Unmanned aerial vehicle on-road parking space inspection system

Abstract

The invention discloses an unmanned aerial vehicle in-road parking space inspection system, and belongs to the technical field of unmanned aerial vehicle inspection, the unmanned aerial vehicle in-road parking space inspection system comprises an unmanned aerial vehicle, an unmanned aerial vehicle cabin space, a control module, a data transmission module and a data processing center which work cooperatively, and compared with a traditional two-wheeled electric vehicle inspection mode, the unmanned aerial vehicle has high-speed flight capability and flexible maneuverability; and the inspection task in a larger range can be completed in an extremely short time. Through actual test and simulation analysis, within the same time, the inspection area covered by the unmanned aerial vehicle is several times that of a two-wheeled electric vehicle, the inspection efficiency can be improved, a modern inspection system based on the unmanned aerial vehicle is introduced, and innovative practice and positive exploration of a city in the field of intelligent traffic management are manifested. Compared with a traditional manual inspection mode, unmanned aerial vehicle inspection shows the characteristics of high technology and high efficiency, the overall image and modernization level of a city are improved, the attraction of the city to residents and investors is enhanced, and highlights are added to smart city construction.

Description

Technical Field

[0001] This invention relates to the field of drone inspection technology, and more specifically, to a drone-based on-street parking space inspection system. Background Technology

[0002] In urban parking management systems, on-street parking space inspections are a crucial link in ensuring the efficient use of parking resources and supporting management decisions. Currently, inspections in this area mainly rely on a traditional two-wheeled electric vehicle-assisted manual labor model. However, this method has many inherent drawbacks that are difficult to overcome and can no longer meet the evolving needs of urban traffic management.

[0003] From the perspective of inspection efficiency, two-wheeled electric vehicles, limited by their speed and range, struggle to achieve comprehensive and rapid coverage when inspecting large areas and high densities of parking spaces in urban areas. In areas with complex geographical environments, such as winding mountain roads and narrow streets, electric vehicles face obstacles or even be unable to enter, resulting in numerous blind spots. Furthermore, traffic conditions significantly interfere with inspection work. During peak hours and other congested periods, electric vehicles are easily trapped in traffic, causing inspection interruptions or severe delays, further reducing the timeliness of inspections. Regarding inspection accuracy, human judgment is greatly affected by subjective factors and environmental conditions. When faced with dim lighting, unclear parking space markings, complex vehicle parking postures, or obstructions, inspectors relying solely on visual inspection are prone to missed or incorrect inspections, leading to distorted parking data and impacting the scientific and effective nature of urban parking management decisions. As cities continue to expand and the number of motor vehicles grows rapidly, the demand for urban parking management is increasingly moving towards refinement and intelligence, placing higher demands on the efficiency, accuracy, and coverage of inspections. The limitations of traditional manual inspection methods are becoming increasingly apparent, and they are far from meeting current management needs. There is an urgent need for new and efficient inspection technologies to fill this gap.

[0004] In recent years, drone technology has made groundbreaking progress, demonstrating significant advantages in mobility, flexibility, and the diversity of equipment it can carry, theoretically providing the possibility of efficient data collection for on-street parking space inspection. However, in practical applications, a complete, mature system and methodology that can fully leverage the characteristics of drone technology has yet to be formed. This makes it difficult to cope with the complex and ever-changing scenarios in urban parking management (such as differences in parking demand in different areas, sudden road conditions, and accurate identification in complex environments) and diverse needs. Consequently, the potential of drone technology has not been effectively translated into actual inspection efficiency. Against this backdrop, developing an on-street parking space inspection system that is adapted to complex urban scenarios and combines high efficiency and accuracy has become an urgent need to solve the current pain points of parking management and promote the intelligent upgrade of urban parking management. Summary of the Invention

[0005] The purpose of this invention is to provide an unmanned aerial vehicle (UAV) roadside parking space inspection system to solve the problems mentioned in the background art.

[0006] A drone-based on-street parking space inspection system includes a collaborative drone, a drone cabin, a control module, a data transmission module, and a data processing center. The control module achieves accurate data collection and management of on-street parking space status through integrated intelligent inspection. The core implementation process is as follows: the control module plans inspection tasks and remotely dispatches drones; after taking off from the drone cabin, the drone performs inspections according to the planned route, collects images through onboard equipment and performs preliminary processing, and transmits the data in real time to the data processing center for in-depth analysis via the data transmission module; after completing the inspection, the drone returns to the drone cabin to charge and undergo status monitoring. The drone is equipped with a high-definition image acquisition and preliminary data processing unit; the drone cabin integrates take-off, landing, parking, charging, and status monitoring functions; the control module has task planning, navigation, and cluster scheduling capabilities; the data transmission module ensures secure and stable data transmission; and the data processing center performs secondary data verification and in-depth analysis.

