Vehicle skylight automatic control method and system based on multi-information fusion

Abstract

The invention discloses a vehicle skylight automatic control method and system based on multi-information fusion. Firstly, a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor and an industrial camera is used for collecting internal and external environment data of a vehicle in real time and obtaining carbon dioxide concentration information, temperature information, vehicle body external wading depth information and living body characteristic information; a first sensing signal, a second sensing signal, a third sensing signal and a fourth sensing signal are respectively obtained; the function of automatically opening and closing the skylight of the vehicle by adopting a dynamic threshold triggering mechanism and a multi-stage response strategy is achieved, millisecond emergency response is achieved under the conditions that the vehicle falls into water, children are forgotten to be in the vehicle at high temperature in summer and other emergency working conditions, the passive safety performance of the vehicle is improved, and the service life of the vehicle is prolonged. And the system also has the characteristics of self-adaption to environment change, low false alarm rate and quick linkage execution, ensures the life safety of passengers, and is suitable for wide popularization and use.

Description

Technical Field

[0001] This invention relates to the field of vehicle sunroof control technology, specifically to an automatic control method and system for vehicle sunroofs based on multi-information fusion. Background Technology

[0002] With the rapid increase in the number of cars, car safety and comfort are becoming increasingly intelligent, and people have higher and higher demands for cars' ability to respond to sudden emergencies.

[0003] Currently, existing vehicles commonly suffer from sunroofs that cannot adaptively control based on the size and severity of emergency events. This can lead to situations where people cannot be rescued in time, and vehicles are also prone to spontaneous combustion or explosion when parked. This not only damages the vehicle but also jeopardizes the safety of passengers. Therefore, there is a need to design an automatic sunroof control method and system based on multi-information fusion. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and to better and more effectively solve the problem that existing vehicles often cannot adaptively control sunroofs according to the size and severity of emergency events, leading to situations where people cannot be rescued in time, and vehicles are prone to spontaneous combustion or explosion when parked. This not only damages the vehicle but also jeopardizes passenger safety. The invention provides a multi-information fusion-based automatic sunroof control method and system. This system achieves automatic opening and closing of the sunroof using a dynamic threshold triggering mechanism and a multi-level response strategy. It also provides millisecond-level emergency response in situations such as a car falling into water, children left inside during hot summer weather, and other emergency conditions. This not only improves the vehicle's passive safety performance but also features adaptive environmental changes, low false alarm rate, and rapid linkage execution, preventing damage to the vehicle and ensuring passenger safety.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] An automatic control method for vehicle sunroof based on multi-information fusion includes the following steps:

[0007] Step A: A sensing layer consisting of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera is used to collect real-time environmental data inside and outside the vehicle and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and living feature information, thereby obtaining the first sensing signal, the second sensing signal, the third sensing signal, and the fourth sensing signal, respectively.

[0008] Step B: Based on carbon dioxide concentration information, temperature information, and wading depth information outside the vehicle body, set preset thresholds for carbon dioxide gas concentration, first temperature, second temperature, and liquid level, respectively.

[0009] Step C: Compare the first sensing signal with the preset threshold for carbon dioxide gas concentration. If the first sensing signal is less than the preset threshold for carbon dioxide gas concentration, the output sunroof control command is a safety control command. If the first sensing signal is not less than the preset threshold for carbon dioxide gas concentration, the output sunroof control command is a hazard control command.

[0010] Step D: Compare the second sensing signal with the first temperature preset threshold. If the second sensing signal is less than the first temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the first temperature preset threshold, the output sunroof control command is a danger control command.

[0011] Step E: Compare the third sensing signal with the preset liquid level threshold. If the third sensing signal is less than the preset liquid level threshold, the output sunroof control command is a safety control command. If the third sensing signal is not less than the preset liquid level threshold, the output sunroof control command is a danger control command.

[0012] Step F: Compare the second sensing signal with the second preset temperature threshold. If the second sensing signal is less than the second preset temperature threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the second preset temperature threshold, detect whether there are living features in the fourth sensing signal. If there are living features, the output sunroof control command is a danger control command. If there are no living features, the output sunroof control command is a safety control command.

[0013] Step G: Operate the sunroof according to the safety control command and the danger control command respectively, and complete the automatic control operation of the vehicle sunroof. If the sunroof control command is a safety control command, the sunroof will remain in its original state. If the sunroof control command is a danger control command, the sunroof will be opened automatically and an alarm will be triggered automatically.

[0014] The aforementioned automatic control method for vehicle sunroof based on multi-information fusion, in step A, uses a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor and an industrial camera to collect real-time environmental data inside and outside the vehicle and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body and living feature information, thereby obtaining a first sensing signal, a second sensing signal, a third sensing signal and a fourth sensing signal, respectively. The carbon dioxide sensor is installed at the bottom of the driver's seat of the vehicle body and serves as the first sensing element. The carbon dioxide concentration information collected by the first sensing element is the first sensing signal.

[0015] The temperature sensor is encapsulated in the metal interlayer on the top of the vehicle body and serves as a second sensing element. The temperature information collected by the second sensing element is the second sensing signal.

