Unmanned aerial vehicle detection tracking adaptive low-altitude economic road picketing and logistics method

By carrying image processing equipment and deep learning models on the drone, combined with radar and TapFlyMission tasks, road violation detection and precise logistics distribution are achieved, solving the problems of insufficient flexibility and high misjudgment rates in the existing technology, and improving the non-resolution of no Comprehensive functional deployment and logistics distribution efficiency of human-machine in low-altitude economy.

CN119068369APending Publication Date: 2024-12-03HARBIN INSTITUTE OF TECHNOLOGY (SHENZHEN) (INSTITUTE OF SCIENCE AND TECHNOLOGY INNOVATION HARBIN INSTITUTE OF TECHNOLOGY SHENZHEN)
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411211173.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Existing drones have problems such as insufficient flexibility, high cost, high misjudgment rate, inability to achieve accurate custom route distribution and poor system linkage in road illegal picketing and logistics distribution.

Method used

UAVs are equipped with image processing equipment, deep learning models are used to process road scene video streams, identify traffic signs and vehicles, combine radar and TapFlyMission tasks to achieve automatic flight and logistics distribution, and use pre-trained models and feature extraction, fusion and detection. technology to realize road violation detection and precise logistics distribution.

Benefits of technology

Improve the flexibility and accuracy of road picking, reduce the misjudgment rate, save costs, realize the comprehensive functional deployment of drones in low-altitude economy, simplify management and maintenance, and improve the convenience and efficiency of logistics and distribution .

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068369A_ABST
    Figure CN119068369A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of unmanned aerial vehicles, and discloses an unmanned aerial vehicle detection and tracking adaptive low-altitude economic road picketing and logistics method, which comprises the following steps: starting an unmanned aerial vehicle and image processing equipment carried by the unmanned aerial vehicle, loading a pre-trained deep learning model to an unmanned aerial vehicle processor or a cloud server, setting the flight route, height and monitoring area of the unmanned aerial vehicle; the method further comprises the following steps that world coordinates of input radar vertical coordinate points are set as target coordinates of TapFly Mission on a DJI Mobile SDK, the unmanned aerial vehicle is controlled to automatically go to the coordinate points, and during the period, a route is automatically planned and obstacles on the way are avoided. The real-time path tracking function of an illegal unit greatly reduces the misjudgment rate and accurately captures illegal vehicles and pedestrians, obtained illegal results are directly uploaded to a background database in real time, an operator can directly receive specific data visually and rapidly, and the law enforcement burden of a manager is relieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicles, and in particular relates to a method for unmanned aerial vehicle detection, tracking and adaptation for low-altitude economic road patrol and logistics. Background Art

[0002] As the commercialization of low-altitude airspace is gradually relaxed, there is huge economic potential hidden in the low-altitude airspace below 1,000 meters in vertical height. Imagine that you can receive express delivery from drones by opening the windows of high-rise buildings every morning. You can easily take a "flying taxi" to commute during the morning rush hour to avoid road congestion. The agricultural products eaten on weekdays are all irrigated and sown by drones. The source is healthy, green and reliable. You can travel freely to all parts of the country by taking large low-altitude aircraft. The development dividends of low-altitude airspace have great hope to benefit every ordinary family and company in the near future. It can be seen that the national government has a strong concern and urgent need to build a modern low-altitude economy, and "low-altitude economy" specifically refers to the related field needs and consumer-level functions that can be undertaken and solved by various low-altitude aircraft controlled by people or unmanned. The present invention uses the environment where low-altitude drone products are located to analyze and design road violations and terminal distribution logistics technology, which helps the development of the frontier of low-altitude economy.

[0003] However, existing drones have certain defects in road violation inspection tasks and transportation logistics distribution: Existing road violation patrol tasks are usually set up on fixed surveillance-level cameras to perform single-view surveillance of intersections. Even though some brands of cameras can manually adjust the view, the flexibility is still unsatisfactory. In order to achieve comprehensive traffic control coverage at important traffic intersections, multiple high-performance cameras are required to detect at the same time, which is costly, complex to deploy and has poor synchronization. Not only that, the camera's capture effect will be greatly reduced due to severe and extreme weather, and such weather is often the peak period for road traffic violations, which is prone to a large number of errors and misjudgments.

