A cloud-edge task cooperative lock control method and system for a shared parking space scene
By using an edge gateway to perform multi-source signal fusion and state compensation, the control delay and state misjudgment problems of the shared parking space lock system in unstable network environments are solved, achieving low-latency security control and state consistency recovery, thus improving the reliability of shared parking space management and user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JIANGSU WUJIE INTELLIGENT TECH CO LTD
- Filing Date
- 2026-05-15
- Publication Date
- 2026-06-12
AI Technical Summary
Existing shared parking space lock systems suffer from latency fluctuations in cellular networks and untimely control in weak network environments, leading to increased collision risks. Furthermore, in the event of a network outage, status misjudgments and billing anomalies affect user experience and parking space turnover efficiency.
By performing multi-source signal fusion and service state differential calculation through the edge gateway, the near-field transient kinematic coupling vector of the parking space and the drift compensation matrix of the cloud-edge asynchronous state machine are generated to realize local anti-collision control and state consistency restoration. Combined with the extended Kalman filter and non-blocking communication mechanism, the timeliness and accuracy of control commands are ensured.
Low-latency security control was achieved in weak network or outage environments, noise interference was reduced, the reliability of parking space status identification and the consistency of system status recovery were improved, and the security and management efficiency of shared parking spaces were enhanced.
Smart Images
Figure CN122204901A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent parking lock management technology, and in particular to a cloud-edge task collaborative parking lock control method and system for shared parking space scenarios. Background Technology
[0002] With the continuous growth of urban motor vehicle ownership, the shortage of parking resources in residential communities, commercial complexes, and office parks is becoming increasingly prominent. To improve the utilization rate of existing parking spaces, the shared parking space operation model is gradually becoming popular. This involves dynamically opening up idle parking spaces to other users through online reservations, time-sharing rentals, and remote authorization. In this scenario, parking space locks, as important execution terminals for parking space occupancy management and vehicle access control, typically need to be linked with a cloud management platform to realize functions such as reservation order management, user identity verification, remote lock control, billing and settlement, and parking space status synchronization.
[0003] In existing technologies, most shared parking space lock systems adopt a centralized cloud control architecture. This means that the parking space lock terminal uploads the on-site status to a cloud server, and the cloud then generates control commands based on business rules and sends them to the terminal for execution. While this solution facilitates unified management, it still has the following shortcomings in actual shared parking space deployment environments: 1. Existing systems typically rely on the cloud for control decisions. Safety-related actions such as a vehicle entering a parking space and the raising or lowering of the parking lock arm require transmission over the public network. When there are latency fluctuations, congestion, or jitter in the cellular network, control commands may not be issued in a timely manner, which can easily lead to the parking lock failing to lower in time when the vehicle approaches the parking space, or the vehicle not coming to a complete stop when the parking lock is raised, thereby increasing the risk of collision.
[0004] 2. Shared parking spaces are often deployed in areas with complex communication conditions, such as underground garages, older communities, and gated parks, where public network signals are easily blocked or attenuated. Existing technologies often cannot receive cloud commands properly under weak or offline network conditions, causing parking locks to be unable to respond promptly to business needs such as scheduled unlocking and status switching, affecting user experience and parking space turnover efficiency.
[0005] 3. Some existing solutions rely solely on information from a single sensor (such as magnetic sensing, limit switches, or simple occupancy detection) to determine the parking space occupancy status. This is easily affected by factors such as environmental noise, short-term vehicle stops, and mechanical errors of the equipment, leading to misjudgments of whether the parking space is empty or occupied, and consequently causing incorrect control actions.
[0006] 4. During the network outage, the on-site parking lock terminal may have already performed local actions, while the cloud platform remains in the old state. After the network is restored, the existing system usually only performs simple state overwrite or re-synchronization, lacking a systematic compensation mechanism for action sequences, time factors and state deviations during the network outage, which can easily cause order status errors, billing anomalies or parking space scheduling conflicts.
[0007] Therefore, there is an urgent need for a shared parking space ground lock control method and system that can take into account real-time on-site control, safety response capabilities, and cloud service consistency, in order to solve problems such as high control latency, poor adaptability to weak networks, and cloud-edge state mismatch in existing technologies. Summary of the Invention
[0008] The purpose of this invention is to provide a cloud-edge task collaborative ground lock control method and system for shared parking space scenarios, so as to solve the problems mentioned in the background art.
