A vehicle-mounted pet retention safety guarantee method, system and vehicle

Abstract

The application provides a kind of vehicle-mounted pet retention safety guarantee method, system and vehicle, it is related to vehicle intelligent control technical field, the method responds pet mode opening instruction, establishes the controlled function subnetwork of logic isolation in vehicle-mounted network, establishes controlled communication path by strategy execution gateway authentication;Based on pet basic information, query breed database and combine correction coefficient to calculate dynamic safety temperature threshold;Obtain the temperature of car outside environment parameter and compartment, iterate calculation temperature prediction sequence by thermodynamic prediction model, predict over-temperature and send refrigeration instruction to HVAC system in advance;Obtain the video stream in car, extract three-dimensional behavior feature vector, normalize weighted fusion and get comprehensive behavior abnormal score, trigger multi-modal pacification instruction by grading determination.The application realizes the safety guarantee to the pet retained in car by the integrated technical scheme of network isolation, predictive temperature control and behavior perception pacification.

Description

Technical Field

[0001] This invention relates to the field of vehicle intelligent control technology, and in particular to a method, system and vehicle for ensuring the safety of pets left in a vehicle. Background Technology

[0002] Pet mode is a vehicle-specific feature designed to maintain a safe temperature environment for pets inside the car when the owner is away for a short period. Currently, some mass-produced models already have a basic pet mode function, which can keep the air conditioning running and monitor the interior temperature after the owner leaves the vehicle.

[0003] However, existing pet-mode solutions have significant safety hazards in practical use: in hot weather or under direct sunlight, the temperature inside the car rises rapidly, and the current solutions can only passively trigger a cooling response after the temperature exceeds the limit, resulting in a lag in temperature regulation. Pets still face the risk of high-temperature exposure during this period. In addition, the current solutions lack effective monitoring of the pet's behavior inside the car. After the owner leaves the vehicle, they cannot promptly detect any abnormalities in the pet or intervene in the pet's discomfort. Summary of the Invention

[0004] In order to solve the technical problems existing in the background art, the present invention provides a method, system and vehicle for ensuring the safety of pets staying in a vehicle.

[0005] To achieve the above objectives, a first aspect of the present invention provides a method for ensuring the safety of pets left in a vehicle, comprising: In response to the pet mode activation command, the system obtains vehicle status information and basic pet information, establishes a logically isolated controlled function subnet in the vehicle network, authenticates cross-domain communication requests through the policy execution gateway, and establishes a controlled communication path between the main control function domain and the controlled function subnet. Based on the pet's basic information, the dynamic safe temperature threshold is obtained by querying the breed database and calculating with a correction factor. The system acquires external environmental parameters and current cabin temperature, and iteratively calculates cabin temperature prediction sequence using a cabin thermodynamic prediction model. When the prediction sequence exceeds the dynamic safety temperature threshold, it sends an advance cooling command to the HVAC system through the controlled communication channel. The video stream from the in-vehicle camera is acquired, and the motion energy features are calculated using optical flow, the posture disorder features are calculated using skeletal keypoint detection, and the cross-union ratio features of the driving area are calculated using target detection to obtain a three-dimensional behavior feature vector. The three-dimensional behavioral feature vectors are normalized and then weighted and fused to obtain a comprehensive behavioral anomaly score. The behavioral anomaly level is determined by classification, and the corresponding multimodal reassurance command is triggered based on the behavioral anomaly level.

[0006] Furthermore, the policy enforcement gateway receives cross-domain communication requests from the controlled functional subnet, extracts the source address, destination address, and instruction type of the request, and matches them one by one with the preset rules in the built-in access control list. When the source address, destination address, and instruction type all match the preset rules, forwarding is allowed and the controlled communication path is established. When any parameter does not match, forwarding is rejected and an authentication failure event is recorded.

[0007] Furthermore, the external environmental parameters include solar radiation heat flux, which is calculated based on the solar transmittance of the vehicle window glass, the vehicle glass area, solar irradiance, and solar incident angle. The solar incident angle is determined based on the vehicle's GPS coordinates, current timestamp, and vehicle orientation, using the solar declination angle, hour angle, and the vehicle's latitude, and then calculated using spherical trigonometric formulas to determine the solar altitude angle and azimuth angle.

[0008] Furthermore, the calculation of the motion energy feature includes: calculating the inter-frame pixel displacement field based on optical flow or other inter-frame motion estimation methods, identifying and obtaining the pet target area in real time through a target detection model, extracting the horizontal and vertical optical flow components of each pixel in the pet target area, calculating the mean value of the optical flow amplitude of each pixel, and obtaining the motion energy feature by normalization processing after passing through a preset upper limit value of motion energy.

[0009] Furthermore, the calculation of the posture disorder feature includes: obtaining the horizontal and vertical displacements of each key point between adjacent frames based on the skeletal key point detection model, calculating the displacement amplitude of each key point and performing a weighted average based on the detection confidence of each key point, and obtaining the posture disorder feature by normalization processing after a preset upper limit value of posture disorder.

[0010] Furthermore, the calculation of the intersection-union ratio feature of the driving area includes: obtaining a pet target detection box and a preset driving area interest box, wherein the driving area interest box is a pre-calibrated rectangular box containing the steering wheel, dashboard and driver's seat area; calculating the ratio of the intersection area to the union area of ​​the pet target detection box and the driving area interest box to obtain the driving area intersection-union ratio feature; when the driving area intersection-union ratio feature is greater than or equal to a preset intrusion judgment threshold, a special alarm for driving area intrusion is triggered.

[0011] Furthermore, the multimodal reassurance instructions include: When the level of abnormal behavior is mild anxiety, an audio reassurance is triggered; When the level of abnormal behavior is moderate stress, light therapy is triggered. When the abnormal behavior level is severe stress, a remote notification is triggered; The effectiveness of reassurance is judged based on the changing trend of the comprehensive abnormal behavior score, and the reassurance strategy is upgraded when the reassurance is ineffective.

[0012] Furthermore, the formula for calculating the dynamic safety temperature threshold is as follows: ,in, To obtain the baseline heat resistance temperature for this variety from the variety database, This is a body shape correction factor determined based on weight ranges. This is a correction factor for physiological maturity determined based on age group.