[0016] The liquid level sensor is encapsulated at the lowest point of the chassis at the bottom of the vehicle body and serves as the third sensing element. The wading depth information outside the vehicle body collected by the third sensing element is the third sensing signal.

[0017] The industrial camera is installed in the center of the roof inside the vehicle body and serves as the fourth sensing element. The liveness feature information collected by the fourth sensing element is the fourth sensing signal.

[0018] In the aforementioned automatic control method for vehicle sunroof based on multi-information fusion, step B involves setting a preset threshold for carbon dioxide gas concentration, a first preset threshold for temperature, a second preset threshold for temperature, and a preset threshold for liquid level based on carbon dioxide concentration information, temperature information, and wading depth information outside the vehicle body, respectively. The preset threshold for carbon dioxide gas concentration is the safe upper limit of carbon dioxide concentration that the human body can tolerate, the first preset threshold for temperature is the critical value for heat radiation damage to human skin, and the second preset threshold for temperature is the critical value for vehicle combustion risk.

[0019] The aforementioned automatic sunroof control method based on multi-information fusion employs a two-level fault-tolerant mechanism in step C during the carbon dioxide gas concentration comparison process. The specific steps are as follows.

[0020] Step C1, the first level of the fault tolerance mechanism, specifically uses the moving average filtering method to calculate the arithmetic mean of data from multiple consecutive sampling periods, as shown in formula (1).

[0021]

[0022] in, C represents the filtered concentration value at the k-th sampling point. k-n The data is a continuous sampling sequence of carbon dioxide gas concentration, where N is the number of continuous sampling sequences of carbon dioxide gas concentration, and n is the loop variable.

[0023] Step C2, the second level of the fault tolerance mechanism, specifically involves the carbon dioxide sensor monitoring the changing trend of the first sensing signal in real time after the output of the hazard control command. If the carbon dioxide gas concentration drops to below the preset threshold and remains there for a longer period than the set time, the sunroof will be closed to the initial position at a preset speed and the audible and visual alarms will be deactivated. At the same time, the sunroof control command will be output as a safety control command.

[0024] In the aforementioned automatic sunroof control method based on multi-information fusion, step D involves using a sliding mean filter and calculating a moving average value for continuous H minutes of data during the comparison of the first preset temperature threshold. When the sliding mean value exceeds the first preset temperature threshold multiple times consecutively, a hazard control command is triggered, as shown in formula (2).

[0025]

[0026] Among them, T valid The temperature value is the result of moving average filtering, where H is the set number of minutes and T is the temperature value. f Let h be the original temperature value of the f-th sampling point, and h be the sampling point number at the current time, where h ≥ H.

[0027] In the aforementioned automatic control method for vehicle sunroofs based on multi-information fusion, the YOLOv11 deep learning model is used for liveness detection in step F. This YOLOv11 deep learning model comprises a backbone network, a Neck layer network, and a Head network. The backbone network includes an input network module, a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, and a fifth convolutional module. The Neck layer network includes an N1 convolutional module and an N2 convolutional module. The N1 convolutional module is connected from top to bottom by a GP module, an Upsample module, an independent C3k2 module, another GP module, and an Upsample module. The N2 convolutional module is connected from top to bottom by a C3k2 module, a Conv module, a DS module, a C3k2 module, a Conv module, a DS module, and a C3k2 module. The Head network includes four segmentation heads, with the inputs of three of these segmentation heads connected to three C3k2 modules within the Neck layer's N2 convolutional module.

[0028] The aforementioned automatic sunroof control method based on multi-information fusion, wherein the input of the first-layer GP module of the N1 convolutional module includes the C3k2 module in the third convolutional module of the backbone network and the first-layer Upsample module of the N1 convolutional module in the Neck network, and the input of the second-layer GP module of the N1 convolutional module includes the C3k2 module in the second convolutional module of the backbone network and the second-layer Upsample module of the N1 convolutional module in the Neck network, and the specific construction steps of the GP module are as follows.

[0029] S1. Define the bipartite graph of the GP module as B = (G, A), and use the output of the C3k2 module in the second or third convolutional module of the backbone network as F. z Then, the output of the first or second upsampled layer of the N1 convolutional module is used as F. x Then F z and Fx Each pixel is treated as a node in a graph architecture, resulting in an image feature node set V. z and the feature node set V of the search region x Then the definitions of G and A are as shown in formula (3).

[0030]

[0031] Where G is the image feature node set V z and the feature node set V of the search region x The union of the image feature nodes, where A is the set of image feature nodes V. z and the feature node set V of the search region x All connections between them, where g is any pixel in set G and a is any pixel in set A;

[0032] S2, for each i∈V z Calculate i for all j∈V x The similarity coefficient is shown in formula (4).

[0033]

[0034] Among them, e ij It is the attention that node i assigns to node j. Let i be the feature vector of node i. Let W be the feature vector of node j. z and W x All are linear transformation matrices, where m is the summation index variable and T is the transpose;

[0035] S3. Weight the features based on the attention coefficients and sum them, then use the attention coefficients as soft weights and the feature node set V of the search region. x The fusion is shown in formula (5).

[0036]

[0037] in, The features are aggregated, σ is the activation function, and W is the feature. v It is a linear transformation matrix.