[0009] To achieve the above objectives, this invention proposes a cloud-edge task collaborative ground lock control method for shared parking space scenarios, comprising the following steps: The system acquires multi-source heterogeneous raw data through edge gateways and parking space lock terminals, converts it into discrete digital signals, and outputs structured data containing parking space perception information, terminal status information, network quality information, and cloud service information. Based on the structured data, the on-site multi-source signal fusion branch and the service state differential calculation branch are executed in parallel through the edge gateway, respectively outputting the parking space near-field transient kinematic coupling vector to characterize the actual state of the parking space, and the cloud-edge asynchronous state machine drift compensation matrix to quantify the degree of cloud-edge state splitting. In a weak network or network outage environment, the edge gateway performs closed-loop control analysis based on the near-field transient kinematic coupling vector of the parking space. When it determines that there is a collision risk, it triggers the local anti-collision safety strategy and sends the edge field control command to the parking space ground lock terminal. After the network is restored, the edge gateway reports the drift compensation matrix of the cloud-edge asynchronous state machine to the cloud service platform, which then performs conflict coverage or incremental merging on the cloud global state machine to achieve global convergence of the system state.
[0010] As a preferred embodiment, the structured data includes: Vehicle microwave radar echo sequence is used to characterize the transient probability distribution and relative radial velocity of the parking space being occupied. The Hall encoder pulse of the ground lock motor is used to characterize the displacement and speed of the ground lock stop arm; Radio frequency link received signal strength indicator, used to characterize the degree of fading and packet loss probability of the communication link; The cloud-based reservation order status vector and global parking space occupancy heatmap matrix are used to represent the global business logic and space topology occupancy probability.
[0011] As a preferred embodiment, the processing of the on-site multi-source signal fusion branch includes: using an extended Kalman filter to perform multi-modal fusion of the vehicle microwave radar echo sequence and the Hall encoder pulse of the ground lock motor, filtering out noise, extracting the vehicle's true occupancy probability and radial micro velocity, and then vector-splitting them with the ground lock mechanical state variables to generate the near-field transient kinematic coupling vector of the parking space.
[0012] As a preferred embodiment, the processing of the service state differential calculation branch includes: real-time monitoring of network quality information; when it is determined that the network is in a weak network or outage state, extracting the locally cached cloud service information and combining it with the local on-site state change sequence that occurred during the outage, performing spatiotemporal differential calculation, and generating the cloud-edge asynchronous state machine drift compensation matrix.
[0013] As a preferred embodiment, the cloud-edge asynchronous state machine drift compensation matrix includes: the duration of network outage, the cumulative difference in cloud-edge state during the outage, the time weight coefficient, and the state weight coefficient.
[0014] As a preferred embodiment, the determination condition for triggering the local anti-collision safety policy is as follows: When the vehicle's actual occupancy probability, the vehicle's radial micro velocity, and the ground lock arm angular displacement in the near-field transient kinematic coupling vector of the parking space all exceed the preset corresponding thresholds, it is determined that the vehicle is approaching the parking space and has not yet come to a complete stop, and a ground lock control command is directly generated and issued.
[0015] As a preferred embodiment, the processing rule for achieving global convergence of the system state is as follows: When the state weight coefficient in the cloud-edge asynchronous state machine drift compensation matrix is greater than the preset value, conflict coverage is performed based on the action sequence reported by the edge side. When the state weight coefficient is less than or equal to a preset value, incremental merging is performed; By minimizing the Frobenius norm of the drift compensation matrix, the state differences between the cloud and the edge are eliminated.
[0016] As a preferred embodiment, the cloud-edge task collaboration follows latency constraints: ; in, For edge-side control link delay, To control link latency in the cloud. To prevent the maximum permissible delay for collision control, It is calculated based on the maximum radial speed of the vehicle entering the parking space and the effective travel of the ground lock arm.
[0017] Furthermore, this invention also discloses a shared parking space management system, which is used to execute the aforementioned cloud-edge task collaborative ground lock control method for shared parking space scenarios. Specifically, the control system includes: The cloud service platform is used to deploy global business scheduling and billing management tasks; An edge gateway, deployed in the parking area, connects to the cloud service platform via a cellular mobile communication network to perform real-time safety control tasks and state drift compensation in network outage scenarios; The parking space lock terminal and sensing unit are connected to the edge gateway via a local area network to collect raw data from the site and execute control commands.