[0013] Furthermore, the advance cooling command includes a target cooling temperature and an expected cooling start advance. The expected cooling start advance is determined based on the estimated time required for the HVAC system to drop from the current temperature to the dynamic safe temperature threshold, so that the HVAC system can complete temperature regulation before the predicted over-temperature moment arrives.

[0014] A second aspect of the present invention provides a vehicle-mounted pet safety protection system, comprising: The logical partitioning module is used to divide the vehicle network into a main control function domain and a controlled function subnet. The main control function domain includes a cluster of electronic control units that are directly related to vehicle driving safety. The policy execution gateway is electrically connected to the main control functional domain and the controlled functional subnet via vehicle Ethernet or CAN bus, respectively. The memory of the policy execution gateway stores an access control list for parsing, authenticating and forwarding cross-domain communication commands. An environmental perception module, deployed within the controlled function subnet, includes a temperature sensor, a humidity sensor, an external environment sensor, a GPS positioning unit, and an in-vehicle camera. The sensors are electrically connected to the environmental perception module to collect data on external ambient temperature, solar irradiance, vehicle parking position, current cabin temperature, and in-vehicle video stream. The environmental perception module communicates with the instruction agent unit via a data connection. The instruction proxy unit is located within the controlled function subnet. The processor of the instruction proxy unit executes the carriage thermodynamic prediction model algorithm and the pet behavior detection module algorithm stored in the memory. It is used to generate a carriage temperature prediction sequence and a comprehensive behavior anomaly score based on the data collected by the environmental perception module, and generate a control request based on a preset threshold. The instruction proxy unit is electrically connected to the interactive alarm module through a signal connection. The interactive alarm module, located within the controlled function subnet, is used to control the in-vehicle display screen to show safety prompts to the outside of the vehicle, control the speaker to play reassuring audio, and trigger a multimodal adaptive reassurance strategy based on the level of abnormal behavior.

[0015] Furthermore, each rule in the access control list includes a source address field, a destination address field, an instruction type field, and an allow or deny flag. The policy enforcement gateway performs rule-by-rule matching on each cross-domain communication request and writes the authentication results and forwarding records into the audit log.

[0016] Furthermore, the thermodynamic prediction model of the carriage in the instruction agent unit is based on the heat balance equation, integrates the solar radiation heat flux, the convective heat transfer power of the carriage outer wall and the cooling power of the HVAC system, and uses a numerical iteration method to perform iterative calculations to generate a predicted sequence of carriage temperature within a preset time period in the future.

[0017] Furthermore, the interactive alarm module supports multimodal adaptive soothing strategies including: triggering audio soothing for mild anxiety, triggering light soothing for moderate stress, and triggering remote notification for severe stress.

[0018] A third aspect of the present invention provides a vehicle including the vehicle-mounted pet-keeping safety system provided in the second aspect of the present invention.

[0019] Furthermore, the vehicle also includes an in-vehicle display screen, a speaker, and a mobile communication module. The in-vehicle display screen is used to display pet safety reminders to the outside of the vehicle, the speaker is used to play soothing audio, and the mobile communication module is used to push an alarm notification containing real-time video footage to the owner's mobile terminal when the pet's abnormal behavior reaches the level of severe stress.

[0020] Compared with existing technologies, the present invention provides a method, system, and vehicle for ensuring the safety of pets stranded in vehicles. By establishing logical isolation between the main control functional domain and the controlled functional subnet in the vehicle network, and by having a policy execution gateway authenticate and filter all cross-domain communication requests, the interference path of non-security domains to the core control domain is cut off at the architectural level. At the same time, based on the thermodynamic prediction model of the vehicle compartment, the temperature trend is predicted in advance, and HVAC cooling is proactively triggered before overheating occurs. Furthermore, the stress state of pets is perceived in real time through three-dimensional behavioral feature fusion scoring, and graded comforting is provided. This achieves proactive, predictive, and end-to-end safety protection for stranded pets. Attached Figure Description

[0021] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0022] Figure 1 This is a flowchart of a method for ensuring the safety of pets staying in a vehicle, as provided in Embodiment 1 of the present invention.

[0023] Figure 2 The flowchart for establishing a logically isolated controlled subnet and authentication communication is provided in Embodiment 1 of the present invention.

[0024] Figure 3 The flowchart for calculating the dynamic safety temperature threshold is provided in Embodiment 1 of the present invention.

[0025] Figure 4 The flowchart of thermodynamic model prediction and advance cooling control provided in Embodiment 1 of the present invention is shown.

[0026] Figure 5 The flowchart for extracting three-dimensional behavioral feature vectors provided in Embodiment 1 of the present invention is shown.

[0027] Figure 6 This is a flowchart of behavioral abnormality scoring and grading determination and multimodal reassurance triggering provided in Embodiment 1 of the present invention.

[0028] Figure 7 This is an architecture diagram of the vehicle-mounted pet safety protection system provided in Embodiment 2 of the present invention.

[0029] Figure 8 This is a control logic diagram of the vehicle-mounted pet safety protection system provided in Embodiment 2 of the present invention. Detailed Implementation

[0030] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the invention. In the description of the embodiments of the present invention, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "this embodiment" should be understood as "at least one embodiment". The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this embodiment, unless otherwise stated, "a plurality of" means two or more.

[0031] Before describing specific embodiments of the present invention, the key terms involved in the present invention will first be explained: 1. Electronic Control Unit (ECU): This refers to the embedded processor and associated circuitry units in a vehicle used to control specific functions or systems. In automotive electronic architecture, different ECUs are responsible for different functions such as engine management, transmission control, anti-lock braking system (ABS), electronic stability control, air conditioning control, window operation, and infotainment. The ECUs communicate with each other through the vehicle network to collaboratively complete various control tasks for the entire vehicle. An ECU typically includes basic components such as a microcontroller, memory, input / output interfaces, and communication interfaces.