[0038] The aforementioned automatic control method for vehicle sunroof based on multi-information fusion, wherein the input of the first layer DS module of the N2 convolutional module includes the first layer Conv module of the N2 convolutional module in the Neck layer network and the C3k2 module of the N1 convolutional module in the Neck layer network, and the input of the second layer DS module of the N2 convolutional module includes the second layer Conv module of the N2 convolutional module in the Neck layer network and the C2PSA module of the fifth convolutional module in the backbone network;

[0039] Specifically, the DS module concatenates the two inputs on the channel using the Concat module, then reduces the width and height using the Pooling operation. Next, it uses a bottleneck structure to increase feature generalization and uses the Upsample module in the N1 convolution module to restore the width and height of the image features. Finally, it uses a Non-local self-attention mechanism to process the features and outputs the result to the C3k2 module in the N2 convolution module.

[0040] The aforementioned automatic sunroof control method based on multi-information fusion includes the following steps in step G: automatically opening the sunroof and triggering an automatic alarm.

[0041] Step G1: Cut off the PID control loop and remove the speed and position limits. Then apply the maximum rated voltage directly to the drive motor and make the motor enter the stall state to output peak torque. At the same time, bypass the soft start protection of the mechanical transmission mechanism and force the sunroof guide rail slider to slide to the terminal mechanical limit point in the shortest time.

[0042] Step G2 activates the alarm, which is simultaneously activated by the buzzer, LED group, and vehicle screen.

[0043] Step G3 activates the alarm information protocol and uses the vehicle-mounted satellite positioning unit to obtain real-time geographic coordinates and automatically alarm. At the same time, it transmits the data to the cloud platform within a set time. After receiving the data, the cloud platform verifies the legality of the digital certificate, parses the risk level label, and matches it with the vehicle owner's preset contact list. If the verification is successful, it sends an SMS alarm to all preset contacts.

[0044] An automatic control system for a vehicle sunroof based on multi-information fusion includes a data acquisition unit, a first signal judgment unit, a second signal judgment unit, a third signal judgment unit, a fourth signal judgment unit, and a sunroof control unit. The data acquisition unit is used to collect real-time environmental data inside and outside the vehicle using a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera, and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and living feature information, thereby obtaining a first sensing signal, a second sensing signal, a third sensing signal, and a fourth sensing signal, respectively.

[0045] The first signal judgment unit is used to compare the first sensing signal with the preset threshold of carbon dioxide gas concentration. If the first sensing signal is less than the preset threshold of carbon dioxide gas concentration, the output sunroof control command is a safety control command. If the first sensing signal is not less than the preset threshold of carbon dioxide gas concentration, the output sunroof control command is a danger control command.

[0046] The second signal judgment unit is used to compare the second sensing signal with the first temperature preset threshold. If the second sensing signal is less than the first temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the first temperature preset threshold, the output sunroof control command is a danger control command.

[0047] The third signal judgment unit is used to compare the third sensing signal with the preset liquid level threshold. If the third sensing signal is less than the preset liquid level threshold, the output sunroof control command is a safety control command. If the third sensing signal is not less than the preset liquid level threshold, the output sunroof control command is a danger control command.

[0048] The fourth signal judgment unit is used to compare the second sensing signal with the second temperature preset threshold. If the second sensing signal is less than the second temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the second temperature preset threshold, the fourth sensing signal is checked for living characteristics. If living characteristics are present, the output sunroof control command is a danger control command. If no living characteristics are present, the output sunroof control command is a safety control command.

[0049] The sunroof control unit is used to operate the sunroof and complete the automatic control of the vehicle sunroof according to the safety control command and the danger control command respectively. If the sunroof control command is a safety control command, the sunroof is kept in its original state. If the sunroof control command is a danger control command, the sunroof is opened automatically and an alarm is triggered automatically.

[0050] The beneficial effects of this invention are as follows: This invention provides a vehicle sunroof automatic control method and system based on multi-information fusion. First, a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera is used to collect real-time data on the vehicle's internal and external environment, obtaining carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and liveness feature information, thereby obtaining a first sensing signal, a second sensing signal, a third sensing signal, and a fourth sensing signal. Then, based on the carbon dioxide concentration information, temperature information, and wading depth information outside the vehicle body, preset thresholds for carbon dioxide gas concentration, first temperature, second temperature, and liquid level are set. Next, the first sensing signal is compared with the preset threshold for carbon dioxide gas concentration, then the second sensing signal is compared with the first preset threshold for temperature, then the third sensing signal is compared with the preset threshold for liquid level, and finally the second sensing signal is compared with the second preset threshold for temperature. If the second sensing signal is less than the second preset threshold for temperature, the output sunroof control command is "safe". The control command, if the second sensing signal is not less than the second preset temperature threshold, detects the presence of living characteristics in the fourth sensing signal, and then operates the sunroof according to the safety control command and the danger control command respectively to complete the automatic control operation of the vehicle sunroof; effectively realizing the automatic control method and system of the vehicle sunroof, which has the function of automatically opening and closing the vehicle sunroof by adopting a dynamic threshold triggering mechanism and a multi-level response strategy, and achieving millisecond-level emergency response in emergency situations such as the car falling into water, forgetting children in the car in high summer temperatures, and other dangerous situations. It not only improves the passive safety performance of the vehicle, but also has the characteristics of adaptive environmental changes, low false alarm rate and rapid linkage execution. At the same time, through the set two-level fault tolerance mechanism, namely the closed-loop control of dynamic monitoring-threshold judgment-rapid response-automatic recovery, it ensures the construction of millisecond-level escape channels when occupants face the risk of suffocation, while minimizing the impact of misoperation on the normal use of the vehicle. It solves the technical defects of traditional vehicle safety systems in complex emergency situations, such as delayed response and low degree of automation, avoids damage to the car itself, and protects the lives of passengers. Attached Figure Description

