Parking control method, device and equipment and storage medium

Abstract

The invention provides a parking control method, device and equipment and a storage medium, and relates to the technical field of automatic driving. The method comprises the steps that a parking command text is determined, and a first image of the environment where a vehicle is located is obtained; the parking command text and the first image are input into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle, and the first parking planning result comprises a plurality of parking track points and vehicle advancing features corresponding to each parking track point; and controlling the vehicle to park based on the first parking planning result. Therefore, an efficient and practical automatic parking motion planning strategy can be generated based on the parking planning model, more complex and detailed parking tasks can be processed, effective reasoning is allowed according to images and texts during operation, and real-time control is achieved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and particularly to a parking control method, device, equipment, and storage medium. Background Art

[0002] Based on algorithms such as sensor data, machine learning, and path planning, the automatic parking technology can enable the vehicle to autonomously complete the parking operation. This technology senses the surrounding environment, extracts features, generates a parking path, and combines vehicle dynamics constraints for control to achieve precise parking.

[0003] However, in complex urban environments, there are challenges such as vehicle recognition, obstacle avoidance, and narrow space handling. In addition, due to factors such as sensor accuracy and model generalization, the robustness of current technologies still needs to be improved in extreme weather, complex road conditions, and other situations. Summary of the Invention

[0004] This application provides a parking control method, device, equipment, and storage medium, aiming to at least solve one of the technical problems in the related technologies to a certain extent.

[0005] In a first aspect, this application provides a parking control method, including:

[0006] Determine a parking command text, and obtain a first image of the environment where the vehicle is located;

[0007] Input the parking command text and the first image into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle, where the first parking planning result includes multiple parking trajectory points and the vehicle travel characteristics corresponding to each parking trajectory point;

[0008] Based on the first parking planning result, control the vehicle to park.

[0009] In a second aspect, this application provides a parking control device, including:

[0010] A determination module, configured to determine a parking command text and obtain a first image of the environment where the vehicle is located;

[0011] An input module, configured to input the parking command text and the first image into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle,

[0012] where the first parking planning result includes multiple parking trajectory points and the vehicle travel characteristics corresponding to each parking trajectory point;

[0013] A control module, configured to control the vehicle to park based on the first parking planning result.

[0014] In a third aspect, the present application provides an electronic device, including: a processor; and a memory for storing instructions executable by the processor; wherein, the processor is configured to execute the instructions to implement a parking control method.

[0015] In a fourth aspect, the present application provides a computer-readable storage medium, when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, enabling the electronic device to execute a parking control method.

[0016] In a fifth aspect, the present application provides a computer program product, including a computer program, and the computer program is executed by a processor to perform a parking control method.

[0017] In an embodiment of the present application, first, a parking command text is determined, and a first image of the vehicle's environment is obtained. Then, the parking command text and the first image are input into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle, where the first parking planning result includes multiple parking trajectory points and the vehicle traveling characteristics corresponding to each parking trajectory point. Finally, based on the first parking planning result, the vehicle is controlled to park. Thus, an efficient and practical automatic parking motion planning strategy can be generated based on the parking planning model, which can handle more complex and detailed parking tasks, allow effective reasoning based on images and texts during operation, achieve real-time control, and improve robustness.

[0018] Additional aspects and advantages of the present disclosure will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0020] Figure 1 is a flowchart showing the parking control method according to the first embodiment of the present application;

[0021] Figure 2 is a flowchart showing the parking control method according to the second embodiment of the present application;

[0022] Figure 3 is a flowchart showing the parking control method according to the third embodiment of the present application;

[0023] Figure 4 is a flowchart showing the parking control method according to the fourth embodiment of the present application;

[0024] Figure 5It is a block diagram of a parking control device shown according to the present application;

[0025] Figure 6 It shows a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present application.

[0026] Through the above-mentioned drawings, specific embodiments of the present application have been shown, and there will be more detailed descriptions hereinafter. These drawings and text descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. Specific Embodiments

[0027] The embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are only used to explain the present application and should not be construed as limiting the present application. On the contrary, the embodiments of the present application include all changes, modifications, and equivalents that fall within the spirit and scope of the appended claims.

[0028] It should be noted that the execution subject of the parking control method in this embodiment can be a parking control device, and this device can be implemented in a software and / or hardware manner and configured in a vehicle.

[0029] Figure 1 It is a schematic flowchart of a parking control method shown according to the first embodiment of the present application, as Figure 1 shown, the method includes:

[0030] S101: Determine the parking command text and obtain a first image of the environment where the vehicle is located.

