A parking lot real-time dynamic parking space navigation method and system

Abstract

The application discloses a kind of parking lot real-time dynamic parking space navigation method and system, and built-in map robot of parking lot plans route scanning parking lot layout, AI big model generates digital parking space map, generates only coding for parking space, marks entrance and equipment facility position;Inspection robot sets frequency scanning parking space idle state by area and confirms whether can park;Cloud end according to parking space idle rate dynamically adjusts inspection frequency and robot quantity, adaptive switching guide strategy;Vehicle enters after car machine end synchronous scanning parking lot of parking lot supplementary data, and parking space state is updated by cross verification;After parking, record parking space number and provide find vehicle navigation service;Special placeholder robot provides reservation and placeholder service;Parking space state and navigation data are opened to third-party map, travel and the like service application with standardized API / SDK by open platform. Realize parking lot parking space management from mapping, state detection, dynamic scheduling, intelligent guide to open service whole process intelligent.

Description

Technical Field

[0001] This invention relates to the field of smart parking and vehicle networking technology, specifically to a real-time dynamic parking space navigation method and system for parking lots. Background Technology

[0002] With the continuous growth of car ownership, the problems of parking difficulties and difficulty in finding cars in large parking lots (especially underground multi-story parking garages) are becoming increasingly prominent. Existing parking management systems mainly have the following shortcomings: 1) Map building relies on manual labor and updates are lagging: Traditional parking lot maps mostly rely on manual surveying or are gradually built based on vehicle SLAM technology, resulting in low mapping efficiency, high cost, and difficulty in responding to changes in parking lot layout.

[0003] 2) Limited and inaccurate parking space status detection methods: Existing parking space status detection mainly relies on geomagnetic induction, ultrasonic sensors, or roadside cameras, which suffer from blind spots, high equipment failure rates, and high maintenance costs. Some solutions use inspection robots for parking space status detection, but these are mostly single robots conducting global patrols, resulting in low efficiency and an inability to achieve high-frequency updates.

[0004] 3) Isolated data sources and lack of cross-validation: The existing solution has a single source of parking space status data and lacks a mechanism for cross-validation of multi-source data. When a single data source has a deviation or failure, the entire parking space status management system will fail.

[0005] 4) Fixed inspection strategy and lack of dynamic scheduling: The existing inspection robots perform fixed inspection frequencies regardless of whether the parking lot is busy or not. This results in energy waste when there are plenty of parking spaces and inability to capture parking space release information in a timely manner when parking spaces are scarce.

[0006] 5) Poor car location navigation experience: Existing car location solutions mostly rely on Bluetooth beacons or mobile phone positioning, which has limited accuracy and lacks deep integration with parking lot maps and parking space coding systems.

[0007] 6) Closed data and lack of open ecosystem: Most existing parking management systems are closed, and parking space status data cannot be output to the outside world, making it difficult to empower third-party applications such as map navigation and travel services. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention proposes a real-time dynamic parking space navigation method and system. It organically integrates technologies such as robot swarm mapping, AI-generated digital maps, dynamic scheduling based on idle rate thresholds, multi-source data cross-verification, unique parking space coding, intelligent car-finding navigation, parking space reservation robots, and an open platform. This achieves intelligent management of parking spaces throughout the entire process, from mapping, status detection, dynamic scheduling, intelligent guidance to open services. This addresses technical problems in existing technologies such as low efficiency in parking map construction, insufficient accuracy in parking space status detection, fixed inspection strategies lacking dynamic scheduling, lack of multi-source data cross-verification, poor car-finding navigation experience, and closed data preventing the empowerment of third-party applications.

[0009] Therefore, the technical solution of the present invention is a real-time dynamic parking space navigation method and system, characterized in that: the present invention adopts a system architecture of "robot cluster mapping + AI-generated digital map + dynamic scheduling based on idle rate threshold + cross-verification of multi-source data + unique parking space code + intelligent navigation + reservation of parking space robot + open platform". Includes the following steps: (1) Map building steps: At least one mapping robot in the parking lot scans the entire parking lot area according to the preset planned route and collects parking lot layout data; the cloud AI big model generates a digital parking space map based on the layout data and generates a unique code for each parking space in the map, while marking the parking lot entrance and exit and the location of equipment and facilities. (2) Parking space status inspection steps: Multiple inspection robots are deployed in different areas. Each inspection robot scans each parking space in its assigned area at a preset frequency, identifies the vacancy status of each parking space, and confirms whether parking is allowed in the parking space. (3) Dynamic scheduling and adaptive guidance steps: The cloud calculates the vacancy rate based on the ratio of the current number of vacant parking spaces to the total number of parking spaces in the parking lot; based on the threshold range of the vacancy rate, the inspection frequency of the inspection robot and the number of active robots are dynamically adjusted, and the vehicle entry guidance strategy is adaptively switched. (4) Multi-source data fusion and update steps: After the vehicle enters the parking lot, the vehicle terminal scans the visible parking spaces in the parking lot through the vehicle-mounted sensors, identifies the vacancy status of the parking spaces and uploads the data to the cloud; the cloud cross-verifies the robot inspection data with the vehicle terminal scanning data, makes a comprehensive judgment and updates the vacancy status of the parking spaces. (5) Parking guidance and recording steps: Based on the updated parking space availability status, push available parking space navigation information to the entering vehicle; after the vehicle parks in the target parking space, the system records the binding relationship between the vehicle and the parking space and the parking space number; (6) Reverse car search navigation steps: In response to the user's car search request, the target parking space location is determined according to the vehicle and parking space binding relationship, and a navigation path from the user's current location to the target parking space is generated in combination with the parking lot map, and real-time navigation guidance is provided; (7) Parking space reservation and occupancy steps: In response to the user's reservation request, the cloud locks the target parking space according to the reservation information; a dedicated occupancy robot is dispatched to the target parking space to perform the occupancy operation and maintains the occupancy status during the reservation time period; the occupancy robot is released after the user arrives and the vehicle is parked in the parking space.

