Door control methods, devices, automobiles, media and program products

Abstract

This application relates to the field of vehicle door control technology, and particularly to a method, device, vehicle, medium, and program product for controlling a vehicle door. The method includes: acquiring user trajectory data based on location information sent by a positioning tag, and extracting corresponding trajectory features from the trajectory data; inputting the trajectory features into a pre-trained LSTM model to output a predicted probability value of the user walking towards a target door in the vehicle; if the predicted probability value is greater than a preset threshold, then calculating the user's dwell time in the corresponding target door area, and if the dwell time exceeds a preset time threshold, generating an opening command for the target door, and controlling the vehicle to open the target door according to the opening command. This solves the problems in related technologies where all actions are initiated by the user, requiring high accuracy and standardization of the actions, easily leading to recognition failures, and being susceptible to environmental interference, affecting the user experience.

Description

Technical Field

[0001] This application relates to the field of vehicle door control technology, and in particular to a vehicle door control method, device, automobile, medium, and program product. Background Technology

[0002] In related technologies, ultrasonic sensors or millimeter-wave radar can be installed below or to the side of the tailgate to emit sound waves or electromagnetic waves and receive reflected signals, thereby detecting specific foot movements of the user to trigger the opening of the car tailgate; alternatively, infrared sensors or camera sensors placed near the tailgate can be used to identify specific user movements to trigger the opening of the tailgate; or the tailgate can be opened and closed by the user actively operating the remote key or mobile APP.

[0003] However, in the relevant technologies, all actions are initiated by the user, which requires high accuracy and standardization of the actions, making it easy to fail to recognize them. Furthermore, these technologies are susceptible to environmental interference, affecting the user experience and urgently need improvement. Summary of the Invention

[0004] This application provides a method, device, automobile, medium, and program product for controlling a vehicle door, in order to solve the problems in related technologies where all related actions are initiated by the user, the accuracy and standardization of the actions are highly required, which can easily lead to recognition failures and are susceptible to environmental interference, thus affecting the user experience.

[0005] The first aspect of this application provides a method for controlling a car door, comprising the following steps: acquiring trajectory data of a user based on location information sent by a positioning tag, and extracting corresponding trajectory features from the trajectory data; inputting the trajectory features into a pre-trained LSTM (Long Short-Term Memory) model to output a predicted probability value of the user walking toward a target car door; if the predicted probability value is greater than a preset threshold, then counting the time the user stays in the area of ​​the corresponding target car door, and if the stay time is greater than a preset time threshold, generating an opening command for the target car door, and controlling the car to open the target car door according to the opening command.

[0006] Optionally, in one embodiment of this application, before inputting the trajectory features into a pre-trained LSTM model, the method further includes: constructing an input layer in the LSTM model based on the target trajectory features; constructing a first bidirectional LSTM layer in the LSTM model according to a preset first number of neurons, wherein the first bidirectional LSTM layer includes at least one forward LSTM sublayer and one backward LSTM sublayer; constructing a random deactivation layer in the LSTM model according to a preset dropout rate; constructing a second bidirectional LSTM layer in the LSTM model according to a preset second number of neurons, wherein the second bidirectional LSTM layer includes at least one forward LSTM sublayer and one backward LSTM sublayer; constructing a fully connected layer in the LSTM model, and constructing an output layer in the LSTM model based on a preset activation function; and constructing the LSTM model based on the input layer, the first bidirectional LSTM layer, the random deactivation layer, the second bidirectional LSTM layer, the fully connected layer, and the output layer.

[0007] Optionally, in one embodiment of this application, before inputting the trajectory features into the pre-trained LSTM model, the method further includes: obtaining target trajectory data of the target user based on the location information sent by the location tag, and extracting the corresponding target trajectory features from the target trajectory data; constructing a training dataset for the LSTM model based on the target trajectory features; and training the LSTM model based on the training dataset until the LSTM model meets the preset training conditions to obtain a trained LSTM model.

[0008] Optionally, in one embodiment of this application, obtaining the user's trajectory data includes: if the trajectory data does not meet preset data conditions, processing the trajectory data until processed trajectory data that meets the preset data conditions is obtained.

[0009] Optionally, in one embodiment of this application, obtaining the user's trajectory data based on the location information sent by the location tag includes: obtaining the communication link status between the location anchor point in the vehicle and the location tag; if the communication link status is in a valid communication link status, calculating the actual distance between the vehicle and the location tag; and obtaining the trajectory data based on the actual distance.