[0032] 2. Automotive Ethernet: This refers to Ethernet communication technology applied in the automotive environment and is the mainstream technology solution for vehicle backbone networks. Automotive Ethernet is based on the IEEE 802.3 standard and is specifically optimized for the electromagnetic compatibility, real-time performance, and reliability requirements of the vehicle environment. Compared to traditional vehicle network technologies, automotive Ethernet has advantages such as high bandwidth (up to 100Mbps, 1Gbps, or even higher), low latency, and support for high-bandwidth data transmission (such as camera video streams). It is widely used in scenarios with high data transmission rate requirements, such as advanced driver assistance systems, infotainment systems, and vehicle diagnostic systems.

[0033] 3. CAN Bus (Controller Area Network): Employing two-wire differential signal transmission, it features strong anti-interference capabilities, good real-time performance, and high reliability. It is particularly suitable for control scenarios with high real-time and safety requirements, such as powertrain control, chassis control, and body control. The CAN bus supports multi-master node communication; any node can actively send data. Bus conflicts are resolved through a non-destructive arbitration mechanism, ensuring that high-priority messages are transmitted first.

[0034] 4. HVAC System (Heating, Ventilation and Air Conditioning): This refers to the comprehensive environmental control system in a vehicle used to regulate the temperature, humidity, cleanliness, and airflow distribution inside the vehicle. An HVAC system typically includes a refrigeration subsystem (compressor, condenser, evaporator, expansion valve, etc.), a heating subsystem (engine waste heat or PTC heater), a ventilation subsystem (blower, ductwork, air vents), and a control subsystem (temperature sensors, mode motor, controller, etc.).

[0035] 5. Optical Flow Method: This refers to a computational method in computer vision used to estimate the motion of pixels in a video image sequence. The basic principle of optical flow is that between two consecutive frames, the position of a point on the same object surface changes in the image. This change forms a two-dimensional vector field, which is the optical flow field. Each pixel corresponds to a motion vector, containing a horizontal component u and a vertical component v.

[0036] 6. Skeleton Keypoint Detection: This refers to the techniques used in computer vision and pose estimation to identify and locate key joints (such as head, shoulder, elbow, wrist, knee, ankle, etc.) on the human or animal body.

[0037] 7. Intersection over Union (IoU): This is an evaluation metric used in object detection to measure the degree of overlap between two bounding boxes. The IoU is calculated as the ratio of the intersection area to the union area of ​​the two bounding boxes, with a value range of [0,1], where 0 indicates that the two boxes are completely separated and 1 indicates that the two boxes completely overlap.

[0038] 8. Discretization Iteration: This refers to the process of converting a differential equation in the continuous time domain into a difference equation in the discrete time domain using numerical methods, and then progressively advancing the calculation.

[0039] Example 1 like Figure 1 This embodiment provides a method for ensuring the safety of pets left in a vehicle, including: S1. Responding to the pet mode activation command, obtain vehicle status information and basic pet information, establish a logically isolated controlled function subnet in the vehicle network, authenticate cross-domain communication requests via the policy execution gateway, and establish a controlled communication path between the main control function domain and the controlled function subnet. The above process of establishing a logically isolated controlled subnet and authenticating communication is as follows: Figure 2 As shown.

[0040] This step first describes the system initialization and subnet construction process. After the vehicle is powered on, the logical partitioning module and the policy execution gateway work together to establish the network boundaries and communication policies of the controlled functional subnets. Specifically, the logical partitioning module broadcasts partition configuration instructions to each ECU in the network through in-vehicle network management protocols (such as SOME / IP or DOIP), dividing network resources into different virtual local area networks (VLANs) or using network namespace technology to achieve logical isolation. The main control functional domain typically contains a cluster of ECUs directly related to vehicle driving safety, including but not limited to: powertrain control unit (EMS, Engine Management System), anti-lock braking system (ABS, Anti-lock Braking System), electronic stability program (ESP, Electronic Stability Program), electric power steering control unit (EPS, Electric Power Steering), and airbag control unit (ACU). These ECUs are classified as the main control functional domain and enjoy the highest network priority and non-bypass safety protection. The controlled function subnet contains the functional nodes required for pet mode, including: the air conditioning control unit (HVAC), the door control module (DCM), the power management system (BMS), and the in-vehicle infotainment control unit (IVI). These ECUs are classified as controlled function subnets, and all cross-domain communication requests issued by them must undergo strict authentication by the policy enforcement gateway.

[0041] When the car owner issues a pet mode activation command via the in-vehicle human-machine interface (central control screen) or mobile terminal application (such as a mobile APP), the system first obtains vehicle status information to determine whether the activation conditions are met. Vehicle status information includes, but is not limited to: vehicle power status (determining if the vehicle is in ON or ACC position), battery charge level (determining if sufficient to support prolonged air conditioning operation, typically requiring a charge greater than 30%), engine status (determining if it is running or if the electric vehicle's high-voltage battery is functioning properly), window status (determining if the windows are fully closed), and door lock status (determining if the doors are locked). Simultaneously, the system obtains basic pet information, including breed, weight, and age. This basic pet information can be entered by the user upon first use and saved in the vehicle's user configuration file, or entered in real-time by the user each time pet mode is activated via the human-machine interface. The system performs condition checks based on the obtained vehicle status information and pet basic information: if the vehicle battery is insufficient or the windows are not closed, the system prompts the user via the human-machine interface or APP about the reason for not meeting the activation conditions and suggests that the user resolve the issue before trying again; if all conditions are met, the system continues with the pet mode activation process.

[0042] Once the conditions are met, the logical partitioning module divides the vehicle network into a main control function domain and a controlled function subnet. This division is achieved through the VLAN configuration function of the vehicle Ethernet switch or the routing table configuration function of the CAN gateway. Specifically, in the vehicle Ethernet architecture, the logical partitioning module issues VLAN configuration commands to the Ethernet switch, assigning ECU ports related to the main control function domain to VLAN ID 1 (main domain VLAN) and ECU ports related to the controlled function subnet to VLAN ID 2 (controlled VLAN). The two VLANs are isolated from each other on the Layer 2 network, and any data packets crossing VLANs must be processed by the Layer 3 routing (i.e., the policy enforcement gateway). In the CAN bus architecture, the logical partitioning module configures the routing table of the CAN gateway to only allow CAN IDs that conform to the whitelist rules to bridge from the CAN bus of the controlled function subnet to the CAN bus of the main control function domain; all other CAN IDs are filtered and blocked. This logical isolation technology ensures that even if an ECU in the controlled functional subnet is maliciously attacked or experiences a software failure, the attack traffic cannot directly penetrate into the main control functional domain, thereby guaranteeing the integrity and reliability of the vehicle's core driving safety functions.

