Vehicle door opening interaction control method and system, medium, equipment and vehicle

Abstract

The invention discloses a vehicle door opening interaction control method and system, a medium, equipment and a vehicle, and relates to the technical field of intelligent interaction. The method comprises the following steps: acquiring user historical preference information and surrounding real-time environment information; learning the historical preference information of the user and the surrounding real-time environment information by using a deep learning network to obtain a preliminary vehicle door opening strategy; the initial vehicle door opening strategy is fed back to the current user, and information interaction is carried out; an interactive feedback network is used for processing the information interaction content, the user historical preference information and the surrounding real-time environment information, and a middle vehicle door opening strategy is obtained; and continuously feeding back the middle vehicle door opening strategy to the current user until the user confirms. According to the invention, interaction with the user can be carried out according to the environment information and historical preference information collected in real time, user demand reasoning is carried out in real time by using the interactive feedback network according to the interaction content, and a door opening strategy capable of balancing the user preference and the actual situation is obtained.

Description

Technical Field

[0001] The present invention relates to the field of intelligent interaction technology, and in particular to a vehicle door opening interactive control method, system, medium, equipment and vehicle. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In modern automotive design, door opening has become a crucial aspect in enhancing the user experience and the vehicle's sense of technology. Currently, some high-end models utilize a combination of scissor doors and swing doors. This hybrid opening method satisfies users' demands for personalization and convenience to a certain extent. However, existing technologies still have many shortcomings in terms of flexibility, safety, intelligence, and user experience of door control systems.

[0004] Existing door control systems have significant limitations in terms of door opening and closing flexibility. Users are unable to instantly switch opening modes or adjust door positions while the door is in motion, which is extremely inconvenient in real-world scenarios. For example, when a user needs to fine-tune the door opening angle in a narrow space to avoid a collision, existing door control systems often fail to meet this need. Furthermore, in emergency situations, some door systems are unable to quickly unlock or switch opening modes, potentially posing a threat to passenger safety.

[0005] While existing power door systems are typically equipped with radar to detect obstacles in the door's opening or closing path, these radars are limited in their application scenarios. Current radar systems primarily focus on simple distance detection and lack the ability to implement more complex functions that enhance the user experience. For example, radar systems cannot accurately determine and respond based on factors such as obstacle type, speed, and direction of movement.

[0006] To improve safety, existing door systems are often equipped with sensors, such as ultrasonic sensors or millimeter-wave radars, to detect obstacles in the door's opening or closing path. However, these sensors still have limitations. For example, one door control system uses millimeter-wave radar to detect obstacles around the vehicle and determine whether the door can be opened based on the distance and speed of the obstacle. However, in complex scenarios, this single criterion often fails to accurately reflect the actual situation, thus affecting the safety and reliability of the door system.

[0007] Some door systems attempt to identify the surrounding environment using cameras and image processing algorithms, but existing technologies are limited in their ability to understand and adapt to complex scenarios. For example, in low-light conditions, complex backgrounds, or scenes with multiple dynamic objects, existing image processing algorithms may not accurately identify obstacles and the surrounding environment, leading to misjudgments or delayed responses by the door system.

[0008] In addition, the current door opening method and angle generation strategy has a single consideration approach, which only considers the surrounding environment or only considers historical user preferences. It cannot interact deeply with the current user based on the actual situation of the surrounding environment, and thus cannot obtain a door opening method that balances user preferences and the actual environment.

[0009] In summary, existing vehicles with multi-mode door opening urgently need a method that can comprehensively obtain surrounding environment information and user preference information and can deeply interact with the current user to generate the best door control strategy to meet user needs. Summary of the Invention

[0010] In response to the shortcomings of the existing technology, the purpose of the present invention is to provide a method, system, medium, equipment and vehicle for interactive control of vehicle door opening, which can interact with the user based on real-time collected environmental information and historical preference information, and use an interactive feedback network to infer user needs in real time based on the interaction content, and obtain a door opening strategy that can balance user preferences and actual conditions.

