Unmanned aerial vehicle-smart window full-link distribution system and method based on dynamic order distribution

Abstract

The invention provides an unmanned aerial vehicle-intelligent window Internet of Things cooperation system. The system comprises an intelligent business order generation module, a dynamic order sending engine, a two-dimensional code navigation system and a biological characteristic closed-loop authorization mechanism. And the merchant side prints an encrypted two-dimensional code quotient list containing the three-dimensional coordinates of the intelligent window, the unmanned aerial vehicle automatically plans a path after scanning the quotient list, and automatically applies for biological feature authorization to trigger opening and closing of the intelligent window when approaching the target for more than or equal to 200 meters. The innovation points are as follows: 1, a business order-window coordinate dynamic binding technology (encrypted two-dimensional codes); 2, a dispatching-charging collaborative algorithm (a benefit model based on electric quantity / distance); 3, an acousto-optic wake-up system (multi-mode activation); 4, an intelligent window complete ID physical fusing and multi-terminal binding mechanism; the central gateway only receives the verification result (yes / no) and does not store any biological characteristic data; the whole-process unmanned distribution is realized, and the safety is improved through risk adaptive multiple verification.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) delivery technology, specifically a drone-smart window end-to-end collaborative system based on dynamic order dispatch, which is particularly suitable for last-mile delivery scenarios in high-density urban areas and rural towns. Background Technology

[0002] Existing technological defects 1. Traditional drone delivery requires users to receive the package outdoors: this poses a safety hazard (refer to patent CN118071232B). 2. Smart window systems often use fixed password verification: this is easily cracked and has security vulnerabilities (refer to IEEE Access vol.11 pp.123-135). 3. Lack of dynamic coordination mechanism between delivery and receiving ends: unable to achieve fully unmanned closed-loop delivery (analysis of technological gaps). Summary of the Invention

[0003] Core system components This invention provides a drone-smart window IoT collaborative system, comprising the following modules: 1. Intelligent Merchant Order Generation Module: Merchants can print encrypted QR code merchant orders containing the three-dimensional coordinates of the intelligent window; 2. Dynamic Dispatch Engine: Calculates weights based on distance, battery level, and urgency to allocate the optimal drone (Weight formula: Weight = 0.6 × (1 / Distance + ε) + 0.3 × Battery Level Coefficient + 0.1 × Urgency, where ε = 1e-5, Battery Level Coefficient = Current Battery Level / 100, Urgency ∈ [0,1] (0: Normal, 0.5: Urgent, 1: Top Urgency)). 3. QR code navigation system: After scanning the order form, the drone automatically parses the encrypted coordinates and plans the route; 4. Biometric closed-loop authorization mechanism: When arriving at the target, initiate multimodal biometric verification (voiceprint / fingerprint / face / password) to trigger the opening and closing of the smart window.

[0004] Innovative technical solutions 1. Dynamic binding of shop order window coordinates - Uses layered encrypted QR codes: - Outer layer: Merchant ID + Order number (plaintext) - Inner layer: Smart window or customer delivery location XYZ coordinates + AES-256 encrypted drone ID - Coordinate acquisition method: The smart window integrates Beidou positioning and a barometer to provide real-time feedback of three-dimensional coordinates (Z-axis accuracy ±0.05m). 2. Order Dispatch-Charging Coordination Algorithm - Dynamic management of drone status: - Battery > 60%: Enter standby mode - Battery level ≤ 30%: Automatically fly to a charging station (path cost = 0.7 × flight distance + 0.3 × (100 - current battery level)) 3. Sound and light wake-up system - Merchant button press triggers multimodal signals: - Visual signal: Red light flashing for 200ms - Auditory signal: 800Hz buzzing tone (3 times) - Positioning signal: Laser beam guides QR code scanning 4. Smart Window Security Mechanism - Unique physical fuse ID: Solidified at the factory via laser etching and OTP memory (format: W_<manufacturer code>_<Beidou serial number>_<check bit>), it becomes invalid upon removal; - Multi-terminal binding: Supports registration of one primary and multiple secondary mobile phone numbers, with decentralized control (the primary number can modify coordinates / unbind, while the secondary number only receives authorization requests). 5. Biometric privacy protection - The original biometric template (voiceprint MFCC coefficients, fingerprint topology data, etc.) is stored only in the encrypted area of ​​the user terminal; - The biometric verification process is performed entirely on the user terminal, and the central gateway only receives the final verification result (pass / reject). - The formula for generating temporary keys has been modified to: Temporary key = SHA-256(UTC timestamp || Smart Window complete ID || random number).

