An intelligent anti-theft system and method for a drone nest

Abstract

The application relates to an intelligent anti-theft system and method for a drone nest, which comprises a heat source sensing module, an image acquisition module, a main control unit, an AI recognition module, a hierarchical decision engine and an execution module. The heat source sensing module is used for detecting whether a heat source appears in a current environment at a preset frequency, and the main control unit is woken up when an environmental heat source is collected. The main control unit comprises the AI recognition module and the hierarchical decision engine. After being woken up, the main control unit starts the image acquisition module to collect image information, carries out face recognition and personnel behavior analysis according to the image information through the AI recognition module, obtains an analysis result, generates a corresponding response strategy according to the analysis result and a hierarchical response mechanism through the hierarchical decision engine, and generates an execution action according to the response strategy of the hierarchical decision engine. The average power consumption of the image acquisition module and the AI recognition module is reduced through the front-end wake-up of the heat source sensing module, AI recognition is improved, the identification capability for legal operation and maintenance personnel is improved, and the identification precision is improved.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) technology, specifically to an intelligent anti-theft system and method for UAV nests. Background Technology

[0002] With the widespread use of drones in logistics, inspection, emergency response, and security, a large number of drone airfields (drone airports) are deployed in unattended areas such as outdoors, urban rooftops, industrial parks, and roadsides. As the hub for drones to stay, charge, take off, land, and communicate, the safety of these airfields directly affects the reliability of drone missions and data security.

[0003] In the field of drone nest anti-theft technology, existing technologies mainly suffer from three major bottlenecks: insufficient security, significant energy efficiency issues, and limited intelligence. Traditional solutions rely on mechanical locks and fixed-angle cameras, triggering local alarms via magnetic door switches. However, mechanical locks are easily damaged by force, the camera's field of view is limited (usually ≤90°), and it lacks AI recognition capabilities, resulting in a high false alarm rate. Another type of solution introduces AI algorithms for facial recognition, but it relies on only a single camera and lacks integrated environmental perception sensors. This leads to a high false alarm rate in scenarios with strong lighting changes or occlusion, and it lacks a physical damage detection module, failing to trigger an effective alarm when the entire drone nest is moved.

[0004] The existing technical solutions mainly have the following problems:

[0005] 1. Traditional cameras that are left on for extended periods and have high power consumption: Traditional cameras often need to be left on for long periods of time, which leads to high power consumption. In addition, the field of view of ordinary cameras is relatively limited, which can easily create blind spots and is not conducive to all-round monitoring.

[0006] 2. Lack of AI recognition capability: Most existing technologies lack AI recognition, resulting in a high false alarm rate. In particular, it is difficult to distinguish between legitimate personnel and illegal intruders in complex environments, and it is impossible to manage whitelisted users through identity recognition.

[0007] 3. Lack of coordination between sensors: In existing solutions, various sensors usually work independently and cannot be effectively linked and coordinated, resulting in insufficient sensitivity and accuracy in protection response.

[0008] 4. Lack of remote collaboration capabilities: Many existing machine nest anti-theft systems cannot achieve remote collaboration between the cloud and the equipment, and cannot achieve real-time remote monitoring and scheduling, which reduces the efficiency of emergency response. Summary of the Invention

[0009] In view of the above problems, this application provides an intelligent anti-theft system and method for drone nests, which solves the problems of existing drone nest anti-theft technologies, such as the long-term operation of ordinary cameras, resulting in high power consumption, blind spots that are not conducive to all-round monitoring, and the lack of AI recognition capabilities, making it difficult to distinguish between legitimate personnel and illegal intruders in complex environments.

[0010] To achieve the above objectives, the inventors provide an intelligent anti-theft system for drone nests, comprising:

[0011] A heat source sensing module is used to detect whether a heat source exists in the current environment at a preset frequency. When an environmental heat source is detected, the main control unit is woken up.

[0012] An image acquisition module, which is used to acquire image information, includes a fisheye camera;

[0013] The main control unit includes an AI recognition module and a hierarchical decision engine. When the main control unit is woken up, it starts the image acquisition module to acquire image information, and performs face recognition and human behavior analysis based on the image information through the AI ​​recognition module to obtain analysis results. The hierarchical decision engine generates corresponding response strategies based on the analysis results and hierarchical response mechanisms.

[0014] The execution module generates execution actions based on the response strategy of the hierarchical decision engine.

[0015] In some embodiments, the main control unit is further configured to use an AI recognition module to collect facial information from image information, compare the facial information with a whitelist database and perform personnel behavior analysis to obtain analysis results. When facial recognition matches the whitelist, a first-level response strategy is generated through a hierarchical decision engine. When facial recognition identifies an unfamiliar face, a second-level response strategy is generated through the hierarchical decision engine. When abnormal behavior is detected, a third-level response strategy is generated through the hierarchical decision engine.