[0011] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0012] A first aspect of the present invention provides a vehicle door opening interactive control method, comprising the following steps:

[0013] Obtain user historical preference information and surrounding real-time environment information;

[0014] A deep learning network is used to learn user historical preference information and surrounding real-time environmental information to obtain a preliminary door opening strategy;

[0015] Feedback the preliminary door opening strategy to the current user for information interaction;

[0016] The interactive feedback network is used to process information interaction content, user historical preference information and surrounding real-time environmental information to obtain the middle door opening strategy;

[0017] The middle door opening strategy is fed back to the current user for confirmation. If the current user confirms, the middle door opening strategy is used as the final door opening strategy. If the current user continues to interact, the new interaction content is re-input into the interactive feedback network for processing until the current user confirms.

[0018] A second aspect of the present invention provides a vehicle door opening interactive control system, comprising:

[0019] A data acquisition module is configured to acquire user historical preference information and surrounding real-time environment information;

[0020] An information learning module is configured to use a deep learning network to learn user historical preference information and surrounding real-time environmental information to obtain a preliminary door opening strategy;

[0021] An information interaction module is configured to feed back a preliminary door opening strategy to the current user for information interaction;

[0022] An interactive feedback module is configured to process information interaction content, user historical preference information, and surrounding real-time environmental information using an interactive feedback network to obtain a center door opening strategy;

[0023] The strategy generation module is configured to continue to feed back the middle door opening strategy to the current user for confirmation. If the current user confirms, the middle door opening strategy will be used as the final door opening strategy. If the current user continues to interact, the new interaction content will be re-input into the interactive feedback network for processing until the current user confirms.

[0024] A third aspect of the present invention provides a medium having a program stored thereon, which, when executed by a processor, implements the steps of the vehicle door opening interactive control method as described in the first aspect of the present invention.

[0025] The fourth aspect of the present invention provides a device comprising a memory, a processor and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the vehicle door opening interactive control method as described in the first aspect of the present invention are implemented.

[0026] A fifth aspect of the present invention provides a vehicle, comprising the vehicle door opening interactive control system described in the first aspect of the present invention.

[0027] One or more of the above technical solutions have the following beneficial effects:

[0028] The present invention discloses a method, system, medium, device and vehicle for interactive control of door opening. By acquiring the user's historical preference information, the method can learn the user's preference habits for door opening modes and opening angles in different scenarios. The learning ability based on historical data enables the door opening strategy to accurately match the user's personalized needs, significantly improving the user experience. Combined with the surrounding real-time environmental information, it can intelligently judge the optimal door opening strategy in the current scenario. The present invention feeds back the preliminary door opening strategy to the user and processes the user's real-time feedback information through an interactive feedback network. The user can adjust the preliminary strategy according to his or her needs, such as requiring a larger opening angle or switching the opening mode. The system can respond to the user's interactive instructions in real time and dynamically update the door opening strategy, so that the user can feel a high degree of autonomy and flexibility during use.

[0029] Through multiple rounds of user interaction, the present invention can gradually optimize the door opening strategy until the user confirms their satisfaction. This multi-round interaction mechanism not only ensures that the final strategy accurately meets user needs, but also further optimizes the system's learning model through real-time user feedback, enhancing the system's intelligence level. Before finalizing the door opening strategy, the system requires user confirmation. This mechanism ensures that the user has complete control over the door opening method, avoiding potential risks caused by system misjudgment or user non-awareness. For example, users can fine-tune the intermediate strategy provided by the system to ensure that the door opening method fully meets the current scenario and their own needs, further improving system reliability and user trust.

[0030] The entire door opening strategy generation and interaction process in this invention is based on the collaborative work of a deep learning network and an interactive feedback network. This process integrates real-time user interaction content, historical preference information, and real-time environmental information to dynamically adjust the door opening strategy. This data-driven intelligent design not only improves system performance but also provides a solid foundation for future functional expansion and optimization.