[0007] Preferably, the high-definition image acquisition unit of the UAV is an ultra-high pixel high-definition camera with autofocus function, and integrates image stabilization technology that combines optical image stabilization principle and electronic image stabilization algorithm; the preliminary data processing unit is a high-performance processor with advanced computing architecture, which has parallel processing capability and can filter and classify the acquired images in real time, extract parking space occupancy status, license plate number and abnormal occupancy information and temporarily cache them.

[0008] Preferably, the drone compartment adopts an integrated sealed design, with the outer shell made of high-strength corrosion-resistant composite material, and the interior space is ergonomically designed to precisely fit the drone's dimensions. The bottom of the drone compartment integrates a wireless charging device based on magnetic resonance coupling technology, which can automatically sense and establish an efficient charging connection after the drone lands, and also has preliminary detection functions for the drone's battery level, body appearance, and the working status of key components.

[0009] Preferably, the control module realizes unified scheduling of the drone swarm through a cloud server. It can accurately set targeted inspection times based on the functional characteristics, parking demand patterns, and traffic control requirements of different urban areas. The inspection times include key inspection periods before weekday peak hours in commercial areas, before peak hours in hospitals, and before school hours.

[0010] Preferably, the control module combines Geographic Information System (GIS) technology, high-precision road map data, and parking space distribution data to support manual planning or automatic generation of optimal inspection routes through intelligent algorithms. During planning, it simultaneously avoids no-fly zones and building obstructions while meeting flight safety distance requirements, and has the dynamic adaptation capability to adjust the route in real time according to changes in road conditions such as road construction and traffic accidents. The control module integrates Global Positioning System (GPS) navigation technology, which can obtain the UAV's position, speed, and attitude information in real time and provide navigation support.

[0011] Preferably, when the drone takes off, the drone cabin automatically opens its door after receiving instructions from the control module. The drone uses its onboard high-precision sensors to monitor environmental information such as wind speed, wind direction, air pressure, and temperature, and adjusts its flight attitude accordingly. During flight, the drone dynamically optimizes its flight altitude and speed based on the surrounding road environment. It increases its altitude and speed on open, unobstructed roads and decreases its altitude and speed in densely built-up areas, narrow streets, or areas with complex obstacles. The camera captures images of parking spaces at fixed intervals of 2-3 seconds.

[0012] Preferably, the drone is equipped with an image recognition algorithm based on deep learning technology. By training a massive number of parking space image samples to build an image analysis model, it can quickly determine the occupancy status of parking spaces, accurately identify license plate numbers, and automatically mark abnormal situations such as debris and non-motorized vehicle occupancy. The image recognition algorithm can automatically optimize parameters according to the quality of the captured image and adjust image enhancement parameters in low light to improve recognition accuracy.

[0013] Preferably, the data transmission module adopts 4G / 5G communication technology, which has the characteristics of high speed and low latency transmission. It also uses multiple encryption technologies to encrypt the transmitted data, and automatically detects and corrects transmission errors caused by signal interference and network fluctuations through error correction technology, so as to ensure the security and integrity of data transmission.

[0014] Preferably, the data processing center is equipped with a high-performance server and uses a multi-model fusion recognition algorithm to perform secondary verification on the transmitted data. Specifically, for license plate numbers, character segmentation, feature extraction, and classification technologies are combined to improve the recognition accuracy. For parking space occupancy status, time series analysis, spatial distribution analysis, and historical data comparison analysis are combined for multi-dimensional verification. The data processing center stores the analysis results in a database and generates daily, weekly, and monthly reports on parking space usage, as well as statistical reports on abnormal occupancy.

[0015] Preferably, after completing the inspection, the UAV adjusts its flight attitude and speed through real-time communication with the UAV cabin according to the landing command from the control module, and lands precisely at the designated location. The UAV cabin uploads the status detection information to the control module, which allows managers to arrange regular maintenance of the UAV based on the information, including replacement of vulnerable parts, battery depth detection and maintenance, and calibration of the flight control system.

[0016] Compared with the prior art, the advantages of this invention are:

[0017] Improving inspection efficiency: Compared to traditional two-wheeled electric vehicle inspection methods, drones, with their high-speed flight capabilities and flexible maneuverability, can complete inspection tasks over a larger area in a very short time. Actual testing and simulation analysis show that, in the same amount of time, drones can cover several times the inspection area of ​​two-wheeled electric vehicles, thus significantly improving inspection efficiency.

[0018] Enhancing the city's image: The introduction of a modern drone-based inspection system showcases the city's innovative practices and proactive exploration in the field of intelligent traffic management. Compared to traditional manual inspection methods, drone inspections are characterized by high technology and high efficiency, helping to improve the city's overall image and modernization level, enhancing its attractiveness to residents and investors, and adding a highlight to smart city construction.

