An AI intelligent automobile safety detection and monitoring method

Abstract

The application relates to the technical field of intelligent automobile safety monitoring, and discloses an AI intelligent automobile safety detection and monitoring method, which comprises the following steps: S1, collecting vehicle surrounding environment information through a quantum laser radar array; S2, collecting physiological state signals of a driver and passengers through a combination of a millimeter wave life radar and a stealth camera; S3, collecting passenger posture change signals, seat belt state signals and seat pressure signals through a friction nano power generation sensing network; S4, inputting the signals collected in steps S1 to S3 into a physical driving decision engine; and S5, performing vehicle control operation according to the safety response instruction. The safety control decision is generated through physical modeling and simulation calculation, the statistical learning or neural network model is avoided, the decision process is completely interpretable and real-time, and the problem that a traditional algorithm based on experience or model training is uncontrollable in an extreme case is solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle safety monitoring technology, specifically to an AI-powered intelligent vehicle safety detection and monitoring method. Background Technology

[0002] With the rapid development of intelligent vehicles and driver assistance technologies, vehicle safety monitoring systems have gradually evolved from passive protection to proactive prevention. Existing vehicle safety detection and monitoring technologies typically rely on visual sensors, radar systems, or onboard control units to analyze external obstacles, driver behavior, and vehicle attitude to achieve collision warnings and braking assistance control. Typical current systems include camera-based driver monitoring systems, millimeter-wave radar-based distance detection systems, and braking control systems incorporating inertial navigation. While these technologies improve vehicle safety to some extent, their decision-making logic is primarily based on empirical formulas, statistical models, or neural network reasoning, resulting in weak system interpretability and insufficient stability and consistency under extreme environmental conditions.

[0003] In existing technologies, decision-making algorithms often rely on training data or empirical thresholds. When vehicles are on low-adhesion surfaces, undergoing high-speed emergency avoidance, or in complex weather conditions, the system may experience response delays or misjudgments, making it difficult to achieve synchronized control of braking and safety constraint actions. Furthermore, existing systems generally lack dynamic reasoning capabilities based on physical models, failing to directly reflect the coupling relationship between vehicle kinematics and occupant biomechanics, resulting in control strategies that cannot reflect the actual force state in real time. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an AI-powered intelligent vehicle safety detection and monitoring method. This method solves the problems of existing technologies, which rely primarily on empirical formulas, statistical models, or neural network reasoning for decision-making logic, resulting in weak system interpretability and insufficient stability and consistency under extreme environmental conditions.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an AI-powered intelligent vehicle safety detection and monitoring method, comprising the following steps: S1. Collect environmental information around the vehicle using a quantum lidar array, wherein the quantum lidar array performs environmental target detection and distance measurement based on entangled photon pairs; S2. The physiological state signals of the driver and passengers are collected by combining millimeter-wave life radar and a hidden camera. The millimeter-wave life radar collects respiratory rate signals and heart rate signals, and the hidden camera collects facial expression signals and posture signals. S3. Acquire occupant posture change signals, seat belt status signals, and seat pressure signals through a triboelectric nano-power generation sensor network. The triboelectric nano-power generation sensor network consists of triboelectric nano-power generation sensor units embedded in the seat, seat belt, and steering wheel. S4. Input the signals collected in steps S1 to S3 into the physical drive decision engine, perform simulation calculations based on the vehicle dynamics model and the occupant biomechanics model, and generate safety response commands. S5. Perform vehicle control operations according to the safety response command, including braking, seat belt tightening and airbag deployment.

[0006] Preferably, the quantum lidar array includes a quantum light source, a quantum modulation unit, and an entangled photon detector. The quantum light source generates entangled photon pairs. The quantum modulation unit modulates the phase of entangled photons. The entangled photon detector receives the reflected light signal and calculates the target distance and relative velocity.

[0007] Preferably, the millimeter-wave life radar transmits electromagnetic waves in the 60GHz to 77GHz frequency band. The millimeter-wave life radar calculates the micro-displacement of the driver's chest cavity based on the Doppler displacement of the reflected signal, and extracts the respiratory rate signal and heartbeat signal from the micro-displacement.

[0008] Preferably, the hidden camera is installed behind the central control display screen and works in conjunction with the infrared light projection module to collect the driver's facial expression signals and eye posture signals in the absence of visible light.

[0009] Preferably, the triboelectric nanogenerator sensing unit in the triboelectric nanogenerator sensing network is composed of a polytetrafluoroethylene thin film layer and a conductive fabric layer. When the occupant comes into contact with the triboelectric nanogenerator sensing unit, a triboelectric charge signal is generated. The triboelectric charge signal is converted into an electrical signal by the sensing unit and then input to the vehicle's central control unit.

[0010] Preferably, the physics-driven decision engine includes a data receiving module, a simulation calculation module, and an execution control module. The data receiving module receives signals from the quantum lidar array, millimeter-wave life radar, and triboelectric nanogenerator sensor network. The simulation calculation module calculates vehicle speed, acceleration, and braking distance based on a vehicle dynamics model, and calculates human body force and displacement parameters based on a occupant biomechanical model. The execution control module generates control commands based on the simulation calculation results and transmits them to the execution mechanism.

[0011] Preferably, the security response command includes the following actions: (a) Execute braking control action when the distance to an obstacle is less than a preset safe distance; (b) When an abnormality is detected in the driver's heartbeat or breathing signal, the vehicle shall decelerate and issue a warning signal; (c) When a seatbelt status signal indicating that the seatbelt is loose is detected, the seatbelt tightening action is performed and a voice prompt is issued.

[0012] Preferably, it also includes a multimodal data fusion step. The multimodal data fusion step performs time synchronization processing to uniformly calibrate the quantum lidar array signal, millimeter-wave life radar signal, and triboelectric nanogenerator sensor network signal and generate comprehensive safety status data.

[0013] Preferably, the physics-driven decision engine performs dynamic threshold update calculations before generating safety response commands. These dynamic threshold update calculations are based on the relationship between vehicle speed v, target distance d, and braking acceleration a. The braking trigger threshold is updated based on the response delay tr.

[0014] Preferably, it also includes data recording and playback steps. The data recording module performs encrypted storage of the quantum lidar array signals, millimeter-wave life radar signals, and triboelectric nanogenerator sensor network signals. Upon detecting a security event, the playback module retrieves the encrypted data and performs event reconstruction calculations.

[0015] Beneficial effects This invention provides an AI-powered intelligent vehicle safety detection and monitoring method. It offers the following advantages: 1. This invention generates safety control decisions through physical modeling and simulation calculations, avoiding reliance on statistical learning or neural network models. This makes the decision-making process fully interpretable and real-time. The method can directly deduce the timing of vehicle braking and seat belt tightening in dynamic environments through mechanical and motion laws, solving the problem of uncontrollable response in extreme situations for traditional algorithms based on experience or model training.