[0010] Further improvements are made in the following steps: In step (1), the AI ​​large model generates a digital parking space map, which specifically includes: identifying lanes, parking space boundaries, pillars, walls and equipment facilities in the parking lot; converting the identification results into structured map data; and assigning a unique combination of area code, floor code and serial number code to each parking space.

[0011] Further improvements are made in the following: In step (2), the deployment of multiple inspection robots in different areas specifically includes: dividing the parking lot into multiple inspection areas according to the parking lot area and layout; allocating at least one inspection robot to each inspection area; and having each inspection robot perform scanning tasks according to a staggered schedule to avoid data conflicts caused by multiple robots scanning the same area at the same time.

[0012] Further improvements are made in the following: In step (3), the cloud dynamically adjusts the inspection frequency and the number of robots according to the threshold range of the idle rate, specifically including: (1) Set a first threshold and a second threshold; when the idle rate is greater than the first threshold, it is determined to be an idle state, the inspection frequency is reduced to 15-20 minutes / time, and the number of active inspection robots is reduced to 1 / 3 to 1 / 2 of the normal number; (2) When the idle rate is less than the second threshold, it is determined to be an idle tension state, the inspection frequency is increased to 2-3 minutes / time, and the number of active inspection robots is increased to all available robots; (3) When the idle rate is between the first threshold and the second threshold, maintain the default inspection frequency of 5-10 minutes / time and the default number of active robots.

[0013] In step (3), the adaptive switching of the vehicle entry guidance strategy specifically includes: (4) When there is ample parking space, implement the "fast and nearest allocation" strategy: based on the current location of the vehicle, match available parking spaces in the area closest to the vehicle, and prioritize guiding the vehicle to the available parking space closest to the entrance or with the shortest path. (5) When the parking space is in a state of high demand and low availability, the “intelligent recommendation matching” strategy is executed: a recommendation list containing multiple candidate parking spaces is pushed to the vehicle terminal, the destination information input by the user is obtained, the walking distance from each candidate parking space to the destination is calculated, the parking space with the shortest walking distance is recommended first, and the unique code and navigation path of the parking space are provided.

[0014] Further improvements are made in the following steps: In step (4), the cloud cross-verifies the robot inspection data with the vehicle terminal scanning data. Specifically, when the robot inspection data and the vehicle terminal scanning data are inconsistent in their judgment of the status of the same parking space, the cloud triggers a secondary confirmation mechanism. The secondary confirmation mechanism includes at least one of the following: retrieving the most recent historical status record of the parking space for comparison, sending an instant verification instruction to the inspection robot in the area, and obtaining the real-time image of the parking space from the fixed monitoring equipment in the parking lot for auxiliary judgment.

[0015] Further improvements are made in the following aspects: In step (5), the system records the binding relationship between the vehicle and the parking space, specifically including: obtaining the precise location of the vehicle after it has been parked in the target parking space through vehicle terminal positioning or parking lot positioning equipment; binding the unique identifier of the vehicle with the unique code of the target parking space and recording the parking start time; the binding relationship is stored in the cloud database for subsequent reverse vehicle search and billing.

[0016] Further improvements are made in the following: In step (6), generating a navigation path from the user's current location to the target parking space specifically includes: obtaining the real-time location of the user's terminal; using the user's real-time location as the starting point and the target parking space location as the ending point, and combining the road network data in the parking lot map, using a path planning algorithm to generate the optimal navigation path; updating the user's location in real time and dynamically adjusting the navigation guidance during the navigation process.

[0017] Further improvements are made in the following steps: In step (7), the dedicated parking space occupancy robot performs the parking space occupancy operation in the following ways: After receiving the dispatch instruction from the cloud, the parking space occupancy robot autonomously navigates and moves to the target parking space; it performs the parking space occupancy action within the target parking space area, and the parking space occupancy action includes at least one of the following: parking in the central area of ​​the parking space and activating the warning sign, deploying the parking space occupancy arm or parking space occupancy barrier to physically block other vehicles from entering; the parking space occupancy robot maintains the parking space occupancy status during the reserved time period until the user arrives or the reservation expires. In step (7), the user's paid reservation and parking space occupancy specifically includes: the user initiates a parking space reservation request through the vehicle terminal or mobile terminal, selects a reservation time period and completes online payment; the cloud locks the target parking space based on the payment confirmation information and dispatches a occupancy robot; if the user arrives and parks in the parking space within the reservation time period, the reservation fee is automatically converted into a parking fee deduction; if the user does not arrive within the reservation time period, the occupancy robot automatically withdraws after the reservation expires, and the reservation fee is handled according to the agreed rules.