[0010] A second aspect of this application provides a vehicle door control device, comprising: a first extraction module, configured to acquire trajectory data of a user based on location information sent by a positioning tag, and extract corresponding trajectory features from the trajectory data; an output module, configured to input the trajectory features into a pre-trained LSTM model to output a predicted probability value of the user walking toward a target door in the vehicle; and a control module, configured to, when the predicted probability value is greater than a preset threshold, count the dwell time of the user in the corresponding target door area, and when the dwell time is greater than a preset time threshold, generate an opening command for the target door, and control the vehicle to open the target door according to the opening command.

[0011] Optionally, in one embodiment of this application, the model further includes: a first construction module, configured to construct an input layer in the LSTM model based on the target trajectory features before inputting the trajectory features into the pre-trained LSTM model; a second construction module, configured to construct a first bidirectional LSTM layer in the LSTM model according to a preset first number of neurons, wherein the first bidirectional LSTM layer includes at least one forward LSTM sublayer and one backward LSTM sublayer; a third construction module, configured to construct a random deactivation layer in the LSTM model according to a preset dropout rate; a fourth construction module, configured to construct a second bidirectional LSTM layer in the LSTM model according to a preset second number of neurons, wherein the second bidirectional LSTM layer includes at least one forward LSTM sublayer and one backward LSTM sublayer; a fifth construction module, configured to construct a fully connected layer in the LSTM model and construct an output layer in the LSTM model based on a preset activation function; and a sixth construction module, configured to construct the LSTM model based on the input layer, the first bidirectional LSTM layer, the random deactivation layer, the second bidirectional LSTM layer, the fully connected layer, and the output layer.

[0012] Optionally, in one embodiment of this application, it further includes: a second extraction module, configured to obtain target trajectory data of the target user based on the location information sent by the location tag before inputting the trajectory features into the pre-trained LSTM model, and extract the corresponding target trajectory features from the target trajectory data; a seventh construction module, configured to construct a training dataset of the LSTM model based on the target trajectory features; and a training module, configured to train the LSTM model based on the training dataset until the LSTM model meets the preset training conditions to obtain a trained LSTM model.

[0013] Optionally, in one embodiment of this application, the first extraction module includes: a processing unit, configured to process the trajectory data when the trajectory data does not meet preset data conditions, until processed trajectory data that meets the preset data conditions is obtained.

[0014] Optionally, in one embodiment of this application, the first extraction module includes: a first acquisition unit, configured to acquire the communication link status between the positioning anchor point in the vehicle and the positioning tag; a calculation unit, configured to calculate the actual distance between the vehicle and the positioning tag when the communication link status is in a valid communication link status; and a second acquisition unit, configured to acquire the trajectory data based on the actual distance.

[0015] A third aspect of this application provides an automobile, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the door control method as described in the above embodiments.

[0016] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described door control method.

[0017] A fifth aspect of this application provides a computer program product, including a computer program that, when executed, implements the above-described door control method.

[0018] This application embodiment can acquire user trajectory data based on location information sent by a location tag, extract corresponding trajectory features from the trajectory data, and then use a pre-trained LSTM model to output a predicted probability value of the user walking towards the target door of a car. When the predicted probability value is greater than a preset threshold, the dwell time of the user in the corresponding target door area is counted. If the dwell time is greater than a preset time threshold, an opening command for the target door is generated, opening the target door. This eliminates the need for the user to actively initiate any specific action. By using location information and the LSTM model to predict the probability of the user walking towards the target door, combined with the dwell time to determine the opening command, the requirements for the accuracy and standardization of the action are reduced, the number of recognition failures is decreased, and it is less susceptible to environmental interference, greatly improving the user experience. Therefore, it solves the problems in related technologies where all actions are actively initiated by the user, requiring high accuracy and standardization of the action, easily leading to recognition failures, and being susceptible to environmental interference, thus affecting the user experience.

[0019] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0020] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the layout of an automotive intelligent sensing system based on UWB (Ultra Wide Band) tags according to an embodiment of this application; Figure 2 This is a flowchart of a door control method according to an embodiment of this application; Figure 3 This is a flowchart illustrating the working principle of a door control method according to an embodiment of this application; Figure 4 This is a block diagram of a door control device provided according to an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a car provided according to an embodiment of this application. Detailed Implementation

[0021] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0022] Before introducing the door control method proposed in the embodiments of this application, the system involving UWB tags in the embodiments of this application will be introduced first.

[0023] Specifically, Figure 1 This is a schematic diagram of the layout of an automotive intelligent sensing system based on a UWB tag according to an embodiment of this application.

[0024] like Figure 1 As shown, the system includes multiple UWB anchor points (first UWB anchor point 101, second UWB anchor point 102, third UWB anchor point 103, fourth UWB anchor point 104 and fifth UWB anchor point 105), ECU (Electronic Control Unit) 106 and tailgate control switch 107.