[0016] 2. This invention uses a triboelectric nano-power generation structure as the core sensing unit for detecting occupant status and seat belt status, realizing attitude recognition and safety constraint status monitoring under conditions without external power supply. This solution integrates energy harvesting and sensing functions into the same material system, breaking through the limitations of existing capacitive or strain gauge sensors on power supply and durability.

[0017] 3. In the data fusion stage, this invention designs a dual alignment process of time synchronization and coordinate system calibration, which enables the fusion of multi-source data from quantum lidar, millimeter-wave radar, optical imaging and nanoscale sensor networks under a unified spatiotemporal framework. This structured fusion method improves the consistency of cross-domain sensor collaboration and solves the monitoring blind spot problem caused by the mismatch between different sampling frequencies and spatial references in traditional systems.

[0018] 4. This invention utilizes quantum light source and entangled photon detection technology to construct a lidar array, achieving high-resolution imaging and distance measurement of environmental targets. This scheme breaks through the sensitivity limitations of traditional pulsed or frequency-modulated continuous wave radar in detecting low-reflectivity objects, and improves detection accuracy and anti-interference capability from the perspective of photon characteristics. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0020] Figure 1 This is a block diagram of the overall system structure of the present invention; Figure 2 This is a block diagram of the quantum lidar array structure of the present invention; Figure 3 This is the millimeter-wave life radar signal processing flow of the present invention; Figure 4 This is a flowchart of the multimodal data fusion process of the present invention; Figure 5 This is a computational logic diagram of the physics-driven decision engine of the present invention; Figure 6 This is a flowchart of the system state machine and self-test of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see the appendix Figure 1 - Appendix Figure 6This invention provides an AI-powered intelligent vehicle safety detection and monitoring method, relating to the field of active vehicle safety control. The system utilizes the collaborative work of multiple sensors and computing modules to comprehensively monitor the vehicle's external environment, driver and occupant status, and generates safety control decisions based on physical modeling and simulation calculations, thereby executing operations such as braking, seatbelt tightening, and airbag deployment.

[0023] The overall system structure consists of a quantum lidar array, a millimeter-wave life radar, a stealth camera and infrared spot projection module, a triboelectric nanogenerator sensor network, a multimodal data fusion module, a physics-driven decision engine, an actuator linkage system, and a data recording and playback module. All modules communicate and coordinate their timing with each other through the vehicle's central control unit.

[0024] Quantum lidar arrays are used to collect information about the vehicle's external environment, including the position, speed, and relative motion parameters of obstacles. Millimeter-wave life radar is used to collect physiological state signals of the driver and occupants, including respiratory rate and heart rate. A stealth camera and infrared spot projection module are used to collect facial expression signals and eye posture signals of the driver. A triboelectric nanogenerator sensor network is used to collect occupant posture change signals, seat belt status signals, and seat pressure signals.

[0025] The multimodal data fusion module performs unified time synchronization, coordinate transformation, and information fusion on the aforementioned multi-source sensor data to generate comprehensive environmental and occupant status data. The physics-driven decision engine performs vehicle dynamics and occupant biomechanical calculations based on the fused data, generating braking, seatbelt tightening, and airbag deployment commands. The actuator linkage system receives and executes the control commands output by the decision engine to achieve corresponding vehicle safety actions. The data recording and playback module encrypts and indexes the multi-source data and control commands during operation to support post-event verification and event reconstruction.

[0026] The system's communication architecture employs a distributed Ethernet topology. Each sensor module connects to the vehicle's central control unit via a communication interface. The central control unit provides a unified time reference for all modules through synchronization signals and a global clock. The multimodal data fusion module and the physical drive decision engine are installed on the vehicle's computing platform, and the actuator linkage system is connected to the central control unit via an independent safety bus to ensure the reliability of critical control links.

[0027] The system's workflow includes the startup phase, data acquisition phase, data fusion phase, calculation phase, execution phase, and recording phase.

[0028] (1) Start-up phase: After each sensor module is powered on, it performs initialization and calibration, and the central control unit completes time synchronization and system status confirmation.

[0029] (2) Acquisition phase: Quantum lidar array acquires information about the environment around the vehicle, millimeter-wave life radar and stealth camera acquire physiological and posture information of the occupants, and triboelectric nano-powered sensor network acquires signals from seat belts and seats.

[0030] (3) Fusion stage: The multimodal data fusion module performs time alignment and spatial calibration on various types of data to generate a comprehensive state data frame.

[0031] (4) Calculation stage: The physics-driven decision engine reads the integrated data frame, performs vehicle dynamics and occupant biomechanics calculations, and generates control commands.

[0032] (5) Execution phase: The actuator linkage system receives control commands, completes braking, seat belt tightening and airbag deployment operations, and feeds back the execution status to the central control unit.

[0033] (6) Recording stage: The data recording and playback module encrypts and stores the data and instructions generated in each stage, and maintains a time index for subsequent analysis.

[0034] The system's modular design ensures the independent operation and reliable collaboration of each functional unit. Data exchange between modules is conducted through standardized protocols, and signal interfaces include time synchronization lines, Ethernet communication interfaces, and power interfaces. All modules are powered by the vehicle's low-voltage power supply and are equipped with independent overcurrent protection and status detection circuits.

[0035] To ensure operational stability, the system is equipped with multi-level self-testing and redundancy mechanisms. Upon power-up, each module performs hardware connectivity and communication function self-tests; during operation, the central control unit periodically reads the module status registers; when a communication interruption or signal abnormality is detected, the system enters a safety degradation mode, limiting vehicle power output and maintaining minimum safety functions.

[0036] Through the above-described overall structure and workflow, this embodiment of the invention achieves comprehensive monitoring of the vehicle's external environment, driver status, and passenger status, and generates control decisions based on a physical model to complete the entire process of active safety control.

[0037] Furthermore, the quantum lidar array in this embodiment of the invention is applied to the acquisition of information about the vehicle's surrounding environment, used to detect the position, speed, and relative motion of obstacles in front of and around the vehicle during vehicle operation. To ensure a clear understanding of the technical solution of this invention, the structural composition, signal acquisition principle, and installation method of the array are described below.

[0038] A quantum lidar array comprises a quantum light source, a quantum modulation unit, an optical emission assembly, an entangled photon detector, a signal receiving circuit, and a control and processing unit. The quantum light source emits a beam of photons through a controlled laser, which is then split by a nonlinear crystal to form entangled photon pairs. The first photon is emitted as a signal light via the optical emission assembly to the region to be measured, while the second photon is retained in the quantum modulation unit as a reference light for coherent comparison.