[0019] Optimizing resource allocation: Through precise data collection and analysis, parking management departments can rationally plan and adjust parking resources based on real-time parking space usage. For example, by dynamically adjusting parking fees according to differences in parking demand in different areas and at different times, they can guide vehicles to park rationally, improve parking space turnover, avoid resource waste or excessive shortages, achieve optimal allocation of parking resources, and improve the utilization efficiency of urban transportation resources.

[0020] Promoting the Development of the Intelligent Transportation Ecosystem: As an important component of the intelligent transportation system, the system and method of this invention generate a large amount of accurate parking data, which can be integrated and analyzed with other traffic data (such as traffic flow and road condition information). This not only helps to further optimize urban traffic planning, but also provides data support for the research and application of other related technologies in the field of intelligent transportation, such as intelligent parking guidance systems and autonomous driving assisted decision-making, thus promoting the coordinated development and improvement of the entire intelligent transportation ecosystem.

[0021] Flexible scalability: With continuous technological advancements and evolving urban management needs, the system boasts excellent scalability. On one hand, software upgrades can continuously optimize drone flight control algorithms, image recognition algorithms, and data processing workflows, enhancing system performance. On the other hand, in terms of hardware, the number of drones can be increased, drone bay facilities upgraded, or data transmission networks expanded as needed, easily adapting to the expansion of city size and the growth of parking management workload, ensuring long-term stable operation and continued high efficiency. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the system module structure of the present invention;

[0023] Figure 2 This is a schematic diagram of the working structure of the system modules of the present invention. Detailed Implementation

[0024] Example 1:

[0025] (I) System Setup

[0026] 1. Unmanned Aerial Vehicle (UAV) Deployment: Through a comprehensive and in-depth assessment of the urban road network, parking demand hotspots, and geographical environment, and utilizing big data analytics and simulation technology, multiple UAV pods are precisely installed in suitable locations along urban roads. For example, on roads surrounding commercial centers, hospitals, schools, and transportation hubs, the spacing between pods is rationally set based on detailed analysis of parking demand density. In areas with high parking demand, the spacing between pods is appropriately reduced to ensure timely response to inspection tasks; in areas with relatively low parking demand, the spacing is appropriately increased to optimize resource allocation. Simultaneously, the flight radius and endurance of the UAVs are fully considered to optimize the pod layout globally, avoiding blind spots and ensuring effective coverage of all on-street parking spaces throughout the city.

[0027] 2. Equipment Debugging and Connection: The drone is meticulously paired and comprehensively debugged with the control and data transmission modules. During debugging, the drone's flight performance is tested across the board, including flight speed, altitude, stability, and maneuverability; the camera's image quality is rigorously tested to ensure clear, high-quality images are acquired under varying lighting and weather conditions; data processing capabilities are stress-tested to verify the drone's efficiency and accuracy in processing large amounts of image data; and the communication stability between modules is tested over extended periods in multiple scenarios to ensure stable and efficient communication connections under complex electromagnetic environments and different network conditions. Through these rigorous debugging processes, it is ensured that the drone can accurately receive commands from the control module, and that the data transmission module can transmit data stably and efficiently. Simultaneously, powerful image recognition and data analysis software is deployed at the data processing center, establishing a stable and reliable connection with the data transmission module via a high-speed network to ensure timely and accurate data transmission to the data processing center for processing.

[0028] 3. System Initialization and Parameter Settings: After the system is built, a comprehensive initialization process is performed. Detailed and accurate road map data, including road names, locations, widths, and slopes, is input into the control module; parking space distribution data, including the specific location, number, and type of each parking space, is also input. Based on actual needs, the drone's flight parameters, such as flight altitude, speed, and shooting interval, as well as relevant parameters for the image recognition algorithm, such as recognition accuracy and classification threshold, are set to ensure the system can perform inspection work according to the predetermined efficient strategy.

[0029] (II) Execution of Inspection Tasks

[0030] 1. Daily Inspection and Scheduling: Taking 7:00 AM as an example, the control module sends a takeoff command to the drone according to the preset task. The drone quickly takes off from its cabin and flies along a pre-planned, precise route to conduct a comprehensive inspection of on-street parking spaces. During the inspection, the inspection strategy is flexibly adjusted according to the characteristics of different areas. In commercial areas, due to the high turnover rate of parking spaces and frequent changes in vehicle parking conditions, the shooting interval is appropriately shortened, such as to every 2 seconds, to increase the data collection frequency and promptly grasp the real-time changes in parking spaces. Around residential areas, based on residents' parking patterns, the focus is on the parking space usage during morning and evening peak hours. The inspection frequency is increased before peak hours to ensure that parking space shortages can be detected in a timely manner, providing better parking services for residents.