[0004] The existing automatic transportation logistics distribution services in low-altitude airspace are limited to point-to-point distribution, that is, drone delivery between logistics warehouse and fixed-point post station. It is impossible to achieve precise customized route delivery within ten meters of the user's specific residence, and the same is true for the delivery to residents of some high-rise buildings. The existing technology only shortens the distance of logistics on a large scale, and the terminal delivery problem still needs to be analyzed and solved in detail. Similarly, the mass of objects that drones can carry for logistics transportation is limited. Usually, the cost-effectiveness of land transportation for logistics warehouse distribution is much higher than that of drone delivery. The low-altitude economy applicable to drones needs to play its advantages of high speed, high controllability and precise control over short-distance and light-weight express items.

[0005] Existing technologies cannot integrate low-altitude logistics and distribution services and road violation monitoring and patrol functions into one system. The reason is that most road patrol functions are deployed on cameras at fixed heights, while logistics and distribution require aircraft to have strong flexibility and operability. The two systems are designed and deployed separately, which has many disadvantages such as high cost, poor linkage, separate fault handling, lack of a unified management platform or tool, and poor continuous operation capabilities. However, with the commercialization of the low-altitude economy, the two functions can be carried on the same low-altitude aircraft at the same time, which not only saves time and labor costs but also facilitates maintenance by managers.

[0006] Therefore, a drone detection and tracking method adapted to low-altitude economic road patrol and logistics is proposed to solve the above problems. Summary of the invention

[0007] The purpose of the present invention is to provide a method for detecting and tracking unmanned aerial vehicles (UAVs) adapted to low-altitude economic road patrol and logistics, so as to solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for detecting and tracking drones and adapting to low-altitude economic road patrol, comprising the following steps: S1: Start the drone and its onboard image processing device, load the pre-trained deep learning model to the drone processor or cloud server, and set the drone’s flight route, altitude, and monitoring area; S2: The drone flies to the monitoring area according to the preset route and keeps hovering or flies according to the planned path. It collects the video stream of the road scene in real time through the high-resolution camera and pre-processes the collected video stream to improve the image quality. S3: Decompose the video stream into continuous image frames, use the Backbone part of the deep learning model to extract features from each frame, and generate feature maps rich in semantic information. In Backbone, the robustness and accuracy of the features are enhanced by re-parameterizing the VGG network and introducing 1×1 convolutional layers. S4: The feature map extracted by Backbone is passed to the Neck layer for feature fusion. In the Neck layer, the Fusion layer is used to perform a depth-separable convolution operation on the two adjacent levels of feature maps to achieve multi-scale feature fusion. The fused feature map is passed to the Head layer for target detection. S5: In the Head layer, the bounding box regressor and classifier are used to detect the target on the fused feature map, identify the traffic signs and vehicles on the road, classify the detected targets, and annotate their bounding boxes; S6: Based on the target classification results, traffic vehicles and traffic signs are detected for violations. For traffic vehicles, their driving trajectory is tracked and their speed is calculated to determine whether they are speeding. For traffic signs, the nearest road is matched to detect speed limit and no-parking violations and record the information of illegal vehicles.

[0009] Preferably, the preprocessing includes denoising and contrast enhancement.

[0010] Preferably, the traffic signs include zebra crossings, traffic lights and speed limit signs.

[0011] Preferably, the transportation vehicles include cars and trucks.

[0012] Preferably, the illegal vehicle information includes an image of the car, license plate information and driving trajectory.

[0013] A method for detecting and tracking drones adapted to low-altitude economic road logistics includes the following steps: A1: On the DJI Mobile SDK, set the input radar vertical coordinate point world coordinate as the target coordinate of TapFlyMission, and control the drone to automatically go to the coordinate point. During the process, the drone will automatically plan the route and avoid obstacles along the way. In order to set the route of the drone, it should stay 20 meters away from non-target residential areas. A2: After arriving near the target area, the order analysis system transmits floor information to the drone, and the drone rises to the corresponding correct height; A3: The drone's built-in floor determination system uses the order information to automatically calculate the specific floor height of the target order user. The drone's operating system then controls the drone to start another vertical TapFly Mission and ascend to the target height.