[0039] The quantum modulation unit modulates the phase of the reference light. This modulation process is driven by an electro-optic modulator driven by a modulation signal generated by the control processing unit. The modulation signal has a fixed frequency and amplitude, and its time reference is synchronized with the emitted light pulse. After being emitted by the optical system, the signal light is reflected upon encountering an obstacle surface, and the reflected light returns to the receiving window of the entangled photon detector. The entangled photon detector includes an optical filter, a single-photon detector array, and a time-correlated counting module, used to receive the reflected signal light and the reference light respectively, and record their arrival times.

[0040] The time-correlated counting module records the time difference between transmission and reception. Assuming the propagation time of the signal light from transmission to reception is t, and the speed of light in air is c, the target distance d is calculated using the following formula: Where d represents the distance between the obstacle and the radar emission point, c represents the speed of light constant, and t represents the time difference between photon emission and reception.

[0041] To obtain the relative velocity of the target, the control processing unit performs time-series differencing on the measurement results of multiple frames. Let the target distances corresponding to the i-th frame and the (i+1)-th frame be respectively... The sampling interval between the two frames is The formula for calculating the relative velocity v of the target is: Where, the symbol v represents the radial velocity of the target relative to the vehicle, and the symbol... This indicates the time interval between two consecutive distance measurements.

[0042] The control processing unit synchronously calibrates the distance and velocity data to generate an environmental target point cloud data structure. This data structure includes the spatial coordinates, reflection intensity, and time label of each target. The resulting point cloud data is transmitted to the physics-driven decision engine via a communication bus for subsequent vehicle dynamics calculations and safety decisions.

[0043] The physical structure of the quantum lidar array is arranged at the front and sides of the vehicle. The front array is mounted on a metal bracket inside the front bumper, which is connected to the vehicle body via shock-absorbing pads to reduce the impact of vehicle vibrations on the optical path. The side arrays are mounted on the rear edge of the front fenders, using an embedded structure and flush with the vehicle body to reduce aerodynamic drag. Each quantum lidar array is encapsulated by a protective glass cover, the outer surface of which has a transparent conductive film for defogging and anti-fogging operations.

[0044] The quantum lidar array is connected to the vehicle's central control unit via a high-speed Ethernet interface. During communication, the control processing unit periodically sends transmission commands and receives echo data. All data is timestamped, with the time reference provided by the vehicle's global clock. During the power-on initialization phase, the system completes light source drive calibration, receiver channel gain calibration, and time synchronization calibration to ensure the temporal consistency of subsequent ranging results.

[0045] After installation, the system performs extrinsic parameter calibration. This calibration process is achieved by measuring the spatial transformation relationship between the quantum lidar array coordinate system and the vehicle coordinate system. This relationship is expressed using a homogeneous transformation matrix. This indicates the spatial coordinates measured by radar. Convert to vehicle coordinates The conversion formula is as follows: Among them, symbols Represents the spatial coordinates in the quantum lidar array coordinate system, with the symbol... Represents the spatial coordinates after transformation to the vehicle coordinate system, symbol This represents the homogeneous transformation matrix from the quantum lidar array coordinate system to the vehicle coordinate system. The elements of this matrix are calculated using known geometric reference points during the calibration process and stored in the system's non-volatile memory.

[0046] During normal operation, the quantum lidar array cyclically performs transmission, reception, and data output operations at a set frequency. After each measurement cycle, the control processing unit increments the time stamp of the current frame and writes it to the buffer for time synchronization with data from other sensors. When the array detects that the echo signal strength is below a set threshold, it performs automatic gain adjustment and sends a status message to the vehicle's central control unit.

[0047] Through the above structure and process, the quantum lidar array can continuously output environmental information data including target distance, velocity and spatial location, providing input support for subsequent multimodal data fusion and physics-driven decision-making.

[0048] Furthermore, the millimeter-wave life radar in this embodiment of the invention is used to collect physiological state signals of the driver and passengers, including respiratory rate signals and heart rate signals. This radar monitors minute displacements of the human chest cavity in a non-contact manner, enabling real-time detection of vital signs.

[0049] The millimeter-wave life support radar includes a transmit link, a receive link, radio frequency control circuitry, signal processing circuitry, and an interface control unit. The transmit link consists of a voltage-controlled oscillator (VCO), a power amplifier, and a transmitting antenna. The VCO generates millimeter-wave signals with a frequency range of 60 GHz to 77 GHz, which are amplified by the power amplifier and then radiated into the vehicle interior by the transmitting antenna. The receive link includes a receiving antenna, a low-noise amplifier, a mixer, and an analog-to-digital converter (ADC). The receiving antenna receives the echo signal reflected from the occupant's chest cavity surface. After amplification by the low-noise amplifier, the signal is frequency-converted with the local oscillator signal in the mixer to output an intermediate frequency (IF) signal.

[0050] The radio frequency control circuit synchronizes the power, frequency, and phase of the transmit and receive links, and synchronizes with the vehicle's global clock via a local clock to ensure time consistency between transmitted and received signals. The signal processing circuit samples and demodulates the intermediate frequency signal, forming a baseband signal which is then input to the interface control unit.

[0051] The interface control unit performs data buffering and preliminary processing. The signal processing includes DC component removal, bandpass filtering, and phase demodulation. The passband of the bandpass filter covers the respiratory frequency band (0.1Hz to 0.5Hz) and the heart rate frequency band (0.8Hz to 3Hz), and the filtered output signal contains displacement information of the periodic movement of the chest cavity.

[0052] Assume the instantaneous displacement of the target thoracic cavity is , The carrier frequency of the transmitted signal is Where c is the speed of light constant. The phase change of the received signal. Caused by displacement, its expression is: The signal processing circuit measures the phase change and then calculates the corresponding displacement signal. .

[0053] The displacement signal is input into the spectrum analysis module, and its spectral distribution is calculated using a Fast Fourier Transform. In the spectrum, the main peak in the low-frequency region corresponds to the respiratory rate, and the main peak in the high-frequency region corresponds to the heart rate. The peak frequencies are extracted separately. and This allows us to obtain the respiratory rate and heart rate of the driver or passenger.

[0054] When there are multiple occupants in the vehicle, the system uses a multi-antenna configuration to distinguish echo signals from different directions. The spatial direction of each receiving channel corresponds to a different location inside the vehicle. The signal processing circuit spatially separates the signals according to the beam direction of the receiving antenna array to obtain the independent vital signs signals of different occupants.

[0055] The millimeter-wave life radar is installed in the center below the vehicle's dashboard, with its antenna facing the driver and passenger areas. The mounting bracket uses a metal frame structure and is fixed to the vehicle body with rubber pads to reduce the impact of mechanical vibration on signal acquisition. The outer surface of the antenna is equipped with an RF transparent cover made of a low-dielectric-loss material to ensure the stability of the electromagnetic wave propagation path.