[0005] End-to-end workflow 1. The user places an order via the app and obtains the 3D coordinates of the smart window; 2. The central gateway executes the order dispatch algorithm and pushes encrypted orders to the merchant's terminal; 3. The merchant prints the invoice and activates the drone, which scans the QR code to parse the target coordinates; 4. When the drone flies to a distance of ≥200 meters from the target: - Reduce speed to ≤3m / s - Initiate a biometric authorization request (voiceprint / fingerprint / face / password); 5. After authorization is approved: - Open the window to a safe opening (60×50cm). - Millimeter-wave radar accurately locates delivery points (error ≤ 0.15m); 6. Upon completion of delivery, the window will automatically close and lock, and the drone will switch tasks according to its battery level.

[0006] Safety Enhancement Design 1. Risk-adaptive verification strategy: Risk Score | Verification Strategy | |---------------|----------------------------------| | <30| Basic Biometric Verification| | 30-70 | Biometrics + Animated Graphics Challenge | | >70| Activate defense protocol (fog glass + alarm recording) | (Risk score = 0.4 × biometric bias + 0.3 × behavioral anomaly + 0.2 × geographical offset + 0.1 × device reliability) Biometric bias = 1 - Match score (voiceprint MFCC cosine similarity / fingerprint feature point overlap rate, etc.) 2. Emergency Communication Protocol: When the signal strength is <-100dBm, switch to the LoRa_470MHz band and use XMSS post-quantum signature verification (effective window ±3 minutes).