[0016] In some embodiments, the abnormal behavior includes: being identified as a stranger, staying for more than a preset time, or possessing a tool.

[0017] In some embodiments, it also includes:

[0018] An attitude detection module is used to collect attitude data of the UAV nest.

[0019] The main control unit is also used to execute a three-level response strategy when the drone nest is determined to be abnormal based on attitude data.

[0020] In some embodiments, the execution module includes a body sound and light alarm and a communication module;

[0021] The main control unit is also used to log and mark locations when executing a level-one response strategy, to issue an audible and visual alarm and record video via an image acquisition module when executing a level-two response strategy, and to send a human intervention request via a communication module and record video via an image acquisition module when executing a level-three response strategy.

[0022] In some embodiments, it also includes:

[0023] The cloud-based AI platform is used to receive structured alarm packets sent by the communication module and perform secondary verification, and generate corresponding handling strategies based on the results of the secondary verification.

[0024] Another technical solution is also provided: a smart anti-theft method for drone nests, which includes the following steps:

[0025] The heat source sensing module detects whether there is a heat source in the current environment at a preset frequency. When an ambient heat source is detected, the main control unit is woken up.

[0026] After the main control unit is woken up, it starts the image acquisition module and acquires image information through the image acquisition module;

[0027] The AI ​​recognition module performs facial recognition and human behavior analysis on image information to obtain analysis results;

[0028] The hierarchical decision engine generates corresponding response strategies based on the analysis results and the hierarchical response mechanism.

[0029] The execution module generates execution actions based on the response strategy of the hierarchical decision engine.

[0030] In some embodiments, the hierarchical decision engine generates a corresponding response strategy based on the analysis results and the hierarchical response mechanism, specifically including the following steps:

[0031] When facial recognition matches the whitelist, a first-level response strategy is generated through a hierarchical decision engine;

[0032] When the face recognition identifies the face as that of an unfamiliar person, a secondary response strategy is generated through a hierarchical decision engine.

[0033] When abnormal behavior is detected, a three-level response strategy is generated through a hierarchical decision engine.

[0034] In some embodiments, the execution module includes an audible and visual alarm and a communication module;

[0035] The execution module generates execution actions based on the response strategy of the hierarchical decision engine, specifically including the following steps:

[0036] When executing the Level 1 response strategy, log entries and location markers are performed;

[0037] When the level-two response strategy is executed, an audible and visual alarm is triggered by the audible and visual alarm device, and video is recorded by the image acquisition module.

[0038] When implementing the three-level response strategy, a human intervention request is sent through the communication module and video is recorded through the image acquisition module.

[0039] In some embodiments, the following steps are also included:

[0040] The structured alarm package, which includes timestamps, location information, image information, and pose data, is sent to the remote AI platform via the communication module.

[0041] The cloud-based AI platform performs secondary verification based on the structured alarm package and implements corresponding handling strategies based on the secondary verification results.

[0042] Unlike existing technologies, the above solution, when no abnormalities occur, only uses a heat source sensing module to detect the presence of heat sources in the current environment at a preset frequency, while the main control unit remains in sleep mode, and the image acquisition module and execution module are turned off. When a heat source is detected in the current environment, the main control unit is awakened. After being awakened, the main control unit starts the image acquisition module to acquire image information, and uses the AI ​​recognition module to perform facial recognition and personnel behavior analysis on the image information to obtain corresponding analysis results. Then, a hierarchical decision engine generates a corresponding response strategy based on the analysis results and hierarchical response mechanism, and the execution module executes the response action according to the generated response strategy. By using the heat source sensing module as the front-end wake-up, the average power consumption of the image acquisition module and AI recognition module is reduced. The use of a fisheye camera enables 180° panoramic image acquisition, increasing the image recognition range. Simultaneously, combined with the AI ​​recognition module's AI recognition capabilities, the ability to identify legitimate maintenance personnel is improved, and the recognition accuracy is increased.

[0043] The above description of the invention is merely an overview of the technical solution of this application. In order to enable those skilled in the art to better understand the technical solution of this application and to implement it based on the description and drawings, and to make the above-mentioned objectives and other objectives, features and advantages of this application easier to understand, the following description is provided in conjunction with the specific embodiments and drawings of this application. Attached Figure Description

[0044] The accompanying drawings are only used to illustrate the principles, implementation methods, applications, features, and effects of specific embodiments of this application and other related content, and should not be considered as limitations on this application.

[0045] In the accompanying drawings of the instruction manual:

[0046] Figure 1This is a schematic diagram of a structure for an intelligent anti-theft system for a drone nest, as described in a specific implementation.