[0056] During system power-up, the millimeter-wave life radar performs automatic gain calibration and phase reference initialization. Automatic gain calibration adjusts the gain of the receiver link amplifier to keep the background noise level within a preset threshold range. Phase reference initialization records the phase reference in the current stationary state, and all subsequent phase changes are measured against this reference.

[0057] After signal acquisition is complete, the interface control unit packages the heart rate and respiratory rate signals by timestamp and transmits them to the multimodal data fusion module via the Ethernet interface. The data packet contains fields for sampling time, channel number, respiratory rate value, and heart rate value.

[0058] During vehicle operation, the millimeter-wave life radar periodically collects data and performs continuous measurements. When the control unit detects that the signal strength is below a set threshold, it executes a self-test procedure, including antenna port connection status detection and echo path monitoring. The self-test results are written to the status register and reported to the vehicle's central control unit.

[0059] Through the above structure and processing flow, millimeter-wave life radar can continuously output the breathing frequency and heartbeat signals of the driver and passengers inside the vehicle, providing input support for subsequent multimodal data fusion and physical-driven decision-making.

[0060] Furthermore, in this embodiment of the invention, the hidden camera and infrared light spot projection module are used to collect the driver's facial expression signals and eye posture signals. This device achieves image acquisition in low-light or no-visible-light conditions through infrared imaging, and provides the image signals to the multimodal data fusion module.

[0061] The stealth camera includes an imaging lens, an infrared filter, an image sensor, a driving circuit, and a communication interface. The infrared light projection module includes an infrared laser emitter, diffusion optics, a power control circuit, and a synchronization trigger circuit. The camera and the infrared light projection module together constitute an infrared active illumination imaging system.

[0062] The camera is installed in a mounting cavity behind the vehicle's central control display screen, with the lens facing the driver's face. An infrared-transmitting panel is located at the front of the mounting cavity; this panel is made of a high-transmittance, uniformly thick infrared transparent material. An infrared light spot projection module is fixed to the side of the camera, projecting infrared light spots evenly onto the driver's facial surface area via an optical diffuser.

[0063] The camera's image sensor is an infrared-sensitive CMOS sensor, with resolution and frame rate set by the system control unit. An infrared filter is mounted at the front of the lens to suppress ambient visible light interference, allowing only infrared light to enter the image sensor. The drive circuit provides the camera with clock signals, bias voltage, and exposure control signals. The communication interface uses a low-voltage differential signal channel to connect to the vehicle's central control unit for transmitting image data frames.

[0064] The infrared laser emitter of the infrared spot projection module operates at a wavelength of 940 nanometers. Diffusion optics disperse the laser beam into an array of light spots covering the driver's face. A power control circuit adjusts the emission power to ensure the illumination intensity remains within a safe range. A synchronization trigger circuit receives the camera drive signal and controls the infrared spot emission to synchronize with the camera exposure via a fixed delay.

[0065] The image signals captured by the camera are converted into digital signals by the driver circuit and then input to the interface module. The interface module performs image buffering, frame numbering, timestamp marking, and data packet encapsulation operations. Image data is stored in the camera buffer according to a fixed frame period and transmitted to the multimodal data fusion module via the Ethernet interface.

[0066] During image processing, the control unit performs grayscale normalization and noise suppression on each frame of the image. A geometric feature-based optical detection method is used to extract the driver's facial region. Within the facial region, the eye position is determined through brightness gradient and contour analysis, and the eye pose angle is calculated. Let the facial coordinate system... The center point of the eyeball is The pupil direction vector is The rotation matrix of the head pose is Then the direction vector of the eye pose in the vehicle coordinate system Calculated by the following formula: Among them, symbols This represents the vector of the eye's line of sight in the vehicle coordinate system, with the symbol... This represents the rotation matrix from the face coordinate system to the vehicle coordinate system, with the symbol... This represents the eye-direction vector. The control unit determines the driver's line-of-sight deviation angle based on the angle between this vector and the vehicle's forward direction.

[0067] Head pose calculation is achieved through geometric solving of facial key points. Based on the coordinates of key points such as the tip of the nose, corners of the eyes, and corners of the mouth in the image plane, combined with the camera intrinsic parameter matrix... and 3D template point coordinate set The PnP algorithm is used to calculate the head pose rotation matrix. Translation vector Its mathematical expression is: Among them, symbols The scaling factor, symbol For image plane coordinates, the symbol is... World coordinates, symbol The intrinsic parameter matrix of the camera, symbol and These represent the rotation matrix and the translation vector, respectively.

[0068] During operation, the camera and infrared light projection module work synchronously at a fixed frequency. In each sampling cycle, the infrared light projection module emits a light spot first, and the camera performs exposure acquisition after receiving the synchronization signal. After the image data is transmitted to the central control unit via the frame buffer, the system immediately executes the next cycle operation. The entire process is managed by timed interrupts from the control unit to ensure periodic stability.

[0069] The hidden camera and infrared beam projection module are powered by the vehicle's low-voltage power system, with a voltage of 12 volts DC. The module incorporates overcurrent protection and temperature monitoring circuitry. When the temperature sensor detects that the internal temperature of the module exceeds a preset threshold, the control circuit reduces the infrared beam emission power and records the status information.

[0070] To ensure the stability of the imaging area, the camera mounting position is calibrated during the assembly phase. During calibration, a standard target is set, and multiple frames of images are acquired while the vehicle is stationary. The transformation matrix TISCS between the camera coordinate system and the vehicle coordinate system is calculated, and the matrix elements are stored in non-volatile memory for subsequent spatial transformation of attitude data.

[0071] During system power-on initialization, the camera performs lens focus self-check and exposure baseline calibration. The lens focus self-check involves fine-tuning the focus using a drive motor to detect the peak position of the image sharpness function. Exposure baseline calibration is performed by measuring dark current noise and adjusting the bias voltage in a dark environment inside the vehicle. After calibration, the camera status register is set to active status.

[0072] During vehicle operation, the hidden camera continuously outputs image frame data. The control unit adds a time tag to each image frame and sends it to the multimodal data fusion module via a communication interface. The multimodal data fusion module synchronizes the data with other sensor data based on the time tags, thereby ensuring the temporal consistency of each data source.

[0073] Through the above structure and processing flow, the hidden camera and infrared light spot projection module can continuously output the driver's facial expression signals and eye posture signals for use by the subsequent fusion and decision-making modules.

[0074] Furthermore, the triboelectric nanogenerator sensing network in this embodiment of the invention is used to collect occupant posture change signals, seat belt status signals, and seat pressure signals. This network generates charge signals through the relative friction between the occupant and the surfaces of the seat and seat belt. After conversion and processing, these signals are transformed into electrical signals for input, reflecting the occupant's real-time status.