[0018] Further improvements include: (1) a data aggregation layer, used to receive and store parking space status inspection data from the parking lot inspection robot, parking space scanning supplementary data from the vehicle terminal, and parking space image data from the parking lot fixed monitoring equipment; the data is fused and processed by the multi-source data cross-verification module to generate highly reliable parking space status data; (2) Data service layer, used to encapsulate the high-reliability parking space status data, parking lot digital map data, parking space unique code data and parking space navigation path data into standardized API interface services; the API interface services include at least: parking space status query interface, which supports querying parking space status by parking lot ID, area code, parking space unique code or geographical coordinate range, and the returned data includes parking space unique code, vacancy status, parking space geographical coordinates, parking space size information and update timestamp; parking space navigation path planning interface, which receives the starting point coordinates and the ending point parking space unique code, and returns the driving navigation path or walking navigation path from the starting point to the target parking space by combining the road network data in the parking lot digital map, including the path node coordinate sequence, turning guidance, distance and estimated travel time; and parking space information subscription interface, which supports real-time push subscription for status changes by parking lot, by area or by specific parking space, and pushes status change notifications to third-party applications through Webhook or message queue; (3) Developer management module, which provides third-party developers with functions such as registration, authentication, application creation and API key management; and performs metering and billing based on the call volume and service level of third-party applications; (4) An open SDK module is provided to provide a software development kit for multiple platforms, encapsulate the calling logic of the API interface services, and allow third-party developers to integrate it into map applications, travel applications, vehicle systems and smart terminal applications; (5) Service output layer, which is used to output standardized parking space status data and navigation services to third-party map service providers, travel service platforms, vehicle system and smart terminal applications. It supports real-time marking and display of parking space vacancy status on third-party maps, as well as driving navigation from the parking lot entrance to the target parking space and walking navigation from the user's current location to the target parking space.

[0019] Further improvements include: the metering and billing function includes: billing based on at least one of the following: the number of API calls from third-party applications, the amount of data subscriptions, and service level agreement requirements; supporting three modes: per-use billing, per-month billing, and tiered billing based on usage; the open platform provides a sandbox testing environment for third-party developers, and the sandbox environment returns simulated data for developers to debug applications without generating actual billing.

[0020] Further improvements include: the generation of the high-confidence parking space status data specifically includes: when the inspection robot data, vehicle-mounted data, and fixed parking lot monitoring data all agree on the status of the same parking space, the status is directly output and marked as high confidence; when the three are not completely consistent, a majority voting mechanism is used to determine the final status, and the sources of inconsistent data and voting results are recorded in the log for auditing; the high-confidence parking space status data is accompanied by a confidence label, which can be selectively used by third-party applications according to their own business needs.

[0021] An architecture for a parking space navigation system, characterized by comprising: (1) Mapping robot: used to traverse and scan the entire parking lot area according to the preset planned route and collect parking lot layout data; (2) Inspection robots: Multiple inspection robots are deployed in different areas to scan each parking space in their respective areas at a preset frequency and identify the vacancy status of the parking spaces; (3) Parking space reservation robot: used to move to the target parking space in response to the reservation instruction and perform the reservation operation; (4) Cloud AI large model platform: used to generate digital parking space maps based on the layout data collected by the mapping robot, generate a unique code for each parking space, dynamically adjust the inspection frequency and number of robots according to the parking space vacancy rate, cross-verify and merge the inspection data and vehicle terminal data, update the parking space status database, and process reservation requests and dispatch instructions. (5) Vehicle terminal: used to scan visible parking spaces in the parking lot and upload data after the vehicle enters the parking lot, receive navigation information, and initiate reservation requests; (6) User terminal: used to initiate vehicle search requests, parking space reservation requests, and receive navigation guidance; It is used to output standardized parking space status data and navigation services to third-party applications.

[0022] The main method flow of this invention is as follows: Map building: The mapping robot traverses and scans the entire parking lot area according to the planned route → The AI ​​large model generates a digital parking space map → A unique code is generated for each parking space → The locations of entrances, exits, and equipment facilities are marked. For example... Figure 6 As shown on the left, the mapping robot has functions such as full-area scanning and mapping, environmental perception and data collection, data uploading to the cloud, autonomous navigation and path planning, autonomous charging and standby.

[0023] Parking space status inspection: Multiple inspection robots are deployed in different areas → scanning each parking space in their respective areas at a set frequency → identifying the vacancy status and confirming whether parking is permitted. Figure 6As shown on the right, the inspection robot has functions such as regional collaborative inspection, parking space status recognition, periodic scanning and data reporting, dynamic scheduling response, self-inspection and fault reporting. The mapping robot and the inspection robot collaborate and share data through the cloud.

[0024] Dynamic scheduling and adaptive guidance: Real-time cloud-based calculation of parking space vacancy rate → Dynamic adjustment of inspection frequency and number of active robots based on the threshold range of vacancy rate → Adaptive switching of entry guidance strategy (quick allocation based on proximity when vacancy is plentiful, and intelligent recommendation and matching when vacancy is scarce).

[0025] Multi-source data fusion: After a vehicle enters the parking lot, the vehicle's terminal scans the visible parking spaces simultaneously → uploads the data to the cloud → the cloud cross-verifies the robot's inspection data with the vehicle's terminal data → makes a comprehensive judgment and updates the parking space status.

[0026] Parking guidance and recording: Push available parking space navigation to entering vehicles → Record parking space number and vehicle binding relationship after the vehicle is parked.

[0027] Reverse car search navigation: Respond to car search requests → Determine the location of the target parking space → Generate navigation route → Real-time navigation guidance.