[0031] Among them, the parking command text describes the parking process and parking location in language. For example, "Park in the first parking space on the left", "Drive forward one meter and then park in the vertical parking space next to the column", etc., which are not limited herein.

[0032] It should be noted that the vehicle can recognize the language command of the user through voice recognition to obtain the parking command text, or the parking command text can also be sent to the vehicle by the user through any terminal device, so that the vehicle can directly obtain the parking command text, or the parking command text can also be input by the user in the display screen of the vehicle, so that the vehicle can obtain the parking command text, etc., which are not limited herein.

[0033] Among them, the first image can be an image of the vehicle surrounding environment for guiding parking planning. Optionally, the first image can be a vehicle environment image acquired by at least one image acquisition device provided on the vehicle. The first image can be an image of the vehicle surrounding environment captured by all the image acquisition devices on the vehicle at a certain moment, or it can also be an image of the vehicle surrounding environment captured by some of the image acquisition devices on the vehicle at a certain moment. Or, the first image can also be an image of the vehicle surrounding environment captured by the image acquisition devices on the vehicle at multiple moments, which is not limited herein.

[0034] Optionally, the device can determine a parking command text corresponding to the parking instruction in response to receiving the parking instruction at the first time.

[0035] Among them, the first time can be the time when the vehicle receives the parking instruction, which can be a certain moment or a relatively short time period, which is not limited herein.

[0036] Among them, the parking instruction can be an instruction for instructing the vehicle to park. Or, the parking instruction can also be transmitted in a specific format or code, and the vehicle resolves it into an understandable parking command text after receiving the instruction for corresponding parking actions.

[0037] As a possible implementation manner, the vehicle can identify the user's voice to determine whether the user's language contains a parking instruction word or a parking instruction statement. If the vehicle identifies a parking instruction word or a parking instruction statement within the first time, it can trigger the vehicle to execute the parking task. The vehicle can obtain the parking command text by identifying the parking instruction word or the parking instruction statement. At the same time, the vehicle can control the image acquisition device installed on the vehicle to collect images of the surrounding environment of the vehicle within the first time until the parking task is completed.

[0038] Optionally, an image of the vehicle's environment at the first time can be collected as the first image.

[0039] For example, if the vehicle receives the user's parking instruction at time T1, it can collect an image of the vehicle's environment at time T1 and use it as the first image, which is not limited herein.

[0040] Or, an image of the vehicle's environment at the second time can also be collected as the first image, and the first time is the historical time of the second time.

[0041] For example, if the vehicle receives a parking instruction from the user at time T1, and as time changes to T2 (the second time), then an image of the vehicle's surrounding environment collected at T2 can be used as the first image, which is not limited here. The image of the vehicle's surrounding environment collected at T2 is used for parking path planning at T2.

[0042] Alternatively, an image of the vehicle's surrounding environment at a third time can also be collected as the first image. The third time includes the first time, the second time, and the time interval between the first time and the second time.

[0043] For example, if the current time is T2, then all the images collected by the vehicle between T1 and T2 (including the images collected at T1 and T2) can be used as the first image.

[0044] It can be understood that automatic parking is a sequential task. Suppose at time t = 0, the rider issues a parking instruction in text form (for example, park in the first bay on the left). At the same time, the image input at t = 0 is I_0. Given the parking instruction and the image input I_0, the embodiments of the present disclosure can then use the parking planning model to give the parking planning result a_0 at t = 0.

[0045] At time t = j, given the parking instruction and the historical image inputs I_0, I_1,..., I_j, the device can then give the motion planning result a_j at t = j. Here, t = j is the second time, and t = 0 is the first time. Alternatively, at time t = j, the device can also give the motion planning result a_j at t = j given the parking instruction and the image input I_j, which is not limited here.

[0046] S102: Input the parking command text and the first image into a pre-constructed parking planning model to obtain the first parking planning result of the vehicle.

[0047] In the embodiments of the present disclosure, the pre-constructed parking planning model is a neural network model that has been trained and is in an available state.

[0048] Optionally, the parking planning model can be constructed based on a pre-trained large image-text model. The large image-text model obtained through pre-training contains effective feature representations learned from large-scale image-text data. Based on the pre-trained model, a stronger transfer learning effect can be obtained, thus achieving better automatic parking performance.