[0047] Figure 2 A schematic diagram of the system architecture of the intelligent anti-theft system for the drone nest described in a specific implementation;

[0048] Figure 3 This is a schematic diagram illustrating the workflow of the intelligent anti-theft system for the drone nest described in a specific implementation.

[0049] Figure 4 This is a flowchart illustrating one specific implementation of the multimodal fusion algorithm.

[0050] Figure 5 This is a flowchart illustrating a specific implementation of the intelligent anti-theft method for the drone nest.

[0051] The reference numerals used in the above figures are explained as follows:

[0052] 110. Heat source sensing module,

[0053] 120. Image acquisition module,

[0054] 130. Main control unit,

[0055] 131. AI Recognition Module

[0056] 132. Hierarchical decision engine

[0057] 140. Execution module;

[0058] 141. Audible and visual alarm device

[0059] 142. Communication module

[0060] 150. Attitude detection module. Detailed Implementation

[0061] To illustrate the possible application scenarios, technical principles, implementable specific solutions, and achievable objectives and effects of this application in detail, the following description, in conjunction with the listed specific embodiments and accompanying drawings, provides a detailed explanation. The embodiments described herein are merely illustrative of the technical solutions of this application and are therefore intended to limit the scope of protection of this application.

[0062] In this document, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The term "embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment, nor does it specifically limit its independence or connection with other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0063] Unless otherwise defined, the technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the use of related terms herein is merely for the purpose of describing particular embodiments and is not intended to limit this application.

[0064] In the description of this application, the term "and / or" is used to describe the logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and A and B exist simultaneously. Additionally, the character " / " in this document generally indicates that the preceding and following objects have an "or" logical relationship.

[0065] In this application, terms such as “first” and “second” are used only to distinguish one entity or operation from another, and do not necessarily require or imply any actual quantity, hierarchy or order relationship between these entities or operations.

[0066] Unless otherwise specified, the use of terms such as “comprising,” “including,” “having,” or other similar expressions in this application is intended to cover non-exclusive inclusion, which does not exclude the presence of additional elements in a process, method, or product that includes the stated elements, such that a process, method, or product that includes a list of elements may include not only those defined elements but also other elements not expressly listed, or elements inherent to such a process, method, or product.

[0067] As understood in the Examination Guidelines, in this application, expressions such as "greater than," "less than," and "exceeding" are understood to exclude the stated number; expressions such as "above," "below," and "within" are understood to include the stated number. Furthermore, in the description of the embodiments in this application, "multiple" means two or more (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups" and "multiple times," unless otherwise explicitly specified.

[0068] In the description of the embodiments of this application, the space-related expressions used, such as "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "vertical," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," indicate the orientation or positional relationship based on the orientation or positional relationship shown in the specific embodiments or drawings. They are only for the purpose of describing the specific embodiments of this application or for the reader's understanding, and do not indicate or imply that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0069] Unless otherwise expressly specified or limited, the terms "installation," "connection," "linking," "fixing," and "setting," as used in the description of the embodiments of this application, should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral setting; it can be a mechanical connection, an electrical connection, or a communication connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be the internal connection of two components or the interaction between two components. For those skilled in the art to which this application pertains, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0070] Please see Figure 1 This embodiment discloses an intelligent anti-theft system for drone nests, comprising:

[0071] A heat source sensing module 110 is used to detect whether a heat source exists in the current environment at a preset frequency. When an environmental heat source is detected, the main control unit 130 is activated. The heat source sensing module 110 employs a PIR sensor: a pyroelectric infrared sensor (detection angle 110°, detection distance 7m), responsible for low-power heat source detection. It is connected to the main control unit 130 via a GPIO interface to achieve milliwatt-level standby power consumption. PIR sensor: Pyroelectric Infrared Sensor.

[0072] Image acquisition module 120 is used to acquire image information. The image acquisition module 120 includes a fisheye camera, which is a 3-megapixel 180° panoramic camera. It is connected to the main control unit 130 through a MIPI interface, supports H.265 encoding, and has wide dynamic range (WDR) and infrared night vision functions.

[0073] The main control unit 130 includes an AI recognition module 131 and a hierarchical decision engine 132. When the main control unit 130 is woken up, it starts the image acquisition module 120 to acquire image information, and performs face recognition and personnel behavior analysis based on the image information through the AI ​​recognition module 131 to obtain analysis results. The hierarchical decision engine 132 generates corresponding response strategies based on the analysis results and hierarchical response mechanisms.

[0074] The execution module 140 generates execution actions according to the response strategy of the hierarchical decision engine 132.