[0028] Parking space reservation and holding: User initiates reservation and completes payment → Target parking space is locked in the cloud → A parking space holding robot is dispatched to hold the space → The parking space holding robot is released upon user arrival. Figure 7 As shown, the parking space occupancy robot possesses four core functions: autonomous navigation and movement, parking space occupancy execution, communication and command response, and parking space occupancy status maintenance and timeout management. The robot has three working states: standby state (parked at a charging standby point, warning lights off, parking space occupancy arm retracted), parking space occupancy execution state (located in the center of the target parking space, warning lights flashing "Reserved," parking space occupancy arm extended to prevent entry), and withdrawal state (park space occupancy arm retracted, warning lights off, vehicle leaves the parking space and returns to the standby point). The robot switches between these states via "receive command →" and "user arrives / timeout →".

[0029] Open platform services: High-reliability parking space status data, digital maps and parking space codes, and navigation route data, verified by cross-validation of multi-source data, will be made available to third-party map service providers, travel service platforms, vehicle systems, and smart terminal applications through standardized APIs / SDKs. Beneficial effects

[0030] (1) Extremely low modification cost: Only signal access and transmission hardware needs to be added (cost less than 200 yuan), without replacing the gate controller or modifying the original line. The traditional modification cost is 3,000-8,000 yuan per lane, while the cost of this invention is 1 / 10 to 1 / 30 of the traditional solution; (2) Extremely short renovation time: a single lane renovation takes no more than 1 hour, while traditional solutions require 2-3 days; (3) The recognition success rate is close to 100%: users actively scan the code, completely avoiding the influence of environmental factors; (4) Significantly improved passage efficiency: The average time from scanning to barrier lifting is less than 3 seconds, while the traditional solution requires more than 30 seconds; (5) Cross-site interconnection: A unified account pool in the cloud enables "one network" overdue payment constraints; (6) High reliability: Three-layer offline caching architecture, the interruption of any layer network will not affect basic passage; manual input of license plate can still allow passage when the network is down; (7) Smooth upgrade: compatible with the original license plate recognition system, the two systems operate in parallel, and the transition is risk-free; (8) Full terminal coverage: Both vehicle-mounted APP and mobile APP can be used, and old vehicles and vehicles without license plates can enter the venue without obstacles; (9) Safety and traceability: The electronic ink screen device captures photos simultaneously to provide evidence for counterfeit license plates and disputes; manual on-duty assistance confirms and prevents accidental release; (10) Improved reservation experience: No need to show QR code, the system will automatically match reservation orders and prioritize the execution of reservation rules upon exit. Attached Figure Description

[0031] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the parking space navigation method of the present invention; Figure 3 This is a flowchart of the multi-source data cross-verification method of the present invention; Figure 4 This is a flowchart illustrating the dynamic scheduling decision-making process based on idle rate thresholds according to the present invention. Figure 5 This is a flowchart of the parking space reservation and occupancy method of the present invention; Figure 6 A functional diagram of a mapping robot and an inspection robot; Figure 7 This is a schematic diagram illustrating the functions and working status of the placeholder robot; Figure 8 This is a flowchart illustrating the open platform system architecture and external service process of this invention. Detailed Implementation Example 1 (System Overall Architecture)

[0032] like Figure 1 As shown, the system of the present invention consists of the following parts: Mapping robot: Equipped with LiDAR, depth camera, inertial measurement unit (IMU), and edge computing module, it traverses and scans the entire parking lot area according to a pre-planned route, collecting parking lot layout data (including lanes, parking space boundaries, pillars, walls, equipment and facility locations, etc.). Figure 6 As shown on the left, the mapping robot has five core functions: full-area scanning mapping, environmental perception and data collection, data uploading to the cloud, autonomous navigation and path planning, autonomous charging and standby.

[0033] Inspection robots: Multiple inspection robots are deployed in different areas. Each robot is equipped with a camera, ultrasonic sensors, and a parking space status recognition module. These robots scan each parking space within their assigned area at a preset frequency to identify the parking space's vacancy status (occupied / empty / obstacles / unavailable). Figure 6 As shown on the right, the inspection robot has five core functions: regional collaborative inspection, parking space status recognition, periodic scanning and data reporting, dynamic scheduling response, self-inspection and fault reporting.

[0034] Parking space reservation robot: Equipped with an autonomous navigation module, a parking space reservation mechanism (warning sign, parking space reservation arm or parking space reservation barrier), and a communication module, it is used to move to the target parking space in response to a reservation instruction to perform the parking space reservation operation. Figure 7 As shown in the upper part, the occupancy robot has four core functions: autonomous navigation and movement, occupancy execution, communication and command response, and occupancy status maintenance and timeout management.

[0035] The cloud-based AI large-scale model platform includes modules for map generation, parking space coding, dynamic scheduling decision-making, data fusion, status management, reservation processing, and instruction scheduling. It generates digital parking space maps based on mapping data, assigns a unique code to each parking space, dynamically adjusts inspection frequency and robot numbers based on parking space vacancy rates, cross-verifies and merges inspection data and vehicle-mounted data, updates the parking space status database, and processes reservation requests and dispatch instructions.

[0036] Vehicle-mounted system: Deployed on the vehicle, integrating onboard sensors (cameras, ultrasonic radar), positioning module and communication module, used to scan visible parking spaces in the parking lot and upload data after the vehicle enters, receive navigation information, and initiate reservation requests.

[0037] User terminals include mobile apps or in-vehicle terminals, used to initiate vehicle search requests, parking space reservation requests, and receive navigation guidance.