[0049] As a possible implementation, a path language and an image model PaLI-X (Pathways Language and Image model) can be selected as the initial parking planning model for training until the model is trained to an available state and can then be used as the parking planning model.

[0050] In the embodiments of the present disclosure, a vision-language model VLM (Vision-Language Model) can be selected as the initial parking planning model for training. The VLM is a model that integrates vision and language and is designed to handle correlation tasks between images and texts. The goal of this model is to understand and generate natural language descriptions related to images, or to generate corresponding images from natural language descriptions. The VLM model is usually based on deep learning methods and combines technologies such as convolutional neural networks and recurrent neural networks. By training a large-scale image-text paired dataset, it learns the semantic relationship between images and texts. During the training process, the VLM model inputs images and texts into the network and performs feature extraction, semantic encoding, and joint representation through different layers and modules, enabling the model to capture the interaction and semantic correlation between images and texts, thereby realizing the functions of image description and image generation.

[0051] Among them, the first parking planning result contains multiple parking trajectory points and the vehicle movement characteristics corresponding to each parking trajectory point. Among them, the parking trajectory points can be the trajectory points between the parking starting point and the parking ending point. Among them, the parking starting point can be the position where the vehicle is currently located, and the parking ending point can be the target position of parking, that is, the final position.

[0052] For example, if the current coordinates of the vehicle are (x1, y1), the first parking planning result can be multiple discrete coordinate points (trajectory points) such as (x2, y2), (x3, y3), (x4, y4), (x5, y5)... (xn, yn), where (xn, yn) is the parking ending point.

[0053] Among them, the vehicle movement characteristics can be the speed, acceleration, angular velocity, turning radius, steering speed, steering angle, and the position of the vehicle relative to the center line of the lane of each parking trajectory point, such as the offset, approaching the left or right edge, trajectory characteristics, etc., which are not limited here.

[0054] It should be noted that the vehicle movement characteristics corresponding to different trajectory points can be different or the same, which are not limited here.

[0055] Among them, the first parking planning result, that is, the output of the model, can be a token representing the motion plan. The device can determine the coordinates of the parking trajectory points and the vehicle movement characteristics corresponding to the trajectory points by parsing the token.

[0056] S103: Based on the first parking planning result, control the vehicle to park.

[0057] Specifically, the device can perform specific parking actions according to the first parking planning result. For example, it can control the driving path of the vehicle according to the coordinates of each parking trajectory point, and can adjust the speed, acceleration, angular velocity, turning radius, steering speed, steering angle, and the position of the vehicle relative to the center line of the lane (such as offset, approaching the left or right edge, trajectory characteristics, etc.) according to the driving characteristics of the vehicle. The details are not limited herein.

[0058] Optionally, during the execution of the first parking planning result, the control system can also monitor the surrounding environment in real time through the vehicle's sensors to avoid collisions, so as to ensure the safety and accuracy of the parking operation. After the parking action is completed, the vehicle system can feedback the execution result to the parking planning system for subsequent adjustment or confirmation.

[0059] In the embodiment of the present application, first, a parking command text is determined, and a first image of the vehicle's environment is obtained. Then, the parking command text and the first image are input into a pre-constructed parking planning model to obtain the first parking planning result of the vehicle. The first parking planning result includes a plurality of parking trajectory points and the driving characteristics of the vehicle corresponding to each parking trajectory point. Finally, based on the first parking planning result, the vehicle is controlled to park. Thus, an efficient and practical automatic parking motion planning strategy can be generated based on the parking planning model, which can handle more complex and detailed parking tasks, allow effective reasoning based on images and texts during operation, and achieve real-time control.

[0060] Figure 2 is a schematic flowchart of a parking control method shown in the second embodiment of the present application, as Figure 2 shown, the method includes:

[0061] S201: Determine a parking command text and obtain a first image of the vehicle's environment.

[0062] It should be noted that the specific implementation manner of step S201 can refer to the above embodiment and will not be elaborated herein.

[0063] S202: Obtain a training data set, which includes a plurality of parking text sets and a plurality of parking driving image sets.

[0064] Each parking text set includes a plurality of parking command texts, and each parking text set and each parking driving image set correspond to a parking task.

[0065] It should be noted that a parking task corresponds to different parking command texts. It should be noted that since different passengers often have differences in the way of describing a parking process or a parking location, while the actual parking process of the vehicle is the same. Therefore, various parking command texts corresponding to different parking tasks can be collected in advance.