[0075] When no abnormalities occur, the heat source sensing module 110 detects the presence of heat sources in the current environment at a preset frequency, while the main control unit 130 remains in sleep mode, and the image acquisition module 120 and execution module 140 are turned off. When a heat source is detected in the current environment, the main control unit 130 is awakened. After being awakened, the main control unit 130 starts the image acquisition module 120 to acquire image information, and the AI ​​recognition module 131 performs facial recognition and personnel behavior analysis on the image information to obtain corresponding analysis results. Then, the hierarchical decision engine 132 generates a corresponding response strategy based on the analysis results and hierarchical response mechanism, and the execution module 140 executes the response action according to the generated response strategy. By using the heat source sensing module 110 as the front-end wake-up, the average power consumption of the image acquisition module 120 and the AI ​​recognition module 131 is reduced. The use of a fisheye camera enables 180° panoramic image acquisition, increasing the image recognition range. At the same time, combined with the AI ​​recognition module 131's AI recognition, the ability to identify legitimate maintenance personnel is improved, and the recognition accuracy is increased.

[0076] Among them, the main control unit 130 is equipped with an RK3568 chip (1TOPS computing power NPU) and has a built-in multimodal fusion algorithm and whitelist database (AES-256 encrypted storage).

[0077] AI Recognition Module 131: Based on the YOLOv8-tiny lightweight model, it achieves face detection (mAP>90%), behavior analysis (loitering / tool ​​holding detection) and abnormal posture recognition.

[0078] Hierarchical Decision Engine 132: Based on AI results and IMU data, dynamically trigger a three-level response mechanism (log recording → audible and visual alarm → emergency call).

[0079] In some embodiments, the main control unit 130 is further configured to collect facial information from image information through the AI ​​recognition module 131, compare the facial information with a whitelist database and perform personnel behavior analysis to obtain analysis results. When the facial recognition matches the whitelist, a first-level response strategy is generated through a hierarchical decision engine. When the facial recognition is a stranger, a second-level response strategy is generated through the hierarchical decision engine. When abnormal behavior is detected, a third-level response strategy is generated through the hierarchical decision engine.

[0080] The AI ​​recognition module 131 collects facial information from the acquired image information and compares it with a whitelist database. Simultaneously, it analyzes the behavior of individuals within the image information. When the facial recognition result matches the whitelist, a hierarchical decision engine generates a first-level response strategy. When the face is identified as an unfamiliar face, a second-level response strategy is generated. When abnormal behavior is detected, a third-level response strategy is generated. Based on different analysis results, corresponding response strategies are generated to address different situations. Abnormal behavior includes: being identified as a stranger staying for more than a preset time or holding a tool. When a stranger is identified staying for more than a preset time, or holding a tool, such as a metal object or a long pole, abnormal behavior is detected, and a third-level response strategy is generated.

[0081] In some embodiments, it also includes:

[0082] The attitude detection module 150 is used to collect attitude data of the UAV nest; the main control unit 130 is also used to execute a three-level response strategy when the attitude data indicates that the UAV nest is abnormal.

[0083] The system also uses an attitude detection module 150 to monitor the drone nest's attitude data in real time. When the attitude data indicates an anomaly in the drone nest, a three-level response strategy is executed. The attitude detection module 150 can employ an IMU module to detect the drone nest's tilt angle and acceleration. If the drone nest's tilt angle is greater than 5° or the acceleration is greater than 0.3g, it is determined to be a movement or impact attack, indicating abnormal behavior and triggering the three-level response strategy. The IMU module integrates a three-axis accelerometer and gyroscope (MPU6050 chip) to monitor the drone nest's attitude data in real time. 2 The C interface transmits attitude data in real time, and the tilt angle > 5° / acceleration > 0.3g is set as the abnormal threshold.

[0084] In some embodiments, the execution module 140 includes an audible and visual alarm 141 and a communication module 142; the audible and visual alarm 141 consists of a 110dB buzzer and an RGB strobe light, controlled via a PWM interface. The communication module 142 is a 4G Cat.1 DTU (supporting MQTT / HTTP protocols) that enables remote reporting of structured alarm packets (JSON format).

[0085] The main control unit 130 is also used to log and mark locations when executing a first-level response strategy, to issue an audible and visual alarm through the audible and visual alarm 141 and to record video through the image acquisition module 120 when executing a second-level response strategy, and to send a human intervention request through the communication module 142 and to record video through the image acquisition module 120 when executing a third-level response strategy.

[0086] When the identified facial information matches the whitelist, the corresponding personnel information is logged and their location is marked. The behavior of the whitelisted personnel is recorded for later traceability. When the Level 2 response strategy is executed, an audible and visual alarm 141 is activated to warn strangers, and video recording is performed by the image acquisition module 120. This recording can be done using a fisheye camera or an additional high-definition camera. When the Level 3 response strategy is executed, a human intervention request is sent through the communication module 142, allowing supervisory personnel to handle the situation. The image acquisition module 120 also records the event. The audible and visual alarm 141 can also be activated when the Level 3 response strategy is executed.