[0025] Among them, the first UWB anchor point 101, the second UWB anchor point 102, the third UWB anchor point 103 and the fourth UWB anchor point 104 are ordinary anchor points with BLE+UWB function. They are deployed in the middle of the front and rear safety and the middle of the left and right side skirts, respectively, for transmitting and receiving UWB signals and for ranging and positioning by interacting with UWB tags.

[0026] The fifth UWB anchor point 105 is the main anchor point. Only one main anchor point with BLE+UWB function is arranged at the vehicle end, which is responsible for coordinating other ordinary anchor points, processing positioning data and communicating with ECU106.

[0027] As the control core, ECU106 can receive signals from the fifth UWB anchor point 105 and the tailgate control switch 107, perform logical judgments, and output control commands to drive the tailgate to perform opening or closing actions.

[0028] The tailgate control switch 107 is installed in the rear area of ​​the vehicle. It can be used as a physical switch to manually trigger the tailgate, or as an auxiliary reference point for UWB positioning to transmit a switch signal to the ECU 106.

[0029] UWB tags are devices such as mobile phones or UWB keys that have a built-in UWB chip. They communicate wirelessly with the vehicle system via Bluetooth to achieve pairing and connection, enabling accurate identification and location tracking.

[0030] The system works as follows: When a user carrying a UWB tag approaches the car, the first UWB anchor point 101, the second UWB anchor point 102, the third UWB anchor point 103, the fourth UWB anchor point 104, and the fifth UWB anchor point 105 measure the distance to the UWB tag. The fifth UWB anchor point 105 calculates the user's location information using a multi-point positioning algorithm and transmits the location information tag to the ECU 106. The ECU 106, combined with the state of the tailgate control switch 107, determines whether the user intends to open the tailgate, thereby automatically opening the tailgate and achieving keyless intelligent control.

[0031] The following description, with reference to the accompanying drawings, outlines a method, apparatus, vehicle, medium, and program product for controlling a car door according to embodiments of this application. Addressing the issues mentioned in the background art, where all related actions are initiated by the user, requiring high accuracy and standardization, easily leading to recognition failures, and being susceptible to environmental interference, thus affecting the user experience, this application provides a method for controlling a car door. In this method, user trajectory data is obtained based on location information sent by a positioning tag, and corresponding trajectory features are extracted from the trajectory data. A pre-trained LSTM model is then used to output a predicted probability value of the user walking towards a target car door. When the predicted probability value is greater than a preset threshold, the user's dwell time in the corresponding target car door area is counted. If the dwell time exceeds a preset time threshold, an opening command for the target car door is generated, opening the target car door. This eliminates the need for the user to initiate a specific action. By using location information and an LSTM model to predict the probability of the user walking towards the target car door, combined with the dwell time, an opening command is generated. This reduces the requirements for action accuracy and standardization, decreases recognition failures, and is less susceptible to environmental interference, greatly improving the user experience. This solves the problems in related technologies where all actions are initiated by the user, requiring high accuracy and standardization of the actions, which easily leads to recognition failure and is susceptible to environmental interference, affecting the user experience.

[0032] Specifically, Figure 2 This is a flowchart of a door control method provided according to an embodiment of this application.

[0033] like Figure 2 As shown, the method for controlling the car door includes the following steps: In step S201, the user's trajectory data is obtained based on the location information sent by the location tag, and the corresponding trajectory features are extracted from the trajectory data.

[0034] It is understood that, in the embodiments of this application, the positioning tag can be understood as a small electronic device that can be attached to a user or object for transmitting wireless signals containing location information in real time, such as UWB, Bluetooth, etc., and this application does not impose any specific limitations.

[0035] Location information can be understood as raw data sent by the location tag, which may include, but is not limited to, timestamps, location coordinates (such as three-dimensional coordinates, etc., which are not specifically limited in this application), etc.

[0036] Trajectory features may include, but are not limited to, dynamic and static features. Dynamic features may include, but are not limited to, instantaneous velocity. Direction angle The specific settings can be configured by those skilled in the art according to the actual situation, and this application does not impose specific limitations. and This data can be collected based on the user's walking habits, distance from the vehicle, etc., and this application does not impose specific limitations; static features may include, but are not limited to, the distance between the UWB tag and the tailgate. The time interval between two adjacent steps of the user This application does not impose specific restrictions, etc.

[0037] In some embodiments, this application can receive location information sent from a location tag and obtain the user's trajectory data based on the location information, thereby determining the user's trajectory characteristics.