[0014] A4: The smart delivery method calls the operating system to control the drone to fly slowly around the floor at the same height, and uses the portrait recognition deep model to continuously search for nearby users. At this time, the target user needs to go to the balcony or open the window. After the drone recognizes the matching user, it calls the ActiveTrack task in the Mobile SDK; A5: When the user takes the product, the drone detects that the logistics is completed, and the order system automatically closes the tracking of this order. The drone activates the automatic return function and returns to the coordinates when the smart delivery method was started, until it returns to the drone management warehouse for maintenance and power monitoring inspection.

[0015] Preferably, in A1, automatically planning a route specifically comprises the steps of: A11: First mark the starting point and target logistics point , connect in the aerial satellite view Search whether there are intersecting or overlapping motor vehicle roads; A12: will intersect and Roads with a line angle less than 45° are stored in containers within, and The left and right endpoints of all roads in the container , stored in container Inside; A13: Traverse the container starting from the starting point A as the origin The endpoint inside determines the origin to the container The shortest distance between the inner endpoints will have the endpoint with the shortest distance The road Add to the final planned path, Containers and The endpoint information of the selected road is deleted from the container, and then the endpoint The corresponding other end point Continue traversing the road as the origin of the next traversal until a complete transportation logistics route is finally connected for travel.

[0016] Preferably, in A3, floor determination specifically includes the following steps: A31: First, perform rough positioning and calculate the rough height according to the number of floors displayed in the order corresponding to the standard floor height of the buildings in the pilot campus; A32: After the drone reaches the rough height, it will detect the floor landmarks on the building's facade. After detecting the landmarks, the drone can manually adjust the hovering position to place the landmarks at the center of the drone's field of view to locate the floor.

[0017] Preferably, the floor landmark objects include balconies and windows. Compared with the prior art, the present invention has the following beneficial effects: (1) The road patrol method provided by the present invention is efficient and flexible, easy to manage, and can be deployed at intersections where fixed-position cameras cannot cover the entire field of view, thus avoiding the high budget cost of deploying multiple cameras at different viewing angles at the same time. The real-time path tracking function for illegal units can greatly reduce the misjudgment rate and accurately capture illegal vehicles and pedestrians. The obtained illegal results can be directly uploaded to the backend database in real time, and operators can directly receive specific data, which is intuitive and fast, thus reducing the law enforcement burden on management personnel.

[0018] (2) The road logistics method provided by the present invention solves the pain point that the existing technology is difficult to complete the "last mile" intelligent delivery by building a drone operating system solution for low-altitude logistics delivery to users on high floors. It realizes unmanned intelligent terminal delivery to reduce labor cost expenditure, improves the convenience and satisfaction of users in using logistics express services, and shortens the terminal delivery cycle to improve efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a structural schematic diagram of the road picket system of the present invention; Figure 2 It is a structural schematic diagram of the logistics transportation system of the present invention.

[0020] Figure 3 is a flow chart of the distribution algorithm of the present invention; DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work should fall within the scope of protection of the present invention.

[0022] Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts disclosed in the present invention.

[0023] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of methods and systems consistent with some aspects of the present invention as detailed in the appended claims.

[0024] See also Figure 1-Figure 3 As shown, the present invention provides the following technical solutions: A method for detecting and tracking drones and adapting to low-altitude economic road patrols includes the following steps: S1: Start the drone and its onboard image processing device, load the pre-trained deep learning model to the drone processor or cloud server, and set the drone’s flight route, altitude, and monitoring area; S2: The drone flies to the monitoring area according to the preset route and keeps hovering or flies according to the planned path. It collects the video stream of the road scene in real time through the high-resolution camera and pre-processes the collected video stream to improve the image quality. S3: Decompose the video stream into continuous image frames, use the Backbone part of the deep learning model to extract features from each frame, and generate feature maps rich in semantic information. In Backbone, the robustness and accuracy of the features are enhanced by re-parameterizing the VGG network and introducing 1×1 convolutional layers.