[0031] 2. Data Acquisition and Preliminary Processing: During flight, the drone captures an image of a parking space every 3 seconds and uses onboard algorithms to initially identify the parking space status. During identification, the algorithm automatically optimizes and adjusts parameters based on the actual image quality and recognition results. For example, in low-light conditions, it automatically adjusts image enhancement parameters to improve image clarity and contrast, thereby increasing recognition accuracy. Simultaneously, the drone performs preliminary processing and packaging of the acquired images and recognition results, storing them according to a specific data format, ready for rapid transmission to the data processing center via the data transmission module.

[0032] 3. Data Transmission and In-Depth Processing: The drone transmits the pre-processed data to the data processing center in real time via its data transmission module. Upon receiving the data, the data processing center utilizes powerful computing resources and complex algorithms for in-depth analysis and verification. The license plate number recognition results are then re-verified to further improve accuracy. Multi-dimensional analysis of parking space occupancy status is performed, combining time series analysis, spatial distribution analysis, and historical data comparison to ensure data accuracy and completeness. The analysis results are stored in a database and relevant reports are generated, such as daily, weekly, and monthly parking space usage reports, and statistical reports on abnormal occupancy, providing parking management departments with intuitive and comprehensive decision support.

[0033] 4. Landing and Maintenance: After completing the inspection mission, the drone returns to the nearest drone bay to land and recharge. While charging, the bay performs a basic visual inspection and fault diagnosis, such as checking for collision marks on the fuselage, overheating or bulging of the battery, and loose connections of key components. The inspection results are uploaded to the control module, allowing administrators to schedule regular maintenance and upkeep, including periodically replacing vulnerable parts, performing in-depth battery testing and maintenance, and calibrating the drone's flight control system. This ensures the drone is always in good operating condition and fully prepared for the next inspection mission.

[0034] (III) Inspection Methods

[0035] 1. Task Planning: Using the control module's functions, managers can comprehensively and meticulously consider various factors, including the unique functional characteristics of different urban areas, the complex and ever-changing parking demand patterns, and strict traffic control requirements, to accurately set the drone's patrol times. For example, in areas surrounding hospitals, since peak patient visits and family visits typically occur between 7:00 AM and 9:00 AM and between 4:00 PM and 6:00 PM, drone patrols can be prioritized before these times to promptly assess real-time parking availability and provide better parking guidance for patients. In areas near schools, based on students' commuting times, drone patrols can be strategically scheduled before and after classes to meet the specific parking management needs around schools.

[0036] Utilizing Geographic Information System (GIS) technology, combined with high-precision road map data and detailed parking space distribution data, the control module allows for both manual planning of drone inspection routes based on practical experience and specific needs, and automatic generation of optimal routes using intelligent algorithms. During planning, no-fly zones (such as near airports and military control zones) are fully considered; building obstructions prevent the drone from obtaining effective images due to building blockages; and safe flight distance requirements ensure the drone maintains a safe distance from obstacles and people. Furthermore, to cope with potential emergencies such as traffic control due to temporary road construction or traffic congestion caused by accidents, the system features real-time route adjustment. When changes in road conditions are detected, it can quickly and automatically regenerate an alternative inspection route based on the latest road information, ensuring the drone can complete its inspection tasks promptly and effectively, unaffected by excessive external interference.

[0037] 2. Takeoff and Inspection: At the precisely set time, after receiving the takeoff command from the control module, the drone's cabin door automatically and smoothly opens, and the drone takes off quickly and smoothly from the cabin according to the preset efficient takeoff procedure. During takeoff, the drone uses its onboard high-precision sensors to monitor the surrounding environment in real time and comprehensively, including wind speed, wind direction, air pressure, and temperature. Based on this real-time environmental data, it automatically and accurately adjusts its flight attitude to ensure a safe and successful ascent.

[0038] The drone flies along a pre-set, precise route, with its flight altitude and speed dynamically optimized based on the surrounding environment and the camera's optimal shooting range. On open, unobstructed roads, the flight altitude and speed can be increased to expand the camera's field of view, improve inspection efficiency, and cover more parking spaces per unit time. In densely built-up areas, narrow streets, or areas with complex obstacles, the flight altitude and speed are reduced to ensure the camera can clearly and accurately capture detailed images of each parking space, avoiding data acquisition quality issues caused by poor shooting angles or blurry images. During flight, the camera captures an image of a parking space every 2-3 seconds at a set interval, ensuring comprehensive and continuous acquisition of parking space status information.

[0039] The image recognition algorithm onboard the drone employs cutting-edge deep learning technology. Through in-depth learning and training on massive amounts of parking space image samples, it has built a powerful image analysis model. This algorithm can quickly and accurately perform comprehensive analysis of captured images. First, it can rapidly determine whether a parking space is occupied. If occupied, it further utilizes advanced character recognition technology and image feature matching algorithms to accurately identify the license plate number of the occupying vehicle. For abnormal occupancy situations, such as parking spaces being occupied by debris or non-motorized vehicles, the algorithm can automatically identify and mark them, while simultaneously recording detailed relevant image information, providing rich and accurate data support for subsequent parking management decisions.