[0038] Parking space vacancy status and parking space navigation open platform: such as Figure 8As shown, it includes a data aggregation layer, a data service layer, a developer management module, an open SDK module, and a service output layer, which are used to output standardized parking space status data and navigation services to third-party map service providers, travel service platforms, vehicle systems, and smart terminal applications. Example 2 (Map Construction and Parking Space Coding)

[0039] like Figure 2 As shown, the specific steps for map construction are as follows: After startup, the mapping robot autonomously moves within the parking lot according to a pre-planned traversal route (such as a bow-shaped scanning path). The robot's onboard LiDAR collects point cloud data of the parking lot environment through 360-degree omnidirectional scanning, while a depth camera simultaneously acquires RGB and depth images. The robot then uses a SLAM algorithm to construct a real-time 3D point cloud map of the parking lot environment.

[0040] After mapping is completed, the robot uploads all collected data to a cloud-based AI big data model platform. The AI ​​big data model performs semantic segmentation and recognition on the point cloud data and image data, identifying elements such as lane areas, boundary lines of each parking space, pillars, walls, fire-fighting facilities, elevator entrances, and exits within the parking lot. The AI ​​big data model then transforms the recognition results into structured map data, including lane topology networks, parking space polygon boundary coordinates, and facility location labels.

[0041] For each identified parking space, the AI ​​model automatically generates a unique code. The code format is: `[Area Code]-[Floor Code]-[Serial Number]`, for example, "A-1-035" represents parking space number 35 on the 1st floor of area A. Simultaneously, the map marks the coordinates of all entrances and exits, as well as the locations of various facilities (charging stations, fire hydrants, elevators, stairwells, etc.).

[0042] The generated digital parking space map is stored in a cloud database and supports real-time updates and dynamic maintenance. Example 3 (Multi-robot collaborative inspection in different areas)

[0043] like Figure 2 As shown, the specific steps for parking space status inspection are as follows: Based on the parking lot's area and layout, the cloud-based management system divides the parking lot into multiple inspection zones. Each zone is assigned at least one inspection robot. The division principles include: balanced area of ​​each zone, clear boundaries (using lanes or pillars as natural boundaries), and roughly the same number of parking spaces in each zone.

[0044] Each inspection robot performs scanning tasks according to the staggered schedule issued by the cloud. The staggered scheduling mechanism ensures that only one robot is scanning the same area at any given time, avoiding data conflicts and resource waste caused by multiple robots scanning simultaneously.

[0045] The inspection robots move along preset patrol routes within their respective areas, capturing images of each parking space using onboard cameras. They then use a parking space status recognition algorithm (based on a deep learning-based object detection model) to determine the status of each space: occupied, vacant (no car), obstructed, or unavailable (e.g., under maintenance). The recognition results are uploaded to the cloud in real time.

[0046] The inspection robot also has a self-inspection function. When it detects a fault in its own sensors or a positioning deviation, it automatically reports to the cloud and returns to the charging / repair point. The cloud then dispatches a backup robot to take over the inspection task in that area. Example 4 (Dynamic scheduling and adaptive guidance based on idle rate threshold)

[0047] like Figure 4 As shown, this embodiment describes in detail the specific implementation of the cloud-based system that dynamically schedules the inspection frequency and the number of robots based on the parking space vacancy rate, and adaptively switches the guidance strategy.

[0048] (1) Idle Rate Calculation and Threshold Setting: The cloud maintains the parking space status database in real time, recording the total number of parking spaces `Total` and the current number of vacant parking spaces `Free`, and calculates the idle rate `R = Free / Total × 100%`. The system presets two key thresholds: the first threshold (high idle threshold) Th1 is set to 60% (configurable), and the second threshold (low idle threshold) Th2 is set to 20% (configurable). When `R > Th1`, it is determined to be "sufficient idle time", when `R < Th2`, it is determined to be "short idle time", and when `Th2 ≤ R ≤ Th1`, it is determined to be "normal".

[0049] (2) Dynamic scheduling of inspection frequency and number of robots:

[0050] (3) Adaptive switching of entry guidance strategy: When there is ample vacancy, the "fast and nearest allocation" strategy is executed, matching the nearest available parking space according to the vehicle's current location; when there is a shortage of vacancy, the "intelligent recommendation matching" strategy is executed, prioritizing the recommendation of the parking space with the shortest walking distance after obtaining the user's destination information; and the "balanced allocation" strategy is executed under normal conditions.

[0051] (4) Smooth transition of state switching: The cloud is set to de-jitter mechanism, and the strategy switching is triggered only when the idle rate is in the same threshold range for three consecutive detection cycles. Example 5 (Cross-verification of data from multiple sources)

[0052] like Figure 3 As shown, the specific steps for multi-source data fusion and updating are as follows: After a vehicle enters the parking lot, the vehicle's infotainment system automatically scans for parking spaces within the visible area using its forward, side, and surround-view cameras. The system then utilizes its onboard AI chip to run a parking space recognition algorithm, identifying the vacancy status of each visible parking space and uploading the results to the cloud.

[0053] After receiving data from the vehicle's infotainment system, the cloud compares it with the data from the most recent robot inspection. For each parking space, the cloud executes the following cross-verification logic: if the data matches, it is directly adopted; if the data does not match, a secondary confirmation mechanism is triggered (retrieving historical records for comparison, real-time verification by the inspection robot, and using fixed monitoring images for auxiliary judgment); if the data is missing, the result from another data source is directly used.