[0075] The triboelectric nanogenerator sensing network consists of triboelectric nanogenerator sensing units, signal conditioning circuits, analog-to-digital converter modules, communication interfaces, and mounting support structures. Each triboelectric nanogenerator sensing unit includes a polytetrafluoroethylene (PTFE) film layer, a conductive fabric layer, a flexible electrode layer, and an encapsulation layer. The PTFE film layer and the conductive fabric layer are arranged opposite each other, forming a triboelectric interface. When an occupant comes into contact with or moves, triboelectric charges are generated at the interface. The flexible electrode layer is connected to the conductive fabric layer and is used to collect the charge signals. The encapsulation layer is made of an elastic polymer material to maintain the mechanical stability and durability of the sensing unit.

[0076] The sensing units are installed on the surface of the seat cushion, backrest, and seat belt. The sensing units on the seat cushion detect the occupant's posture and weight distribution, the sensing units on the backrest detect changes in upper body posture, and the sensing units on the seat belt detect the tightness of the seat belt. All sensing units are connected to the signal conditioning circuit via flexible ribbon cables, with a shielding layer on the outer layer of the cables to reduce electromagnetic interference.

[0077] The signal conditioning circuit includes a charge amplifier, a low-pass filter, and a bias circuit. The charge amplifier converts the triboelectric signal into a voltage signal, and the low-pass filter removes high-frequency noise components. The conditioned signal is then input to the analog-to-digital converter (ADC), with the sampling rate set by the control unit. The ADC outputs a digital signal to the communication interface.

[0078] The communication interface connects to the vehicle's central control unit via a standard serial bus. During each sampling cycle, the multiple signals acquired by the analog-to-digital converter are packaged into frames containing channel numbers, sampling times, and voltage values. The communication interface then sequentially sends these frame data to the vehicle's central control unit.

[0079] Occupant posture change identification is based on time series analysis of voltage signals. Let the sampling time interval be... The voltage value of the i-th sample is Then the rate of change of the signal Defined as: Among them, symbols Represents the rate of change of voltage, symbol Indicates the first The voltage value sampled next time, symbol This indicates the sampling interval. The control unit determines this interval by... Determine whether the occupant posture change event exceeds the set threshold.

[0080] The seatbelt status signal is obtained by detecting the average voltage value of the seatbelt's sensing unit. Let the average voltage at the current moment be... The threshold is when When the seatbelt is deemed to be unfastened; when When the seatbelt is in a properly worn state, it is determined that the seatbelt is in a properly worn state. This status information is sent to the multimodal data fusion module along with the data frame.

[0081] The seat pressure signal is calculated from the outputs of multiple sensing units in the seat cushion area. Let the number of sensing units be n, and the output voltage of the k-th sensing unit be... The representative value of seat pressure distribution. It can be expressed by the following formula: Among them, symbols Represents the average pressure signal value, symbol Represents the output voltage of the k-th sensing unit, symbol This indicates the number of sensor units involved in the calculation. This value is used to determine whether the occupant is seated.

[0082] The signal conditioning circuit and the analog-to-digital conversion module share a single power supply, powered by the vehicle's low-voltage power supply with a rated voltage of 5 volts DC. The power input is equipped with filtering and overvoltage protection circuits. Upon power-up, the system executes a self-test program, including channel connectivity detection and static noise testing. After passing the tests, the system sets the status register to normal.

[0083] The triboelectric nanogenerator sensing unit employs a modular design for installation. The sensing unit in the seat cushion area is embedded between the foam layer and the surface fabric layer, while the backrest is fixed to the back of the interior panel via an adhesive layer. The seatbelt strap is connected to the webbing via stitching. After installation, the system performs initial calibration. During calibration, a static voltage value is collected as a baseline in an unattended state, and subsequent signals are referenced to this baseline value as zero.

[0084] During vehicle operation, the triboelectric nanogenerator sensor network continuously collects and outputs digital signals. The control unit periodically reads data from all channels and calculates attitude change events and seatbelt status. The results are timestamped, packaged into data frames, and sent to the multimodal data fusion module via the communication interface.

[0085] When a signal anomaly is detected (including communication errors, power supply anomalies, or excessively low signal amplitude), the control unit records an error code and reports the status information to the vehicle's central control unit. The error code is used for subsequent fault diagnosis.

[0086] Through the above structure and processing flow, the triboelectric nanogenerator sensor network can provide occupant posture change signals, seat belt status signals, and seat pressure signals, providing input support for subsequent fusion processing and decision calculation.

[0087] Furthermore, in this embodiment of the invention, the multimodal data fusion module is used to synchronize, calibrate, and fuse the multi-source signals output by the quantum lidar array, millimeter-wave life radar, stealth camera and infrared light spot projection module, and triboelectric nanogenerator sensor network, so as to generate unified environmental and occupant status data and provide input for the physics-driven decision engine.

[0088] The multimodal data fusion module consists of a time synchronization unit, a coordinate system calibration unit, a data fusion unit, and an output interface. The time synchronization unit is responsible for the temporal alignment of data from different sensors; the coordinate system calibration unit performs spatial coordinate transformation; the data fusion unit performs comprehensive data generation and data quality management; and the output interface is responsible for transmitting the fusion results to subsequent modules.

[0089] The time synchronization unit maintains consistency with the vehicle's global clock, employing both hardware synchronization signals and software timestamps to ensure time-domain consistency. Each sensor module appends a high-precision timestamp when acquiring data. The time synchronization unit receives all sensor output data, calculates time deviations, and performs interpolation correction. Let the sampling time of a certain sensor be... The global clock reference time is The time deviation is Then the synchronized corrected time stamp Determine by the following formula: Among them, symbols Indicates the original sampling time, symbol Represents global clock time, symbol Indicates the time offset, symbol This represents the corrected timestamp. The correction results are written into a unified time index table for subsequent fusion calculations.

[0090] The coordinate system calibration unit performs spatial coordinate transformation based on the transformation relationship between each sensor and the vehicle coordinate system. Let the quantum lidar array coordinate system be LS, the stealth camera coordinate system be IS, and the vehicle coordinate system be CS. Then, the spatial points under the sensor coordinate system... Convert to vehicle coordinates The calculation formula is: Among them, symbols This represents the homogeneous transformation matrix from the i-th sensor coordinate system to the vehicle coordinate system. Each transformation matrix is ​​calculated and stored in non-volatile memory during the system initialization and calibration phase. The coordinate system calibration unit performs coordinate transformations based on the matrix parameters in each operating cycle, outputting environmental and occupant status data in a unified coordinate system.

[0091] The data fusion unit receives multimodal data after time synchronization and coordinate transformation, and performs information fusion operations. The fusion process consists of three steps: data matching, state determination, and comprehensive generation.

[0092] During the data matching phase, the system matches data frames from different sensors based on time stamps. Let the fusion time of the k-th frame be... Then the system is in the time window Search for data frames from each sensor and extract them. The closest frame is used as the fusion input, where This is the time tolerance constant.