[0087] In some embodiments, it also includes:

[0088] The cloud-based AI platform is used to receive structured alarm packets sent by the communication module 142 and perform secondary verification, and generate corresponding handling strategies based on the results of the secondary verification.

[0089] When the main control module is activated, it generates a structured alarm package via the AI ​​recognition module 131 to perform corresponding identification and classification strategies. This alarm package is then sent to the cloud AI platform for secondary verification. Based on the verification results, a corresponding response strategy is generated. For example, in the event of a low-risk incident, the alarm is pushed to the management personnel's app. In the event of a high-risk incident, the system automatically triggers drone countermeasures (such as verbal warnings or the deployment of barricades) or remotely locks the drone's nest door. Low-risk incidents include identifying strangers, while high-risk incidents include abnormal behavior by strangers or abnormal drone nest status.

[0090] In some embodiments, an intelligent anti-theft system for unmanned aerial vehicle (UAV) nests is provided, comprising a three-layer architecture of perception layer, decision layer, and execution layer, as shown in the system architecture below. Figure 2 As shown.

[0091] Perception layer

[0092] PIR sensor: Employs a pyroelectric infrared sensor (detection angle 110°, detection distance 7m) to detect low-power heat sources. It connects to the main control unit via a GPIO interface to achieve milliwatt-level standby power consumption.

[0093] Fisheye camera: 3-megapixel 180° panoramic camera, connected to the main control unit via MIPI interface, supports H.265 encoding, and features wide dynamic range (WDR) and infrared night vision.

[0094] IMU module: integrates a three-axis accelerometer and gyroscope (MPU6050 chip), via I... 2 The C interface transmits attitude data in real time, and the tilt angle > 5° / acceleration > 0.3g is set as the abnormal threshold.

[0095] decision-making level

[0096] Main control unit: Equipped with RK3568 chip (1TOPS computing power NPU), with built-in multimodal fusion algorithm and whitelist database (AES-256 encrypted storage).

[0097] AI recognition module: Based on the YOLOv8-tiny lightweight model, it achieves face detection (mAP>90%), behavior analysis (loitering / tool ​​holding detection) and abnormal posture recognition.

[0098] Hierarchical decision engine: Based on AI results and IMU data, dynamically trigger a three-level response mechanism (log recording → audible and visual alarm → emergency call).

[0099] Execution layer

[0100] Audible and visual alarm: 110dB buzzer + RGB strobe light, controlled via PWM interface.

[0101] Communication module: 4G Cat.1 DTU (supports MQTT / HTTP protocol) to realize remote reporting of structured alarm packets (JSON format).

[0102] If no anomalies are detected within a specified time, the device enters sleep mode, with only the PIR sensor operating. The PIR sensor detects a dynamic heat source, triggering the device to wake up and the system begins operation. The fisheye camera starts and captures images. The AI ​​world module detects anomalies in the image. If no obvious anomalies are detected and other detection devices issue warnings, it is marked as a false alarm, and the current false alarm image and information are uploaded. If no obvious anomalies are detected, the system enters sleep mode. If the AI ​​detects an anomaly in the image, and it is a non-human entity, it initiates an audio-visual deterrent, uploading the current image and information. If the AI ​​detects a human presence, it checks a whitelist. If the human is on the whitelist, a level one response is triggered, recording the access log and uploading the current image and information. If the human is not on the whitelist, a level two response is triggered, issuing an audio-visual warning and continuously recording the current image. If the human leaves within a specified time without other anomalies, the system enters sleep mode. If the human remains for an extended period, a human intervention request is sent, allowing a human to take over the system. When the IMU detects movement or abnormal acceleration, an audio-visual warning is triggered directly, the camera continuously records the current image, and a human intervention request is sent. In summary, the main method is: the system is woken up by the PIR sensor, the camera records the current image, and the AI ​​intelligently analyzes and makes decisions.

[0103] Specifically, the workflow of this system is as follows: Figure 3 As shown, a four-stage protection logic is adopted: "PIR wake-up → visual confirmation → physical verification → cloud-edge collaboration".

[0104] Hibernation / Standby:

[0105] The system defaults to low power mode, with only the PIR sensor polling the ambient heat source at a frequency of 0.5Hz (power consumption <5mW).

[0106] Function: To avoid continuous power consumption of the camera and AI module, the actual standby time is extended to 72 hours (traditional solution <24 hours).