[0007] Instruction manual illustrations Figure 1: System Hardware Architecture ┌─────────────┐┌─────────────┐┌─────────────┐ │User App│←4G / 5G→│Central Gateway│←Encrypted Channel→│Merchant Terminal│ └──────┬──────┘└──────┬──────┘└──────┬──────┘ │Biometric Authentication Data│Order Dispatch Instructions│Thermal Printing Coordinate Encryption | Wake-up Signal ┌──────▼──────┐┌──────▼──────┐┌──────▼──────┐ │Smart Window Terminal│←Infrared Positioning→│Registered Drone│←Audio and Light Wake-up←│Wake-up Button│ └─────────────┘└─────────────┘└─────────────┘ Component Description: 1. User App: - Biometric authentication interface (voiceprint / fingerprint / face) - Smart Window Binding Module 2. Central Gateway: - Dispatch engine (distance / battery level / urgency weight calculation) - Coordinate Encryption Service (AES-256) 3. Merchant Terminal: - Thermal printer (outputs encrypted QR code business orders) - Wake-up button (triggers sound and light wake-up protocol) 4. Registered drones: - QR code scanner (parses encrypted coordinates) - Millimeter-wave radar (precise positioning, error ≤0.15m) - Sound and light wake-up receiver (responding to merchant commands) 5. Smart Window Terminal: - BeiDou positioning + barometer (to obtain three-dimensional coordinates) - Infrared positioning module (for interaction with drones) - Physical fuse ID chip (tamper-proof unique identifier) Figure 2: End-to-end workflow 1. User places order via APP 2. The system obtains the window's three-dimensional coordinates (X, Y, Z). 3. The central gateway executes a dispatch algorithm to calculate the optimal drone. 4. Merchant terminals print encrypted invoices. 5. Merchants can wake up the drone by pressing a button. 6. Drones scan the QR codes on merchant invoices. 7. The drone executes path planning and flies to the target location (reducing its speed to ≤3m / s when 200 meters from the target). 8. Initiate a multimodal biometric authorization request └─Authorize → 9. Automatically open a window to execute delivery 9. After delivery, close and lock the window. Figure 3: Sound and light wake-up sequence Merchant button trigger signal sequence: 1. Visual signal: Red light flashes for 200ms (wake-up confirmation) 2. Auditory signal: 800Hz buzzer sound for 3 consecutive times (system activated) 3. Positioning signal: Laser positioning beam activated (wavelength 940nm, guiding QR code scanning) Figure 4: Smart Window Activation and Binding Flowchart 1. Users scan the device's QR code using the app. 2. The APP obtains the unique ID of the smart window. 3. The system automatically locates the three-dimensional coordinates of the window. 4. User enters primary mobile phone number 5. Users can add a secondary mobile phone number (optional) 6. The binding information is encrypted and uploaded to the central gateway. 7. Generate blockchain-based evidence records. 8. Activation complete and feedback provided to user. Figure 5: Multi-barrier architecture ┌─────────────────┐ │ Biometric Verification Layer │◀─ Fingerprint / Voiceprint / Face / Password └────────┬──────┘ │(Liveness detection results) ┌──────────────▼──────────────┐ │Risk Decision Engine│◀─── Environmental Validator (GPS / Signal Strength) │ Real-time risk score calculation = 0.4 × biometrics │ | Deviation + 0.3 × Behavioral Anomaly Degree + 0.2 × Geographic Offset | └──────────────┬──────────────┘ (Dynamic strategy adjustment) | ┌──────────────┴──────────────┐ │Security Challenge Response Layer│ │ Dynamically Generate Graphs / Topology Challenge (Complexity increases with risk score) │ └──────────────┬──────────────┘ (Device binding verification)│ ┌──────────────▼──────────────┐ │Device Fingerprint Authentication Layer│ │ Physical circuit breaker ID + timestamp encrypted binding + multi-terminal authentication │ └─────────────────────────────┘ Core Interactions: 1. Four layers of defense from top to bottom: - Biometrics Layer: Multimodal Liveness Detection - Risk decision-making level: Real-time scoring and dynamic adjustment of verification strategies - Security Challenge Layer: Dynamic Challenges for Generating Anti-Robots - Device fingerprint layer: Hardware-level ID binding and time lock 2. Horizontal collaboration: - The environment validator provides real-time data such as GPS / signal strength. - Risk score thresholds strictly correspond: Risk score > 70 or 3 consecutive failures → Activate defense protocol Figure 6: Anti-spy fogging mechanism 1. The system detected a verification anomaly. 2. Trigger the smart window fogging start command 3. The light transmittance of the smart glass is reduced to 5%. 4. To achieve visual shielding of the view outside the window. 5. Simultaneously start alarm video recording. 6. Send alarm information to both primary and secondary mobile phones. Figure 7: Flowchart of Multimodal Biometric Verification 1. The system receives the authorization request. 2. Select verification mode: - Voiceprint mode → Vocal cord vibration detection - Fingerprint mode → Skin conductivity detection - Face Mode → Micro-expression Analysis - Password Mode → Quantum Encryption Verification 3. After verification, generate the encrypted window opening command locally. 4. Encrypted transmission of window opening commands 5. Perform the window opening operation. Figure 8: Schematic diagram of the safety barrier for liveness detection Input signal ↓ Multimodal liveness detection firewall: 1. Voiceprint: Fundamental frequency fluctuation detection (80-280Hz) 2. Fingerprint: Blood flow signal detection (>0.5mV) 3. Face: Infrared depth map (error < 0.1mm) 4. Password: Quantum-resistant hash chain verification ↓ Generate verification result record ↓ Complete authorization verification Figure 9: Updated Claims Structure Core claim 6 ├─→ Claim 7 (Face Recognition Protection) → Example 6 ├─→ Claim 8 (XMSS cryptography) → Example 8 ├─→ Claim 9 (Blockchain Evidence) → Example 5 └─→ Claim 10 (Emergency Protocol) → Example 7. Detailed Implementation