[0038] For example, such as Figure 1 As shown in the embodiment of this application, multiple UWB anchor points supporting BLE+UWB functionality can be installed on the vehicle. When a user carrying a UWB tag approaches the vehicle, the user's location information can be obtained in real time through the UWB anchor points.

[0039] Furthermore, embodiments of this application can calculate tag coordinates in real time (accuracy ±5cm) using positioning information to generate user trajectory data, which can be represented as follows: ,in, For the first Frame timestamp For the UWB tag The three-dimensional coordinates relative to the vehicle coordinate system at frame time. The total number of frames for trajectory data (determined by the sampling frequency and time span, e.g., a sampling frequency of 10Hz for 10 seconds). wait).

[0040] Furthermore, embodiments of this application can extract the user's instantaneous speed from trajectory data. Direction angle Distance between the UWB label and the tailgate The time interval between two adjacent steps of the user Trajectory characteristics.

[0041] It should be noted that in certain scenarios, such as when the signal transmission between the key and the vehicle is blocked by a person, the UWB tag may not be able to effectively open the car door.

[0042] Optionally, in one embodiment of this application, obtaining user trajectory data includes: if the trajectory data does not meet preset data conditions, processing the trajectory data until processed trajectory data that meets the preset data conditions is obtained.

[0043] In some embodiments, the present application can detect whether the trajectory data meets preset data conditions, and if not, process the trajectory data until the preset data conditions are met. The preset data conditions can be set by those skilled in the art according to actual circumstances, and the present application does not impose specific limitations.

[0044] For example, in combination Figure 1 As shown, in this embodiment of the application, when the trajectory data does not meet the preset data conditions, the ECU can be controlled to perform Kalman filtering on the trajectory data to remove abnormal points caused by occlusion or multipath interference, thereby obtaining smooth trajectory data that meets the preset data conditions.

[0045] Optionally, in one embodiment of this application, obtaining user trajectory data based on the location information sent by the location tag includes: obtaining the communication link status between the location anchor point in the vehicle and the location tag; if the communication link status is in a valid communication link status, calculating the actual distance between the vehicle and the location tag; and obtaining trajectory data based on the actual distance.

[0046] It is understood that, in the embodiments of this application, the communication link status can be understood as the status between the vehicle-side UWB anchor point and the UWB positioning tag carried by the user. It may include, but is not limited to, a valid communication link status, a signal loss communication link status, an interrupted communication link status, an authentication failure communication link status, etc., and this application does not impose specific limitations.

[0047] In some embodiments, the communication link status between the vehicle and the positioning tag can be detected first, and if the communication link status is in a valid communication link status, the actual distance between the vehicle and the positioning tag can be calculated to obtain the user's trajectory data.

[0048] Optionally, in one embodiment of this application, before inputting the trajectory features into a pre-trained LSTM model, the method further includes: constructing an input layer in the LSTM model based on the target trajectory features; constructing a first bidirectional LSTM layer in the LSTM model according to a preset first number of neurons, wherein the first bidirectional LSTM layer includes at least one forward LSTM sublayer and one backward LSTM sublayer; constructing a random deactivation layer in the LSTM model according to a preset dropout rate; constructing a second bidirectional LSTM layer in the LSTM model according to a preset second number of neurons, wherein the second bidirectional LSTM layer includes at least one forward LSTM sublayer and one backward LSTM sublayer; constructing a fully connected layer in the LSTM model, and constructing an output layer in the LSTM model based on a preset activation function; and constructing an LSTM model based on the input layer, the first bidirectional LSTM layer, the random deactivation layer, the second bidirectional LSTM layer, the fully connected layer, and the output layer.

[0049] In some embodiments, the LSTM model constructed in this application includes an input layer, a first bidirectional LSTM layer, a random deactivation layer, a second bidirectional LSTM layer, a fully connected layer, and an output layer. Specifically, in this application, the target trajectory features are used as the input values ​​of the LSTM model, that is, the input layer of the LSTM model includes four features.

[0050] Furthermore, the first bidirectional LSTM layer constructed in this application includes a predetermined number of neurons, such as 64 neurons, and employs a bidirectional mechanism: the forward LSTM sublayer follows... Extract features of how the user approaches (such as increasing speed, stable orientation angle, etc., this application does not impose specific limitations); the inverse LSTM sublayer follows... Extract the user's possible future trends (e.g., continuing at the current speed, reaching the tailgate in 5 seconds, etc., this application does not impose specific limitations); and concatenate the outputs of the forward LSTM sublayer and the backward LSTM sublayer to obtain the first layer of hidden states. (This application does not impose specific limitations on the basic characteristics of the past and future.)