[0066] For example, "Park in the first bay on the left", "Park in bay 1", "Park in the leftmost bay", and "Park at the leftmost position" may all indicate parking at the same position. The parking process and parking location of the vehicle when executing the above parking command texts are the same. Therefore, it can be considered that the above several parking command texts correspond to the same parking task. In the embodiments of the present disclosure, multiple parking command texts for describing the same parking task can be collected in advance to form a parking text set for this parking task, so that the initial parking planning model can obtain a stronger transfer learning effect and improve the generalization ability of the model.

[0067] Among them, the parking driving image set can be collected in advance, which is the surrounding environment image of the vehicle when the driver executes any parking task. It should be noted that the parking task is a time-sequence task, and the surrounding environment images of the vehicle are usually different at different times. Multiple drivers can be pre-arranged to continuously execute parking tasks within a relatively long time period, such as 3 years, so as to obtain the surrounding environment images of the vehicle corresponding to different parking tasks at different times.

[0068] It should be noted that in order to enable the subsequently trained parking planning model to obtain good generalization, it is necessary to require that the training data set has a certain scale, richness, and diversity, covering various parking tasks and learning objectives. Therefore, it is necessary for the driver to pre-execute a wide range of parking tasks in a complex and changeable environment to collect sufficient and rich training data.

[0069] S203: Based on the training data set, train the initial parking planning model until the initial parking planning model meets the preset available conditions, and determine that the initial parking planning model is the trained parking planning model.

[0070] Optionally, the parking planning model can be constructed based on a pre-trained large image-text model. The large image-text model obtained through pre-training contains effective feature representations learned from large-scale image-text data. Based on the pre-trained model, a stronger transfer learning effect can be obtained, thereby achieving better automatic parking performance.

[0071] As a possible implementation, the path language and the image model PaLI-X (Pathways Language and Image model) can be selected as the initial parking planning model for training until the model is trained to an available state and can then be used as the parking planning model.

[0072] In the embodiments of the present disclosure, the vision-language model VLM (Vision-Language Model) can be selected as the initial parking planning model for training. The VLM is a model that integrates vision and language and aims to handle correlation tasks between images and texts. The goal of this model is to understand and generate natural language descriptions related to images, or to generate corresponding images from natural language descriptions. The VLM model is usually based on deep learning methods and combines technologies such as convolutional neural networks and recurrent neural networks. By training a large-scale image-text paired dataset, it learns the semantic relationship between images and texts. During the training process, the VLM model inputs images and texts into the network and performs feature extraction, semantic encoding, and joint representation through different layers and modules, enabling the model to capture the interaction and semantic correlation between images and texts, thereby realizing the functions of image description and image generation.

[0073] In the embodiments of the present disclosure, the initial parking planning model can be trained based on the parking text set and the parking driving image set corresponding to the parking task until the model is trained to an available state and can then be used as the parking planning model.

[0074] For example, if there are tasks Task 1, Task 2, and Task 3 for the parking task, the parking text sets respectively correspond to text sets Text Set 1, Text Set 2, and Text Set 3, and the parking driving image sets respectively have image sets Image Set 1, Image Set 2, and Image Set 3. Among them, Text Set 1 corresponds to Image Set 1 and Task 1, Text Set 2 corresponds to Image Set 2 and Task 2, and Text Set 3 corresponds to Image Set 3 and Task 3.

[0075] Taking Text Set 1 as an example, if Text Set 1 contains parking command texts Parking Command Text 1, Parking Command Text 2, and Parking Command Text 3, then the initial parking planning model can be trained respectively based on Parking Command Text 1 and the image corresponding to Image Set 1 at each moment, based on Parking Command Text 2 and the image corresponding to Image Set 1 at each moment, and based on Parking Command Text 3 and the image corresponding to Image Set 1 at each moment. The same applies to Text Set 2 and Text Set 3 until the initial parking planning model is trained to an available state and can then be used as the parking planning model.

[0076] Among them, the preset available condition is used to determine whether the model training is completed.

[0077] Optionally, if the initial parking planning model meets the preset available conditions, it can usually be judged in the following ways:

[0078] Convergence of the loss function: During the training process, the value of the loss function of the model on the validation set or test set can be monitored. When the value of the loss function gradually converges and tends to be stable, it can be considered that the training of the model has been completed. The convergence of the loss function indicates that the model has gradually improved its performance on the training data and can better generalize to new data.