[0008] Example 1: Fresh Produce Delivery Scenario 1. When a customer selects seafood in the app, the system automatically obtains the coordinates of the smart window in the residence (X:121.4737, Y:31.2304, Z:83.5m). 2. The dispatch engine selects the DJ_08F drone with 92% battery (distance coefficient 0.87). 3. The merchant prints a receipt (QR code containing encrypted coordinate data), and presses a button to trigger the drone wake-up protocol. 4. After scanning the QR code, the drone plans its route and sends a request for one of the following methods (voiceprint / fingerprint / face / password) to the customer's mobile phone during flight. 5. After the customer says "Confirm Receipt": - The system verifies the fundamental frequency (80-280Hz) of the voiceprint and content matching. - Window opening preset safe opening degree (60×50cm) - Millimeter-wave radar corrects the window position in real time (dynamic compensation on XYZ axes, compensation algorithm: Δx=k×wind speed, k is preset to 0.1-0.3 based on the UAV's aerodynamic parameters). 6. If verification fails: Initiate the security challenge layer (as shown in Figure 5). 7. After delivery, close the window and fly the drone to the charging station (41% battery remaining). Example 2: Cooperative Charging Protocol When the drone's battery level is ≤30%: 1) The central gateway automatically assigns the user to the nearest charging station (path priority algorithm). 2) The magnetic charging port is coupled to the bottom electrode of the drone (positioning accuracy ±2mm). 3) Marked as "Unavailable for dispatch" during charging. Example 3: Smart Window Activation and Binding Process 1. Users open the app and scan the smart window's QR code (which contains the device's unique ID). 2. The app automatically obtains the window's position and provides a notification: [System Prompt] Please confirm the window installation location: X: 121.4737, Y: 31.2304, Z: 83.5m (based on BeiDou differential positioning + barometer fusion data) Coordinate encryption method: AES_Encrypt(coordinates||timestamp, device public key) 3. User settings for binding mobile phone number: - Primary mobile number: 1 (required) - Secondary mobile phone numbers: 0 or more 4. The central gateway generates device binding credentials and stores them on the blockchain: { "device_id": "W_DIF_08B7X2_9A", "coord": [121.4737, 31.2304, 83.5], "master_tel": "+86138****1234", "backup_tels": ["+86139****5678"] } Example 4: Triple Authorization Protocol (Anti-hacking Scenario) 1. Unauthorized acquisition of customer identity credentials to initiate authorization requests 2. The system detected an anomaly: - Behavioral analysis revealed abnormal operation speed (>500px / s) - GPS location shows abnormal distance (distance from window > 4000km) 3. Trigger a high-risk challenge response: [System Challenge] Please draw a trajectory in the coordinate system that matches the preset path (path data is only stored on the user terminal). ███ \│ / ●─●─● (Generate a topology map based on pre-stored fingerprints) 4. Hackers cannot complete the drawing (it requires matching a preset topology). 5. When the risk score is >70 or three consecutive verifications fail, activate the fogging glass and alarm recording: - Activate window fogging mode (reduces light transmittance to 5%). - The smart window camera activates night vision to capture intruders. - Send an alert to the main phone: "Abnormal app login detected! Blocked and recorded as evidence." Example 5: Multimodal Biometric Verification and Dynamic Key Generation step: 1. The system selects the verification mode (voiceprint / fingerprint / face / password) based on the risk level. 2. Perform biometric verification on the user terminal and output a Boolean result. 3. Generate a temporary key after successful verification: Temporary key = SHA-256(UTC timestamp || Smart Window complete ID || random number) 4. Encrypt the window opening command with a temporary key: Ciphertext = AES_CTR_Encrypt(window command, temporary key) 5. Generate blockchain evidence (following the structure of claim 9) Example 6: High-risk face verification When the risk score is >70, the triple anti-counterfeiting protocol described in claim 13 is activated: 1. Depth detection: Infrared sensor generates millimeter-level depth maps (error <0.1mm) 2. Action Challenge: - Randomly generate instructions (such as "Please turn your head to the left and blink twice"). - Verify the consistency of action execution (by verifying that the 3D keypoint trajectory matching degree of consecutive frames is ≥95%). 3. Reflectance Analysis: The scattering characteristics of skin tissue were detected using 850nm infrared spectroscopy. After verification: - Calculate the overall confidence score for liveness detection (range 0-100). - Generate blockchain evidence (following the structure of claim 9), evidence content: { "result": true, "timestamp": "2025-08-01T14:30:05Z", "device_hash":"0x9a8b..."} Example 7: LoRa Emergency Communication Protocol When a drone enters an area with strong electromagnetic interference: 1. The communication module monitors signal strength in real time (threshold: -100dBm). 2. Automatically switch to the LoRa_470MHz band when the threshold is reached. 3. The user's mobile phone generates an XMSS emergency key pair based on the Smart Window's complete ID. 4. Sign the delivery instruction with a timestamp (format: EMERGENCY_DELIVERY|coordinates|time) 5. The smart window controller verifies the signature validity period (valid window: ±3 minutes). Note: Emergency protocol trigger conditions: - Three consecutive communication timeouts - And GPS signal loss > 60 seconds Example 8: Emergency Communication Verification Process When the signal strength is < -100dBm: - Generate an XMSS key pair on the user's mobile phone: (SK_e, PK_e) = XMSS_keygen(Smart Window Complete ID) - Signing instruction: σ_e = XMSS_sign("EMERGENCY_DELIVERY", SK_e) - Smart windows verify signatures using pre-stored public keys. Example 9: Implementation of the Sound and Light Wake-up Protocol 1. After the merchant presses the wake-up button: - Terminal generates signal sequence: -Signal = Merchant ID_HMAC + Timestamp_AES 2. Drone receiver verification: a) Red light flashing (200ms) → Visual confirmation b) 800Hz buzzing sound (3 times) → Auditory activation c) Laser positioning beam activated → Guided scanning 3. When QR code scanning fails: - Retry mechanism: ≤2 times (5-second interval) - If it still fails, then report that the merchant terminal wake-up has been aborted.