[0043] The policy enforcement gateway, acting as the sole communication hub between the master functional domain and the controlled functional subnets, has a built-in Access Control List (ACL) to authenticate all cross-domain communication requests. The policy enforcement gateway can be a standalone hardware module (such as an in-vehicle gateway controller) or integrated into the domain controller. The ACL is stored as a rule table in the gateway's persistent storage (such as EEPROM or Flash). Each rule contains four fields: source address, destination address, instruction type, and allow / deny flag. The source and destination addresses can be the ECU's network layer address (such as an IP address) or data link layer address (such as a CAN ID). The instruction type field describes the type of operation requested, such as "start HVAC cooling," "set target temperature to 22°C," or "request slight window lowering." The allow / deny flag indicates whether to allow forwarding or reject and discard the request if the rule is matched. The rule table configuration is preset by the vehicle manufacturer at the factory and can also be updated via secure OTA remote upgrades. The authentication process of the policy enforcement gateway is as follows: When the gateway receives a cross-domain communication request from the controlled function subnet, it first parses the request and extracts three characteristic parameters: source address, destination address, and instruction type. Then, it iterates through the rules in the access control list in order of priority (usually from most specific to most general), matching the extracted three parameters with the corresponding fields of the rules. If the source address, destination address, and instruction type fields of a rule all match the parameters in the request (or conform to wildcard matching rules), the gateway executes the corresponding forwarding or denial operation according to the rule's allow / deny flag, and records the authentication result (including timestamp, request content, matched rule ID, and execution action) to the audit log. If no completely matching rule is found after iterating through all rules, the request is processed according to the default security policy (usually denial and recording). For example, when the instruction agent unit needs to start HVAC cooling, it sends a request packet with the source address "Instruction Agent Unit IP", the destination address "HVAC Control Unit IP", and the instruction type "StartCooling". Upon receiving the packet, the policy enforcement gateway matches it against preset rules in the access control list. Rule 1 is "Source Address = Command Agent Unit, Target Address = HVAC Unit, Command Type = StartCooling, Action = Allow". If all three match, the gateway allows the request and forwards it to the HVAC control unit. Conversely, if a malicious program attempts to send an illegal request from the infotainment system with a source address of "Infotainment System IP", a target address of "Brake Control Unit IP", and a command type of "DisableBrake", the gateway will reject the request and record the authentication failure event because there is no matching rule in the access control list (and the target address belongs to a safety-critical ECU in the main control functional domain). This may also trigger a vehicle safety alarm.This architecture achieves logical isolation between the pet mode function and the vehicle's core safety domain, fundamentally cutting off the interference path of the non-safety domain to the core control domain, and solving the technical problem of the coupling risk of functional safety and information security pointed out in the background technology.

[0044] Furthermore, the policy enforcement gateway supports writing authentication results and forwarding records to audit logs. Audit logs are stored in the gateway's local storage, and each log entry includes the following fields: log sequence number (auto-incrementing integer), timestamp (accurate to milliseconds), request source address, request destination address, instruction type, matching rule ID, execution action (allow / deny), packet length, and checksum. Audit logs can be periodically uploaded to the cloud-based security operations center via the vehicle's telematics unit (T-BOX) for post-incident security auditing, abnormal behavior analysis, and attack tracing. When the vehicle enters maintenance mode, authorized technicians can also read the audit logs via the OBD interface to assist in troubleshooting network communication anomalies.

[0045] S2. Based on the pet's basic information, the dynamic safe temperature threshold is obtained by querying the breed database and calculating using a correction factor. The process for calculating the dynamic safe temperature threshold is as follows: Figure 3 As shown.

[0046] Specifically, pets of different breeds, weights, and ages exhibit significant differences in their tolerance to high temperatures, making fixed temperature thresholds unsuitable for individual variations. This step retrieves the baseline heat tolerance temperature for the corresponding breed from a breed database and performs personalized calculations using weight and age correction factors to obtain a dynamic safe temperature threshold for the current individual pet. The calculation formula is:

[0047] in, The reference heat tolerance temperature for this pet breed, in °C, is obtained from a table in the breed database. The body type-age comprehensive correction factor is 0.25 in °C·kg, which reflects the effect of body weight on heat tolerance (the smaller the body weight, the worse the heat dissipation, and the lower the threshold). This refers to the pet's weight, in kg. This refers to the pet's actual age, in years. This refers to the reference adult age of this variety, in years. This dynamic threshold reflects the impact of a pet's physiological maturity on its heat tolerance. It provides a precise, individualized reference for subsequent overheating predictions, avoiding misjudgments or missed detections caused by fixed thresholds.

[0048] The core of this step lies in achieving personalized temperature threshold adaptation. Pets of different breeds, weights, and ages exhibit significant differences in their tolerance to high temperatures. For example, short-nosed breeds (such as French Bulldogs and Pugs) have poor heat dissipation due to their unique respiratory structure, making them far less tolerant of high temperatures than long-nosed breeds. Young and senior pets have relatively weak thermoregulation abilities, so their safe temperature threshold should be appropriately lowered. Larger pets, due to their smaller surface area to volume ratio, dissipate heat relatively slowly and require a more conservative temperature threshold. A fixed temperature threshold (e.g., simply set at 30℃) cannot accommodate these individual differences, potentially leading to insufficient protection for some pets or excessive cooling and wasted energy when conditions permit. This step retrieves the benchmark heat tolerance temperature for the corresponding breed from a breed database and performs personalized calculations using weight and age correction factors to obtain a dynamic safe temperature threshold for the current individual pet.

[0049] S3. Obtain external environmental parameters and the current cabin temperature. Iteratively calculate the cabin temperature prediction sequence using the cabin thermodynamic prediction model. When the prediction sequence exceeds the dynamic safety temperature threshold, send an advance cooling command to the HVAC system via the controlled communication channel. The above thermodynamic model prediction and advance cooling control process is as follows: Figure 4 As shown.