[0018] As a preferred embodiment, the edge gateway uses an asynchronous non-blocking mechanism to monitor the downlink in the cloud and caches the parsed global business data in a built-in non-volatile memory.
[0019] Compared with the prior art, the present invention has the following beneficial effects: (1) Deploy global business tasks such as appointment scheduling and billing management in the cloud, and push high real-time tasks such as anti-collision control and on-site status analysis to the edge gateway for execution, so that different tasks are processed on suitable computing nodes, taking into account both centralized management capabilities and on-site response efficiency.
[0020] (2) In the case of weak network or network outage, the edge gateway performs local closed-loop control based on the near-field transient kinematic coupling vector of the parking space. When a vehicle is detected approaching and there is a risk of interference with the parking lock, the edge gateway can directly issue on-site control commands without waiting for cloud decision-making, thereby shortening the control link delay and improving the timeliness of safety control.
[0021] (3) By performing multi-source fusion processing on the vehicle microwave radar echo sequence and the Hall encoder pulse of the ground lock motor, and combining the filtering algorithm to extract the vehicle's true occupancy probability, vehicle radial speed and ground lock mechanical state, compared with a single sensing method, noise interference can be reduced and the reliability of vehicle occupancy and motion state recognition can be improved.
[0022] (4) The edge side can cache cloud service information and continuously execute field control logic in combination with local state change sequence during network outage, so that the system still has basic business operation capabilities when the network is abnormal, reducing service interruption caused by communication interruption.
[0023] (5) By constructing a cloud-edge asynchronous state machine drift compensation matrix, the duration of network outage, cumulative state difference and weight parameters are quantified, and conflict coverage or incremental merging is performed accordingly, which helps to more accurately correct the state deviation generated during the network outage and improve the consistency of cloud-edge state recovery.
[0024] (6) After the cloud-based global state machine completes the compensation convergence, it can more accurately reflect the real-time occupancy status of parking spaces, order execution status and billing information, thereby improving the reliability of shared parking space scheduling management and commercial operation.
[0025] (7) The proposed solution can be implemented based on edge gateways, parking space lock terminals and existing cellular communication networks. It does not require complete reliance on high-quality public network connections and is applicable to various shared parking space scenarios such as underground garages, residential communities and park parking lots. It has good promotion and application value. Attached Figure Description
[0026] Figure 1 This is a flowchart of a cloud-edge task collaborative ground lock control method for a shared parking space scenario, as shown in the embodiment.
[0027] Figure 2 This is a structural block diagram of the shared parking space management system in the embodiment.
[0028] Figure 3 This is a timing diagram of cloud-edge collaboration in this invention. Detailed Implementation
[0029] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.
[0030] See Figure 1 This embodiment discloses a cloud-edge task collaborative parking lock control method for shared parking space scenarios, applied to a shared parking space management system including a cloud service platform, edge gateway, parking lock terminal, and microwave radar sensing unit (see...). Figure 2 As shown in the figure, the edge gateway is deployed on-site in the parking area and is connected to multiple parking space lock terminals and microwave radar sensing units in the same parking area through local area network communication. The edge gateway establishes a communication link with the cloud service platform through the cellular mobile communication network. The parking space lock terminal has a built-in microcontroller MCU, motor drive module and Hall encoder. The microwave radar sensing unit adopts a 24GHz frequency band frequency modulated continuous wave radar to collect target echo signals in the parking space.
[0031] The core idea of this method is to decompose tasks between the cloud and the edge, delegating high-real-time security control tasks to the edge for execution, while deploying global business scheduling and billing management tasks to the cloud. Simultaneously, a state drift compensation mechanism is established for network outage scenarios to achieve a balance between control real-time performance and business consistency. The cloud-edge collaborative timing diagram of this invention is shown below. Figure 3 As shown.