[0031] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0033] Figure 1 This is a flow chart of the door opening interactive control method in the first embodiment of the present invention;

[0034] Figure 2 This is a flow chart of the user preference learning process in Example 1 of the present invention. DETAILED DESCRIPTION

[0035] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0036] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations;

[0037] Example 1:

[0038] The first embodiment of the present invention provides a door opening interactive control method, such as Figure 1 As shown, the following steps are included:

[0039] Step 1: Obtain user historical preference information and surrounding real-time environment information.

[0040] Step 1.1: Obtain real-time surrounding environment information.

[0041] In this embodiment, data acquisition equipment is installed on the vehicle to obtain real-time information about the surrounding environment, including the location, size, speed, and acceleration of obstacles. Specifically, the data acquisition equipment is responsible for real-time monitoring of the vehicle's surroundings, particularly obstacles in the door opening path. The data acquisition equipment utilizes high-precision ultrasonic, millimeter-wave, or other sensors, as well as cameras, to ensure accurate detection in a variety of environmental conditions.

[0042] The radar detection module is used for obstacle detection and distance measurement. It identifies obstacles in the door opening path, such as pedestrians, bicycles, and walls. It accurately measures the distance between the obstacle and the door, providing data support for mode switching. The radar detection module maintains stable detection performance in various weather conditions and lighting environments.

[0043] Step 1.2: Obtain user historical preference information.

[0044] This embodiment collects and analyzes user door-opening habits to learn user preferences. This module leverages machine learning algorithms to conduct in-depth analysis of user behavior data to predict and adapt to user needs. User preference information includes historical door opening modes, historical opening angles, and user preference analysis. Door opening modes include scissor doors and sliding doors.

[0045] The process of obtaining user historical preference information is as follows: Figure 2 As shown, the user can manually select the opening method. When the user manually selects, the system records and analyzes the user's selection to optimize future automatic mode switching. This manual selection can override the automatic mode switching to meet the user's needs in specific situations. When the user manually selects the opening method, it counts and compares the number of times different modes have been manually selected in the past. The mode that has been selected more times is recorded as the habitual mode. When the user manually opens the car door, the current number and the current angle recorded value are compared with the historical record. The smaller value is recorded as the minimum angle that the user can exit the door in the current opening mode. This process helps the system better understand the user's habits and needs.

[0046] Step 2: Use the deep learning network to learn the user's historical preference information and the surrounding real-time environment information to obtain a preliminary door opening strategy.

[0047] Step 2.1: Build a deep learning network and train the deep learning network.

[0048] Step 2.1.1: The deep learning network consists of a convolutional neural network (CNN) and a recurrent neural network (RNN).

[0049] The convolutional neural network consists of a combination of multiple convolutional layers (Conv2D) and pooling layers (MaxPooling2D) to extract features from the image. It also includes a flattening layer (Flatten) and a fully connected layer (Dense). The flattening layer is used to convert the multi-dimensional convolutional feature map into a one-dimensional vector for input to the subsequent fully connected layer. The fully connected layer is used to further process the image features and output a feature vector related to the environment. The convolutional neural network extracts key features from the camera image, such as the presence of obstacles around the car door, the type of obstacles, and their location. The complex information in the image is converted into a feature vector that can be processed by the subsequent network, providing an environmental perception basis for the generation of the door opening strategy.

[0050] In this embodiment, multiple convolutional layers and pooling layers are combined into three layers. The first convolutional layer is used to extract low-level features such as obstacle edges and textures. The second pooling layer is used to reduce feature dimensionality and reduce computational complexity. The third convolutional layer is used to further extract high-level features such as obstacle shapes.

[0051] The input to the recurrent neural network includes encoded user preference information, such as default activation mode and historical preferred angles, radar data, and camera image features extracted by a convolutional neural network. The user preference information vector, the environmental feature vector output by the CNN, and the radar data vector are concatenated to form a comprehensive input vector.

[0052] GRUs are used as recurrent layers to process time series data. Radar data changes over time, such as the speed and direction of obstacles. GRU layers are able to capture the dynamic changes in these time series data. The GRU's internal state remembers information from previous time steps, allowing it to better understand the context of the current state. This is crucial for handling door opening strategies in dynamic environments.