[0050] The instruction agent unit has a built-in thermodynamic prediction model for the train car. Based on Newton's law of cooling and the principle of thermal balance in the train car, the following dynamic prediction equation for the train car temperature is established:

[0051] The parameters are defined as follows: for Temperature inside the carriage at any given time, in °C; for Solar radiation heat flux at any given time, in W, is calculated by the following formula:

[0052] in, is the solar transmittance coefficient of the car window glass, with a value range of [0, 1]. The total area of ​​the vehicle's glass is expressed in square meters (m²). for Horizontal solar irradiance at any time, in W / m², obtained from online meteorological services or vehicle-mounted light sensors; The solar incidence angle is measured in rad. The current latitude is obtained from the vehicle's GPS coordinates. The solar declination and hour angle are calculated based on the current timestamp. Then, based on the vehicle's orientation (heading angle), the solar altitude angle and azimuth angle are calculated using spherical trigonometry formulas. Finally, the angle between the sunlight and the vehicle's glass normal is obtained as the solar incidence angle.

[0053] for The total convective heat transfer power of the outer wall of the carriage at any given time, in W, is calculated by the following formula:

[0054] in, The comprehensive convective heat transfer coefficient of the vehicle body exterior wall is expressed in W / (m²·℃), taking into account the effects of natural convection and forced convection (wind speed). The total area of ​​the vehicle's outer surface, in m²; for The ambient temperature outside the vehicle at any given time, in °C. for The cooling capacity of the HVAC system at any given time is expressed in watts (W). The rated cooling capacity is used when the system is operating normally. The value is 0 when the system is not started or is malfunctioning. The equivalent thermal mass of the carriage is expressed in kg. The equivalent specific heat capacity of the carriage is expressed in J / (kg·℃).

[0055] The instruction proxy unit adopts a step size The forward Euler method (1 minute) is used to discretize and iterate the above differential equation, generating the future... ( Temperature prediction sequence of step 1:

[0056] When the predicted sequence contains At that time, the system in advance (Unit: seconds) Triggering a cooling command enables proactive, predictive protection. The advance cooling command includes the target cooling temperature and the expected cooling start advance. The expected cooling start advance is determined based on the estimated time required for the HVAC system to drop from the current temperature to the dynamic safe temperature threshold. This ensures that the HVAC system completes temperature regulation before the predicted over-temperature moment arrives, eliminating the temperature protection blind spot caused by HVAC response delay and realizing a shift from passive response to proactive prevention.

[0057] If the environmental perception module detects that the main HVAC system has failed, causing the temperature to continue to deteriorate, the instruction agent unit will activate the redundancy strategy, generate a request for slightly lowering the windows for ventilation, and execute it after being reviewed by the strategy execution gateway, thus forming a combined primary and backup environmental protection system.

[0058] The environmental perception module continuously collects the external ambient temperature (obtained through external sensors or connected meteorological data), solar irradiance, vehicle parking location (determined by GPS combined with timestamps and shading models to determine direct sunlight / shade status), and current cabin temperature, and sends the data to the command agent unit.

[0059] Redundancy safety mechanism: If the environmental perception module detects that the main HVAC system has failed and the temperature continues to deteriorate, the instruction agent unit will activate the redundancy policy, generate a "slightly lower the window for ventilation" request, and execute it after being reviewed by the policy execution gateway, forming a main and backup combined environmental protection system.

[0060] S4. Acquire the video stream from the in-vehicle camera, calculate motion energy features using optical flow, calculate posture disorder features using skeletal keypoint detection, and calculate the intersection-over-union (IoU) ratio of the driving area using target detection to obtain a 3D behavior feature vector. The extraction process of the above 3D behavior feature vector is as follows: Figure 5 As shown.

[0061] Three-dimensional behavioral feature vectors include motion energy features Postural disorder characteristics and regional intrusion characteristics The specific calculation method is as follows: (1) Characteristics of kinetic energy Based on the optical flow method, the inter-frame pixel displacement field is calculated, and the horizontal optical flow component of each pixel in the pet target area is extracted. and vertical optical flow components Calculate the normalized mean motion amplitude:

[0062] in, This represents the total number of pixels within the pet target detection bounding box. For the first The optical flow component in the horizontal direction of each pixel, in units of pixel / frame; For the first The optical flow component in the vertical direction of each pixel, measured in pixels per frame. Perform maximum value normalization to obtain the features. :

[0063] in, This represents the upper limit of motion energy calibrated by the system, measured in pixels per frame. This feature reflects the average level of motion activity within the pet's area, has low computational overhead, and is suitable for real-time processing.

[0064] (2) Characteristics of postural disorder Based on the skeletal keypoint detection model, the coordinate displacement of each keypoint between adjacent frames is obtained, and the confidence-weighted mean is calculated.

[0065] in, This represents the total number of skeletal key points. For the first The inter-frame displacement of key points in the horizontal direction, in pixels; For the first The inter-frame displacement of key points in the vertical direction, in pixels; The confidence weights for keypoint detection, output by the pose estimation model, are used. The displacement changes of high-confidence keypoints contribute more to the score, effectively reducing the impact of keypoint detection errors on feature quality. Perform maximum value normalization to obtain the features. :

[0066] in, This represents the upper limit of the attitude disorder calibrated by the system, in pixels.

[0067] (3) Regional invasion characteristics Define the driving area interest box (A pre-defined rectangular bounding box containing the steering wheel, dashboard, and driver's seat areas) Calculate the pet target detection box. and Intersection over Union (IoU):

[0068] in, ;when ( When the preset intrusion detection threshold (default value 0.3) is reached, a special alarm for intrusion into the driving area is triggered, a high-priority notification is immediately pushed to the car owner's mobile application, and a warning message is displayed on the external display screen, which solves the safety hazard that pets may interfere with driving operations if they accidentally enter the driving area.

[0069] The above three types of features characterize the pet's behavioral state from three perspectives: exercise intensity, abnormal posture, and locational danger, respectively, forming a three-dimensional behavioral feature vector. ].