[0054] After cross-verification is completed, the cloud platform makes a comprehensive judgment and updates the parking space status database. All data fusion processes are logged. Example 6 (Parking Guidance and Parking Space Recording)

[0055] like Figure 2 As shown, the specific steps for parking guidance and recording are as follows: After a vehicle enters the parking lot, the vehicle's infotainment system retrieves the latest parking space availability data from the cloud. Based on the vehicle's current location and destination, and combined with the parking lot map and parking space availability data, the cloud system plans a navigation route from the vehicle's current location to the nearest available or recommended parking space.

[0056] After the vehicle parks in the target parking space, the vehicle's infotainment system uses onboard sensors to confirm that the vehicle has come to a complete stop and is within the parking space boundaries. The system then binds the vehicle's unique identifier (VIN code or user account ID) to the unique code of the target parking space, records the parking start time, and uploads the binding relationship to a cloud database for storage. Example 7 (Reverse Vehicle Navigation)

[0057] like Figure 2 As shown, the specific steps for reverse vehicle location navigation are as follows: When a user needs to locate their car, they can initiate a car-finding request via a mobile app or the vehicle's infotainment system. The system obtains the user's current location through the phone's GPS / Bluetooth / UWB positioning or a positioning base station within the parking lot. Using the user's current location as the navigation starting point and the target parking space as the navigation endpoint, the system combines lane network and pedestrian access data from the parking lot map, employing either Algorithm A or Dijkstra's algorithm to generate the optimal navigation route. Example 8 (Parking Space Reservation and Space Holding Robot)

[0058] like Figure 5 and Figure 7 As shown, the specific steps for reserving and holding a parking space are as follows: Users can check the availability of parking spaces through the in-vehicle system or mobile app, select a target parking space and initiate a reservation request, choose a reservation time slot and complete online payment.

[0059] After receiving the reservation request, the cloud will: (1) verify whether the parking space is available during the reservation period; (2) if it is available, mark the parking space as "reserved"; (3) send a dispatch instruction to the nearest available occupant robot.

[0060] like Figure 7 As shown in the upper part, the parking space occupancy robot has four core functions: autonomous navigation and movement, parking space occupancy execution, communication and command response, parking space occupancy status maintenance and timeout management. After receiving the command, the parking space occupancy robot autonomously navigates to the target parking space. Upon arrival, it performs the parking space occupancy operation: (1) stops in the central area of ​​the parking space and activates the warning sign; (2) if equipped with a physical parking space occupancy mechanism, deploys the mechanism to physically block other vehicles from entering.

[0061] like Figure 7 As shown in the lower half, the parking space occupancy robot has three working states: standby state (parked at the charging standby point, warning light off, occupancy arm retracted), occupancy execution state (located in the center of the target parking space, warning light flashing "Reserved", occupancy arm extended to block entry), and withdrawal state (occupancy arm retracted, warning light off, leaves the parking space and returns to the standby point). The robot switches between these states via "Receive Instruction →" and "User Arrival / Timeout →".

[0062] The parking space reservation robot will remain in a reservation state for the scheduled time period. Once the user arrives and clicks "Arrival Confirmation," the cloud sends a withdrawal command to the robot, which then withdraws, allowing the vehicle to enter the parking space normally. If the user does not arrive within the scheduled time period, the robot will automatically withdraw after the reservation expires, and the reservation fee will be processed according to the agreed rules. Example 9 (Diagram of robot function)

[0063] like Figure 6 As shown in the figure, this embodiment describes in detail the core functions of the mapping robot and the inspection robot.

[0064] On the left are the five core functional modules of the mapping robot (full-area scanning mapping, environmental perception and data acquisition, data upload to the cloud, autonomous navigation and path planning, autonomous charging and standby); on the right are the five core functional modules of the inspection robot (regional collaborative inspection, parking space status recognition, periodic scanning and data reporting, dynamic scheduling response, self-inspection and fault reporting); in the middle, a double-headed arrow indicates the collaborative operation and data sharing relationship between the two robots. Example 10 (Detailed Implementation of the Open Platform)

[0065] like Figure 8As shown in the figure, this embodiment describes in detail the system architecture and external service process of the parking space vacancy status and parking space navigation open platform.

[0066] The data aggregation layer is responsible for receiving and storing parking space status inspection data from the inspection robot, supplementary parking space scanning data from the vehicle-mounted terminal, and parking space image data from the parking lot's fixed monitoring equipment. The multi-source data cross-verification module compares and merges status data of the same parking space from different data sources to generate highly reliable parking space status data.

[0067] The data service layer encapsulates high-reliability parking space status data, parking lot digital map data, unique parking space code data, and parking space navigation route data into standardized API interface services. Core interfaces include: a parking space status query interface (supporting queries by parking lot ID / area code / parking space code / geographic coordinate range), a parking space navigation route planning interface (supporting both driving and walking navigation modes), and a parking space information subscription interface (supporting real-time push notifications for status changes).

[0068] The developer management module provides third-party developers with functions such as registration, real-name authentication, application creation, and API key management, and supports three billing modes: per use, per month, and tiered billing based on usage.

[0069] The open SDK module provides a software development kit for multiple platforms (Web / Android / iOS / vehicle infotainment system), encapsulating the authentication, calling, retry, and error handling logic of API interfaces.

[0070] The sandbox testing environment provides third-party developers with a simulated data debugging environment isolated from the production environment, and does not generate actual billing.