[0025] S4: The feature map extracted by Backbone is passed to the Neck layer for feature fusion. In the Neck layer, the Fusion layer is used to perform a depth-separable convolution operation on the two adjacent levels of feature maps to achieve multi-scale feature fusion. The fused feature map is passed to the Head layer for target detection. S5: In the Head layer, the bounding box regressor and classifier are used to detect the target on the fused feature map, identify the traffic signs and vehicles on the road, classify the detected targets, and annotate their bounding boxes; S6: Based on the target classification results, traffic vehicles and traffic signs are detected for violations. For traffic vehicles, their driving trajectory is tracked and their speed is calculated to determine whether they are speeding. For traffic signs, the nearest road is matched to detect speed limit and no-parking violations and record the information of illegal vehicles.

[0026] The preprocessing includes denoising and contrast enhancement.

[0027] The traffic signs include zebra crossings, traffic lights and speed limit signs.

[0028] The transportation vehicles include cars and trucks.

[0029] The illegal vehicle information includes the image, license plate information and driving track of the car.

[0030] In addition, the present invention provides another embodiment: B1: System deployment and initialization: A drone equipped with a high-resolution camera and advanced image processing equipment is selected, and a pre-trained deep learning model optimized for urban traffic scenarios is pre-installed on the drone.

[0031] The same deep learning model is loaded on the cloud server to ensure efficient processing even when the drone is offline or the network is unstable.

[0032] The drone’s flight route is set to cover the key monitoring areas of the entire main road, the flight altitude is set to 50 meters, which is safe and does not affect traffic, and the monitoring area is accurately located through GPS.

[0033] B2: Real-time monitoring and image preprocessing: The drone flies over the monitoring area according to the preset route, remains in a hovering state, and collects real-time video streams of road scenes through high-definition cameras.

[0034] The collected video stream is preprocessed by denoising and enhancing contrast to improve image quality and provide clear and accurate image data for subsequent target detection.

[0035] B3: Feature extraction and target detection: The video stream is decomposed into continuous image frames, and the Backbone part of the deep learning model is used to extract features from each frame to generate a feature map rich in semantic information.

[0036] By introducing a re-parameterized VGG network and a 1×1 convolutional layer, the robustness and accuracy of the features are enhanced, and the ability to recognize targets under different weather and lighting conditions is improved.

[0037] The extracted feature map is passed to the Neck layer for feature fusion, and multi-scale feature fusion is realized using depthwise separable convolution to improve the accuracy and efficiency of target detection.

[0038] B4: Detection and recording of illegal activities: In the Head layer, bounding box regressors and classifiers are used to perform target detection on the fused feature maps to accurately identify traffic signs (such as zebra crossings, traffic lights, speed limit signs) and traffic vehicles (such as cars and trucks) on the road.

[0039] For detected traffic vehicles, the driving trajectory is tracked and the speed is calculated to determine in real time whether they are speeding. Once speeding is detected, the illegal vehicle information is immediately recorded, including the car's image, license plate information and driving trajectory.

[0040] At the same time, it matches the nearest road, detects compliance with traffic signs such as speed limits and no parking, and records information on vehicles that violate traffic regulations.

[0041] B5: Data transmission and subsequent processing: The detected violation data is transmitted to the traffic management center in real time for further analysis and processing by managers.

[0042] After completing one round of monitoring tasks, the drone automatically returns to the starting point and prepares for the next round of monitoring tasks, ensuring all-weather, uninterrupted road monitoring.

[0043] like Figure 1 As shown, this embodiment includes a DJI drone 1, a drone operation module 2, an image acquisition module 3, an image processing module 4, a target recognition module 5, an illegal detection module 6 and a data server 7.