[0040] 3. Data Processing and Transmission: During the UAV's inspection mission, the acquired images undergo preliminary processing to extract key information. This data is then transmitted in real-time and rapidly to the data processing center via the data transmission module. The data processing center is equipped with high-performance servers, possessing powerful computing resources and advanced data analysis capabilities. The center utilizes more complex and accurate image recognition algorithms to perform secondary verification and in-depth analysis of the data transmitted by the UAV.

[0041] For example, in license plate number recognition, the data processing center employs an advanced multi-model fusion algorithm, combining character segmentation, feature extraction, and classification techniques to further improve recognition accuracy. By comprehensively analyzing and comparing license plate images from different angles and under varying lighting conditions, the center effectively solves the misidentification problems that easily occur in traditional recognition algorithms. For determining parking space occupancy, a multi-dimensional comprehensive analysis is conducted by comparing image data from different time periods, combined with the occupancy status of surrounding parking spaces, historical parking data, and real-time traffic flow information, ensuring the accuracy and reliability of the data. Simultaneously, this data provides comprehensive and accurate support to parking management departments, assisting them in making scientific and rational decisions.

[0042] 4. Landing and Charging: After completing the inspection mission, the UAV automatically and accurately returns to the nearest UAV bay based on the landing command sent by the control module. During landing, the UAV adjusts its flight attitude and speed in real time through communication with the bay to ensure an accurate and smooth landing at the designated location. Once the charging device in the bay detects the UAV's landing, it automatically initiates the charging process for rapid charging. Simultaneously, the bay also has basic status detection functions, capable of performing preliminary checks on the UAV's battery level, whether there is any damage to the fuselage, and the operational status of key components, recording relevant information and uploading it to the control module. Based on this information, management personnel can promptly schedule regular maintenance and upkeep of the UAV, ensuring it is always in good operating condition and fully prepared for the next inspection mission.

[0043] Example 2:

[0044] I. Implementation Scenario Setting

[0045] This embodiment selects a central urban area of ​​a city as the application scenario. The area covers a total area of ​​50 square kilometers and includes a core commercial area, a cluster of top-tier hospitals, key primary and secondary school districts, densely populated residential areas, and areas surrounding transportation hubs, with a total of 20,000 on-street parking spaces planned. The area has typical urban parking management pain points such as significant spatial and temporal differences in parking demand (daytime peak hours in commercial centers, morning and evening peak hours at hospitals, and short-term peak hours for school commuting), narrow road sections (streets and alleys in the old city are less than 5 meters wide), and frequent traffic congestion (traffic efficiency on main roads drops by 60% during morning and evening peak hours). The traditional two-wheeled electric vehicle inspection mode can only complete two rounds of full-coverage inspections per day, with a missed inspection rate of 5% and a data update delay of more than 1 hour, which cannot meet the needs of refined management.

[0046] II. System Setup and Implementation Details

[0047] (a) Deployment of unmanned aerial vehicle (UAV) cabins

[0048] Based on a heat map of regional parking demand density and the drone flight radius (maximum 1.5 km), big data simulation was used to optimize the layout, deploying a total of 40 drone pods:

[0049] Core commercial area (3 square kilometers): One parking space will be deployed every 500 meters along the main commercial area roads, for a total of 8 spaces, to meet the high turnover parking demand;

[0050] Hospital Cluster Area (2 square kilometers): 3 hospitals are deployed around each of the 3 top-tier hospitals, for a total of 9 hospitals, covering the hospital entrances and surrounding auxiliary roads;

[0051] Key primary and secondary school areas (4 square kilometers): 2 parking spaces will be deployed within 500 meters of the gates of 6 schools, for a total of 12 spaces, to match the short-term parking peaks during school hours;

[0052] Densely populated areas and suburban roads (41 square kilometers): 11 units will be deployed at 1-kilometer intervals to balance coverage and resource investment.

[0053] The cabin is installed using a modular fixing method, selecting the edge of the sidewalk on both sides of the road (without occupying passage space). The outer shell is made of carbon fiber composite material and has an IP67 waterproof and dustproof rating, which is suitable for the city's rainy and windy and sandy climate in spring and autumn.

[0054] (II) Equipment Selection and Commissioning

[0055] Core equipment parameters:

[0056] Drone: Selected as a multi-rotor industrial-grade drone, equipped with a 48-megapixel high-definition camera (focal length 24-70mm, autofocus response time ≤0.1 seconds), integrating optical image stabilization + electronic image stabilization dual modules, built-in Snapdragon 8 Gen3 processor (8-core parallel computing architecture), equipped with two 20000mAh lithium batteries, with a single battery providing 45 minutes of flight time;

[0057] Wireless charging device: adopts magnetic resonance coupling technology, output power of 60W, charging efficiency ≥85%, sensing distance of 0-10cm, and supports automatic alignment within ±5cm of drone landing deviation;

[0058] Control module: A cloud-based scheduling platform built on Alibaba Cloud servers, supporting the management of a cluster of 50 drones;

[0059] Data transmission module: adopts a 5G+4G dual-mode communication module, with a peak transmission rate of 1Gbps and a latency of ≤20ms;

[0060] Data processing center: Deployed with 4 high-performance servers (each configured with 32 CPU cores, 128GB memory, and 4TB SSD storage), equipped with a self-developed multi-model fusion image recognition system.