[0070] S5. After normalizing the three-dimensional behavioral feature vectors, a weighted fusion is performed to obtain a comprehensive behavioral anomaly score. A behavioral anomaly level is then determined through grading, and a corresponding multimodal reassurance command is triggered based on the behavioral anomaly level. The process of the behavioral anomaly score grading and multimodal reassurance triggering described above is as follows: Figure 6 As shown.

[0071] This step fuses the three-dimensional feature vectors obtained in step S4 to obtain a comprehensive score that can fully assess the pet's stress level, and then implements a graded soothing strategy based on the score. The weighted fusion method avoids the limitations of single features: for example, relying solely on kinetic energy features cannot distinguish whether a pet is playing happily or struggling in pain; relying solely on postural instability features may miss abnormalities when the pet is resting; and relying solely on area intrusion features can only trigger a response when the pet has entered a danger zone, failing to provide early intervention. By weightedly fusing these three features, the system can comprehensively judge the pet's overall state, achieving a more scientific and humane soothing approach.

[0072] The comprehensive behavioral abnormality score is derived by weighted fusion of three dimensionless features:

[0073] in, , , The fusion weights for each feature satisfy... , ; The weights can be adaptively adjusted based on the characteristics of the pet breed; for example, the weights can be appropriately reduced for sporty dogs. ,improve To more accurately distinguish between normal activity and stress.

[0074] System basis Values ​​and configurable grading thresholds , , (satisfy Implement a graded response:

[0075] in, , , For configurable classification thresholds, satisfying .

[0076] Specifically: When the pet experiences mild anxiety, the interactive alarm module plays pre-recorded audio messages from the owner through the vehicle's speakers to provide audio reassurance. The recordings can include the owner calling the pet's name, comforting words (such as "Don't be afraid, baby, I'll be right back"), or familiar commands (such as "Sit, sweetie"). The pre-recorded audio files are stored in the user configuration folder in the vehicle's memory, and users can update the recordings at any time via a mobile app or the vehicle's interface. The advantage of audio reassurance lies in the pet's natural affinity and soothing effect on their owner's voice, effectively alleviating mild anxiety.

[0077] During periods of moderate stress, the system adjusts the brightness and color temperature of the interior lights for visual soothing. Research shows that lighting conditions significantly impact an animal's emotional state: overly bright or glaring light can exacerbate a pet's anxiety, while soft, low-brightness light and a warm color temperature (e.g., 2700K–3000K) help create a safe and comfortable atmosphere. The interactive alarm module sends a light adjustment command to the vehicle control unit, dimming the interior reading lights and ambient lights to a set brightness (e.g., 30% of maximum brightness) and switching the color temperature to a warm tone (if the ambient lights support RGB color adjustment). If the vehicle is equipped with a panoramic sunroof or sunshade, the system can also automatically close the sunshade to further reduce interior brightness. Visual soothing, as a non-contact method of comforting, requires no additional hardware and is suitable for use when pets experience moderate stress.

[0078] In cases of severe stress, the system immediately triggers a remote notification. Specifically, the command agent unit pushes an alarm message to the cloud service platform via the vehicle's mobile communication module (T-BOX, Telematics BOX). The cloud service platform then sends the notification to the owner's linked mobile terminal APP via application push services (such as APNs, Firebase Cloud Messaging). The notification includes: alarm type ("severe pet stress"), vehicle location (GPS coordinates and map link), current cabin temperature, pet's comprehensive abnormality score, and an embedded playback link containing real-time video footage. After clicking the notification, the owner can view the real-time video stream from the in-vehicle camera on the mobile APP and use the APP's two-way voice intercom function to speak to the pet inside the vehicle to comfort it, or decide whether to return to the vehicle immediately. The remote notification ensures that the owner can be informed immediately and take intervention measures when the pet exhibits severe stress, maximizing the pet's safety and health.

[0079] The system simultaneously The system dynamically assesses the soothing effect based on the changing trend of the pet's behavior. When soothing is ineffective (i.e., the score does not decrease or continues to rise), it automatically upgrades the soothing strategy level, achieving adaptive closed-loop soothing control. For example, if the pet is currently at a mild anxiety level and the system has played audio soothing but the score has not improved after 2 minutes, the system will proactively upgrade to light soothing at a moderate stress level, while continuing to use audio soothing (overlay soothing). This adaptive closed-loop soothing control mechanism ensures that the system can dynamically adjust intervention measures based on the pet's real-time feedback, avoiding a continuous deterioration of the pet's condition due to the failure of fixed strategies.

[0080] Meanwhile, the interactive alarm module dynamically displays the following information on the external display screen (usually a small LED screen inside the windshield or displayed on the back of the central control screen facing outwards): real-time cabin temperature (e.g., "Interior temperature: 24℃"), current timestamp (e.g., "14:35"), and reassuring message (e.g., "Pet is safe, air conditioning is on, please do not disturb"). This information effectively conveys a safety signal to people outside the vehicle (such as kind passersby or parking lot attendants) that "the pet inside the vehicle is protected," eliminating misunderstandings and unnecessary emergency interventions (such as breaking windows to rescue the pet) caused by public lack of awareness. At the same time, the pet's status images and environmental data (including temperature, humidity, HVAC operating status, etc.) captured by the environmental perception module are encrypted and sent to the owner's mobile application via the remote communication unit, giving the owner full access to information and control. The owner can check the pet's real-time status, adjust the air conditioning temperature, interact with the pet via voice, and even remotely terminate the pet mode when necessary through the app.

[0081] The conditions for exiting pet mode include the following three situations: (1) User remote command: The car owner sends the command "Exit pet mode" through the mobile terminal APP. After receiving the command, the system executes the exit process - stops the HVAC air conditioner, restores the window status (closes the slightly lowered ventilation gap), turns off the external display screen prompts, stops the video stream push, and restores the vehicle to normal standby status.

[0082] (2) Vehicle unlocking action: When the owner unlocks the vehicle using a smart key, NFC card or mobile phone Bluetooth key, the system detects that the door is opened, automatically exits pet mode, and sends a reminder to the owner's APP that "Pet mode has been exited, please check pet status".