[0093] During the status determination phase, the system performs joint analysis based on environmental and occupant data. A quantum lidar array provides spatial coordinates and velocity information of obstacles, a millimeter-wave life radar provides occupant physiological status signals, a stealth camera provides facial and eye posture information, and a triboelectric nanometer sensor network provides posture and seatbelt status signals. Based on data source identifiers and time stamps, the system combines environmental and occupant information from the same moment into a linked record.

[0094] During the synthesis and generation phase, the data fusion unit generates a structured data frame. This data frame includes the following fields: time label, obstacle coordinate set, obstacle velocity set, driver heart rate, driver breathing rate, driver head posture matrix, driver gaze direction vector, occupant posture status, seat belt status, and seat pressure signal. Each field is accompanied by a validity flag. For missing data, the system fills in the gaps using linear interpolation or the nearest-value hold method.

[0095] To ensure the reliability of the fused data, the data fusion unit performs a data quality assessment. Let the data validity weight of sensor i be . Its current data confidence level is The formula for calculating the overall confidence level Q is: Where, the symbol N represents the number of sensors participating in the fusion, and the symbol Represents sensor weights, symbol This indicates the current confidence level, and the symbol Q represents the overall confidence level of the fused data. If Q is lower than a set threshold, the system marks the corresponding field as invalid in the fusion result and notifies the central control unit to perform diagnostics.

[0096] After fusion, the output interface transmits the integrated data frames to the physical drive decision engine. The transmission uses an Ethernet interface and a fixed frame structure. Each frame contains a data type identifier, data length, checksum, and timestamp. Before each transmission, the output interface performs a data integrity check, including CRC check and frame sequence number continuity detection.

[0097] During system operation, the multimodal data fusion module periodically performs synchronization, calibration, and fusion operations at a fixed frequency. If a sensor is detected to be offline or experiencing a communication anomaly, the module skips the data from that sensor in the current cycle and records the abnormal status, ensuring that the remaining data fusion processes proceed normally.

[0098] Through the above structure and process, the multimodal data fusion module achieves unified temporal and spatial alignment of output data from different sensors, generating comprehensive environmental and occupant status data, providing complete input support for subsequent physics-driven decision calculations.

[0099] Furthermore, the physics-driven decision engine in this embodiment of the invention is used to perform simulation calculations and decision generation based on vehicle external environment data and occupant state data, and output control commands to the actuators. The core design idea of ​​this engine is to generate safety response decisions based on a physical model through dynamic and biomechanical calculation processes, thereby achieving active safety control of the vehicle.

[0100] The physics-driven decision engine comprises a data receiving module, a simulation calculation module, an execution control module, and an internal communication bus. The data receiving module receives unified data frames from the multimodal data fusion module; the simulation calculation module performs vehicle dynamics and occupant biomechanics calculations; and the execution control module generates and issues control commands.

[0101] The data receiving module connects to the vehicle communication bus and periodically reads fused data frames. Each frame contains fields such as environmental target coordinates, target speed, driver and occupant physiological state signals, attitude status, and seat belt status. The module performs data integrity checks, including frame header detection, checksum verification, and timestamp continuity checks. Verified data is buffered for the simulation calculation module to read.

[0102] The simulation calculation module includes a vehicle dynamics calculation unit and an occupant biomechanics calculation unit. The vehicle dynamics calculation unit is used to calculate the motion relationship between the vehicle and external targets. Assuming the vehicle speed is v, the vehicle acceleration is a, and the target distance is d, the time required for the vehicle to brake is calculated. Defined by the following formula: Where, the symbol v represents the vehicle speed, the symbol a represents the braking acceleration, and the symbol... Indicates the time required for the vehicle to decelerate from its current speed to zero. Braking distance. The calculation formula is: Among them, symbols This indicates the distance the vehicle has traveled during braking. The system uses the current target distance *d* and the braking distance as a reference. The difference in values ​​determines whether braking control is triggered. At that time, the system enters an early warning state, among which... This is the safety distance compensation coefficient.

[0103] The occupant biomechanics calculation unit is used to calculate the force state of the occupant under potential collision conditions. It assumes the occupant's mass is... The vehicle deceleration is Seatbelt restraint force is The displacement of the occupants is Then the net force on the occupants for Among them, symbols Indicates the forces acting on the occupants, symbol Indicates passenger mass, symbol Indicates vehicle deceleration, symbol This indicates the seatbelt restraint force. The control unit, based on... The size of the seatbelt is used to calculate the trigger condition for the seatbelt tightening action. When Exceeding the set threshold At that time, the system generates a seatbelt tightening command. To synchronize braking and seatbelt tightening actions, a time-constrained model is established for the system. Let the response delay be... The target distance is d, and the vehicle speed is v. If the braking acceleration is 'a', the response delay is calculated using the following formula: Among them, symbols This indicates the available time window from when the system detects an action to when it can perform it. Control Unit Compare with the internal calculation cycle, when When the time is less than the preset cycle threshold, the braking action and seat belt tightening action are performed in advance.

[0104] The simulation calculation module updates the vehicle state variables and occupant state variables in each calculation cycle. The calculation process includes four steps: input reading, state update, condition judgment, and output generation.

[0105] Input Read: Reads the latest fused data frame from the buffer.

[0106] Status update: Update velocity, acceleration, and force parameters based on vehicle dynamics equations and occupant biomechanics equations.

[0107] Condition judgment: Logically judge the distance and mechanical conditions to determine the type of safety action that needs to be performed.

[0108] Output generation: Generates corresponding control commands and transmits them to the execution control module.

[0109] The execution control module generates braking, seatbelt tightening, and airbag deployment commands based on simulation calculations. The command message structure includes command type, execution parameters, priority, and timestamp fields. If multiple actions are triggered within the same cycle, the system arbitrates the commands according to a preset priority. The priority, from highest to lowest, is: airbag deployment, braking, and seatbelt tightening. After generating the commands, the execution control module verifies and confirms them before sending them to the vehicle's central control unit.

[0110] An internal communication bus enables data transmission between modules within the physical drive decision engine. The bus employs a bidirectional isolation interface to ensure independent signal transmission from each module and prevent interference. At the end of each cycle, the control unit records all input and output parameters and stores them in non-volatile memory.

[0111] To ensure stable system operation, the physical drive decision engine performs a self-test during power-on. The self-test includes memory read / write checks, bus communication checks, and computational unit verification. The self-test results are stored as status codes and reported to the vehicle's central control unit.

[0112] During normal operation, the decision engine executes calculation cycles periodically at a fixed frequency. When a change in the external environment or an abnormal occupant status is detected, it immediately enters a high-priority calculation mode, shortens the calculation cycle, and recalculates the control output.