[0051] Furthermore, by using a randomly deactivated layer with a preset dropout rate of 0.3, overfitting of the LSTM layer is prevented, improving model generalization and reducing redundant computation. The preset dropout rate of 0.3 means that during training, each neuron has a 30% probability of being temporarily deactivated (outputting 0), with the remaining 70% of neurons participating in the current computation. During testing, all neurons are restored, but the output is weighted according to the retention probability.

[0052] It should be noted that the reason for using a random deactivation layer in this application embodiment is that because LSTM has a memory function, it is easy to overfit the input features and lose generalization, such as only being applicable to the data of adult car owners while losing the walking data of the elderly and children, and losing certain proximity patterns.

[0053] Furthermore, in this embodiment, the output of the randomly deactivated layer is input to the second bidirectional LSTM layer. This second bidirectional LSTM layer includes a predetermined second number of neurons, such as 32 neurons, and employs a bidirectional mechanism: the forward LSTM sublayer receives... Abstracting higher-order features of continuous approach (such as increasing velocity and unchanged direction, etc., this application does not impose specific limitations); receiving from the inverse LSTM sublayer. This paper abstracts stable features without sudden changes in direction (such as low trajectory curvature, which are not specifically limited in this application); and concatenates the outputs of the forward LSTM sub-layer and the backward LSTM sub-layer to obtain the second hidden state. (Features that include the probability of intent).

[0054] In summary, the embodiments of this application output through a two-layer bidirectional LSTM layer and a random deactivation layer. The process is as follows: →Random deactivation layer (0.3)→ The structural basis of LSTM: neurons, or memory units. The core of LSTM is the LSTM unit, each of which is a memory module that manages the storage and forgetting of information through gating mechanisms (input gate, forget gate, output gate). In an LSTM layer, the number of neurons refers to the number of LSTM units contained in that layer. Each unit independently processes a time step in the time series and maintains its own cellular state (long-term memory) and hidden state (short-term memory).

[0055] Furthermore, in this embodiment, the predicted probability value is output through a fully connected layer. The corresponding working principle is as follows: Since the fully connected layer is one of the most basic layer types in a neural network, its core feature is that each neuron in the current layer is connected to all neurons in the previous layer through weights. In the intention prediction task, the main function of the fully connected layer is: (1) Feature aggregation: compressing the high-dimensional hidden state output by the LSTM into low-dimensional features (such as compressing 32-dimensional hidden states into low-dimensional features). (1) Compressed to 1 dimension); (2) Classification decision: Through linear transformation and preset activation functions, such as the Sigmoid function, the aggregated features are mapped to probability values. The main content can be: input The parameters of the fully connected layer are designed as 32-dimensional input → 1-dimensional output. Then, the output value is compressed to the [0,1] interval through the Sigmoid function of the output layer, which yields the prediction probability value, such as the prediction probability value of heading towards the tailgate.

[0056] The preset first quantity, preset discard rate, and preset second quantity can be set by those skilled in the art according to the actual situation, and this application does not impose specific restrictions.

[0057] Optionally, in one embodiment of this application, before inputting the trajectory features into the pre-trained LSTM model, the method further includes: obtaining the target trajectory data of the target user based on the location information sent by the location tag, and extracting the corresponding target trajectory features from the target trajectory data; constructing a training dataset for the LSTM model based on the target trajectory features; and training the LSTM model based on the training dataset until the LSTM model meets the preset training conditions to obtain a trained LSTM model.

[0058] As one possible implementation, embodiments of this application can obtain target trajectory data of a target user by parsing the location information sent by the location tag, extract the corresponding target trajectory features from it, construct a training dataset suitable for an LSTM model, and iteratively train the LSTM model until it meets preset training conditions, thereby obtaining a trained LSTM model. The preset training conditions can be set by those skilled in the art according to actual conditions, and this application does not impose specific limitations.

[0059] For example, in this application embodiment, an LSTM model can be trained by collecting 1,000 target trajectory data (including two types of samples: "opening the tailgate" and "passing by without stopping") from 100 target users (with different heights and step counts). After the test accuracy of the LSTM model reaches 92%, it is determined that the LSTM model meets the preset training conditions, and thus a trained LSTM model is obtained.

[0060] In step S202, the trajectory features are input into a pre-trained LSTM model to output the predicted probability value of the user walking toward the target door in the car.

[0061] It is understood that, in the embodiments of this application, the target vehicle door may include, but is not limited to, the front door, the rear door, and the tailgate, etc., and this application does not impose specific limitations.