[0071] The service output layer outputs parking space status data and navigation services to third-party map service providers (such as Gaode Maps and Baidu Maps), travel service platforms (such as Didi Chuxing and Meituan), vehicle systems, and smart terminal applications through standardized APIs / SDKs.

Claims

1. A method and system for real-time dynamic parking space navigation in a parking lot, characterized in that, Includes the following steps: (1) Map building steps: At least one mapping robot in the parking lot scans the entire parking lot area according to the preset planned route and collects parking lot layout data; The cloud-based AI model generates a digital parking space map based on the layout data, and generates a unique code for each parking space on the map, while also marking the location of parking lot entrances and exits and equipment facilities; (2) Parking space status inspection steps: Multiple inspection robots are deployed in different areas. Each inspection robot scans each parking space in its assigned area at a preset frequency, identifies the vacancy status of each parking space, and confirms whether parking is allowed in the parking space. (3) Dynamic scheduling and adaptive guidance steps: The cloud calculates the vacancy rate based on the ratio of the current number of vacant parking spaces to the total number of parking spaces in the parking lot; based on the threshold range of the vacancy rate, the inspection frequency of the inspection robot and the number of active robots are dynamically adjusted, and the vehicle entry guidance strategy is adaptively switched. (4) Multi-source data fusion and update steps: After the vehicle enters the parking lot, the vehicle terminal scans the visible parking spaces in the parking lot through the vehicle-mounted sensors, identifies the vacancy status of the parking spaces, and uploads the data to the cloud. The cloud platform cross-verifies the robot inspection data with the vehicle's scanning data to comprehensively judge and update the parking space vacancy status; (5) Parking guidance and recording steps: Based on the updated parking space availability status, push available parking space navigation information to the entering vehicle; after the vehicle parks in the target parking space, the system records the binding relationship between the vehicle and the parking space and the parking space number; (6) Reverse car search navigation steps: In response to the user's car search request, the target parking space location is determined according to the vehicle and parking space binding relationship, and a navigation path from the user's current location to the target parking space is generated in combination with the parking lot map, and real-time navigation guidance is provided; (7) Parking space reservation and occupancy steps: In response to the user's reservation request, the cloud locks the target parking space based on the reservation information; A dedicated parking space reservation robot is dispatched to the target parking space to perform a reservation operation and maintains the reservation status during the scheduled time period. Once the user arrives, the placeholder robot is released, and the vehicle is parked in the parking space.

2. The parking lot real-time dynamic parking space navigation method and system according to claim 1, characterized in that, In step (1), the AI ​​large model generates a digital parking space map, which specifically includes: identifying lanes, parking space boundaries, pillars, walls and equipment facilities in the parking lot; converting the identification results into structured map data; and assigning a unique combined code containing area code, floor code and serial number code to each parking space.

3. The parking lot real-time dynamic parking space navigation method and system according to claim 1, characterized in that, In step (2), the deployment of multiple inspection robots in different areas specifically includes: dividing the parking lot into multiple inspection areas according to the parking lot area and layout; assigning at least one inspection robot to each inspection area; and having each inspection robot perform scanning tasks according to a staggered schedule to avoid data conflicts caused by multiple robots scanning the same area at the same time.

4. The parking lot real-time dynamic parking space navigation method and system according to claim 1, characterized in that, In step (3), the cloud dynamically adjusts the inspection frequency and the number of robots according to the threshold range of the idle rate, specifically including: (1) Preset a first threshold and a second threshold; When the idle rate is greater than the first threshold, it is determined to be an idle state, the inspection frequency is reduced to 15-20 minutes / time, and the number of active inspection robots is reduced to 1 / 3 to 1 / 2 of the normal number; (2) When the idle rate is less than the second threshold, it is determined to be an idle tension state, the inspection frequency is increased to 2-3 minutes / time, and the number of active inspection robots is increased to all available robots; (3) When the idle rate is between the first threshold and the second threshold, maintain the default inspection frequency of 5-10 minutes / time and the default number of active robots; (4) When there is ample parking space, implement the "fast and nearest allocation" strategy: based on the current location of the vehicle, match available parking spaces in the area closest to the vehicle, and prioritize guiding the vehicle to the available parking space closest to the entrance or with the shortest path. (5) When the parking space is in a state of high demand and low availability, the "intelligent recommendation matching" strategy is executed: a recommendation list containing multiple candidate parking spaces is pushed to the vehicle terminal, the destination information input by the user is obtained, the walking distance from each candidate parking space to the destination is calculated, the parking space with the shortest walking distance is recommended first, and the unique code and navigation path of the parking space are provided.

5. The parking lot real-time dynamic parking space navigation method and system according to claim 1, characterized in that, In step (4), the cloud cross-verifies the robot inspection data with the vehicle terminal scanning data. Specifically, when the robot inspection data and the vehicle terminal scanning data are inconsistent in their judgment of the status of the same parking space, the cloud triggers a secondary confirmation mechanism. The secondary confirmation mechanism includes at least one of the following: retrieving the most recent historical status record of the parking space for comparison, sending an instant verification instruction to the inspection robot in the area, and obtaining the real-time image of the parking space from the fixed monitoring equipment in the parking lot for auxiliary judgment.