[0107] Dynamic wake-up:

[0108] When the PIR detects the movement of the heat source, it wakes up the main control unit via an interrupt signal (response time <200ms).

[0109] Significance: This solves the problem of false triggering in traditional solutions, activating high-power devices only when an actual intrusion occurs.

[0110] Visual verification:

[0111] The main control unit activates the fisheye camera to acquire 180° panoramic images, and the AI ​​module performs the following analysis:

[0112] Face detection: Identifies individuals and compares them with a whitelist database (supports 1:N comparison, response time <1s).

[0113] Behavioral analysis: Detects loitering (staying for more than 30 seconds), tool holding (recognition of metal / long objects), and destructive behavior (detection of violent movements).

[0114] Hierarchical processing:

[0115] Whitelist matching → Level 1 response (log recording + location identification);

[0116] Unfamiliar face → Level 2 response (audio-visual alarm + video enhancement);

[0117] Abnormal behavior → Level 3 response (continuous recording + emergency alarm).

[0118] Physical verification:

[0119] The IMU module monitors the nest attitude in real time.

[0120] If the tilt angle is greater than 5° or the acceleration is greater than 0.3g, it is judged as a moving or impact attack;

[0121] The system synchronously triggers a three-level response, uploading acceleration curves and attitude angle change data via the communication module.

[0122] Cloud-edge collaboration:

[0123] Structured alarm packets (including timestamps, GPS coordinates, image hash values, and IMU data) are reported to the cloud via the 4G network;

[0124] The cloud-based AI platform performs secondary verification and generates a response strategy.

[0125] Low-risk events → pushed to the administrator's APP;

[0126] High-risk events → Automatically trigger drone countermeasures (shouting warnings / throwing barricades) or remotely lock the drone's hatch doors.

[0127] In some embodiments, such as Figure 4The multimodal fusion algorithm architecture shown uses multimodal perception fusion: weighted fusion of PIR heat source detection (low-power wake-up), visual recognition (identity authentication), and IMU attitude monitoring (physical damage determination), and Kalman filtering algorithm to eliminate sensor noise, reducing the false alarm rate to less than 1%. Specifically, after waking up via heat source detection, the visually acquired image information is used for face recognition and behavior analysis via AI based on the YOLOv8-tiny lightweight model. IMU data is collected through the IMU module. When the confidence level of AI recognition is greater than a preset value (e.g., 0.8), a visual score is calculated as Score_Vision = Conf * (1 - IOU_loss). The collected IMU data is then scored according to the IMU scoring formula: Score_IMU = 1 - (|θ |θ_max + |a |a_max) / 2. Finally, the total score Score_Total = 0.7Score_Vision + 0.3Score_IMU is calculated based on the visual score and the IMU score to obtain the threat score. A response strategy is then generated based on the threat score. For example, if the threat score is greater than 0.8, a level 3 response strategy is generated; if it is between 0.5 and 0.8, a level 2 response strategy is generated; and if it is less than 0.5, a level 1 response strategy is generated.

[0128] This system achieves low-power, highly reliable, and intervention-enabled anti-theft system for drone nests by combining multiple sensors and AI intelligent recognition technologies. It integrates multi-modal perception fusion: weighted fusion of PIR heat source detection (low-power wake-up), visual recognition (identity authentication), and IMU attitude monitoring (physical damage assessment), and uses a Kalman filter algorithm to eliminate sensor noise, reducing the false alarm rate to less than 1%. Simultaneously, a tiered response mechanism is employed to dynamically adjust the response strategy based on the threat level.

[0129]

[0130] Cloud-edge collaborative architecture: Real-time inference is achieved at the edge (NPU computing power supports 5 frames of image processing per second), while the cloud undertakes big data analysis and strategy generation tasks, and bidirectional command issuance (such as remote reset and mode switching) is achieved through the MQTT protocol.

[0131] This paper presents a low-power, highly reliable, and timely human-intervention-enabled anti-theft system for drone nests to ensure the safety and long-term operation of outdoor drone nests. Compared with existing technologies, it has at least the following advantages:

[0132] 1) Employing a multimodal hierarchical collaborative mechanism significantly reduces the false alarm rate;

[0133] 2) Using PIR as the front-end wake-up reduces the average power consumption of the camera and AI modules;

[0134] 3) Improve the ability to identify legitimate operation and maintenance personnel through AI recognition and whitelist mechanisms;

[0135] 4) Effective identification of moving, prying, and destructive behaviors is achieved through IMU detection;

[0136] 5) Supports cloud-based linkage and remote alarms, improving operational efficiency and security levels.

[0137] Please see Figure 5 In another embodiment, a smart anti-theft method for drone nests is applied to the smart anti-theft system for drone nests in the above embodiments. The method includes the following steps:

[0138] Step S510: The heat source sensing module detects whether there is a heat source in the current environment at a preset frequency. When an environmental heat source is detected, the main control unit is woken up.