[0053] Capturing the dynamic relationship between user preferences and environmental information. By combining user preferences with environmental features, the recurrent layer can learn user behavior patterns in different scenarios. For example, a user's preference for door opening angle varies in different environments (such as parking lots and narrow streets).

[0054] Finally, a fully connected layer (Dense) is used to further process the recurrent layer output, extracting key features and converting the recurrent layer output into a higher-level feature representation. The processed feature vector is then passed to the fusion layer to be combined with the output of the CNN module to obtain a preliminary door opening strategy.

[0055] Step 2.1.2: Obtain environmental data and user preference data from historical records or public datasets to form a dataset, and preprocess the data in the dataset.

[0056] In this embodiment, the preprocessing step includes encoding the user preference information into a numerical vector. For example, the scissor door mode is encoded as [1, 0, 0], the swing door mode is encoded as [0, 1, 0], and the combination mode is encoded as [0, 0, 1]. The angle preference is encoded according to intervals.

[0057] Normalize the radar data, including the distance, speed, and direction of obstacles. Perform preprocessing operations such as cropping, scaling (e.g., 224×224 pixels), and normalization on the camera image.

[0058] Step 2.1.3: Divide the dataset into a training set and a test set, use the training set to train the deep learning network, and use the test set to test the trained deep learning network.

[0059] In this embodiment, a deep learning network is trained using training data. The network weights are adjusted using a backpropagation algorithm to ensure that the network output is as close as possible to the actual door opening strategy. The prediction error of the opening pattern is measured using a cross-entropy loss function, while the prediction error of the opening angle is measured using a mean squared error loss function.

[0060] Step 2.2: Use deep learning networks to learn user historical preference information and surrounding real-time environment information.

[0061] Deep learning is performed on the user's manual door opening position, analyzing the angle data when the user manually opens the door, determining the minimum angle required for the user to enter and exit, and recording the most frequently selected mode as the preferred mode. This strategy helps the system continuously learn and adapt to user habits, providing more accurate and personalized services in future use.

[0062] In a specific implementation, radar can be used to detect the possible opening angles under different opening modes. User preferences are primarily responsible for recording the user's preferred door opening mode and the angles that are convenient for the user to enter and exit. These two factors work together to enhance the user experience. Deep learning is performed on the user's manual door opening position, analyzing the angle data when the user manually opens the door to determine the minimum angle required for the user to enter and exit. The minimum angle is then used to determine if there are obstacles in the parking location and control the door opening method.

[0063] Specifically, radar data is processed to extract key obstacle features, such as distance, speed, and direction of motion (e.g., approaching, moving away, or stationary). Camera image information is processed using a convolutional neural network (CNN) to extract key objects in the environment (e.g., pedestrians, vehicles, etc.) and their positional relationships. User preference information features, radar features, and image features are fused. Weighted summation, splicing, or more complex fusion strategies can be employed, adjusting the weights of each component based on actual needs, allowing the network to comprehensively consider user preferences and environmental factors.

[0064] A recurrent neural network (RNN) is used to process the fused feature sequence. This is because the door opening process requires consideration of time series information, such as the user's continuous interaction commands and the dynamic changes of obstacles. At the end of the decision layer, a fully connected layer outputs the probability distribution of the door opening mode and angle.

[0065] Step 3: Feedback the preliminary door opening strategy to the current user for information interaction.

[0066] In a specific embodiment, the preliminary door opening strategy generated by the system does not necessarily fully meet the actual door opening requirements, and does not necessarily meet the needs of the current user (different users or the user's current mood, etc.). For example, the system recommends using a swing door, but the user finds that the surrounding space is small and requires switching to a scissor door mode. Or the system recommends using a scissor door based on the surrounding environment, but the user wants to use a swing door in the current scenario to make it easier to get on and off the vehicle. Or the preliminary door opening angle generated by the system does not meet the user's needs, and the user requires increasing or decreasing the opening angle. And in different weather conditions, users have different needs. For example, on rainy days, users want a larger door opening angle to quickly get on and off the vehicle and reduce rainwater entering the vehicle.