[0062] In some embodiments, the present application can use trajectory features as input and then use a pre-trained LSTM model to output the predicted probability value of a user walking toward the target door in the car.

[0063] In step S203, if the predicted probability value is greater than a preset threshold, the user's dwell time in the corresponding target door area is counted, and if the dwell time is greater than a preset time threshold, an opening command for the target door is generated, and the car is controlled to open the target door according to the opening command.

[0064] It is understood that in the embodiments of this application, the target door area can be understood as an electronic fence area defined with the car door as the center, used to detect whether the user is within the effective door opening range. It can be understood as an area within 0 to 0.5 meters. This application does not make specific limitations. For example, the tailgate area can be defined as a cylindrical space with a radius of 0.5 meters centered on the door core.

[0065] In some embodiments, this application can, when the predicted probability value is greater than a preset threshold, count the time a user spends in the corresponding target door area, and when the time spent exceeds a preset time threshold, generate an opening command for the target door, thereby opening the corresponding target door. The preset threshold and preset time threshold can be set by those skilled in the art according to actual conditions, and this application does not impose specific limitations.

[0066] For example, in this embodiment, a preset threshold can be set to 0.9, a preset time threshold can be set to 2 seconds, and the target door can be set to the tailgate. When the predicted probability value within the next 5 days is greater than 0.9, a timer is started to count the time the user stays in the tailgate area. When the stay time is greater than 2 seconds, the tailgate is triggered to open, and the timer is reset to zero. In addition, when the vehicle detects that the UWB tag is not in the tailgate area and t > 10 seconds, the tailgate is closed, without any manual operation required throughout the process.

[0067] The working principle of the door control method proposed in this application will be described below with reference to a specific embodiment.

[0068] in, Figure 3 This is a flowchart illustrating the working principle of a door control method according to an embodiment of this application.

[0069] Step S301: The user approaches the car with the UWB tag.

[0070] Step S302: Obtain the communication link status between the UWB anchor point and the UWB tag in the vehicle.

[0071] In this embodiment of the application, step S303 can be executed if the communication link is in a valid communication link state; otherwise, step S301 can be executed.

[0072] Step S303: Make a preliminary estimate of the actual distance.

[0073] Step S304: Identify the target door.

[0074] In this embodiment of the application, the user's trajectory data can be obtained based on the actual distance, and then the corresponding target door, such as the tailgate, can be determined based on the trajectory data.

[0075] Step S305: Analyze the trajectory data to obtain the corresponding trajectory features.

[0076] Trajectory features may include, but are not limited to, the following: , , and This application does not impose specific restrictions, etc.

[0077] Step S306: Use the pre-trained LSTM model to output the predicted probability value of the user walking towards the tailgate of the car.

[0078] The construction and training processes of the LSTM model are as described above, and will not be elaborated further here.

[0079] Step S307: When the user enters the tailgate area, start the timer and record the dwell time.

[0080] Step S308: When the predicted probability value is greater than 0.9 and the dwell time is greater than 2s, generate an opening command for the car tailgate.

[0081] Step S309: If the user is not in the tailgate area and stays there for more than 10 seconds, close the car tailgate.

[0082] The door control method proposed in this application can acquire user trajectory data based on positioning information sent by a positioning tag, extract corresponding trajectory features from the trajectory data, and then use a pre-trained LSTM model to output a predicted probability value of the user walking towards the target door in the car. When the predicted probability value is greater than a preset threshold, the user's dwell time in the corresponding target door area is counted. If the dwell time is greater than a preset time threshold, an opening command for the target door is generated, opening the target door. This eliminates the need for the user to actively initiate any specific action. By using positioning information and the LSTM model to predict the probability of the user walking towards the target door, combined with the dwell time, the opening command is generated. This reduces the requirements for the accuracy and standardization of actions, decreases recognition failures, and is less susceptible to environmental interference, greatly improving the user experience. Therefore, it solves the problems in related technologies where all actions are actively initiated by the user, requiring high accuracy and standardization of actions, easily leading to recognition failures, and being easily affected by environmental interference, thus impacting the user experience.

[0083] Next, the control device for a car door according to an embodiment of this application is described with reference to the accompanying drawings.

[0084] Figure 4 This is a block diagram of a door control device provided according to an embodiment of this application.

[0085] like Figure 4 As shown, the control device 10 of the car door includes: a first extraction module 100, an output module 200 and a control module 300.

[0086] The first extraction module 100 is used to obtain the user's trajectory data based on the location information sent by the location tag, and extract the corresponding trajectory features from the trajectory data.

[0087] Output module 200 is used to input trajectory features into a pre-trained LSTM model to output the predicted probability value of a user walking toward the target door in a car.