[0044] Drone 1 is DJI mini4 pro, the drone operation module is connected to the SDK operation authority and interface of drone 1, the image acquisition module is connected to the video stream signal provided by drone operation module 2, the image processing module is connected to the video image of drone operation module 2, the target recognition module performs target detection and recognition on the image transmitted by module 4, the violation detection module obtains the location and category information of pedestrian and vehicle traffic signs obtained by target recognition module 5 to perform corresponding violation inspection work, and the obtained illegal unit information will be sent to the data server 7 for background storage and display.

[0045] like Figure 2 As shown, this embodiment includes a DJI drone 1, a drone operation module 2, a logistics order receiving terminal 3, an order parsing module 4, an intelligent distribution algorithm module 5, an identity recognition module 6 and an automatic return module 7.

[0046] Drone 1 is DJI mini4 pro. The drone operation module is connected to the SDK operation authority and interface of drone 1. The logistics order receiving terminal 3 is responsible for receiving the logistics order information in the area in real time. The order parsing module is connected to the logistics order receiving terminal 3 and parses the order information into address location and user information. The intelligent delivery algorithm module is simultaneously connected to the drone fixed-point flight command authority of the drone operation module 2 and the address user information of the order parsing module 4. It automatically plans the shortest route and searches for users with the same identity when arriving to complete the intelligent delivery task. After the intelligent delivery algorithm module 5 completes the task, the automatic return module controls the drone to return to the starting point to end the overall process.

[0047] A method for detecting and tracking drones adapted to low-altitude economic road logistics includes the following steps: A1: On the DJI Mobile SDK, set the input radar vertical coordinate point world coordinate as the target coordinate of TapFlyMission, and control the drone to automatically go to the coordinate point. During the process, the drone will automatically plan the route and avoid obstacles along the way. In order to set the route of the drone, it should stay 20 meters away from non-target residential areas. A2: After arriving near the target area, the order analysis system transmits floor information to the drone, and the drone rises to the corresponding correct height; A3: The drone's built-in floor determination system uses the order information to automatically calculate the specific floor height of the target order user. The drone's operating system then controls the drone to start another vertical TapFly Mission and ascend to the target height.

[0048] A4: The smart delivery method calls the operating system to control the drone to fly slowly around the floor at the same height, and uses the portrait recognition deep model to continuously search for nearby users. At this time, the target user needs to go to the balcony or open the window. After the drone recognizes the matching user, it calls the ActiveTrack task in the Mobile SDK; A5: When the user takes the product, the drone detects that the logistics is completed, and the order system automatically closes the tracking of this order. The drone activates the automatic return function and returns to the coordinates when the smart delivery method was started, until it returns to the drone management warehouse for maintenance and power monitoring inspection.

[0049] In A1, the automatic route planning specifically includes the following steps: A11: First mark the starting point and target logistics point , connect in the aerial satellite view Search whether there are intersecting or overlapping motor vehicle roads; A12: will intersect and Roads with a line angle less than 45° are stored in containers within, and The left and right endpoints of all roads in the container , stored in container Inside; A13: Traverse the container starting from the starting point A as the origin The endpoint inside determines the origin to the container The shortest distance between the inner endpoints will have the endpoint with the shortest distance The road Add to the final planned path, Containers and The endpoint information of the selected road is deleted from the container, and then the endpoint The corresponding other end point Continue traversing the road as the origin of the next traversal until a complete transportation logistics route is finally connected for travel.

[0050] In A3, floor determination specifically includes the following steps: A31: First, perform rough positioning and calculate the rough height according to the number of floors displayed in the order corresponding to the standard floor height of the buildings in the pilot campus; A32: After the drone reaches the rough height, it will detect the floor landmarks on the building facade. After detecting the landmarks, you can manually adjust the hovering position to place the landmarks at the center of the drone's field of view to locate the floor.

[0051] The floor landmarks include balconies and windows.

[0052] In addition, in the present invention, if Figure 3 As shown in the figure, the specific intelligent distribution algorithm is: B1. By receiving the address information of module 4, the starting point and the target delivery point are marked on the radar map.

[0053] B2. Call the delivery path algorithm to automatically plan the shortest route, preset the flight mission on the DJI Mobile SDK and implement it.