[0061] Debugging process:

[0062] Flight performance testing: Verify the drone's flight speed (30km / h, reduced to 15km / h in dense areas) and altitude stability (±0.5 meters) on open roads, and test attitude control accuracy in a level 5 gust of wind to ensure that image acquisition is blur-free;

[0063] Image acquisition test: Under three scenarios—sunny day with strong sunlight, cloudy day, and nighttime street lighting—parking spaces with different parking angles (0°-45°) and different degrees of occlusion (unobstructed, partially obstructed, and severely obstructed) were photographed to verify the image clarity of the camera and ensure that the license plate character recognition rate is ≥95%.

[0064] Communication stability test: Continuous transmission test for 72 hours in the vicinity of transportation hubs with strong electromagnetic interference (near subway entrances and signal towers), with a data packet loss rate of ≤0.1% and an error correction technology response time of ≤10ms;

[0065] Module linkage test: Simulate the entire process inspection to verify the coordination of control module command issuance, drone response, door opening and closing, charging start-up, and data transmission, with a single linkage success rate of ≥99.8%.

[0066] (III) System Initialization Settings

[0067] Data entry: Import 1:1000 high-precision road map data (including road width, slope, and no-fly zone coordinates) and precise coordinates, numbers, and types (small vehicles / new energy vehicles) of 20,000 parking spaces into the control module to create electronic files for parking spaces;

[0068] Parameter configuration:

[0069] Flight parameters: Open roads: flight altitude 50 meters, speed 30 km / h; narrow streets and alleys (width < 8 meters): flight altitude 20 meters, speed 15 km / h; densely built-up areas (building spacing < 20 meters): flight altitude 30 meters, speed 20 km / h.

[0070] Shooting parameters: 3-second shooting interval in regular areas, 2-second shooting interval in commercial centers and around schools;

[0071] Recognition algorithm parameters: license plate recognition threshold 0.85, parking space occupancy status judgment threshold 0.9, abnormal occupancy recognition includes three types: debris, non-motorized vehicles, and parking over the line.

[0072] III. Inspection Task Execution Process

[0073] (I) Daily inspection and dispatch

[0074] Time period planning:

[0075] Business core area: Inspections will be conducted once daily at 8:30, 11:30, and 17:30, covering data collection before the peak parking hours of 10:00-14:00 and 19:00-22:00 on weekdays;

[0076] Hospital cluster area: Inspections will be conducted once each at 6:30, 12:30, and 16:30 daily to coincide with peak patient visit times of 7:00-9:00 and 16:00-18:00;

[0077] School district: Inspections will be conducted once each at 6:30, 11:00, and 15:00 on weekdays, covering the school arrival and departure times of 7:30-8:30 and 16:30-17:30;

[0078] In densely populated areas, inspections will be conducted once a day at 7:00, 14:00, and 20:00 to match residents' travel patterns.

[0079] Route planning: The control module combines GIS technology with real-time traffic data to automatically generate the optimal inspection route. For example, in the core commercial area, the route avoids congested sections of main roads, prioritizing coverage of side roads and parking spaces around parking lot entrances and exits; in the narrow streets and alleys of the old city, the route uses a zigzag flight trajectory to ensure no blind spots. When road construction is detected (such as a main road being closed for construction from 9:00 AM to 12:00 PM), the system automatically replans a detour route within 5 seconds to ensure that the inspection coverage rate is not affected.

[0080] (II) Data Collection and Preliminary Processing

[0081] Takeoff Execution: The control module sends takeoff commands to the corresponding UAV cabin at preset times. The cabin door opens automatically within 2 seconds. After the UAV initiates a self-check (battery power, camera status, communication signal), it completes a smooth takeoff within 3 seconds. During takeoff, environmental data is monitored in real time by the onboard temperature and humidity sensor (measurement range -20℃-60℃), barometric pressure sensor (accuracy ±1hPa), and wind speed sensor (range 0-20m / s). When a gust of wind speed of 8m / s is detected, the flight speed is automatically reduced to 10km / h, and the attitude is adjusted to maintain stability.

[0082] Image acquisition and preliminary recognition: The drone flies along a planned route, the camera captures images at set intervals, and the onboard processor performs preliminary processing in real time.

[0083] The parking space occupancy status and license plate character outlines are directly extracted from clear images and temporarily cached to a 128GB high-speed storage module;

[0084] For images taken at night or in low-light conditions, an image enhancement algorithm is automatically activated (increasing brightness by 15%-30% and contrast by 20%) before feature extraction is performed.