[0051] Figure 1 This is an overall flowchart of an automatic control method for vehicle sunroof based on multi-information fusion according to the present invention;

[0052] Figure 2 This is a schematic diagram of the dynamic threshold triggering mechanism and multi-level response strategy of the present invention;

[0053] Figure 3 This is a schematic diagram of the automatic control architecture for vehicle sunroofs according to the present invention;

[0054] Figure 4 This is a schematic diagram of the YOLOv11 deep learning model of the present invention;

[0055] Figure 5 This is a schematic diagram of the DS module architecture of the present invention;

[0056] Figure 6 This is a schematic diagram illustrating the liveness detection effect in an embodiment of the present invention. Detailed Implementation

[0057] The present invention will now be further described with reference to the accompanying drawings.

[0058] like Figure 1 As shown, the present invention provides an automatic control method for a vehicle sunroof based on multi-information fusion, comprising the following steps:

[0059] Step A involves using a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera to collect real-time environmental data inside and outside the vehicle and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and liveness feature information, thereby obtaining a first sensing signal, a second sensing signal, a third sensing signal, and a fourth sensing signal, respectively. The carbon dioxide sensor is installed at the bottom of the driver's seat of the vehicle body and serves as the first sensing element. The carbon dioxide concentration information collected by the first sensing element is the first sensing signal.

[0060] The temperature sensor is encapsulated in the metal interlayer on the top of the vehicle body and serves as a second sensing element. The temperature information collected by the second sensing element is the second sensing signal.

[0061] The liquid level sensor is encapsulated at the lowest point of the chassis at the bottom of the vehicle body and serves as the third sensing element. The wading depth information outside the vehicle body collected by the third sensing element is the third sensing signal.

[0062] The industrial camera is installed in the center of the roof inside the vehicle body and serves as the fourth sensing element. The liveness feature information collected by the fourth sensing element is the fourth sensing signal.

[0063] Step B involves setting a preset threshold for carbon dioxide gas concentration, a first preset threshold for temperature, a second preset threshold for temperature, and a preset threshold for liquid level based on carbon dioxide concentration information, temperature information, and wading depth information outside the vehicle body. The preset threshold for carbon dioxide gas concentration is the safe upper limit of carbon dioxide concentration that the human body can tolerate. The first preset threshold for temperature is the critical value for heat radiation damage to human skin. The second preset threshold for temperature is the critical value for vehicle combustion risk.

[0064] like Figure 2As shown, in step C, the first sensing signal is compared with a preset threshold for carbon dioxide gas concentration. If the first sensing signal is less than the preset threshold, the output sunroof control command is a safety control command; if the first sensing signal is not less than the preset threshold, the output sunroof control command is a hazard control command. A two-level fault-tolerant mechanism is used in the carbon dioxide gas concentration comparison process. The specific steps are as follows.

[0065] Step C1, the first level of the fault tolerance mechanism, specifically uses the moving average filtering method to calculate the arithmetic mean of data from multiple consecutive sampling periods, as shown in formula (1).

[0066]

[0067] in, C represents the filtered concentration value at the k-th sampling point. k-n The data is a continuous sampling sequence of carbon dioxide gas concentration, where N is the number of continuous sampling sequences of carbon dioxide gas concentration, and n is the loop variable.

[0068] Step C2, the second level of the fault tolerance mechanism, specifically involves the carbon dioxide sensor monitoring the changing trend of the first sensing signal in real time after the output of the hazard control command. If the carbon dioxide gas concentration drops to below the preset threshold and remains there for a longer period than the set time, the sunroof will be closed to the initial position at a preset speed and the audible and visual alarms will be deactivated. At the same time, the sunroof control command will be output as a safety control command.

[0069] The system employs a two-tiered fault-tolerance mechanism—a closed-loop control system consisting of dynamic monitoring, threshold determination, rapid response, and automatic recovery—to ensure the construction of millisecond-level escape routes when occupants face the risk of suffocation, while minimizing the impact of misoperation on the normal use of the vehicle.

[0070] Step D involves comparing the second sensing signal with the first preset temperature threshold. If the second sensing signal is less than the first preset temperature threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the first preset temperature threshold, the output sunroof control command is a hazard control command. The comparison process for the first preset temperature threshold uses a sliding average filter and calculates a moving average value for continuous H minutes of data. When the sliding average value exceeds the first preset temperature threshold multiple times consecutively, a hazard control command is triggered, as shown in formula (2).

