft bus of the open-source HarmonyOS, constructing a parking space sensing resource pool. The parking space sensing devices include at least cameras, geomagnetic sensors, and Bluetooth beacons within the parking lot. One of the following steps is to acquire abnormal information within the current parking lot to determine the target perception range based on the abnormal information; wherein, the abnormal information includes at least one of the abnormal information of the basic parking space sensing equipment and abnormal information of parking behavior; based on the target perception range, multiple reference parking space sensing devices are selected from the parking space sensing resource pool to form a collaborative sensing network to collect collaborative parking space sensing data for the target perception range; the collected collaborative parking space sensing data is fused and analyzed, and the weight of each parking space sensing data in the decision-making is dynamically adjusted based on the real-time status of each data source to generate collaborative parking space sensing information corresponding to the target perception range.

[0065] This application provides a non-volatile computer storage medium storing computer-executable instructions. These instructions are configured to: dynamically network multiple parking space sensing devices within a parking lot using the distributed soft bus of the open-source HarmonyOS to construct a parking space sensing resource pool; wherein the parking space sensing devices include at least one of the following: a camera, a geomagnetic sensor, and a Bluetooth beacon; acquire abnormal information within the current parking lot to determine the target sensing range based on the abnormal information; wherein the abnormal information includes at least one of the following: abnormal information from basic parking space sensing devices and abnormal parking behavior information; based on the target sensing range, select multiple reference parking space sensing devices from the parking space sensing resource pool to form a collaborative sensing network for collaborative parking space sensing data collection within the target sensing range; perform fusion analysis on the collected collaborative parking space sensing data, and dynamically adjust the weight of each parking space sensing data in the decision-making process based on the real-time status of each data source to generate collaborative parking space sensing information corresponding to the target sensing range.

[0066] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0067] The above descriptions are merely embodiments of this application and are not intended to limit the scope of this application. For those skilled in the art, various modifications and variations can be made to the embodiments of this application. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions in the embodiments of this application.

Claims

1. A dynamic collaborative parking space perception method based on open-source HarmonyOS, characterized in that, The method includes: By using the distributed soft bus of open-source HarmonyOS, multiple parking space sensing devices in the parking lot are dynamically networked to build a parking space sensing resource pool; wherein, the parking space sensing devices include at least one of the following: camera equipment, geomagnetic sensor and Bluetooth beacon in the parking lot; Obtain abnormal information within the current parking lot to determine the target sensing range based on the abnormal information; wherein, the abnormal information includes at least one of the abnormal information of the basic parking space sensing device and abnormal information of parking behavior; Based on the target sensing range, multiple reference parking space sensing devices are selected from the parking space sensing resource pool to form a collaborative sensing network, so as to collect collaborative parking space sensing data within the target sensing range. The collected collaborative parking space perception data is fused and analyzed, and the weight of each parking space perception data in the decision-making is dynamically adjusted based on the real-time status of each data source, so as to generate collaborative parking space perception information corresponding to the target perception range.

2. The dynamic collaborative parking space perception method based on open-source HarmonyOS according to claim 1, characterized in that, The process involves dynamically networking multiple parking space sensing devices within a parking lot using the open-source HarmonyOS distributed soft bus to construct a parking space sensing resource pool. Specifically, this includes: Through the distributed soft bus of the open-source HarmonyOS, multiple parking space sensing devices that support the open-source HarmonyOS protocol in the parking lot are automatically acquired, and the discovered multiple parking space sensing devices are dynamically networked to build a device network. The heterogeneous sensing capabilities of each parking space sensing device in the device network are encapsulated into a sensing service instance with a unified calling format; wherein, the sensing service instance is decoupled from the native interface of the parking space sensing device. Each of the encapsulated perception service instances and their corresponding device metadata are registered in a metadata database maintained by a distributed data management service; wherein the device metadata includes at least one of the following: device location, device capability type, and device real-time status. The registered perception service instances are aggregated to construct the parking space perception resource pool, and a unified service call interface decoupled from the parking space perception device is provided for upper-layer perception applications.

3. The dynamic collaborative parking space perception method based on open-source HarmonyOS according to claim 1, characterized in that, The step of acquiring abnormal information within the current parking lot, and determining the target perception range based on the abnormal information, specifically includes: If the abnormal information is abnormal information of the basic parking space sensing device, the target sensing range is determined based on the location information of the basic parking space sensing device and the sensing range corresponding to the basic parking space sensing device. When the abnormal information is the parking behavior abnormal information, a vehicle outline rectangle is generated based on the real-time location coordinates of the target vehicle. The spatial geometric relationship between the vehicle outline rectangle and the boundary coordinates of the vacant parking spaces in the parking lot digital map is calculated, and the target perception range corresponding to the target vehicle is determined based on the calculation results.