[0139] Step S520: After the main control unit is woken up, it starts the image acquisition module and acquires image information through the image acquisition module;

[0140] Step S530: The AI ​​recognition module performs face recognition and personnel behavior analysis on the image information to obtain the analysis results;

[0141] Step S540: The hierarchical decision engine generates corresponding response strategies based on the analysis results and the hierarchical response mechanism;

[0142] Step S550: The execution module generates execution actions based on the response strategy of the hierarchical decision engine.

[0143] Under normal circumstances, the system only uses the heat source sensing module to detect the presence of heat sources in the current environment at a preset frequency, while the main control unit remains in sleep mode, and the image acquisition module and execution module are turned off. When a heat source is detected, the main control unit is awakened. Upon awakening, the main control unit activates the image acquisition module to acquire image information. The AI ​​recognition module then performs facial recognition and personnel behavior analysis on the image information to obtain corresponding analysis results. A hierarchical decision engine then generates a corresponding response strategy based on the analysis results and a hierarchical response mechanism. Finally, the execution module executes the response action according to the generated response strategy. By using the heat source sensing module as the front-end wake-up point, the average power consumption of the image acquisition module and AI recognition module is reduced. The use of a fisheye camera enables 180° panoramic image acquisition, increasing the image recognition range. Simultaneously, the AI ​​recognition module's AI recognition capabilities improve the ability to identify legitimate maintenance personnel, thereby increasing recognition accuracy.

[0144] In some embodiments, the hierarchical decision engine generates a corresponding response strategy based on the analysis results and the hierarchical response mechanism, specifically including the following steps:

[0145] When facial recognition matches the whitelist, a first-level response strategy is generated through a hierarchical decision engine;

[0146] When the face recognition identifies the face as that of an unfamiliar person, a secondary response strategy is generated through a hierarchical decision engine.

[0147] When abnormal behavior is detected, a three-level response strategy is generated through a hierarchical decision engine.

[0148] The AI ​​recognition module collects facial information from the acquired images and compares it with a whitelist database. Simultaneously, it analyzes the behavior of individuals within the images. When the facial recognition result matches the whitelist, a hierarchical decision engine generates a first-level response strategy. When the face is identified as an unfamiliar face, a second-level response strategy is generated. When abnormal behavior is detected, a third-level response strategy is generated. Based on different analysis results, corresponding response strategies are generated to address different situations. Abnormal behavior includes being identified as a stranger staying for more than a preset time or holding a tool. If a stranger is identified staying for more than a preset time or holding a tool, such as a metal object or a long stick, an abnormal behavior is detected, and a third-level response strategy is generated.

[0149] In some embodiments, the execution module includes an audible and visual alarm and a communication module;

[0150] The execution module generates execution actions based on the response strategy of the hierarchical decision engine, specifically including the following steps:

[0151] When executing the Level 1 response strategy, log entries and location markers are performed;

[0152] When the level-two response strategy is executed, an audible and visual alarm is triggered by the audible and visual alarm device, and video is recorded by the image acquisition module.

[0153] When implementing the three-level response strategy, a human intervention request is sent through the communication module and video is recorded through the image acquisition module.

[0154] When a facial recognition match is found on the whitelist, the corresponding person's information is logged and their location is marked. The behavior of whitelisted individuals is recorded for later traceability. When a Level 2 response strategy is executed, an audible and visual alarm is activated to warn strangers, and video recording is performed via an image acquisition module. This recording can be done using a fisheye camera or an additional high-definition camera. When a Level 3 response strategy is executed, a human intervention request is sent via a communication module, allowing supervisory personnel to handle the situation. Video recording is also performed via the image acquisition module. An audible and visual alarm can also be activated when a Level 3 response strategy is executed.

[0155] In some embodiments, the following steps are also included:

[0156] The structured alarm package, which includes timestamps, location information, image information, and pose data, is sent to the remote AI platform via the communication module.

[0157] The cloud-based AI platform performs secondary verification based on the structured alarm package and implements corresponding handling strategies based on the secondary verification results.

[0158] When the main control module is activated, it generates a structured alarm package and sends it to the cloud AI platform for secondary verification. Based on the verification results, a corresponding response strategy is generated. For example, in the event of a low-risk incident, a notification is sent to the management personnel's app. In the event of a high-risk incident, automatic drone countermeasures (such as verbal warnings or the deployment of barricades) or remote locking of the drone's nest door are triggered. Low-risk incidents include the detection of strangers, while high-risk incidents include abnormal behavior by strangers or abnormal drone nest status.