[0113] Through the above structure and process, the physics-driven decision engine completes multi-source data reading, physical simulation calculation, action decision and control command generation in each operating cycle, providing real-time decision input for the vehicle's active safety actuators.

[0114] Furthermore, the actuator linkage system in this embodiment of the invention is used to receive control commands generated by the physical drive decision engine and execute corresponding vehicle braking actions, seat belt tightening actions, and airbag deployment actions. The actuator includes a braking system, a seat belt pretensioner, an airbag controller, a drive interface circuit, and an execution monitoring module.

[0115] The braking system is electrically connected to the vehicle's central control unit and is used to perform vehicle deceleration or emergency braking operations. The braking system consists of an electronically controlled braking unit, hydraulic actuators, and pressure sensors. The electronically controlled braking unit receives braking commands from the decision engine and converts them into drive signals to control the brake motors of the hydraulic actuators. The hydraulic actuators are connected to the wheel brake calipers via the master cylinder and generate braking force under the action of the drive signal. The pressure sensors detect the hydraulic pressure and transmit feedback signals back to the central control unit for closed-loop control.

[0116] The seatbelt pretensioner is used to tighten the seatbelt when a potential collision risk is detected. The seatbelt pretensioner consists of a drive motor, a reel assembly, a torque sensor, and a release mechanism. Control signals are transmitted via the vehicle's central control unit to the pretensioner drive module, which controls the motor's rotation direction and speed according to the command parameters. The motor drives the reel to rotate via a gear set, thereby tightening the seatbelt belt. The torque sensor detects the axial torque of the reel and feeds the detected value back to the central control unit to determine if tightening is complete.

[0117] The airbag controller includes an ignition control unit, an ignition circuit, an acceleration sensor, and a backup power module. The ignition control unit receives the airbag deployment command from the physical drive decision engine, verifies it, and then outputs ignition current to the ignition circuit. The acceleration sensor detects the vehicle's longitudinal acceleration in real time and compares the measured value with a preset threshold to trigger a redundancy protection mechanism. The backup power module uses a capacitor energy storage structure, which can still provide the energy required for ignition when the vehicle's main power is disconnected, ensuring the reliability of the airbag deployment.

[0118] The execution monitoring module monitors the operational status of all actuators. This module collects signals such as brake pressure, pretensioner torque, and airbag ignition current, and compares them with the command parameters output by the decision engine. If the detected deviation from the execution result exceeds a threshold, the module generates an error code and reports it to the vehicle's central control unit.

[0119] In terms of control procedures, the actuator linkage system operates in the following sequence: (1) Receiving stage: The vehicle central control unit receives the control command frame issued by the physical drive decision engine.

[0120] (2) Analysis stage: The central control unit analyzes the command type and extracts the commands for braking, seat belt tightening or airbag deployment.

[0121] (3) Execution phase: The central control unit distributes the parsed commands to the corresponding execution modules, and each module executes the actions synchronously according to the timestamp.

[0122] (4) Feedback phase: Each execution module returns the actual execution signal to the monitoring module, and the monitoring module compares the target value with the actual value.

[0123] (5) Confirmation phase: If the actual execution result meets the set conditions, the central control unit records the execution completion status; if the deviation exceeds the allowable range, the system generates an error report and issues a prompt signal.

[0124] The control signal for braking is an analog voltage signal, ranging from 0V to 5V. The mapping relationship between the control voltage and the braking force is determined through a calibration curve. Let the braking force be... The control voltage is Then the relationship between the two satisfies: Among them, symbols Indicates the actual braking force, symbol Indicates control voltage, symbol This is the proportional coefficient of the braking system.

[0125] The seatbelt tightening control signal is a PWM (Pulse Width Modulation) signal with a fixed period, and the duty cycle is calculated and determined by the control unit. Let the duty cycle be D, and the maximum motor voltage be... The average output voltage is ,but Among them, symbols This represents the average voltage at the motor terminals, with the symbol D indicating the PWM duty cycle. This indicates the motor's rated voltage. The control unit controls the tightening speed by adjusting the value of D.

[0126] The airbag deployment control signal is a single-pulse trigger signal, and the signal width is determined by the duration of the ignition current. Ignition current With ignition time The relationship is Among them, symbols Indicates ignition energy, symbol Indicates ignition current, symbol Indicates the duration of the current flow. The control unit ensures... The ignition threshold must be greater than the ignition threshold to ensure effective airbag deployment.

[0127] To prevent conflicting commands from multiple actuators, the central control unit sets up an action priority table. When multiple commands arrive simultaneously, they are executed in the order of airbag deployment > braking action > seatbelt tightening action. If a higher-priority action is being executed, lower-priority commands enter a waiting queue until the higher-priority action is completed or times out.

[0128] During the power-on phase, the actuator linkage system performs a self-test. The self-test items include brake pressure sensor calibration, pretensioner motor connectivity detection, and airbag ignition circuit resistance detection. All test results are recorded in the status register, and any abnormalities trigger an alarm output.

[0129] During vehicle operation, the actuator linkage system periodically communicates with the physical drive decision engine. When the monitoring module detects an execution timeout, abnormal signal, or execution failure, the system immediately stops the action and reports the fault information.

[0130] Through the above structure and process, the actuator linkage system realizes the coordinated execution of braking, seat belt pretensioning and airbag control, ensuring that the control commands generated by the physical drive decision engine are reliably executed at the vehicle level.

[0131] Furthermore, in this embodiment of the invention, the data recording and playback module is used to encrypt and store, time-index, and perform event reconstruction calculations on the data generated by the quantum lidar array, millimeter-wave life radar, stealth camera, triboelectric nanogenerator sensor network, and multimodal data fusion module, so as to support the post-event analysis and security verification of the system.

[0132] The data recording and playback module includes a data caching unit, an encrypted storage unit, a time indexing unit, a playback calculation unit, and an interface control unit. Each unit is connected via an internal bus and communicates with the vehicle's central control unit.

[0133] The data caching unit receives and temporarily stores real-time data from various sensors and fusion modules during system operation. The caching adopts a ring-shaped storage structure, divided into a high-speed cache and a slow cache. The high-speed cache stores the most recent raw data frames, while the slow cache is used for batch writing to the encrypted storage unit. The cache size and refresh cycle are set by system parameters.

[0134] The encrypted storage unit encrypts the cached data and writes it to a non-volatile storage medium. The encryption algorithm uses a symmetric encryption method. Let the input data block be... The key is K, and the encryption function is... Then the encrypted data block Defined by the following formula: Among them, symbols Represents the encrypted data block, symbol Represents the original data block, symbol This represents an encryption function that takes a key K as a parameter. The key is generated by the vehicle's central control unit and distributed to the data logging module via a secure channel.