[0071]

[0072] Among them, T valid The temperature value is the result of moving average filtering, where H is the set number of minutes and T is the temperature value. fLet h be the original temperature value of the f-th sampling point, and h be the sampling point number at the current moment, where h ≥ H.

[0073] Preferably, H is 15 minutes, and the first temperature preset threshold is 40°C.

[0074] Step E: Compare the third sensing signal with the preset liquid level threshold. If the third sensing signal is less than the preset liquid level threshold, the output sunroof control command is a safety control command. If the third sensing signal is not less than the preset liquid level threshold, the output sunroof control command is a danger control command.

[0075] like Figure 4 As shown, in step F, the second sensing signal is compared with the second preset temperature threshold. If the second sensing signal is less than the second preset temperature threshold, the output sunroof control command is a safety control command; if the second sensing signal is not less than the second preset temperature threshold, the presence of liveness features in the fourth sensing signal is detected. If liveness features are present, the output sunroof control command is a danger control command; if no liveness features are present, the output sunroof control command is a safety control command. The liveness feature detection process uses a YOLOv11 deep learning model, which includes a backbone network, a Neck layer network, and a Head network. The backbone network includes... The input network module comprises a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, and a fifth convolutional module. The Neck layer network includes an N1 convolutional module and an N2 convolutional module. The N1 convolutional module is connected from top to bottom by a GP module, an Upsample module, an independent C3k2 module, a GP module, and an Upsample module. The N2 convolutional module is connected from top to bottom by a C3k2 module, a Conv module, a DS module, a C3k2 module, a Conv module, a DS module, and a C3k2 module. The Head network includes four segmentation heads, with the inputs of three of the segmentation heads respectively connected to the three C3k2 modules in the N2 convolutional module of the Neck layer.

[0076] The inputs of the first-layer GP module of the N1 convolutional module include the C3k2 module in the third convolutional module of the backbone network and the first-layer Upsample module of the N1 convolutional module in the Neck network. The inputs of the second-layer GP module of the N1 convolutional module include the C3k2 module in the second convolutional module of the backbone network and the second-layer Upsample module of the N1 convolutional module in the Neck network. The specific construction steps of the GP module are as follows.

[0077] S1. Define the bipartite graph of the GP module as B = (G, A), and use the output of the C3k2 module in the second or third convolutional module of the backbone network as F. zThen, the output of the first or second upsampled layer of the N1 convolutional module is used as F. x Then F z and F x Each pixel is treated as a node in a graph architecture, resulting in an image feature node set V. z and the feature node set V of the search region x Then the definitions of G and A are as shown in formula (3).

[0078]

[0079] Where G is the image feature node set V z and the feature node set V of the search region x The union of the image feature nodes, where A is the set of image feature nodes V. z and the feature node set V of the search region x All connections between them, where g is any pixel in set G and a is any pixel in set A;

[0080] S2, for each i∈V z Calculate i for all j∈V x The similarity coefficient is shown in formula (4).

[0081]

[0082] Among them, e ij It is the attention that node i assigns to node j. Let i be the feature vector of node i. Let W be the feature vector of node j. z and W x All are linear transformation matrices, where m is the summation index variable and T is the transpose;

[0083] S3. Weight the features based on the attention coefficients and sum them, then use the attention coefficients as soft weights and the feature node set V of the search region. x The fusion is shown in formula (5).

[0084]

[0085] in, The features are aggregated, σ is the activation function, and W is the feature. v Let be a linear transformation matrix.

[0086] like Figure 5As shown, the input of the first layer DS module of the N2 convolutional module includes the first layer Conv module of the N2 convolutional module in the Neck layer network and the C3k2 module of the N1 convolutional module in the Neck layer network. The input of the second layer DS module of the N2 convolutional module includes the second layer Conv module of the N2 convolutional module in the Neck layer network and the C2PSA module of the fifth convolutional module in the backbone network.

[0087] Specifically, the DS module concatenates the two inputs on the channel using the Concat module, then reduces the width and height using the Pooling operation. Next, it uses a bottleneck structure to increase feature generalization and uses the Upsample module in the N1 convolution module to restore the width and height of the image features. Finally, it uses a Non-local self-attention mechanism to process the features and outputs the result to the C3k2 module in the N2 convolution module.

[0088] like Figure 3 As shown, in step G, the sunroof is operated and the automatic control operation of the vehicle sunroof is completed according to the safety control command and the hazard control command respectively. If the sunroof control command is a safety control command, the sunroof remains in its original state. If the sunroof control command is a hazard control command, the sunroof is automatically opened and the alarm is automatically triggered. The specific steps for automatically opening the sunroof and triggering the alarm are as follows.

[0089] Step G1: Cut off the PID control loop and remove the speed and position limits. Then apply the maximum rated voltage directly to the drive motor and make the motor enter the stall state to output peak torque. At the same time, bypass the soft start protection of the mechanical transmission mechanism and force the sunroof guide rail slider to slide to the terminal mechanical limit point in the shortest time.

[0090] Step G2 activates the alarm, which is simultaneously activated by the buzzer, LED group, and vehicle screen.