[0079] Performance metrics meet the standards: In addition to the loss function, some other performance metrics such as accuracy, recall, F1-score, etc. can also be monitored. When these metrics reach the preset thresholds or meet specific performance requirements, it can be considered that the training of the model has been completed.

[0080] Good performance on the validation set: Use the validation set to evaluate the performance of the model. When the model performs well on the validation set and there is no overfitting, it can be considered that the training of the model has been completed. Good performance on the validation set indicates that the model has good generalization ability.

[0081] Hyperparameter tuning is completed: During the training process, it may be necessary to tune the hyperparameters of the model, such as the learning rate, regularization coefficient, etc. When the hyperparameter tuning is completed and the model reaches the best performance on the validation set, it can be considered that the training of the model has been completed.

[0082] The training time / number of epochs reaches the set value: Set a maximum training time or number of epochs. When this set value is reached, it can be considered that the training of the model has been completed.

[0083] It should be noted that the above methods are usually used in combination rather than using only one of them. In addition, for different tasks and models, the criteria for judging the completion of training will also be different. Therefore, in the actual training process, it may be necessary to comprehensively consider multiple factors according to the specific situation to judge whether the training of the model is completed.

[0084] S204: Input the parking command text and the first image into the pre-constructed parking planning model to obtain the first parking planning result of the vehicle. Among them, the first parking planning result contains multiple parking trajectory points and the vehicle driving characteristics corresponding to each parking trajectory point.

[0085] S205: Control the vehicle to park based on the first parking planning result.

[0086] It should be noted that the specific implementation methods of steps S204 and S205 can refer to the above embodiments and will not be elaborated here.

[0087] In the embodiments of the present disclosure, first, a parking command text is determined, and a first image of the vehicle's environment is obtained. Then, a training data set is obtained. The training data set includes multiple parking text sets and multiple parking driving image sets. Each parking text set contains multiple parking command texts. Each parking text set and each parking driving image set correspond to a parking task. Then, based on the training data set, an initial parking planning model is trained until the initial parking planning model meets the preset available conditions, and the initial parking planning model is determined as the trained parking planning model. After that, the parking command text and the first image are input into the pre-constructed parking planning model to obtain the first parking planning result of the vehicle. The first parking planning result includes multiple parking trajectory points and the vehicle movement characteristics corresponding to each parking trajectory point. Finally, based on the first parking planning result, the vehicle is controlled to park. Thus, by combining the pre-training of the vision-text model with the image data in the automatic parking scenario, a parking planning model is trained. After fine-tuning, it can output an automatic parking motion planning result represented in the form of text tokens. It not only generates an efficient and practical automatic parking motion planning strategy, but also significantly improves the generalization performance and emergence ability of the model based on large-scale vision-language pre-training.

[0088] Figure 3 is a schematic flowchart of the parking control method shown in the third embodiment of the present application, as Figure 3 shown, the method includes:

[0089] S301: Determine the parking command text and obtain the first image of the vehicle's environment.

[0090] S302: Obtain the training data set. The training data set includes multiple parking text sets and multiple parking driving image sets,

[0091] where each parking text set contains multiple parking command texts, and each parking text set and each parking driving image set correspond to a parking task.

[0092] It should be noted that the specific implementation manners of steps S301 and S302 can refer to the above embodiments and will not be elaborated here.

[0093] S303: Based on any parking task, obtain the second image corresponding to any moment from the corresponding parking driving image set, obtain any parking command text from the corresponding parking text set, and obtain the second parking planning result corresponding to any moment from the corresponding token planning data set.

[0094] Any moment can be a certain moment within the time period when the vehicle is performing any parking task.

[0095] Among them, the second image can be an image of the vehicle's surrounding environment collected at any moment, or the second image can also include historical images collected before any moment. For example, if any moment is T3 and the initial moment is T1, the second image corresponding to any moment can be an image of the vehicle's surrounding environment collected at T3, or it can also be all the environment images collected during the time period from T1 to T3. This is not limited here.

[0096] Among them, each parking driving image set contains images of the vehicle's surrounding environment corresponding to any parking task collected in a historical period. The training data set also includes multiple token planning data sets, and each parking task corresponds to a token planning data set.

[0097] Among them, the token planning data set contains motion planning tokens corresponding to each parking task. The motion planning token can be a discretized token, and the discretized token is a symbol or identifier obtained by discretizing a continuous numerical space.