[0088] The control module 300 is used to count the time a user stays in the corresponding target door area when the predicted probability value is greater than a preset threshold, and generate an opening command for the target door when the stay time is greater than a preset time threshold, and control the car to open the target door according to the opening command.

[0089] Optionally, in one embodiment of this application, it further includes: a first building module, a second building module, a third building module, a fourth building module, a fifth building module, and a sixth building module.

[0090] The first building module is used to construct the input layer of the LSTM model based on the target trajectory features before inputting the trajectory features into the pre-trained LSTM model.

[0091] The second building module is used to build the first bidirectional LSTM layer in the LSTM model according to a preset first number of neurons, wherein the first bidirectional LSTM layer includes at least one forward LSTM sub-layer and one backward LSTM sub-layer.

[0092] The third building module is used to construct the random deactivation layer in the LSTM model according to the preset dropout rate.

[0093] The fourth building module is used to build the second bidirectional LSTM layer in the LSTM model according to the preset second number of neurons. The second bidirectional LSTM layer includes at least one forward LSTM sub-layer and one backward LSTM sub-layer.

[0094] The fifth building module is used to build the fully connected layers in the LSTM model and to build the output layer in the LSTM model based on the preset activation function.

[0095] The sixth building block is used to construct an LSTM model based on the input layer, the first bidirectional LSTM layer, the random deactivation layer, the second bidirectional LSTM layer, the fully connected layer, and the output layer.

[0096] Optionally, in one embodiment of this application, it further includes: a second extraction module, a seventh construction module, and a training module.

[0097] The second extraction module is used to obtain the target user's target trajectory data based on the location information sent by the location tag before inputting the trajectory features into the pre-trained LSTM model, and to extract the corresponding target trajectory features from the target trajectory data.

[0098] The seventh building block is used to construct the training dataset for the LSTM model based on the target trajectory features.

[0099] The training module is used to train the LSTM model based on the training dataset until the LSTM model meets the preset training conditions, so as to obtain a trained LSTM model.

[0100] Optionally, in one embodiment of this application, the first extraction module 100 includes a processing unit.

[0101] The processing unit is used to process the trajectory data when the trajectory data does not meet the preset data conditions until processed trajectory data that meets the preset data conditions is obtained.

[0102] Optionally, in one embodiment of this application, the first extraction module 100 includes: a first acquisition unit, a calculation unit, and a second acquisition unit.

[0103] The first acquisition unit is used to acquire the communication link status between the positioning anchor point and the positioning tag in the vehicle.

[0104] The calculation unit is used to calculate the actual distance between the vehicle and the positioning tag when the communication link is in a valid communication link state.

[0105] The second acquisition unit is used to acquire trajectory data based on the actual distance.

[0106] It should be noted that the foregoing explanation of the control method embodiment for the car door also applies to the control device for the car door in this embodiment, and will not be repeated here.

[0107] The door control device proposed in this application can acquire user trajectory data based on positioning information sent by a positioning tag, extract corresponding trajectory features from the trajectory data, and then use a pre-trained LSTM model to output a predicted probability value of the user walking towards the target door in the car. When the predicted probability value is greater than a preset threshold, the device counts the user's dwell time in the corresponding target door area. If the dwell time is greater than a preset time threshold, an opening command for the target door is generated, thus opening the target door. This eliminates the need for the user to actively initiate any specific action. By using positioning information and an LSTM model to predict the probability of the user walking towards the target door, combined with the dwell time, the opening command is generated. This reduces the requirements for the accuracy and standardization of actions, decreases recognition failures, and is less susceptible to environmental interference, greatly improving the user experience. Therefore, it solves the problems in related technologies where all actions are actively initiated by the user, requiring high accuracy and standardization of actions, easily leading to recognition failures, and being easily affected by environmental interference, thus impacting the user experience.

[0108] Figure 5 This is a schematic diagram of a vehicle structure according to an embodiment of this application. The vehicle may include: The memory 501, the processor 502, and the computer program stored on the memory 501 and capable of running on the processor 502.

[0109] When the processor 502 executes the program, it implements the door control method provided in the above embodiments.

[0110] Furthermore, automobiles also include: Communication interface 503 is used for communication between memory 501 and processor 502.

[0111] The memory 501 is used to store computer programs that can run on the processor 502.

[0112] Memory 501 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0113] If the memory 501, processor 502, and communication interface 503 are implemented independently, then the communication interface 503, memory 501, and processor 502 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0114] Optionally, in a specific implementation, if the memory 501, processor 502, and communication interface 503 are integrated on a single chip, then the memory 501, processor 502, and communication interface 503 can communicate with each other through an internal interface.