[0091] Step G3 activates the alarm information protocol and uses the vehicle-mounted satellite positioning unit to obtain real-time geographic coordinates and automatically alarm. At the same time, it transmits the data to the cloud platform within a set time. After receiving the data, the cloud platform verifies the legality of the digital certificate, parses the risk level label, and matches it with the vehicle owner's preset contact list. If the verification is successful, it sends an SMS alarm to all preset contacts.

[0092] like Figure 6As shown, an automatic control system for a vehicle sunroof based on multi-information fusion includes a data acquisition unit, a first signal judgment unit, a second signal judgment unit, a third signal judgment unit, a fourth signal judgment unit, and a sunroof control unit. The data acquisition unit is used to collect real-time environmental data inside and outside the vehicle using a perception layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera, and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and liveness feature information, thereby obtaining the first perception signal, the second perception signal, the third perception signal, and the fourth perception signal, respectively.

[0093] The first signal judgment unit is used to compare the first sensing signal with the preset threshold of carbon dioxide gas concentration. If the first sensing signal is less than the preset threshold of carbon dioxide gas concentration, the output sunroof control command is a safety control command. If the first sensing signal is not less than the preset threshold of carbon dioxide gas concentration, the output sunroof control command is a danger control command.

[0094] The second signal judgment unit is used to compare the second sensing signal with the first temperature preset threshold. If the second sensing signal is less than the first temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the first temperature preset threshold, the output sunroof control command is a danger control command.

[0095] The third signal judgment unit is used to compare the third sensing signal with the preset liquid level threshold. If the third sensing signal is less than the preset liquid level threshold, the output sunroof control command is a safety control command. If the third sensing signal is not less than the preset liquid level threshold, the output sunroof control command is a danger control command.

[0096] The fourth signal judgment unit is used to compare the second sensing signal with the second temperature preset threshold. If the second sensing signal is less than the second temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the second temperature preset threshold, the fourth sensing signal is checked for living characteristics. If living characteristics are present, the output sunroof control command is a danger control command. If no living characteristics are present, the output sunroof control command is a safety control command.

[0097] The sunroof control unit is used to operate the sunroof and complete the automatic control of the vehicle sunroof according to the safety control command and the danger control command respectively. If the sunroof control command is a safety control command, the sunroof is kept in its original state. If the sunroof control command is a danger control command, the sunroof is opened automatically and an alarm is triggered automatically.

[0098] In summary, the present invention provides a vehicle sunroof automatic control method and system based on multi-information fusion. First, a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera is used to collect real-time environmental data inside and outside the vehicle, obtaining carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and liveness feature information, thereby obtaining a first sensing signal, a second sensing signal, a third sensing signal, and a fourth sensing signal. Then, based on the carbon dioxide concentration information, temperature information, and wading depth information outside the vehicle body, preset thresholds for carbon dioxide gas concentration, first temperature, second temperature, and liquid level are set. Next, the first sensing signal is compared with the preset threshold for carbon dioxide gas concentration, then the second sensing signal is compared with the first preset threshold for temperature, then the third sensing signal is compared with the preset threshold for liquid level, and finally the second sensing signal is compared with the second preset threshold for temperature. If the second sensing signal is less than the second preset threshold for temperature, the output sunroof control command is a safety control command. If the second sensing signal is not less than the second preset temperature threshold, the system detects whether the fourth sensing signal has any living characteristics. Then, based on the safety control command and the danger control command, it operates the sunroof and completes the automatic control operation of the vehicle sunroof. This effectively realizes that the automatic control method and system of the vehicle sunroof has the function of automatically opening and closing the vehicle sunroof by adopting a dynamic threshold triggering mechanism and a multi-level response strategy. It can also achieve millisecond-level emergency response in emergency situations such as the car falling into water, forgetting children in the car in high summer temperatures, and other dangerous situations. This not only improves the passive safety performance of the vehicle, but also has the characteristics of adaptive environmental changes, low false alarm rate, and rapid linkage execution. At the same time, through the set two-level fault tolerance mechanism, namely the closed-loop control of dynamic monitoring-threshold judgment-rapid response-automatic recovery, it ensures the construction of a millisecond-level escape route when the occupants face the risk of suffocation. At the same time, it minimizes the impact of misoperation on the normal use of the vehicle, solves the technical defects of traditional vehicle safety systems in complex emergency situations such as slow response and low degree of automation, avoids damage to the car itself, and protects the lives of passengers.