[0009] Hardware Design Specification 1. Core Hardware Specifications: | Equipment | Key Components | Technical Specifications | |------|----------|----------| | Merchant Terminal | Industrial-grade Thermal Printer | Printing speed ≥ 80mm / s, resolution 8 dots / mm | | Drone Scanner | Global Shutter CMOS + Laser-Assisted Focus | Scanning Distance 0.1-1.2m, Response Time <0.3s | | Smart Window Positioning Module | Beidou-3 + MS5611 Barometer | Z-axis accuracy ±0.05m, temperature drift compensation ±0.5% | | Sound and light wake-up receiver | 940nm infrared receiver tube + microphone array | Wake-up sensitivity ≤-80dBm | 2. Biometric Module Technical Table: | Module | Anti-counterfeiting technology | |------------|--------------------------| | Voiceprint | Vocal cord vibration spectrum analysis | | Fingerprint | Capacitive + Optical Dual-Mode Detection | | Face | 3D Dynamic Liveness Detection | | Cryptography | Post-quantum cryptography coprocessor | 3. Anti-infringement hardware module list: | Module | Technical Solution | Anti-counterfeiting Features | |------|----------|----------| | Behavior Sensor | Mobile Phone Gyroscope + Accelerometer | Detects Abnormal Operational Inertia (e.g., Robot Fixed Patterns) | | Challenge Generator | FPGA Graphics Rendering Chip | Real-time Generation of Non-Repeating Topology Graphics | | Anti-peeping actuator | PDLC smart dimming glass | Switches between atomization states within 10ms | 4. Mechanical Structure Innovation: Folding cargo hold design: Closed state: Cabin thickness ≤ 5cm Activation mechanism: (1) When the window opening command is received, the servo drives the four-link linkage. (2) Door opening angle θ = min(130°, window opening / 2) (3) Built-in weighing sensor provides feedback signal indicating delivery completion 5. ID Curing Process Table: | Process | Technical Implementation | Anti-tampering Features | |------|----------|------------| Laser etching | Writing ID on PCB copper layer | Physical damage visible | | OTP Memory | One-time Programmable ROM Stores ID | Cannot be rewritten after fuse blown | | Beidou chip binding | ID is hardware-associated with Beidou module MAC address | Disassembly will disable it |.