[0159] Finally, it should be noted that although the above embodiments have been described in the text and drawings of this application, this should not limit the scope of patent protection of this application. Any technical solutions that are based on the essential concept of this application and utilize the content described in the text and drawings of this application, resulting in equivalent structural or procedural substitutions or modifications, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of patent protection of this application.

Claims

1. An intelligent anti-theft system for drone nests, characterized in that, include: A heat source sensing module is used to detect whether a heat source exists in the current environment at a preset frequency. When an environmental heat source is detected, the main control unit is woken up. An image acquisition module, which is used to acquire image information, includes a fisheye camera; The main control unit includes an AI recognition module and a hierarchical decision engine. When the main control unit is woken up, it starts the image acquisition module to acquire image information, and performs face recognition and human behavior analysis based on the image information through the AI ​​recognition module to obtain analysis results. The hierarchical decision engine generates corresponding response strategies based on the analysis results and hierarchical response mechanisms. The execution module generates execution actions based on the response strategy of the hierarchical decision engine.

2. The intelligent anti-theft system for unmanned aerial vehicle (UAV) nests according to claim 1, characterized in that, The main control unit is also used to collect facial information from image information through the AI ​​recognition module, compare the facial information with the whitelist database and perform personnel behavior analysis to obtain the analysis results. When the facial recognition matches the whitelist, a first-level response strategy is generated through the hierarchical decision engine. When the facial recognition is a stranger, a second-level response strategy is generated through the hierarchical decision engine. When abnormal behavior is detected, a third-level response strategy is generated through the hierarchical decision engine.

3. The intelligent anti-theft system for unmanned aerial vehicle (UAV) nests according to claim 2, characterized in that, The abnormal behavior includes: being identified as a stranger, staying for more than a preset time, or possessing a tool.

4. The intelligent anti-theft system for unmanned aerial vehicle (UAV) nests according to claim 2, characterized in that, Also includes: An attitude detection module is used to collect attitude data of the UAV nest. The main control unit is also used to execute a three-level response strategy when the drone nest is determined to be abnormal based on attitude data.

5. The intelligent anti-theft system for unmanned aerial vehicle (UAV) nests according to claim 2, characterized in that, The execution module includes a body sound and light alarm and a communication module; The main control unit is also used to log and mark locations when executing a level-one response strategy, to issue an audible and visual alarm and record video via an image acquisition module when executing a level-two response strategy, and to send a human intervention request via a communication module and record video via an image acquisition module when executing a level-three response strategy.

6. The intelligent anti-theft system for unmanned aerial vehicle (UAV) nests according to claim 5, characterized in that, Also includes: The cloud-based AI platform is used to receive structured alarm packets sent by the communication module and perform secondary verification, and generate corresponding handling strategies based on the results of the secondary verification.

7. A smart anti-theft method for drone nests, characterized in that, Includes the following steps: The heat source sensing module detects whether there is a heat source in the current environment at a preset frequency. When an ambient heat source is detected, the main control unit is woken up. After the main control unit is woken up, it starts the image acquisition module and acquires image information through the image acquisition module; The AI ​​recognition module performs facial recognition and human behavior analysis on image information to obtain analysis results; The hierarchical decision engine generates corresponding response strategies based on the analysis results and the hierarchical response mechanism. The execution module generates execution actions based on the response strategy of the hierarchical decision engine.

8. The intelligent anti-theft method for the drone nest according to claim 7, characterized in that, The hierarchical decision engine generates corresponding response strategies based on the analysis results and the hierarchical response mechanism, specifically including the following steps: When facial recognition matches the whitelist, a first-level response strategy is generated through a hierarchical decision engine; When the face recognition identifies the face as that of an unfamiliar person, a secondary response strategy is generated through a hierarchical decision engine. When abnormal behavior is detected, a three-level response strategy is generated through a hierarchical decision engine.

9. The intelligent anti-theft method for the drone nest according to claim 8, characterized in that, The execution module includes an audible and visual alarm and a communication module; The execution module generates execution actions based on the response strategy of the hierarchical decision engine, specifically including the following steps: When executing the Level 1 response strategy, log entries and location markers are performed; When the level-two response strategy is executed, an audible and visual alarm is triggered by the audible and visual alarm device, and video is recorded by the image acquisition module. When implementing the three-level response strategy, a human intervention request is sent through the communication module and video is recorded through the image acquisition module.

10. The intelligent anti-theft method for the drone nest according to claim 9, characterized in that, It also includes the following steps: The structured alarm package, which includes timestamps, location information, image information, and pose data, is sent to the remote AI platform via the communication module. The cloud-based AI platform performs secondary verification based on the structured alarm package and implements corresponding handling strategies based on the secondary verification results.

Images