[0054] B3. Arrive near the target address and calculate the rough floor height in the address information. Preset the point flight mission to vertically locate to the approximate height, and then locate the precise position based on the inter-floor information such as balconies or windows.

[0055] B4. Search for user candidates around the floor, and terminate the search only when the identity recognition result is a matching user or the search times out, otherwise repeat the search task.

[0056] B5. After locking the user, it will automatically track their location and slowly approach and deliver the goods.

[0057] B6. After the delivery process is completed, the automatic return program is started to return to the drone's starting point, ending the algorithm.

[0058] In addition, the present invention provides another embodiment: C1: Drone automatically plans routes: Using the DJI Mobile SDK, the input radar vertical coordinate point world coordinates are set as the target coordinates of the TapFly Mission, and the drone automatically plans the flight route from the logistics center to the target dormitory building.

[0059] When planning a route, the drone first searches for motor vehicle roads between the starting point and the target point through overhead satellite images, excluding intersecting roads with an angle greater than 45° to ensure the shortest and safest route.

[0060] By continuously traversing the road endpoints in the container, the shortest path is selected until a complete logistics route is formed, and it is ensured that the drone stays 20 meters away from non-target residential areas during the flight.

[0061] C2: Accurate positioning and floor recognition: After the drone arrives near the target area, the order analysis system transmits the floor information through the communication module carried by the drone.

[0062] The drone automatically rises to the corresponding height based on the floor information, and calculates the specific floor height of the target user through the built-in floor judgment system combined with the order information.

[0063] Based on the rough positioning, the drone uses a high-definition camera to detect the floor landmarks (such as balconies and windows) on the building's facade, and further fine-tunes the hovering position to ensure that it stays accurately on the target floor.

[0064] C3: Intelligent delivery and user identification: The drone flies slowly at the target floor height and uses the portrait recognition deep model to search for nearby users.

[0065] When the user appears on the balcony or opens the window, the drone confirms the identity of the target user through facial recognition technology and starts the ActiveTrack task to ensure that the drone stably tracks the user.

[0066] After the user takes the product, the drone detects that the logistics is completed, automatically closes the order tracking, and activates the automatic return function to return to the drone management warehouse.

[0067] C4: Security Monitoring and Maintenance: During the entire delivery process, the drone transmits flight status and location information to the control center in real time to ensure flight safety.

[0068] After returning to the management warehouse, the drone undergoes comprehensive maintenance and power monitoring checks to prepare for the next delivery mission.

[0069] In addition, to achieve the above-mentioned purpose, an embodiment of the present invention further proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the drone detection, tracking, adaptation, low-altitude economic road patrol and logistics method of an embodiment of the present invention.

[0070] It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable vehicles (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0071] The embodiments of the present invention are described with reference to flowcharts and / or block diagrams of methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or multiple boxes.

[0072] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction system, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0073] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0074] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. "" and / or "" indicate that either one of the two can be selected, or both can be selected. Moreover, the terms "" include "" "" contain "" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "" include a..."" does not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the above embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.

[0076] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting and tracking drones and adapting to low-altitude economic road patrol, characterized by: The steps include: S1: Start the drone and its onboard image processing device, load the pre-trained deep learning model to the drone processor or cloud server, and set the drone’s flight route, altitude, and monitoring area; S2: The drone flies to the monitoring area according to the preset route and keeps hovering or flies according to the planned path. It collects the video stream of the road scene in real time through the high-resolution camera and pre-processes the collected video stream to improve the image quality. S3: Decompose the video stream into continuous image frames, use the Backbone part of the deep learning model to extract features from each frame, and generate feature maps rich in semantic information. In Backbone, the robustness and accuracy of the features are enhanced by re-parameterizing the VGG network and introducing 1×1 convolutional layers. S4: The feature map extracted by Backbone is passed to the Neck layer for feature fusion. In the Neck layer, the Fusion layer is used to perform a depth-separable convolution operation on the two adjacent levels of feature maps to achieve multi-scale feature fusion. The fused feature map is passed to the Head layer for target detection. S5: In the Head layer, the bounding box regressor and classifier are used to detect the target on the fused feature map, identify the traffic signs and vehicles on the road, classify the detected targets, and annotate their bounding boxes; S6: Based on the target classification results, traffic vehicles and traffic signs are detected for violations. For traffic vehicles, their driving trajectory is tracked and their speed is calculated to determine whether they are speeding. For traffic signs, the nearest road is matched to detect speed limit and no-parking violations and record the information of illegal vehicles.