[0067] In this embodiment, the user sends a corresponding demand instruction through the interactive interface in the vehicle display screen, and the vehicle control system regenerates a new door opening strategy based on the instruction sent by the user and the previous door opening strategy.

[0068] Step 4: Use the interactive feedback network to process the information interaction content, user historical preference information and surrounding real-time environmental information to obtain the middle door opening strategy.

[0069] Step 4.1: Encode user real-time interaction instructions.

[0070] Step 4.2: By introducing a lightweight attention mechanism and feature embedding, an interactive feedback network is used to perform multimodal feature fusion processing on radar data, image features, user preference features, and user interaction instructions.

[0071] In a specific embodiment, to ensure the real-time performance of the interaction and door opening processes, the interactive feedback network embeds data from different modalities into a shared feature space for easy fusion. A lightweight attention mechanism is also introduced to reduce computational complexity while dynamically assigning importance weights to data from different modalities. The lightweight attention mechanism improves the real-time performance of the system by reducing the computational complexity of the attention mechanism. At the same time, dynamically assigning importance weights to data from different modalities ensures that the system can flexibly adjust based on the current scenario and user interaction. Finally, a fully connected layer is used to further process the fused feature vector. Ultimately, the probability distributions for scissor doors, swing doors, and combined modes are output, as well as the predicted value of the door opening angle.

[0072] Preprocessed radar data, image features, user preference features, and user interaction commands are embedded into a shared feature space through fully connected layers, reducing computational complexity. A lightweight attention mechanism dynamically assigns importance to data of different modalities. For example, when user interaction commands are clear, the system pays more attention to them; when obstacles are present in the environment, the system pays more attention to radar data and image features.

[0073] Specifically, the purpose of the lightweight attention mechanism is to dynamically assign the importance weights of different modal data without increasing too much computational complexity. The specific structure is as follows:

[0074] The feature vectors of different modalities are taken as input, and the feature vectors of different modalities include radar data features Fr, image features Fi, user preference features Fu and user interaction command features Fui.

[0075] Then, the feature vectors of different modalities are embedded into a shared feature space through a fully connected layer to obtain the embedded feature vectors: radar data feature vector Er, image feature vector Ei, user preference feature vector Eu and user interaction instruction feature vector Eui.

[0076] The weights of different modalities are calculated through a lightweight fully connected layer, denoted as Wa. The weight calculation formula is as follows:

[0077] Wa=softmax(Da·Concat(Er,Ei,Eu,Eui)).

[0078] Among them, Da is the weight matrix of the fully connected layer, and Concat means concatenating the embedded feature vectors of different modalities.

[0079] According to the calculated weight Wa, the embedded feature vectors of different modes are weighted and summed to obtain the comprehensive feature vector Ff:

[0080] Ff=Wa·Concat(Er,Ei,Eu,Eui).

[0081] Data from different modalities is fused through weighted summation to generate a comprehensive feature vector. This fused feature vector is further processed through a fully connected layer to extract key features. The output layer generates the final door opening strategy, including the opening mode and angle.

[0082] This embodiment reduces the computational complexity of multimodal information fusion and significantly improves the real-time performance of the system by introducing a lightweight attention mechanism and feature embedding. The lightweight attention mechanism can dynamically adjust the importance weights of different modal data based on the current scenario and user interaction, ensuring that the system can flexibly adapt to user needs and environmental changes. By optimizing multimodal information fusion and dynamic weight allocation, the system can more quickly generate door opening strategies that meet user needs and enhance user trust in the system. This improved network structure and data processing process can significantly improve the real-time and intelligent level of the system, and is a key technology for achieving dynamic optimization and personalized adjustment of door opening strategies.

[0083] Step 5: The middle door opening strategy is fed back to the current user for confirmation. If the current user confirms, the middle door opening strategy is used as the final door opening strategy. If the current user continues to interact, the new interaction content is re-input into the interactive feedback network for processing until the current user confirms.