[0032] First, we derive the constraints for task decomposition, assuming the end-to-end latency of the cloud control link is... The end-to-end latency of the edge control link is The maximum allowable delay for the anti-collision control of the ground lock is Then the constraints for task splitting are: ,in The preferred value is 50ms. This value is calculated based on the maximum radial velocity of the vehicle entering the parking space and the effective travel of the parking lock arm. The preferred maximum radial velocity of the vehicle entering the parking space is 5km / h, which is equivalent to 1.39m / s. The effective travel of the parking lock arm is 0.15m. The shortest time for the vehicle to collide with the parking lock arm is... Therefore, take This serves as an upper limit for control latency, ensuring the effectiveness of collision avoidance control.
[0033] The implementation process of this method consists of four processing steps: The first stage involves acquiring multi-source heterogeneous raw data through the sensing and communication modules of the edge gateway and the parking lock terminal, providing input for subsequent processing. The raw signals input in this stage include simulated electromagnetic wave reflection signals from the parking space, simulated magnetic field change signals, radio frequency signals from the communication base station, and service data packets sent from the cloud service platform. The simulated electromagnetic wave reflection signals are generated from the echo signals emitted and received by the microwave radar sensing unit; the simulated magnetic field change signals are generated from the induced electromotive force signal generated by the Hall encoder of the parking lock motor; the radio frequency signals from the communication base station are received by the cellular communication module of the edge gateway; and the cloud service data packets are sent by the cloud service platform via the TCP / IP protocol. Subsequently, the input continuous analog signal is converted into a discrete digital signal via the analog-to-digital converter (ADC) module built into the edge gateway and the ground lock MCU. The preferred sampling bit depth of the ADC module is 12 bits, and the preferred sampling frequency is 1kHz. Simultaneously, the edge gateway's cellular communication module obtains the current network channel fading level through RF baseband analysis. During the network connectivity window, the edge gateway employs an asynchronous non-blocking mechanism, using an epoll model to monitor the downlink communication link of the cloud service platform, parsing and caching the global service data sent from the cloud. The cached storage medium uses the edge gateway's built-in non-volatile memory, and the preferred update period for the cached data is 100ms. This stage outputs multiple sets of structured data: the first set is the vehicle microwave radar echo sequence. This parameter is of floating-point type and has a dimension of ,in The number of receiving antenna channels for the microwave radar is preferably 4. This represents the number of sampling points within a single sampling period, with a preferred value of 256. The first group is the sampling timestamp. Each element of this parameter represents the echo signal amplitude of the corresponding sampling time and the corresponding receiving channel. After processing with Fast Fourier Transform, the distance and radial velocity of the target in the parking space can be obtained, thus representing the transient probability distribution and relative radial velocity of the parking space being occupied. The second group is the Hall encoder pulse of the ground lock motor. This parameter is an unsigned integer type with a dimension of . , is a scalar parameter. The sampling timestamp parameter represents the number of pulses output by the Hall encoder of the ground lock motor within a unit sampling period. The preferred value for the number of pulses output per revolution of the motor is 12, and the preferred reduction ratio of the ground lock motor is 1:180. Therefore, the rotation angle of the motor output shaft can be calculated using this parameter, which in turn characterizes the three-dimensional angular displacement and instantaneous angular velocity of the ground lock stop arm in vertical space. The formula for calculating the angular displacement is as follows: ,in for The formula for calculating the instantaneous angular displacement and instantaneous angular velocity of the locking arm at any given moment is: ,in The sampling period is 1ms, which is preferred; the third group is the RF link received signal strength indication. This parameter is of floating-point type and has a dimension of , is a scalar parameter. This is the sampling timestamp, measured in dBm, with a range of -120dBm to 0dBm. It characterizes the channel fading level and packet loss probability of the current cellular communication link. The values are negatively correlated, and the probability of packet loss in the communication link increases with... The value decreases and then increases; the fourth group is the cloud-based reservation order status vector. This parameter is a composite structure type with dimensions of [missing information]. ,in The number of business status dimensions is 6, with each dimension corresponding to a unique user identifier, reservation start time stamp, reservation end time stamp, unique parking space identifier, order payment status, and order execution status. This represents a discrete set of business logic dimensions. The fifth group is the global parking space occupancy heatmap matrix. This parameter is of floating-point type and has a dimension of ,in This refers to the number of rows of parking spaces within the parking area. The matrix represents the number of parking spaces within the parking area. Each element in the matrix has a value ranging from 0 to 1, representing the probability of the corresponding parking space node being occupied at the current moment. A value of 0 indicates that the parking space is completely vacant, while a value of 1 indicates that the parking space is completely occupied. The overall matrix represents the spatial topology occupancy probability of each parking space node within a specific parking area.