[0010] List of Patent Document Attachments 1. Encrypted QR code generation algorithm code (Python implementation) 2. Voiceprint authorization engine test dataset (.wav format).

[0011] Key points of legal statement 1. The "window or customer-specified location coordinates XYZ" in the claims specifically refers to the fused data of the BeiDou geodetic coordinate system and the barometric altimeter system. 2. "Voiceprint authorization" includes two-factor verification: voice content recognition and voiceprint biometric features. 3. The audio-visual communication protocol between the merchant terminal and the drone complies with the Radio Management Regulations and the SRRC (2023) certification standards. 4. The "unique ID" complies with the Level 3 security requirements of GB / T 37036-2018. 5. The smart window's physical fuse ID binding method complies with Article 22 of the Patent Law regarding inventiveness (achieving a technological breakthrough through hardware binding and dynamic verification). 6. The multimodal biometric system complies with the principle requirements of ISO / IEC 30107-3. 7. The voiceprint liveness detection technology employs the fundamental frequency fluctuation analysis method (see Example 5 in the instruction manual for details). 8. The biometric verification process is executed entirely on the user terminal. The central gateway only receives Boolean verification results (pass / reject) and does not access any biometric data, which conforms to: - Section 51 of the Personal Information Protection Act and Section 25 of the GDPR - ISO / IEC 24745:2022 Standard for the Protection of Biometric Information - GB / T 35273-2020 Personal Information Security Specification 9. The operating frequency band of the UAV millimeter-wave radar is limited to 76-81GHz within China, which complies with the Ministry of Industry and Information Technology's specification No. 189 of 2022. 10. The wavelength of the laser positioning beam is limited to 940nm, which complies with the IEC 60825-1 Class 1 safety standard.

Claims

1. A drone-intelligent window delivery hardware system, characterized in that... include: (1) Merchant terminal: integrates thermal printer, wake-up button, and order management screen; (2) Unmanned aerial vehicles (UAVs): equipped with a QR code scanner, millimeter-wave radar (working frequency band 76-81GHz), laser positioning receiver, sound and light wake-up receiver, intelligent path planning system, automatic application for intelligent window opening and closing authorization during flight, low-latency communication module, and cargo delivery and collection components; (3) Intelligent window terminal: including a uniquely identified positioning module, a height measurement unit and a multi-terminal communication interface; (4) Central gateway: Deploy the dispatch engine (calculate weights based on distance, power consumption, and urgency) Order dispatching algorithm (drone queue, orders): a) Selected drones = argmax(drone queue, key=weight) Note: Weight calculation formula: Weight = 0.6 × (1 / distance + ε) + 0.3 × power coefficient + 0.1 × urgency, where ε = 1e-5 (to prevent division by zero error), power coefficient = current power / 100, distance = straight-line distance from merchant to target window (km), urgency value rules: 0 (normal order), 0.5 (urgent order), 1 (extremely urgent order); b) Generate an encrypted order (order number, smart window coordinates XYZ, drone ID).