[0085] For parking spaces partially obscured by trees or billboards, edge detection algorithms are used to reconstruct the parking space outline and determine its occupancy status.

[0086] (III) Data Transmission and Deep Processing

[0087] Real-time transmission: The drone uploads the pre-processed data (including original images, preliminary identification results, shooting time and coordinates) in real time via the 5G communication module. The transmission time for a single image is ≤0.5 seconds, and each drone transmits approximately 8GB of data per inspection (45 minutes). Data transmission uses the AES-256 encryption algorithm. If the 5G signal is interrupted during transmission, it automatically switches to the 4G network to ensure no data loss.

[0088] In-depth analysis: After receiving the data, the data processing center initiates multi-dimensional verification:

[0089] License plate recognition: A CNN+LSTM multi-model fusion algorithm is used, combined with character segmentation and tilt correction technology, to perform secondary recognition on license plates at different angles (0°-45°). The final recognition accuracy reaches 99.2%, which is 3.5% higher than the initial recognition accuracy of airborne recognition.

[0090] Occupancy status verification: By comparing three adjacent images of the same parking space, and combining the occupancy status of five surrounding parking spaces with historical data from the same period, false judgments caused by passing vehicles are eliminated, and the accuracy rate of status judgment reaches 99.5%.

[0091] Anomaly Handling: For the 12 identified abnormal occupancy incidents (8 non-motorized vehicle occupancy incidents and 4 incidents of debris accumulation), the location coordinates and images are automatically marked, and early warning information is generated and pushed to the mobile terminal of the management personnel.

[0092] Data output: Three types of reports are generated daily: "Daily Report on Parking Space Usage" (including parking space usage rate, peak hours, and turnover rate in each area), "Statistical Report on Abnormal Occupation" (categorized by abnormal type and area), and "Report on Equipment Operation Status" (including drone battery wear, cabin charging efficiency, etc.). The data is synchronized to the city traffic management platform.

[0093] (iv) Landing and Maintenance

[0094] Automatic Return-to-Home: After completing its inspection mission (automatically triggering return-to-home when battery power is ≤30%), the drone returns based on the coordinates of the nearest cabin from the control module, combined with GPS navigation (positioning accuracy ±1 meter). During landing, it uses ultrasonic sensors to measure distance to the cabin in real time and dynamically adjusts its attitude, with a landing deviation ≤3cm. Once the cabin door detects the drone's arrival, it closes within 1 second and initiates wireless charging, allowing for a full charge in 40 minutes.

[0095] Condition detection and maintenance: The cabin has a built-in infrared detection module and vision sensor, which can perform three checks while charging:

[0096] Battery status: Detects voltage and temperature. If the temperature exceeds 45℃ or the voltage fluctuates abnormally, it will automatically stop charging and send an alert.

[0097] Exterior: Identify any signs of impact, scratches, or propeller damage;

[0098] Key components: Detecting the cleanliness of the camera lens and the connection status of the communication antenna. Detection data is uploaded to the control module in real time. Administrators perform a comprehensive maintenance on the drone weekly: replacing worn propellers, cleaning the camera lens, and calibrating the flight control system. A deep charge-discharge test is conducted on the battery monthly to ensure an annual failure rate of ≤3%.

[0099] IV. Implementation Results Verification

[0100] Inspection efficiency: This system completes an average of 6 rounds of full-area, full-coverage inspections per day, with each round taking 2 hours. This is 100% more efficient than traditional two-wheeled electric vehicle inspections (4 hours per round). The average daily data update frequency has been increased to 6 times, and the data latency has been reduced to within 15 minutes.

[0101] Data accuracy: The false negative rate dropped from 5% in the traditional model to 0.3%, and the false positive rate dropped from 3% to 0.2%, providing accurate data support for parking management departments. The turnover rate of parking spaces in commercial centers increased by 25%, and the number of complaints about illegal parking around hospitals decreased by 40%.

[0102] Resource optimization: Based on inspection data, the management department implemented differentiated pricing during peak hours (50% increase in fees from 10:00-14:00 and 19:00-22:00) in the core area of ​​the commercial center to encourage vehicles to stop briefly and leave quickly; preferential pricing was implemented in densely populated residential areas at night (20:00-7:00 the next day) to improve parking space utilization, resulting in an overall 30% improvement in parking resource allocation efficiency.

[0103] Scalability and adaptability: Six months after implementation, due to the expansion of the new urban area, 5,000 new on-street parking spaces were added. By adding 10 drone cabins and upgrading the control module scheduling algorithm, the system can achieve full coverage inspection without large-scale modification, demonstrating good adaptability.