[0034] The second stage is the core computing power processing stage of the method. Through the computing unit of the edge gateway, the multi-source heterogeneous signals output from the first stage are processed separately to obtain the corresponding output parameters. The input parameter of this stage is the vehicle microwave radar echo sequence output from the first stage. , floor lock motor Hall encoder pulse RF link received signal strength indication Cloud-based reservation order status vector and global parking space occupancy heatmap matrix The parameter transformation and combination process is divided into two parallel branches. The first branch is the on-site multi-source signal fusion process, in which a lightweight extended Kalman filter (EKF) is called at the edge side to... and To perform multimodal fusion, the system's state equation and observation equation are first established. The state equation is: ,in for The system state vector at any given time contains four state variables: vehicle distance, radial velocity, angular displacement of the ground lock arm, and angular velocity. The state transition matrix has dimension 1. , To control the input matrix, the dimension is... , To control the input amount, Let be the process noise vector, and let be the covariance matrix of the process noise. The preferred value is a diagonal matrix, with the diagonal elements being: , , , The observation equation is ,in for The observation vector at time t is given by The processed vehicle distance, radial velocity, and The processed angular displacement and angular velocity are composed of... The observation matrix has dimensions of . , The observation noise vector, and the covariance matrix of the observation noise. The preferred value is a diagonal matrix, with the diagonal elements being: , , , By using the prediction and update steps of the extended Kalman filter, environmental noise and measurement noise are filtered out, and the true occupancy probability and radial micro velocity of the vehicle are extracted. The formula for calculating the true occupancy probability of the vehicle is as follows: ,in This is a reference value for parking space depth, with 5m being the preferred value. for The vehicle distance output by the time-of-flight filter. The standard deviation of the distance measurement is preferably set to 0.2m. The obtained vehicle occupancy probability, radial micro velocity, and ground lock mechanical angular displacement and angular velocity are vector-concatenated to complete the on-site multi-source signal fusion process. The second branch is the service status differential calculation process, which is monitored in real time by the edge gateway. The value, the pre-set weak network threshold , The preferred value is -95dBm, when If the network is currently weak or offline, the edge gateway freezes its long-lived connection to the cloud, stops receiving real-time business data from the cloud, and then retrieves the cached data from its local non-volatile storage. and By combining the local state changes that occurred during the network outage, spatiotemporal difference calculations are performed to quantify the degree of cloud-edge state fragmentation. The start time of the network outage is denoted as... The current time is recorded as The network outage time range is During this time interval, the sequence of ground lock status changes recorded on the edge side is as follows: The expected state sequence sent by the cloud before the network outage is as follows: The spatiotemporal difference calculation process involves aligning the two state sequences along their time dimensions, with a preferred alignment time step of 100ms. Then, for each aligned time step, the state difference is calculated. This step outputs two sets of parameters: the first set is the near-field transient kinematic coupling vector of the parking space. This parameter is of floating-point type and has a dimension of , At the current moment, the three elements of the vector represent the actual occupancy probability of the vehicle, in order. Vehicle radial micro velocity Angular displacement of the ground lock arm The first vector represents the fused local state parameters, independent of the cloud business logic, and represents the actual state of the parking space obtained from the edge side; the second group is the cloud-edge asynchronous state machine drift compensation matrix. This parameter is a composite floating-point matrix with dimensions of . , The element in the first row and first column of the matrix at the current moment. The duration of the internet outage, in milliseconds, is represented by the element in the first row and second column. This represents the cumulative difference in cloud-edge status during the network outage, ranging from 0 to 1. (Element in the second row, first column) This is the time weighting coefficient, with a value of [value missing]. ,in The preset maximum network outage compensation duration is preferably 3600000ms, and the element in the second row and second column is... These are the state weight coefficients, with values ranging from 1 to 2. This matrix quantifies the difference between the expected state in the cloud and the execution state at the edge during the network outage, providing a compensation reference for state synchronization after the network is restored.