[0084] A practical scenario simulation is now provided to facilitate understanding of the method of this embodiment:

[0085] When a user drives into a narrow underground parking lot, the system needs to generate a door opening strategy based on the user's historical preferences and the current environment. After entering the parking lot, the user may adjust the door opening mode and angle based on the actual situation.

[0086] The default opening mode is swing. Historically preferred angle: In confined spaces, an opening angle of 30°-45° is typically selected. The radar detected other vehicles on either side of the door, relatively close (approximately 0.5 meters). The camera image indicates limited space around the door. The vehicle is detected in an underground parking lot.

[0087] The interaction process is as follows:

[0088] First round of interaction: The system generates a preliminary door opening strategy:

[0089] Opening mode: swing door (based on user historical preference)

[0090] Opening angle: 30° (based on user historical preferences and current narrow environment)

[0091] Feedback to users: The system prompts users through the in-car display: "It is recommended to use a flat door with an opening angle of 30°."

[0092] User feedback:

[0093] Users provide feedback via the in-car control panel or mobile app: "The opening angle is too small, I need a wider angle to get in and out of the car."

[0094] The system records the interaction content:

[0095] The user requested a larger opening angle, but did not specify a specific angle.

[0096] Second round of interaction: Interactive feedback network processing: Combining user feedback (request to increase the angle), historical preferences (sliding door) and current environment (narrow space), the interactive feedback network regenerates the middle door opening strategy.

[0097] Adjusted strategy: Opening mode: Swing door (remains unchanged because the user did not request to switch modes)

[0098] Opening angle: 45° (adjusted based on user feedback and current environment)

[0099] System feedback to users:

[0100] The system prompts the user through the in-car display: "Based on your feedback, the opening angle has been adjusted to 45°.

[0101] Please confirm whether you are satisfied.

[0102] User feedback:

[0103] The user confirmed: "The angle is OK, but there are vehicles approaching. I'm worried about a collision. Please reduce the angle."

[0104] The system records the interaction content:

[0105] The user confirms the angle but requests that it be reduced to avoid a collision.

[0106] The third round of interaction: Interactive feedback network processing: Combining user feedback (fear of collision), historical preferences (opening doors), and the current environment (narrow space with approaching vehicles), the interactive feedback network regenerates the middle door opening strategy.

[0107] Adjusted strategy:

[0108] Opening mode: swing door (remain unchanged)

[0109] Opening angle: 35° (dynamically adjusted based on user feedback and obstacles detected by radar)

[0110] System feedback to users:

[0111] The system prompts the user through the in-car display: "The opening angle has been adjusted to 35° to avoid collision.

[0112] Please confirm whether you are satisfied.

[0113] User feedback:

[0114] The user confirmed: "The angle is correct, no other issues."

[0115] The system records the interaction content:

[0116] The user confirms the final policy.

[0117] Final door opening strategy:

[0118] Opening mode: swing door

[0119] Opening angle: 35°

[0120] System execution: The door control system performs the opening operation according to the final strategy.

[0121] Example 2:

[0122] A second embodiment of the present invention provides a vehicle door opening interactive control system, comprising:

[0123] A data acquisition module is configured to acquire user historical preference information and surrounding real-time environment information;

[0124] An information learning module is configured to use a deep learning network to learn user historical preference information and surrounding real-time environmental information to obtain a preliminary door opening strategy;

[0125] An information interaction module is configured to feed back a preliminary door opening strategy to the current user for information interaction;

[0126] An interactive feedback module is configured to process information interaction content, user historical preference information, and surrounding real-time environmental information using an interactive feedback network to obtain a center door opening strategy;

[0127] The strategy generation module is configured to continue to feed back the middle door opening strategy to the current user for confirmation. If the current user confirms, the middle door opening strategy will be used as the final door opening strategy. If the current user continues to interact, the new interaction content will be re-input into the interactive feedback network for processing until the current user confirms.

[0128] Example 3:

[0129] A third embodiment of the present invention provides a medium on which a program is stored. When the program is executed by a processor, the steps of the vehicle door opening interactive control method as described in the first embodiment of the present invention are implemented.

[0130] Example 4:

[0131] Embodiment 4 of the present invention provides a device comprising a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in the vehicle door opening interactive control method as described in embodiment 1 of the present invention are implemented.