[0104] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A drone-based on-street parking space inspection system, comprising a collaborative drone, a drone cabin, a control module, a data transmission module, and a data processing center, characterized in that: The control module achieves accurate data collection and management of on-street parking space status through integrated intelligent inspection. The core implementation process is as follows: the control module plans inspection tasks and remotely dispatches drones. After taking off from the drone bay, the drone performs inspections according to the planned route, collects images through onboard equipment and performs preliminary processing, and transmits the data in real time to the data processing center for in-depth analysis via the data transmission module. After completing the inspection, the drone returns to the drone bay to charge and undergo status detection. The drone is equipped with a high-definition image acquisition and preliminary data processing unit. The drone bay integrates take-off, landing, parking, charging, and status detection functions. The control module has task planning, navigation, and cluster scheduling capabilities. The data transmission module ensures secure and stable data transmission. The data processing center performs secondary data verification and in-depth analysis.

2. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 1, characterized in that: The high-definition image acquisition unit of the drone is an ultra-high pixel high-definition camera with autofocus, and integrates image stabilization technology that combines optical image stabilization principle and electronic image stabilization algorithm; the preliminary data processing unit is a high-performance processor with advanced computing architecture, which has parallel processing capability and can filter and classify the acquired images in real time, extract parking space occupancy status, license plate number and abnormal occupancy information and temporarily cache them.

3. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 1, characterized in that: The drone cabin adopts an integrated sealed design, with the outer shell made of high-strength corrosion-resistant composite material. The interior space is ergonomically designed to precisely fit the drone's dimensions. The bottom of the drone cabin integrates a wireless charging device based on magnetic resonance coupling technology, which can automatically sense and establish an efficient charging connection after the drone lands. It also has preliminary detection functions for the drone's battery level, body appearance, and the working status of key components.

4. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 1, characterized in that: The control module achieves unified scheduling of drone swarms through a cloud server. It can accurately set targeted inspection times based on the functional characteristics, parking demand patterns, and traffic control requirements of different urban areas. The inspection times include key inspection periods before weekday peak hours in commercial areas, before peak hours in hospitals, and before school hours.

5. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 4, characterized in that: The control module combines Geographic Information System (GIS) technology, high-precision road map data, and parking space distribution data to support manual planning or automatic generation of optimal inspection routes through intelligent algorithms. During planning, it simultaneously avoids no-fly zones and building obstructions while meeting flight safety distance requirements. It also has the dynamic adaptation capability to adjust the route in real time according to changes in road conditions such as road construction and traffic accidents. The control module integrates Global Positioning System (GPS) navigation technology, which can obtain the UAV's position, speed, and attitude information in real time and provide navigation support.

6. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 1, characterized in that: When the drone takes off, the drone cabin automatically opens its door after receiving instructions from the control module. The drone uses its high-precision sensors to monitor environmental information such as wind speed, wind direction, air pressure, and temperature and adjusts its flight attitude accordingly. During flight, the drone dynamically optimizes its flight altitude and speed based on the surrounding road environment. It increases its altitude and speed on open, unobstructed roads and decreases its altitude and speed in densely built-up areas, narrow streets, or areas with complex obstacles. The camera takes images of parking spaces at fixed intervals of 2-3 seconds.

7. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 6, characterized in that: The drone is equipped with an image recognition algorithm based on deep learning technology. It builds an image analysis model by training on a large number of parking space image samples. It can quickly determine the occupancy status of parking spaces, accurately identify license plate numbers, and automatically mark abnormal situations such as debris and non-motorized vehicles occupying the space. The image recognition algorithm can automatically optimize parameters according to the quality of the captured image and adjust image enhancement parameters in low light to improve recognition accuracy.

8. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 1, characterized in that: The data transmission module adopts 4G / 5G communication technology, which has the characteristics of high speed and low latency transmission. It also uses multiple encryption technologies to encrypt the transmitted data. At the same time, it uses error correction technology to automatically detect and correct transmission errors caused by signal interference and network fluctuations, ensuring the security and integrity of data transmission.

9. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 8, characterized in that: The data processing center is equipped with a high-performance server and uses a multi-model fusion recognition algorithm to perform secondary verification on the transmitted data. Specifically, it combines character segmentation, feature extraction, and classification techniques to improve the recognition accuracy of license plate numbers, and combines time series analysis, spatial distribution analysis, and historical data comparison analysis to perform multi-dimensional verification of parking space occupancy status. The data processing center stores the analysis results in a database and generates daily, weekly, and monthly reports on parking space usage, as well as statistical reports on abnormal occupancy.

10. The unmanned aerial vehicle (UAV) roadside parking space inspection system according to claim 3, characterized in that: After completing the inspection, the UAV adjusts its flight attitude and speed through real-time communication with the UAV cabin according to the landing command from the control module, and lands precisely at the designated location. The UAV cabin uploads the status detection information to the control module, which allows managers to arrange regular maintenance of the UAV based on the information, including replacement of vulnerable parts, battery depth detection and maintenance, and calibration of the flight control system.

Images