[0035] The third stage is executed in a weak network or network outage environment. It achieves low-latency closed-loop control through the local computing unit of the edge gateway. The input parameter for this stage is the near-field transient kinematic coupling vector of the parking space output by the second stage. The parameter transformation and combination process is as follows: the nonlinear control law decoder inside the edge gateway... Perform analysis to extract the vehicle's radial micro-velocities from the vector. Vehicle actual occupancy probability and the angular displacement of the ground lock arm Pre-set occupancy probability threshold The preferred value is 0.7, which is the speed threshold. The preferred value is 0.2 m / s, angular displacement threshold. The preferred value is 5°, when the following conditions are met. , and When the system determines that a vehicle is approaching a parking space and has not come to a complete stop, posing a risk of collision with the parking lock arm, and that the parking lock arm is not currently fully lowered, the edge gateway bypasses the cloud-based business logic and triggers the local anti-collision safety policy, generating corresponding control commands. The parameters output in this step are the edge-side control commands. This parameter is an integer enumeration type with dimensions of 1. , is a scalar parameter. This is the timestamp for the command issuance. The enumerated values of this parameter include 1, 2, and 3, where 1 corresponds to the ground lock arm raising action, 2 corresponds to the ground lock arm lowering action, and 3 corresponds to the ground lock motor emergency stop action. The edge gateway directly issues this command to the ground lock MCU via the local area network. After receiving the command, the ground lock MCU immediately executes the corresponding action, compressing the end-to-end delay of the control link to within 50ms, which meets the delay requirements of anti-collision control.
[0036] The fourth step is executed after network recovery. It is used to eliminate the state differences accumulated during the network outage and achieve global consistency of business logic. The input parameter for this step is the cloud-edge asynchronous state machine drift compensation matrix generated in the second step. The first stage continuously monitors the received signal strength indication of the radio frequency link. And the cloud-based global state machine currently being maintained in the cloud. ,in It is a composite structure located in a high-dimensional discrete space, containing global business information such as the real-time status of all parking spaces within the parking area, the execution status of all orders, and billing data. The parameter transformation and combination process involves continuous monitoring by the edge gateway. The value, preset network recovery security threshold , The preferred value is -85dBm, when And the duration exceeds the preset stable duration. , When the preferred value is 1000ms, it is determined that the current network has recovered to a stable connectivity state. The edge gateway re-establishes the communication link with the cloud service platform, including the sequence of all on-site actions during the network outage. The data is reported to the cloud service platform. After receiving the matrix, the cloud service platform analyzes the dynamic compensation weights in the matrix, i.e., the time weight coefficients. With state weight coefficients To handle state changes that occur during a network outage, the handling rule is as follows: when At that time, based on the on-site action sequence reported by the edge side, conflict overwrite is performed on the corresponding state in the cloud global state machine. At that time, the on-site action sequence reported by the edge side and the expected state sequence in the cloud are incrementally merged. The optimization objective set by the system is to minimize the Frobenius norm of the drift matrix, i.e. The formula for calculating the Frobenius norm is: , For matrix The Middle Line number The elements of the column, through this optimization process, gradually eliminate the state differences between the cloud and the edge. The output parameters of this step are the updated cloud global state machine. The updated It is fully synchronized with the on-site status at the edge, completing the global convergence of system status and closed-loop compensation of business logic, ensuring the accuracy of order billing and parking space scheduling.
[0037] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.
Claims
1. A cloud-edge task collaborative ground lock control method for shared parking space scenarios, characterized in that, Includes the following steps: The system acquires multi-source heterogeneous raw data through edge gateways and parking space lock terminals, converts it into discrete digital signals, and outputs structured data containing parking space perception information, terminal status information, network quality information, and cloud service information. Based on the structured data, the on-site multi-source signal fusion branch and the service state differential calculation branch are executed in parallel through the edge gateway, respectively outputting the parking space near-field transient kinematic coupling vector to characterize the actual state of the parking space, and the cloud-edge asynchronous state machine drift compensation matrix to quantify the degree of cloud-edge state splitting. In a weak network or network outage environment, the edge gateway performs closed-loop control analysis based on the near-field transient kinematic coupling vector of the parking space. When it determines that there is a collision risk, it triggers the local anti-collision safety strategy and sends the edge field control command to the parking space ground lock terminal. After the network is restored, the edge gateway reports the drift compensation matrix of the cloud-edge asynchronous state machine to the cloud service platform, which then performs conflict coverage or incremental merging on the cloud global state machine to achieve global convergence of the system state.