[0115] Processor 502 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0116] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described door control method.

[0117] This application also provides a computer program product, including a computer program that, when executed, implements the above-described door control method.

[0118] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0119] Furthermore, 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 technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0120] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0121] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0122] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or more of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0123] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0124] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0125] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A control method of a vehicle door, characterized by, Includes the following steps: Based on the location information sent by the location tag, the user's trajectory data is obtained, and the corresponding trajectory features are extracted from the trajectory data. The trajectory features are input into a pre-trained Long Short-Term Memory (LSTM) network model to output the predicted probability value of the user walking toward the target door in the car. If the predicted probability value is greater than a preset threshold, the user's dwell time in the corresponding target door area is counted, and if the dwell time is greater than the preset time threshold, an opening command for the target door is generated, and the car is controlled to open the target door according to the opening command.

2. The method of claim 1, wherein, Before inputting the trajectory features into the pre-trained LSTM model, the following steps are also included: Based on the target trajectory features, the input layer of the LSTM model is constructed; Construct the first bidirectional LSTM layer in the LSTM model according to a preset first number of neurons, wherein the first bidirectional LSTM layer includes at least one forward LSTM sub-layer and one reverse LSTM sub-layer; Construct a random deactivation layer in the LSTM model according to a preset dropout rate; Construct a second bidirectional LSTM layer in the LSTM model according to a preset second number of neurons, wherein the second bidirectional LSTM layer includes at least one forward LSTM sublayer and one reverse LSTM sublayer; Construct the fully connected layer in the LSTM model, and construct the output layer in the LSTM model based on the preset activation function; The LSTM model is constructed based on the input layer, the first bidirectional LSTM layer, the random deactivation layer, the second bidirectional LSTM layer, the fully connected layer, and the output layer.

3. The method according to claim 2, characterized in that, Before inputting the trajectory features into the pre-trained LSTM model, the following steps are also included: Based on the location information sent by the location tag, the target user's target trajectory data is obtained, and the corresponding target trajectory features are extracted from the target trajectory data. Based on the target trajectory features, a training dataset for the LSTM model is constructed; Based on the training dataset, the LSTM model is trained until it meets the preset training conditions to obtain a trained LSTM model.

4. The method according to claim 1, characterized in that, The acquisition of the user's trajectory data includes: If the trajectory data does not meet the preset data conditions, the trajectory data is processed until processed trajectory data that meets the preset data conditions is obtained.

5. The method according to claim 1, characterized in that, The acquisition of the user's trajectory data based on the location information sent by the location tag includes: Obtain the communication link status between the positioning anchor point in the vehicle and the positioning tag; If the communication link is in a valid communication link state, calculate the actual distance between the vehicle and the positioning tag; The trajectory data is obtained based on the actual distance.

6. A control device for a vehicle door, characterized in that, include: The first extraction module is used to obtain the user's trajectory data based on the location information sent by the location tag, and extract the corresponding trajectory features from the trajectory data. The output module is used to input the trajectory features into a pre-trained LSTM model to output the predicted probability value of the user walking towards the target door in the car. The control module is used to count the time the user stays in the corresponding target door area when the predicted probability value is greater than a preset threshold, and generate an opening command for the target door when the stay time is greater than a preset time threshold, and control the car to open the target door according to the opening command.

7. The apparatus according to claim 6, characterized in that, Also includes: The first construction module is used to construct the input layer of the LSTM model based on the target trajectory features before inputting the trajectory features into the pre-trained LSTM model; The second construction module is used to construct the first bidirectional LSTM layer in the LSTM model according to a preset first number of neurons, wherein the first bidirectional LSTM layer includes at least one forward LSTM sub-layer and one reverse LSTM sub-layer. The third construction module is used to construct the random deactivation layer in the LSTM model according to a preset dropout rate; The fourth construction module is used to construct the second bidirectional LSTM layer in the LSTM model according to a preset second number of neurons, wherein the second bidirectional LSTM layer includes at least one forward LSTM sub-layer and one backward LSTM sub-layer. The fifth construction module is used to construct the fully connected layer in the LSTM model and to construct the output layer in the LSTM model based on a preset activation function; The sixth construction module is used to construct the LSTM model based on the input layer, the first bidirectional LSTM layer, the random deactivation layer, the second bidirectional LSTM layer, the fully connected layer, and the output layer.

8. A car, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the door control method as described in any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the door control method as described in any one of claims 1-5.

10. A computer program product, characterized in that, Includes a computer program, which, when executed, is used to implement the door control method as described in any one of claims 1-5.

Images