6. The real-time dynamic parking space navigation method and system for parking lots according to claim 1, characterized in that, In step (5), the system records the binding relationship between the vehicle and the parking space, specifically including: obtaining the precise location of the vehicle after it has been parked in the target parking space through vehicle-mounted positioning or parking lot positioning equipment; binding the unique identifier of the vehicle with the unique code of the target parking space and recording the parking start time; the binding relationship is stored in the cloud database for subsequent reverse vehicle search and billing.

7. The parking lot real-time dynamic parking space navigation method and system according to claim 1, characterized in that, In step (6), generating a navigation path from the user's current location to the target parking space specifically includes: obtaining the real-time location of the user's terminal; using the user's real-time location as the starting point and the target parking space location as the ending point, and combining the road network data in the parking lot map, using a path planning algorithm to generate the optimal navigation path; updating the user's location in real time and dynamically adjusting the navigation guidance during the navigation process.

8. The parking lot real-time dynamic parking space navigation method and system according to claim 1, characterized in that, In step (7), the dedicated parking space occupancy robot performs the parking space occupancy operation in the following ways: after receiving the dispatch instruction from the cloud, the parking space occupancy robot autonomously navigates and moves to the target parking space; it performs the parking space occupancy action within the target parking space area, the parking space occupancy action including at least one of the following: parking in the central area of ​​the parking space and activating the warning sign, deploying the parking space occupancy arm or parking space occupancy barrier to physically block other vehicles from entering; the parking space occupancy robot maintains the parking space occupancy status during the reserved time period until the user arrives or the reservation expires. In step (7), the user's paid reservation and parking space occupancy specifically includes: the user initiates a parking space reservation request through the vehicle terminal or mobile terminal, selects a reservation time period and completes online payment; the cloud locks the target parking space based on the payment confirmation information and dispatches a occupancy robot; if the user arrives and parks in the parking space within the reservation time period, the reservation fee is automatically converted into a parking fee deduction; if the user does not arrive within the reservation time period, the occupancy robot automatically withdraws after the reservation expires, and the reservation fee is handled according to the agreed rules.

9. The real-time dynamic parking space navigation method and system for parking lots according to claim 1, characterized in that, include: (1) Data aggregation layer, used to receive and store parking space status inspection data from the inspection robot in the parking lot, parking space scanning supplementary data from the vehicle terminal, and parking space image data from the fixed monitoring equipment in the parking lot; the data is fused and processed by the multi-source data cross-verification module to generate highly reliable parking space status data; (2) Data service layer, used to encapsulate the high-reliability parking space status data, parking lot digital map data, parking space unique code data and parking space navigation path data into standardized API interface services; The API interface services include at least: a parking space status query interface, which supports querying parking space status by parking lot ID, area code, unique parking space code, or geographical coordinate range, and returns data including the unique parking space code, vacancy status, parking space geographical coordinates, parking space size information, and update timestamp; a parking space navigation route planning interface, which receives the starting point coordinates and the unique parking space code of the destination, and returns a driving or walking navigation route from the starting point to the target parking space by combining road network data in the digital parking lot map, including the path node coordinate sequence, turning instructions, distance, and estimated travel time; and a parking space information subscription interface, which supports real-time push subscription for status changes by parking lot, area, or specific parking space, and pushes status change notifications to third-party applications via Webhook or message queue. (3) Developer management module, which provides third-party developers with functions such as registration, authentication, application creation and API key management; and performs metering and billing based on the call volume and service level of third-party applications; (4) An open SDK module is provided to provide a software development kit for multiple platforms, encapsulate the calling logic of the API interface services, and allow third-party developers to integrate it into map applications, travel applications, vehicle systems and smart terminal applications; (5) Service output layer, which is used to output standardized parking space status data and navigation services to third-party map service providers, travel service platforms, vehicle system and smart terminal applications. It supports real-time marking and display of parking space vacancy status on third-party maps, as well as driving navigation from the parking lot entrance to the target parking space and walking navigation from the user's current location to the target parking space.

10. The real-time dynamic parking space navigation method and system for parking lots according to claim 1, characterized in that, The generation of the high-confidence parking space status data specifically includes: when the inspection robot data, vehicle terminal data, and parking lot fixed monitoring data all agree on the status of the same parking space, the status is directly output and marked as high confidence; when the three are not completely consistent, a majority voting mechanism is used to determine the final status, and the sources of inconsistent data and voting results are recorded in the log for auditing; the high-confidence parking space status data is accompanied by a confidence label, which can be selectively used by third-party applications according to their own business needs.

11. An architecture for a parking space navigation system, characterized in that, include: (1) Mapping robot: used to traverse and scan the entire parking lot area according to the preset planned route and collect parking lot layout data; (2) Inspection robots: Multiple inspection robots are deployed in different areas to scan each parking space in their respective areas at a preset frequency and identify the vacancy status of the parking spaces; (3) Parking space reservation robot: used to move to the target parking space in response to the reservation instruction and perform the reservation operation; (4) Cloud AI large model platform: used to generate digital parking space maps based on the layout data collected by the mapping robot, generate a unique code for each parking space, dynamically adjust the inspection frequency and number of robots according to the parking space vacancy rate, cross-verify and merge the inspection data and vehicle terminal data, update the parking space status database, and process reservation requests and dispatch instructions. (5) Vehicle terminal: used to scan visible parking spaces in the parking lot and upload data after the vehicle enters the parking lot, receive navigation information, and initiate reservation requests; (6) User terminal: used to initiate vehicle search requests, parking space reservation requests and receive navigation guidance; used to output standardized parking space status data and navigation services to third-party applications.

Images