[0083] (3) Power supply voltage is below the threshold: The system continuously monitors the power of the vehicle's battery (fuel vehicle) or high-voltage power battery (electric vehicle). When the power is below the safety threshold (such as the battery voltage is below 11.5V or the high-voltage battery SOC is below 10%), in order to ensure that the vehicle can start normally, the system automatically exits the pet mode and pushes an alarm notification to the owner's APP: "Insufficient power, pet mode has been automatically exited, please return as soon as possible".

[0084] Example 2 like Figure 7 and Figure 8 As shown, this embodiment provides a vehicle-mounted pet safety protection system, including: The logical partitioning module is used to divide the vehicle network into a main control functional domain (100) and a controlled functional subnet (200). The main control functional domain includes a cluster of electronic control units directly related to vehicle driving safety, such as the engine control ECU (101), chassis control ECU (102), and driving safety ECU (103). This module achieves logical isolation between the main control functional domain and the controlled functional subnet through network segmentation or virtual local area network technology, ensuring that the functional nodes required by the pet mode operate in an independent network area and do not share network resources with the vehicle's core safety domain.

[0085] The policy enforcement gateway (600) is electrically connected to the main control functional domain (100) and the controlled functional subnet (200) via an in-vehicle Ethernet or CAN bus, respectively. The memory of the policy enforcement gateway (600) stores an access control list (601) for parsing, authenticating, and forwarding cross-domain communication commands. Each rule in the access control list (601) includes a source address field, a destination address field, a command type field, and an allow / deny flag. The policy enforcement gateway (600) performs rule-by-rule matching on each cross-domain communication request and writes the authentication results and forwarding records to the audit log (700), providing complete data records for post-event security auditing and anomaly tracing.

[0086] An environmental perception module (300), deployed within the controlled functional subnet (200), integrates a temperature sensor (301), a humidity sensor (302), a millimeter-wave radar vital signs monitor, an external environmental sensor (303), a GPS positioning unit (304), and an in-vehicle optical camera (305). The temperature sensor (301) and humidity sensor (302) are used to collect environmental parameters inside the vehicle compartment; the external environmental sensor (303) (or via connected meteorological services) is used to obtain the external temperature and solar irradiance; the GPS positioning unit (304) is used to obtain the vehicle's parking location and, combined with timestamps and shading models, determine whether the vehicle is in direct sunlight or shade; the in-vehicle optical camera (305) is used to collect in-vehicle video streams and monitor pet behavior in real time. The environmental perception module (300) communicates with the command agent unit via a data connection, continuously providing multi-dimensional environmental perception data to the command agent unit.

[0087] The instruction agent unit (400), located within the controlled function subnet (200), incorporates a carriage thermodynamic prediction model (401) and a pet behavior detection module (402). The carriage thermodynamic prediction model (401) is based on the heat balance equation. Integrating solar radiation heat flux convective heat transfer power of the outer wall of the carriage and HVAC system cooling power Using step size The forward Euler method is used for discretization and iterative calculation to generate a predicted temperature sequence for the next 30 minutes, providing a quantitative basis in the time dimension for advance cooling decisions. The pet behavior detection module (402) is based on optical flow kinetic energy. Postural disorder of skeletal key points Intersection with driving area Three-dimensional features are used to calculate a comprehensive anomaly score. The current behavior level of the pet is determined by comparing the prediction results with the classification threshold. The instruction agent unit (400) generates a corresponding control request based on the prediction results and behavior detection results, and sends it to the corresponding execution unit after authentication by the policy execution gateway (600).

[0088] The interactive alarm module (500), located within the controlled function subnet (200), integrates three soothing methods: audio soothing, light soothing, and remote notification. This module controls the in-vehicle display screen (501) to display real-time temperature, timestamps, and soothing messages to the outside of the vehicle; controls the speaker (502) to play pre-recorded audio recordings from the owner for audio soothing; supports adjusting the brightness / color temperature of the in-vehicle lights for visual soothing; and pushes an alarm notification containing real-time video footage to the owner's mobile terminal via the mobile communication module (503) during severe stress. This module can automatically select the corresponding soothing strategy based on the abnormal behavior level output by the instruction agent unit (400), achieving precise matching of soothing intensity with the pet's stress level. The vehicle is also equipped with an in-vehicle display screen (501), a speaker (502), and a mobile communication module (503), which respectively support the functions of safety prompt display, audio soothing playback, and severe stress alarm push, providing a hardware foundation for the implementation of multimodal soothing strategies.

[0089] Example 3 Embodiment 3 of the present invention provides a vehicle including the vehicle-mounted pet-keeping safety protection system described in Embodiment 2.

[0090] Specifically, the vehicle also includes an in-vehicle display screen, a speaker, and a mobile communication module. The in-vehicle display screen is used to display pet safety tips to the outside of the vehicle, the speaker is used to play soothing audio, and the mobile communication module is used to push an alarm notification containing real-time video footage to the owner's mobile terminal when the pet's abnormal behavior reaches a level of severe stress.

[0091] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A method for ensuring the safety of pets left in a vehicle, characterized in that, include: In response to the pet mode activation command, the system obtains vehicle status information and basic pet information, establishes a logically isolated controlled function subnet in the vehicle network, authenticates cross-domain communication requests through the policy execution gateway, and establishes a controlled communication path between the main control function domain and the controlled function subnet. Based on the pet's basic information, the dynamic safe temperature threshold is obtained by querying the breed database and calculating with a correction factor. The system acquires external environmental parameters and current cabin temperature, and iteratively calculates cabin temperature prediction sequence using a cabin thermodynamic prediction model. When the prediction sequence exceeds the dynamic safety temperature threshold, it sends an advance cooling command to the HVAC system through the controlled communication channel. The video stream from the in-vehicle camera is acquired, and the motion energy features are calculated using optical flow, the posture disorder features are calculated using skeletal keypoint detection, and the cross-union ratio features of the driving area are calculated using target detection to obtain a three-dimensional behavior feature vector. The three-dimensional behavioral feature vectors are normalized and then weighted and fused to obtain a comprehensive behavioral anomaly score. The behavioral anomaly level is determined by classification, and the corresponding multimodal reassurance command is triggered based on the behavioral anomaly level.