2. A full-chain delivery method, characterized in that... Including the following steps: (1) After installing the smart window, the user activates it by scanning the smart window's QR code through the APP and then executes: a) Enter the primary mobile number and multiple secondary mobile numbers b) Automatically obtain the 3D coordinates (X, Y, Z) of the smart window and associate it with a unique ID. c) Binding information is encrypted and stored in the central gateway. (2) When a user submits an order via the APP, the system automatically obtains the smart window's BeiDou coordinates (X,Y) and barometer altitude (Z); (3) The central gateway executes the order dispatch algorithm and pushes orders containing three-dimensional coordinates to the merchant terminal; (4) The merchant presses a button to wake up the drone, and the drone performs the following actions: a) Scan the merchant's QR code to obtain the target coordinates b) If the scan fails, automatically retry ≤2 times (with a 5-second interval). (5) When the UAV is ≥200 meters away from the target location, it initiates at least one multimodal biometric authorization request (voiceprint / fingerprint / face / password at least one), the specific verification mode is dynamically allocated by the risk decision engine, and the flight speed is reduced to ≤3m / s; (6) After authorization is approved: a) Open the window to the preset safe opening degree b) The UAV uses millimeter-wave radar to accurately locate the delivery point (error ≤ 0.15m) (7) The window automatically locks after delivery, and the drone switches states based on battery level: a) Enter standby mode when battery level > 60% b) When the battery level is ≤30%, it will automatically fly to the charging station.

3. The system according to claim 1, characterized in that: (1) The QR code adopts a layered encryption structure: a) Outer layer: Merchant ID + Order Number (plaintext) b) Inner layer: Smart window or customer delivery location XYZ coordinates + AES-256 encrypted drone ID; (2) The unique ID is solidified during the production of the smart window through laser etching and OTP memory. It becomes invalid upon removal. The ID format is: W_<Manufacturer Code>_<BeiDou Chip Serial Number>_<Check Digit> (3) The UAV is equipped with a folding cargo compartment, and the door opening angle θ = min(130°, window opening / 2) (4) Biometric privacy protection mechanisms meet the following requirements: a) The original biometric template (including voiceprint MFCC coefficients and fingerprint topology data) is stored only in the encrypted area of ​​the user terminal; b) The biometric verification process is performed entirely on the user terminal; c) The central gateway only accepts Boolean authentication results (pass / reject). d) The central gateway does not store any raw biometric data, templates, or hash values. (5) The sound and light wake-up system includes: a) Visual signal: Red light flashes for 200ms; b) Auditory signal: 800Hz buzzing sound (3 times); c) Positioning signal: Laser beam guides QR code scanning.

4. The method according to claim 2, characterized in that... Also includes: (1) Coordinate Encryption Service: Converts the XYZ coordinates of the smart window or customer delivery location into a business order QR code, following the following: Ciphertext = AES_Encrypt(Order Number||Drone ID||Coordinates, Device Public Key) (2) Correction to the voiceprint authorization engine execution process: Audio input → Execute locally on the user terminal: Wavelet denoising → MFCC feature extraction → Comparison with local template → Output verification results to gateway (3) Charging scheduling module: Generates charging paths based on real-time location: a) Path cost = 0.7 × flight distance + 0.3 × (100 - current battery level) b) Charging scheduling module: Generates charging paths based on real-time location. Path cost = 0.7 × flight distance + 0.3 × (100 - current battery level) (4) The anti-fraud decision engine dynamically adjusts the verification strategy: a) Risk score <30: Basic voiceprint verification b) Risk score 30-70: Voiceprint + Graphic Challenge c) Risk score > 70: Activate defense protocol Note: Real-time calculation of authorization risk score: - Risk Score = 0.4 × Biometric Bias Value + 0.3 × Behavioral Anomaly Degree + 0.2 × Geographic Offset + 0.1 × Equipment Reliability - Biometric bias = 1 - Match score (voiceprint MFCC cosine similarity / fingerprint feature point overlap rate, etc.) (5) Electronic Evidence Preservation System: The evidence only includes: a) Validation result (pass / reject) b) Timestamp c) Smart Window complete ID hash value.