4. The dynamic collaborative parking space perception method based on open-source HarmonyOS according to claim 1, characterized in that, Based on the target sensing range, multiple reference parking space sensing devices are selected from the parking space sensing resource pool to form a cooperative sensing network, specifically including: Based on the target sensing range, and the location information and device functional attribute information of each parking space sensing device, the degree of overlap between the sensing coverage range of each parking space sensing device and the target sensing range is determined. Obtain the operating status information corresponding to each of the parking space sensing devices; wherein, the operating status information is related to the load status and operating health status of the parking space sensing devices; The operating status information corresponding to each parking space sensing device is weighted and processed to determine the optimal score of each parking space sensing device; Based on the preferred score, the parking space sensing devices are sorted, and reference parking space sensing devices are selected in sequence according to the sorting order to construct a reference device set; Based on the overlap between each of the reference parking space sensing devices and the target sensing range, the sensing coverage blind zone of the reference device set for the target sensing range is determined. If the area of ​​the sensing coverage blind zone is less than a preset blind zone area threshold, the cooperative sensing network is constructed based on the reference device set.

5. The dynamic collaborative parking space perception method based on open-source HarmonyOS according to claim 1, characterized in that, The process of fusing and analyzing the collected parking space perception data, and dynamically adjusting the weight of each parking space perception data point in the decision-making process based on the real-time status of each data source, to generate collaborative parking space perception information corresponding to the target perception range, specifically includes: Spatiotemporal alignment of multi-source parking space sensing data; wherein, the multi-source parking space sensing data includes at least one of image data, geomagnetic data, and Bluetooth signals; The weights of the multi-source parking space sensing data are dynamically adjusted based on the real-time status and data quality of each data source; wherein, the real-time status refers to the operating status of each parking space sensing device that acquires the parking space sensing data, and the data quality refers to the accuracy of the parking space sensing data. Based on the synthesis rules of evidence theory, conflict resolution and decision synthesis are performed on the multi-source parking space perception data after dynamic weight adjustment to generate collaborative parking space perception information corresponding to the target perception range.

6. The dynamic collaborative parking space perception method based on open-source HarmonyOS according to claim 5, characterized in that, The synthesis rule based on evidence theory performs conflict resolution and decision synthesis on the multi-source parking space perception data after dynamic weight adjustment to generate collaborative parking space perception information corresponding to the target perception range, specifically including: The multi-source parking space sensing data from different parking space sensing devices are transformed into a basic probability allocation function in evidence theory to construct parking space status evidence bodies corresponding to each data source. Based on the distance and orientation similarity of each parking space status evidence in the vector space, the conflict degree factor between the evidence is determined. When a conflict is detected between pieces of evidence, the evidence of the parking space status is fused based on the conflict degree factor and the preset evidence conflict synthesis strategy to obtain a comprehensive support score for multiple reference parking space statuses. Based on the comprehensive support score, the collaborative parking space perception information corresponding to the target perception range is determined, and based on the distribution characteristics of the comprehensive support score, the confidence level corresponding to the state information is determined, so that if the confidence level is greater than a preset threshold, the collaborative parking space perception information corresponding to the target perception range is determined; wherein, the ratio between the distribution characteristics and the comprehensive support level is related to the information entropy corresponding to the comprehensive support level.

7. The dynamic collaborative parking space perception method based on open-source HarmonyOS according to claim 1, characterized in that, After selecting multiple reference parking space sensing devices from the parking space sensing resource pool to form a collaborative sensing network, and then collecting collaborative parking space sensing data for the target sensing range, the method further includes: The reference parking space sensing device is monitored in real time; If a fault is detected in the reference parking space sensing device, the optimization target corresponding to the screening of the reference parking space sensing device will be transformed so that the alternative parking space sensing device corresponding to the faulty reference parking space sensing device can be screened in the parking space sensing resource pool through the transformed optimization target. The alternative parking space sensing devices are combined and adjusted so that the sensing range of the combined alternative parking space sensing devices covers the sensing range of the fault reference device.

8. The dynamic collaborative parking space perception method based on open-source HarmonyOS according to claim 7, characterized in that, The alternative parking space sensing devices corresponding to the faulty reference parking space sensing devices are selected from the parking space sensing resource pool, specifically including: In the parking space sensing resource pool, a first group of parking space sensing devices with the same sensing capability as the fault reference parking space sensing device is selected. In the first parking space sensing device group, a second parking space sensing device group that overlaps with the sensing range of the fault reference parking space sensing device is selected. In the second parking space sensing device group, the substitution value score of each sensing device is determined based on the real-time load index of each sensing device and the communication cost in the network topology. Based on the substitution value score, devices are selected sequentially in the second parking space sensing device group until the comprehensive sensing range corresponding to the selected device covers the sensing range corresponding to the fault reference parking space sensing device, thereby generating a preferred device set. The preferred set of devices is subjected to collaborative readiness verification, and the devices that pass the verification are used as the alternative parking space sensing devices.

9. A dynamic collaborative parking space sensing device based on the open-source HarmonyOS, characterized in that, The device includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to perform the method described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are capable of performing the method described in any one of claims 1-8.

Images