[0135] The time index unit generates a unique time identifier for each frame of data. The time identifier consists of a timestamp provided by the global clock and a sequence number. Let the global clock time be... The serial number is n, and the time is... The definition is as follows: Among them, symbols Indicates time index, symbol This indicates the current clock time, and the symbol 'n' represents the frame number. Indicates the data sampling frequency. The time index is used to quickly locate the target data segment during playback.

[0136] The playback calculation unit is used to reconstruct the event process after a security incident occurs. Upon receiving a playback command, the module extracts data for the corresponding time period from the encrypted storage unit based on the time index. The extracted data is then decrypted using a function. Restore to the original data frame: Among them, symbols This represents the decryption function corresponding to the encryption function. The decrypted data frames are input into the simulation calculation module in chronological order to re-execute dynamic and biomechanical calculations for event reconstruction.

[0137] During event reconstruction, the system maintains the same computational logic as in real-time operation. The simulation calculation module sequentially calculates vehicle speed, acceleration, braking command, seatbelt tightening signal, and airbag deployment signal based on the extracted historical data. The calculation results are compared with the recorded data to verify the correctness of the system response.

[0138] The interface control unit manages data input and output. This unit includes a communication interface, control logic, and a data buffer. The communication interface uses the Ethernet physical layer to exchange data with the vehicle's central control unit and diagnostic tools. The interface control unit performs data integrity checks, including CRC checks and frame length verification, before writing to encrypted storage. If the checks fail, the current data frame is discarded and an error log is generated.

[0139] Under normal vehicle operation, the data recording and playback module performs data writing at fixed intervals. The write interval is configurable, with a default value of 100 milliseconds. After each write operation, the system updates the write pointer and records the latest time index. When the storage space reaches the preset limit, the system automatically overwrites the oldest data segment.

[0140] In the event of a security incident, the central control unit triggers emergency storage mode. In emergency storage mode, the system freezes the most recent 30 seconds of historical data and forces it to be written to non-volatile storage media. The frozen data is marked as an event segment to prevent it from being overwritten by subsequent writes.

[0141] Playback is triggered by the vehicle maintenance system or diagnostic terminal. Maintenance personnel input a time range or event number, and the system reads the corresponding data index and executes the playback. The playback calculation module outputs data frame by frame at the same time step. The output includes vehicle dynamic parameters, occupant status parameters, and control signals.

[0142] To prevent unauthorized access, the encrypted storage unit is equipped with an access control mechanism. Read operations require authorization from the central control unit. After authorization, the interface control unit generates a temporary access key, which automatically expires after playback ends.

[0143] During the power-on phase, the system performs a self-test, checking the encryption module, memory read / write functionality, and interface communication connectivity. If any test fails, the system will disable the data logging function and report a fault code.

[0144] Through the above structure and operation process, the data recording and playback module can realize encrypted recording, time indexing and reconstructed playback of multi-source data during vehicle operation and safety events, providing technical support for system verification and post-event analysis.

[0145] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An AI-powered intelligent vehicle safety detection and monitoring method, characterized in that, Includes the following steps: S1. Collect environmental information around the vehicle using a quantum lidar array, wherein the quantum lidar array performs environmental target detection and distance measurement based on entangled photon pairs; S2. The physiological state signals of the driver and passengers are collected by combining millimeter-wave life radar and a hidden camera. The millimeter-wave life radar collects respiratory rate signals and heart rate signals, and the hidden camera collects facial expression signals and posture signals. S3. Acquire occupant posture change signals, seat belt status signals, and seat pressure signals through a triboelectric nano-power generation sensor network. The triboelectric nano-power generation sensor network consists of triboelectric nano-power generation sensor units embedded in the seat, seat belt, and steering wheel. S4. Input the signals collected in steps S1 to S3 into the physical drive decision engine, perform simulation calculations based on the vehicle dynamics model and the occupant biomechanics model, and generate safety response commands. S5. Perform vehicle control operations according to the safety response command, including braking, seat belt tightening and airbag deployment.

2. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, The quantum lidar array includes a quantum light source, a quantum modulation unit, and an entangled photon detector. The quantum light source generates entangled photon pairs. The quantum modulation unit modulates the phase of entangled photons. The entangled photon detector receives the reflected light signal and calculates the target distance and relative velocity.

3. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, The millimeter-wave life radar emits electromagnetic waves in the 60GHz to 77GHz frequency band. The millimeter-wave life radar calculates the micro-displacement of the driver's chest cavity based on the Doppler displacement of the reflected signal, and extracts the respiratory rate signal and heartbeat signal from the micro-displacement.

4. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, The hidden camera is installed behind the central control display screen and works in conjunction with the infrared light spot projection module to collect the driver's facial expression signals and eye posture signals in the absence of visible light.

5. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, The triboelectric nanogenerator sensing unit in the triboelectric nanogenerator sensing network is composed of a polytetrafluoroethylene thin film layer and a conductive fabric layer. When the occupant comes into contact with the triboelectric nanogenerator sensing unit, a triboelectric charge signal is generated. The triboelectric charge signal is converted into an electrical signal by the sensing unit and then input to the vehicle's central control unit.

6. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, The physics-driven decision engine includes a data receiving module, a simulation calculation module, and an execution control module. The data receiving module receives signals from the quantum lidar array, millimeter-wave life radar, and triboelectric nanogenerator sensor network. The simulation calculation module calculates vehicle speed, acceleration, and braking distance based on a vehicle dynamics model, and calculates human body force and displacement parameters based on a occupant biomechanical model. The execution control module generates control commands based on the simulation calculation results and transmits them to the execution mechanism.

7. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, The security response command includes the following actions: (a) Execute braking control action when the distance to an obstacle is less than a preset safe distance; (b) When an abnormality is detected in the driver's heartbeat or breathing signal, the vehicle shall decelerate and issue a warning signal; (c) When a seatbelt status signal indicating that the seatbelt is loose is detected, the seatbelt tightening action is performed and a voice prompt is issued.

8. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, It also includes a multimodal data fusion step. The multimodal data fusion step performs time synchronization processing to uniformly calibrate the quantum lidar array signal, millimeter-wave life radar signal, and triboelectric nanogenerator sensor network signal and generate comprehensive safety status data.

9. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, The physics-driven decision engine performs dynamic threshold update calculations before generating safety response commands. These calculations are based on the relationship between vehicle speed v, target distance d, and braking acceleration a. The braking trigger threshold is updated based on the response delay tr.

10. The AI-powered intelligent vehicle safety detection and monitoring method according to claim 1, characterized in that, It also includes data recording and playback steps. The data recording module performs encrypted storage of the quantum lidar array signals, millimeter-wave life radar signals, and triboelectric nanogenerator sensor network signals. Upon detecting a security event, the playback module retrieves the encrypted data and performs event reconstruction calculations.

Images