2. A method for detecting and tracking drones and adapting to low-altitude economic road patrol according to claim 1, characterized in that: The preprocessing includes denoising and contrast enhancement.

3. The method for detecting and tracking drones and adapting to low-altitude economic road patrol according to claim 1 is characterized by: The traffic signs include zebra crossings, traffic lights and speed limit signs.

4. The method for detecting and tracking drones and adapting to low-altitude economic road patrol according to claim 1 is characterized by: The transportation vehicles include cars and trucks.

5. The method for detecting and tracking drones and adapting to low-altitude economic road patrol according to claim 1 is characterized by: The illegal vehicle information includes the image, license plate information and driving track of the car.

6. A method for detecting and tracking drones adapted to low-altitude economic road logistics, characterized in that: The steps include: A1: On the DJI Mobile SDK, set the input radar vertical coordinate point world coordinate as the target coordinate of TapFly Mission, and control the drone to automatically go to the coordinate point. During the process, the drone will automatically plan the route and avoid obstacles along the way. The route of the drone is set to stay 20 meters away from non-target residential areas. A2: After arriving near the target area, the order analysis system transmits floor information to the drone, and the drone rises to the corresponding correct height; A3: The drone is equipped with a gimbal camera with an embedded floor judgment system that automatically calculates the specific floor height of the target order user based on the order information. The drone operating system controls the drone to start another vertical TapFly Mission and ascend to the target height. A4: The smart delivery method calls the operating system to control the drone to fly slowly around the floor at the same height, and uses the portrait recognition deep model to continuously search for nearby users. At this time, the target user needs to go to the balcony or open the window. After the drone recognizes the matching user, it calls the ActiveTrack task in the Mobile SDK; A5: When the user takes the product, the drone detects that the logistics is completed, and the order system automatically closes the tracking of this order. The drone activates the automatic return function and returns to the coordinates when the smart delivery method was started, until it returns to the drone management warehouse for maintenance and power monitoring inspection.

7. A method for detecting and tracking drones and adapting to low-altitude economic road logistics according to claim 6, characterized in that: In A1, the automatic route planning specifically includes the following steps: A11: First mark the starting point and target logistics point , connect in the aerial satellite view Search whether there are intersecting or overlapping motor vehicle roads; A12: will intersect and Roads with a line angle less than 45° are stored in containers within, and The left and right endpoints of all roads in the container , stored in container Inside; A13: Traverse the container starting from the starting point A as the origin The endpoint inside determines the origin to the container The shortest distance between the inner endpoints will have the endpoint with the shortest distance The road Add to the final planned path, Containers and The endpoint information of the selected road is deleted from the container, and then the endpoint The corresponding other end point Continue traversing the road as the origin of the next traversal until a complete transportation logistics route is finally connected for travel.

8. The method for detecting and tracking drones and adapting to low-altitude economic road logistics according to claim 6 is characterized by: In A3, floor determination specifically includes the following steps: A31: First, perform rough positioning and calculate the rough height according to the number of floors displayed in the order corresponding to the standard floor height of the buildings in the pilot campus; A32: After the drone reaches the rough altitude, it will detect the floor landmark objects on the building facade.

9. A method for detecting and tracking drones and adapting to low-altitude economic road logistics according to claim 6, characterized in that: The floor landmark objects include balconies and windows.

10. A method for detecting and tracking drones and adapting to low-altitude economic road logistics according to claim 8, characterized in that: After the marker is detected, the hovering position can be manually adjusted to place the marker at the origin position in the center of the drone's field of view to achieve the purpose of locating the floor.

Citation Information

Cited By

  • Unmanned aerial vehicle detection tracking adaptive low-altitude economic road picketing and logistics method

    CN120708393A