2. The cloud-edge task collaborative ground lock control method for shared parking space scenarios according to claim 1, characterized in that, The structured data includes: Vehicle microwave radar echo sequence is used to characterize the transient probability distribution and relative radial velocity of the parking space being occupied. The Hall encoder pulse of the ground lock motor is used to characterize the displacement and speed of the ground lock stop arm; Radio frequency link received signal strength indicator, used to characterize the degree of fading and packet loss probability of the communication link; The cloud-based reservation order status vector and global parking space occupancy heatmap matrix are used to represent the global business logic and space topology occupancy probability.
3. The cloud-edge task collaborative ground lock control method for shared parking space scenarios according to claim 2, characterized in that, The processing of the on-site multi-source signal fusion branch includes: using an extended Kalman filter, multimodal fusion of the vehicle microwave radar echo sequence and the Hall encoder pulse of the ground lock motor is performed. After filtering out noise, the vehicle's true occupancy probability and radial micro velocity are extracted and vector-concatenated with the ground lock mechanical state variables to generate the near-field transient kinematic coupling vector of the parking space.
4. The cloud-edge task collaborative ground lock control method for shared parking space scenarios according to claim 1, characterized in that, The processing steps of the business state differential calculation branch include: real-time monitoring of network quality information; when it is determined that the network is in a weak or out-of-network state, extracting the cloud business information cached locally, and combining it with the local on-site state change sequence that occurred during the out-of-network period to perform spatiotemporal differential calculation and generate the cloud-edge asynchronous state machine drift compensation matrix.
5. The cloud-edge task collaborative ground lock control method for shared parking space scenarios according to claim 1 or 4, characterized in that, The cloud-edge asynchronous state machine drift compensation matrix includes: the duration of network outage, the cumulative difference of cloud-edge states during the outage, the time weight coefficient, and the state weight coefficient.
6. The cloud-edge task collaborative ground lock control method for shared parking space scenarios according to claim 1 or 3, characterized in that, The conditions for triggering the local anti-collision safety policy are as follows: When the vehicle's actual occupancy probability, the vehicle's radial micro velocity, and the ground lock arm angular displacement in the near-field transient kinematic coupling vector of the parking space all exceed the preset corresponding thresholds, it is determined that the vehicle is approaching the parking space and has not yet come to a complete stop, and a ground lock control command is directly generated and issued.
7. The cloud-edge task collaborative ground lock control method for shared parking space scenarios according to claim 1 or 4, characterized in that, The processing rule for achieving global convergence of the system state is as follows: When the state weight coefficient in the cloud-edge asynchronous state machine drift compensation matrix is greater than the preset value, conflict coverage is performed based on the action sequence reported by the edge side. When the state weight coefficient is less than or equal to a preset value, incremental merging is performed; By minimizing the Frobenius norm of the drift compensation matrix, the state differences between the cloud and the edge are eliminated.
8. The cloud-edge task collaborative ground lock control method for shared parking space scenarios according to claim 1, characterized in that, The cloud-edge task collaboration follows latency constraints: ; in, For edge-side control link delay, To control link latency in the cloud. To prevent the maximum permissible delay for collision control, It is calculated based on the maximum radial speed of the vehicle entering the parking space and the effective travel of the ground lock arm.
9. A shared parking space management system, characterized in that, The control system is used to execute the cloud-edge task collaborative parking lock control method for shared parking space scenarios as described in any one of claims 1 to 8, the control system comprising: The cloud service platform is used to deploy global business scheduling and billing management tasks; An edge gateway, deployed in the parking area, connects to the cloud service platform via a cellular mobile communication network to perform real-time safety control tasks and state drift compensation in network outage scenarios; The parking space lock terminal and sensing unit are connected to the edge gateway via a local area network to collect raw data from the site and execute control commands.
10. The shared parking space management system according to claim 9, characterized in that, The edge gateway uses an asynchronous non-blocking mechanism to monitor the downlink in the cloud and caches the parsed global business data in its built-in non-volatile memory.