2. The method according to claim 1, characterized in that, The policy enforcement gateway receives cross-domain communication requests from the controlled functional subnet, extracts the source address, destination address, and instruction type of the request, and matches them one by one with the preset rules in the built-in access control list. When the source address, destination address, and instruction type all match the preset rules, forwarding is allowed and the controlled communication path is established. When any parameter does not match, forwarding is rejected and an authentication failure event is recorded.

3. The method according to claim 1, characterized in that, The external environmental parameters include solar radiation heat flux, which is calculated based on the solar transmittance of the vehicle window glass, the vehicle glass area, solar irradiance, and solar incident angle. The solar incident angle is determined based on the vehicle's GPS coordinates, current timestamp, and vehicle orientation, using the solar declination angle, hour angle, and vehicle latitude, and then calculated using spherical trigonometric formulas to determine the solar altitude angle and azimuth angle.

4. The method according to claim 1, characterized in that, The calculation of the motion energy feature includes: calculating the inter-frame pixel displacement field based on optical flow or other inter-frame motion estimation methods; identifying and obtaining the pet target region in real time through a target detection model; extracting the horizontal and vertical optical flow components of each pixel within the pet target region; calculating the mean value of the optical flow amplitude of each pixel; and normalizing the result by a preset upper limit value of motion energy to obtain the motion energy feature.

5. The method according to claim 1, characterized in that, The calculation of the posture disorder feature includes: obtaining the horizontal and vertical displacements of each key point between adjacent frames based on the skeletal key point detection model, calculating the displacement amplitude of each key point and performing a weighted average based on the detection confidence of each key point, and obtaining the posture disorder feature by normalization after passing through a preset upper limit value of posture disorder.

6. The method according to claim 1, characterized in that, The calculation of the intersection-union ratio feature of the driving area includes: obtaining a pet target detection box and a preset driving area interest box, wherein the driving area interest box is a pre-calibrated rectangular box containing the steering wheel, dashboard and driver's seat area; calculating the ratio of the intersection area to the union area of ​​the pet target detection box and the driving area interest box to obtain the driving area intersection-union ratio feature; when the driving area intersection-union ratio feature is greater than or equal to a preset intrusion judgment threshold, a special alarm for driving area intrusion is triggered.

7. The method according to claim 1, characterized in that, The multimodal reassurance instructions include: When the level of abnormal behavior is mild anxiety, an audio reassurance is triggered; When the level of abnormal behavior is moderate stress, light therapy is triggered. When the abnormal behavior level is severe stress, a remote notification is triggered; The effectiveness of reassurance is judged based on the changing trend of the comprehensive abnormal behavior score, and the reassurance strategy is upgraded when the reassurance is ineffective.

8. The method according to claim 1, characterized in that, The formula for calculating the dynamic safety temperature threshold is: ,in, To obtain the baseline heat resistance temperature for this variety from the variety database, This is a body shape correction factor determined based on weight ranges. This is a correction factor for physiological maturity determined based on age group.

9. The method according to claim 1, characterized in that, The advance cooling command includes a target cooling temperature and an expected cooling start advance. The expected cooling start advance is determined based on the estimated time required for the HVAC system to drop from the current temperature to the dynamic safe temperature threshold, so that the HVAC system can complete temperature regulation before the predicted over-temperature moment arrives.

10. A vehicle-mounted pet safety protection system, characterized in that, include: The logical partitioning module is used to divide the vehicle network into a main control function domain and a controlled function subnet. The main control function domain includes a cluster of electronic control units that are directly related to vehicle driving safety. The policy execution gateway is electrically connected to the main control functional domain and the controlled functional subnet via vehicle Ethernet or CAN bus, respectively. The memory of the policy execution gateway stores an access control list for parsing, authenticating and forwarding cross-domain communication commands. An environmental perception module, deployed within the controlled function subnet, includes a temperature sensor, a humidity sensor, an external environment sensor, a GPS positioning unit, and an in-vehicle camera. The sensors are electrically connected to the environmental perception module to collect data on external ambient temperature, solar irradiance, vehicle parking position, current cabin temperature, and in-vehicle video stream. The environmental perception module communicates with the instruction agent unit via a data connection. The instruction proxy unit is located within the controlled function subnet. The processor of the instruction proxy unit executes the carriage thermodynamic prediction model algorithm and the pet behavior detection module algorithm stored in the memory. It is used to generate a carriage temperature prediction sequence and a comprehensive behavior anomaly score based on the data collected by the environmental perception module, and generate a control request based on a preset threshold. The instruction proxy unit is electrically connected to the interactive alarm module through a signal connection. The interactive alarm module, located within the controlled function subnet, is used to control the in-vehicle display screen to show safety prompts to the outside of the vehicle, control the speaker to play reassuring audio, and trigger a multimodal adaptive reassurance strategy based on the level of abnormal behavior.

11. The system according to claim 10, characterized in that, Each rule in the access control list includes a source address field, a destination address field, an instruction type field, and an allow or deny flag. The policy enforcement gateway performs rule-by-rule matching on each cross-domain communication request and writes the authentication results and forwarding records into the audit log.

12. The system according to claim 10, characterized in that, The thermodynamic prediction model of the carriage in the instruction agent unit is based on the heat balance equation, integrates the solar radiation heat flux, the convective heat transfer power of the carriage outer wall and the cooling power of the HVAC system, and uses a numerical iteration method to perform iterative calculations to generate a predicted sequence of carriage temperature within a preset time period.

13. The system according to claim 10, characterized in that, The interactive alarm module supports multimodal adaptive soothing strategies including: triggering audio soothing for mild anxiety, triggering light soothing for moderate stress, and triggering remote notification for severe stress.

14. A vehicle, characterized in that, Including the vehicle-mounted pet safety protection system as described in any one of claims 10 to 13.

15. The vehicle according to claim 14, characterized in that, The vehicle also includes an in-vehicle display screen, a speaker, and a mobile communication module. The in-vehicle display screen is used to display pet safety tips to the outside of the vehicle, the speaker is used to play soothing audio, and the mobile communication module is used to push an alarm notification containing real-time video footage to the owner's mobile terminal when the pet's abnormal behavior reaches the level of severe stress.

Images