5. The method according to claim 2, characterized in that: (1) The voiceprint / fingerprint / face / password authorization includes dynamic noise filtering and performs verification under 90dB ambient noise; (2) The mobile phone number binding adopts a hierarchical control strategy: a) Primary mobile number: Location can be modified / device unbound. b) Secondary mobile number: Only accepts delivery authorization requests; (3) The graphical complexity of the security challenge response increases dynamically with the risk level; (4) The device fingerprint comparison includes a timestamp encryption binding mechanism; (5) Execute when the risk score is >70 or after 3 consecutive failures: a) Activate the smart window privacy film (frosted glass). b) Send real-time alarm video from the smart window camera to the primary and secondary mobile phone numbers.

6. A biometric verification method based on dynamic keys, characterized in that... Including the following steps: (1) When responding to an authorization request, activate at least one verification mode (voiceprint / fingerprint / face / password) according to the drone application format; (2) The liveness detection unit executes the corresponding anti-counterfeiting protocol: a) Voiceprint pattern: Detects the vocal cord vibration spectrum (80-280Hz) The user terminal performs audio noise reduction and MFCC feature extraction. b) Fingerprint mode: Simultaneously acquires capacitance and blood flow signals. c) Face mode: Perform 3D depth map continuity analysis d) Cryptographic mode: Invoking the post-quantum cryptographic coprocessor (3) After successful verification, a temporary key bound to the device ID is generated: Temporary key = SHA-256(UTC timestamp || Smart Window complete ID || random number) (4) Encrypt the window opening command with a temporary key: Ciphertext = AES_CTR_Encrypt(window opening command, temporary key).

7. The method according to claim 6, characterized in that: The face pattern verification includes a triple anti-spoofing protocol: (1) Depth detection: Infrared sensor generates millimeter-level depth map (2) Action Challenge: Randomly generate head-turning / blinking commands and verify the consistency of execution. (3) Reflectance analysis: Detecting the scattering characteristics of skin tissue to a specific spectrum. Note: Verify the consistency of action execution (by verifying that the 3D keypoint trajectory matching degree of consecutive frames is ≥95%). The scattering characteristics of skin tissue were detected using 850nm infrared spectroscopy.

8. The method according to claim 6, characterized in that: The cryptographic mode employs a post-quantum encryption algorithm, specifically implementing the XMSS signature verification protocol, which includes: (1) The client generates a one-time key pair: (SK, PK) = XMSS_keygen(random seed) (2) Sign the challenge command with the private key SK: σ = XMSS_sign("OPEN_WINDOW", SK) (3) The server uses the public key PK to verify the validity of the signature.

9. The method according to claim 6, characterized in that: Once the temporary key is generated, a blockchain evidence record is created simultaneously, including: (1) Validation result (pass / reject) (2) Timestamp (3) Smart Window Complete ID Hash Value.

10. The method according to claim 6, characterized in that: When the drone detects a signal strength < -100dBm, it executes the emergency communication protocol, and the smart window controller needs to pre-store the XMSS public key. (1) Automatically switch to the frequency band approved by the Ministry of Industry and Information Technology (470-510MHz / 779-787MHz) (2) Biometric verification is converted to offline digital signature verification mode (3) In emergency mode, the window opening command uses XMSS signature instead of temporary key encryption: - The user's mobile terminal generates an emergency key pair: (SK_e, PK_e) = XMSS_keygen(Smart Window Device ID). -Signature delivery command: σ_e = XMSS_sign("EMERGENCY_DELIVERY", SK_e) - The smart window controller uses the pre-stored public key PK_e to verify the signature.