[0132] Embodiment 5:

[0133] A fifth embodiment of the present invention provides a vehicle, comprising the door opening interactive control system described in the second embodiment of the present invention.

[0134] The steps involved in the above embodiments 2, 3, 4 and 5 correspond to those in the method embodiment 1. For the specific implementation methods, please refer to the relevant description part of the embodiment 1.

[0135] Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using a general-purpose computer device. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. Alternatively, they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0136] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A door opening interactive control method, characterized in that: The following steps are involved: Obtain user historical preference information and surrounding real-time environment information; A deep learning network is used to learn user historical preference information and surrounding real-time environmental information to obtain a preliminary door opening strategy; Feedback the preliminary door opening strategy to the current user for information interaction; The interactive feedback network is used to process information interaction content, user historical preference information and surrounding real-time environmental information to obtain the middle door opening strategy; The middle door opening strategy is fed back to the current user for confirmation. If the current user confirms, the middle door opening strategy is used as the final door opening strategy. If the current user continues to interact, the new interaction content is re-input into the interactive feedback network for processing until the current user confirms.

2. The door opening interactive control method according to claim 1, characterized in that: The specific steps of using deep learning networks to learn user historical preference information and surrounding real-time environment information are as follows: Build and train a deep learning network, which consists of a convolutional neural network (CNN) and a recurrent neural network (RNN). The CNN is used to extract features from images, while the RNN is used to capture the dynamic relationship between user preferences and environmental information. Use deep learning networks to learn user historical preference information and surrounding real-time environment information.

3. The door opening interactive control method according to claim 1, characterized in that: The specific steps for training a deep learning network are: Obtain environmental data and user preference data from historical records or public data sets to form a data set, and preprocess the data in the data set; The dataset is divided into a training set and a test set. The training set is used to train the deep learning network, and the test set is used to test the trained deep learning network.

4. The door opening interactive control method according to claim 1, characterized in that: During the information interaction process, the user sends corresponding demand instructions through the interactive interface on the vehicle display screen. The vehicle control system regenerates a new door opening strategy based on the instructions sent by the user and the previous door opening strategy.

5. The door opening interactive control method according to claim 1, wherein: The specific process of using the interactive feedback network to process information interaction content, user historical preference information and surrounding real-time environmental information is as follows: Encode user real-time interaction instructions; By introducing a lightweight attention mechanism and feature embedding, an interactive feedback network is used to perform multimodal feature fusion processing on radar data, image features, user preference features and user interaction instructions.

6. The door opening interactive control method according to claim 5, characterized in that: Radar data, image features, user preference features, and user interaction instructions are embedded into a shared feature space through fully connected layers, and the importance weights of different modal data are dynamically assigned through a lightweight attention mechanism.

7. A door opening interactive control system, characterized in that: include: A data acquisition module is configured to acquire user historical preference information and surrounding real-time environment information; An information learning module is configured to use a deep learning network to learn user historical preference information and surrounding real-time environmental information to obtain a preliminary door opening strategy; An information interaction module is configured to feed back a preliminary door opening strategy to the current user for information interaction; An interactive feedback module is configured to process information interaction content, user historical preference information, and surrounding real-time environmental information using an interactive feedback network to obtain a center door opening strategy; The strategy generation module is configured to continue to feed back the middle door opening strategy to the current user for confirmation. If the current user confirms, the middle door opening strategy will be used as the final door opening strategy. If the current user continues to interact, the new interaction content will be re-input into the interactive feedback network for processing until the current user confirms.

8. A computer-readable storage medium, characterized in that A plurality of instructions are stored therein, and the instructions are suitable for being loaded by a processor of a terminal device and executed by the vehicle door opening interactive control method according to any one of claims 1-6.

9. A terminal device, characterized in that: The method comprises a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, wherein the instructions are suitable for being loaded by the processor and executing the vehicle door opening interactive control method according to any one of claims 1 to 6.

10. A vehicle comprising the door opening interactive control system according to claim 7.

Images