[0099] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for automatic control of a vehicle sunroof based on multi-information fusion, characterized in that: Includes the following steps, Step A: A sensing layer consisting of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera is used to collect real-time environmental data inside and outside the vehicle and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and living feature information, thereby obtaining the first sensing signal, the second sensing signal, the third sensing signal, and the fourth sensing signal, respectively. Step B: Based on carbon dioxide concentration information, temperature information, and wading depth information outside the vehicle body, set preset thresholds for carbon dioxide gas concentration, first temperature, second temperature, and liquid level, respectively. Step C: Compare the first sensing signal with the preset threshold for carbon dioxide gas concentration. If the first sensing signal is less than the preset threshold for carbon dioxide gas concentration, the output sunroof control command is a safety control command. If the first sensing signal is not less than the preset threshold for carbon dioxide gas concentration, the output sunroof control command is a hazard control command. Step D: Compare the second sensing signal with the first temperature preset threshold. If the second sensing signal is less than the first temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the first temperature preset threshold, the output sunroof control command is a danger control command. Step E: Compare the third sensing signal with the preset liquid level threshold. If the third sensing signal is less than the preset liquid level threshold, the output sunroof control command is a safety control command. If the third sensing signal is not less than the preset liquid level threshold, the output sunroof control command is a danger control command. Step F: Compare the second sensing signal with the second preset temperature threshold. If the second sensing signal is less than the second preset temperature threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the second preset temperature threshold, detect whether there are living features in the fourth sensing signal. If there are living features, the output sunroof control command is a danger control command. If there are no living features, the output sunroof control command is a safety control command. Step G: Operate the sunroof according to the safety control command and the danger control command respectively, and complete the automatic control operation of the vehicle sunroof. If the sunroof control command is a safety control command, the sunroof will remain in its original state. If the sunroof control command is a danger control command, the sunroof will be opened automatically and an alarm will be triggered automatically.

2. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 1, characterized in that: Step A involves using a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera to collect real-time environmental data inside and outside the vehicle and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and liveness feature information, thereby obtaining a first sensing signal, a second sensing signal, a third sensing signal, and a fourth sensing signal, respectively. The carbon dioxide sensor is installed at the bottom of the driver's seat of the vehicle body and serves as the first sensing element. The carbon dioxide concentration information collected by the first sensing element is the first sensing signal. The temperature sensor is encapsulated in the metal interlayer on the top of the vehicle body and serves as a second sensing element. The temperature information collected by the second sensing element is the second sensing signal. The liquid level sensor is encapsulated at the lowest point of the chassis at the bottom of the vehicle body and serves as the third sensing element. The wading depth information outside the vehicle body collected by the third sensing element is the third sensing signal. The industrial camera is installed in the center of the roof inside the vehicle and serves as the fourth sensing element. The liveness feature information collected by the fourth sensing element is the fourth sensing signal.

3. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 1, characterized in that: Step B involves setting a preset threshold for carbon dioxide gas concentration, a first preset threshold for temperature, a second preset threshold for temperature, and a preset threshold for liquid level based on carbon dioxide concentration information, temperature information, and wading depth information outside the vehicle body. The preset threshold for carbon dioxide gas concentration is the safe upper limit of carbon dioxide concentration that the human body can tolerate. The first preset threshold for temperature is the critical value for heat radiation damage to human skin. The second preset threshold for temperature is the critical value for vehicle combustion risk.

4. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 1, characterized in that: Step C employs a two-stage fault-tolerance mechanism in the comparison of carbon dioxide gas concentrations. The specific steps are as follows. Step C1, the first level of the fault tolerance mechanism, specifically uses the moving average filtering method to calculate the arithmetic mean of data from multiple consecutive sampling periods, as shown in formula (1). in, C represents the filtered concentration value at the k-th sampling point. k-n The data is a continuous sampling sequence of carbon dioxide gas concentration, where N is the number of continuous sampling sequences of carbon dioxide gas concentration, and n is the loop variable. Step C2, the second level of the fault tolerance mechanism, specifically involves the carbon dioxide sensor monitoring the changing trend of the first sensing signal in real time after the output of the hazard control command. If the carbon dioxide gas concentration drops to below the preset threshold and remains there for a longer period than the set time, the sunroof will be closed to the initial position at a preset speed and the audible and visual alarms will be deactivated. At the same time, the sunroof control command will be output as a safety control command.

5. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 1, characterized in that: In step D, during the comparison of the first preset temperature threshold, a sliding average filter is used, and a moving average is calculated for continuous H minutes of data. When the sliding average exceeds the first preset temperature threshold multiple times consecutively, a hazard control command is triggered, as shown in formula (2). Among them, T valid The temperature value is the result of moving average filtering, where H is the set number of minutes and T is the temperature value. f Let h be the original temperature value of the f-th sampling point, and h be the sampling point number at the current moment, where h ≥ H.

6. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 1, characterized in that: In step F, the liveness detection process uses a YOLOv11 deep learning model. This model comprises a backbone network, a Neck layer network, and a Head network. The backbone network includes an input network module, a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, and a fifth convolutional module. The Neck layer network includes an N1 convolutional module and an N2 convolutional module. The N1 convolutional module, from top to bottom, consists of a GP module, an Upsample module, an independent C3k2 module, another GP module, and an Upsample module connected together. The N2 convolutional module, from top to bottom, consists of a C3k2 module, a Conv module, a DS module, a C3k2 module, a Conv module, a DS module, and a C3k2 module connected together. The Head network contains four segmentation heads, with the inputs of three of these head segments connected to three C3k2 modules within the N2 convolutional module of the Neck layer.

7. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 6, characterized in that: The inputs of the first-layer GP module of the N1 convolutional module include the C3k2 module in the third convolutional module of the backbone network and the first-layer Upsample module of the N1 convolutional module in the Neck network. The inputs of the second-layer GP module of the N1 convolutional module include the C3k2 module in the second convolutional module of the backbone network and the second-layer Upsample module of the N1 convolutional module in the Neck network. The specific construction steps of the GP module are as follows. S1. Define the bipartite graph of the GP module as B = (G, A), and use the output of the C3k2 module in the second or third convolutional module of the backbone network as F. z Then, the output of the first or second upsampled layer of the N1 convolutional module is used as F. x Then F z and F x Each pixel is treated as a node in a graph architecture, resulting in an image feature node set V. z and the feature node set V of the search region x Then the definitions of G and A are as shown in formula (3). Where G is the image feature node set V z and the feature node set V of the search region x The union of the image feature nodes, where A is the set of image feature nodes V. z and the feature node set V of the search region x All connections between them, where g is any pixel in set G and a is any pixel in set A; S2, for each i∈V z Calculate i for all j∈V x The similarity coefficient is shown in formula (4). Among them, e ij It is the attention that node i assigns to node j. Let i be the feature vector of node i. Let W be the feature vector of node j. z and W x All are linear transformation matrices, where m is the summation index variable and T is the transpose; S3. Weight the features based on the attention coefficients and sum them, then use the attention coefficients as soft weights and the feature node set V of the search region. x The fusion is shown in formula (5). in, The features are aggregated, σ is the activation function, and W is the feature. v Let be a linear transformation matrix.

8. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 6, characterized in that: The input of the first layer DS module of the N2 convolutional module includes the first layer Conv module of the N2 convolutional module in the Neck layer network and the C3k2 module of the N1 convolutional module in the Neck layer network. The input of the second layer DS module of the N2 convolutional module includes the second layer Conv module of the N2 convolutional module in the Neck layer network and the C2PSA module of the fifth convolutional module in the backbone network. Specifically, the DS module concatenates the two inputs on the channel using the Concat module, then reduces the width and height using the Pooling operation. Next, it uses a bottleneck structure to increase feature generalization and uses the Upsample module in the N1 convolution module to restore the width and height of the image features. Finally, it uses a Non-local self-attention mechanism to process the features and outputs the result to the C3k2 module in the N2 convolution module.

9. The automatic control method for a vehicle sunroof based on multi-information fusion according to claim 1, characterized in that: The specific steps for automatically opening the sunroof and triggering the alarm in step G are as follows: Step G1: Cut off the PID control loop and remove the speed and position limits. Then apply the maximum rated voltage directly to the drive motor and make the motor enter the stall state to output peak torque. At the same time, bypass the soft start protection of the mechanical transmission mechanism and force the sunroof guide rail slider to slide to the terminal mechanical limit point in the shortest time. Step G2 activates the alarm, which is simultaneously activated by the buzzer, LED group, and vehicle screen. Step G3 activates the alarm information protocol and uses the vehicle-mounted satellite positioning unit to obtain real-time geographic coordinates and automatically alarm. At the same time, it transmits the data to the cloud platform within a set time. After receiving the data, the cloud platform verifies the legality of the digital certificate, parses the risk level label, and matches it with the vehicle owner's preset contact list. If the verification is successful, it sends an SMS alarm to all preset contacts.

10. A vehicle sunroof automatic control system based on multi-information fusion, wherein the specific control process of the vehicle sunroof automatic control system is based on the vehicle sunroof automatic control method according to any one of claims 1-9, characterized in that: It includes a data acquisition unit, a first signal judgment unit, a second signal judgment unit, a third signal judgment unit, a fourth signal judgment unit, and a sunroof control unit. The data acquisition unit is used to collect real-time environmental data inside and outside the vehicle using a sensing layer composed of a carbon dioxide sensor, a temperature sensor, a liquid level sensor, and an industrial camera, and obtain carbon dioxide concentration information, temperature information, wading depth information outside the vehicle body, and living feature information, thereby obtaining the first sensing signal, the second sensing signal, the third sensing signal, and the fourth sensing signal, respectively. The first signal judgment unit is used to compare the first sensing signal with the preset threshold of carbon dioxide gas concentration. If the first sensing signal is less than the preset threshold of carbon dioxide gas concentration, the output sunroof control command is a safety control command. If the first sensing signal is not less than the preset threshold of carbon dioxide gas concentration, the output sunroof control command is a danger control command. The second signal judgment unit is used to compare the second sensing signal with the first temperature preset threshold. If the second sensing signal is less than the first temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the first temperature preset threshold, the output sunroof control command is a danger control command. The third signal judgment unit is used to compare the third sensing signal with the preset liquid level threshold. If the third sensing signal is less than the preset liquid level threshold, the output sunroof control command is a safety control command. If the third sensing signal is not less than the preset liquid level threshold, the output sunroof control command is a danger control command. The fourth signal judgment unit is used to compare the second sensing signal with the second temperature preset threshold. If the second sensing signal is less than the second temperature preset threshold, the output sunroof control command is a safety control command. If the second sensing signal is not less than the second temperature preset threshold, the fourth sensing signal is checked for living characteristics. If living characteristics are present, the output sunroof control command is a danger control command. If no living characteristics are present, the output sunroof control command is a safety control command. The sunroof control unit is used to operate the sunroof and complete the automatic control of the vehicle sunroof according to the safety control command and the danger control command respectively. If the sunroof control command is a safety control command, the sunroof is kept in its original state. If the sunroof control command is a danger control command, the sunroof is opened automatically and an alarm is triggered automatically.

Images