[0098] In the automatic parking task, the discretized tokens can be used to represent different states or actions of the vehicle during movement. Among them, the discretized tokens can be a set of finite symbols or identifiers, representing different intervals or values in the continuous numerical space. For example, if the vehicle's motion trajectory is discretized, the discretized tokens can be defined as a series of symbols representing different positions or actions, and these symbols can be used by the model to learn and predict the vehicle's motion planning. In the automatic parking task, the discretized tokens may represent the vehicle's states at different positions and directions, or represent specific actions to be performed, such as turning left, turning right, parking, etc. By discretizing the continuous motion planning information into these symbols or identifiers, the model can better understand and predict the vehicle's behavior during automatic parking.

[0099] Among them, since the parking task is a sequential task, there are corresponding different vehicle environment images, trajectory features, etc. for different moments. Therefore, when training the initial parking planning model, it can be trained based on the image data corresponding to each moment. For example, if any parking task is A, the second image corresponding to any moment T can be obtained from the parking driving image set corresponding to A, the parking command text corresponding to any parking can be obtained from the parking text set corresponding to A, and the second parking planning result corresponding to any moment T can be obtained from the token planning data set corresponding to A.

[0100] Among them, the second parking planning result can be a parking planning result for reference. The second parking planning result can include discrete reference trajectory points and reference motion planning tokens.

[0101] S304: Input any parking command text corresponding to any parking task and the second image into the initial parking planning model to obtain a third parking planning result corresponding to any moment.

[0102] Among them, the third parking planning result can be a predicted parking planning result. The third parking planning result may include discrete predicted trajectory points and motion planning tokens. Among them, the motion planning token can be the motion planning token corresponding to each predicted trajectory point. The motion planning token corresponding to each predicted trajectory point represents the predicted vehicle traveling characteristics of the vehicle at each predicted trajectory point, such as speed, acceleration, turning radius, etc., which are reflected in the form of symbols or identifiers.

[0103] S305: Modify the initial parking planning model according to the differences between the third parking planning results and the second parking planning results corresponding to each parking task at each moment until the initial parking planning model meets the preset available conditions, and determine that the initial parking planning model is the trained parking planning model.

[0104] It should be noted that by comparing the predicted parking planning result and the reference parking planning result, that is, the difference between the third parking planning result and the second parking planning result, and then modifying the initial parking planning model. For example, the model parameters of the initial parking planning model can be adjusted, training data can be increased, or the algorithm can be improved to enhance the performance of the model. Then, if the difference between the third parking planning result and the second parking planning result is less than the preset value, it can be considered that the initial parking planning model meets the preset available conditions.

[0105] Specifically, by comparing the differences between the third parking planning result and the second parking planning result, the situation of the vehicle's position deviation, trajectory path deviation, speed and acceleration differences, etc. can be considered. Through comparative analysis, determine the difference between the third parking planning result and the second parking planning result, and find out the parts that need to be corrected, such as the path planning algorithm, control strategy, or environmental perception module. Then, the parts that need to be corrected can be determined according to the differences, and the initial parking planning model can be adjusted and optimized accordingly to reduce the difference between the third parking planning result and the second parking planning result. Finally, the corrected parking planning model can be used to regenerate the third parking planning result for verification and evaluation to ensure that the corrected model can meet the preset available conditions. If the corrected parking planning model still cannot meet the preset available conditions, compare the third parking planning result and the second parking planning result, find the differences and continue to correct until the requirements are met. When the corrected parking planning model can meet the preset available conditions, it can be determined as the trained parking planning model.

[0106] S306: Input the parking command text and the first image into a pre-constructed parking planning model to obtain the first parking planning result of the vehicle. The first parking planning result includes a plurality of parking trajectory points and the vehicle traveling characteristics corresponding to each parking trajectory point.

[0107] S307: Control the vehicle to park based on the first parking planning result.

[0108] It should be noted that the specific implementation manners of steps S306 and S307 can refer to the above embodiments and will not be elaborated here.

[0109] In the embodiments of the present disclosure, first, a parking command text is determined, and a first image of the environment where the vehicle is located is obtained. Then, a training data set is obtained. The training data set includes multiple parking text sets and multiple parking driving image sets. Each parking text set includes multiple parking command texts. Each parking text set and each parking driving image set correspond to a parking task. Then, based on any parking task, a second image corresponding to any moment is obtained from the corresponding parking driving image set, any parking command text is obtained from the corresponding parking text set, and a second parking planning result corresponding to any moment is obtained from the corresponding token planning data set. Each parking driving image set includes images of the vehicle's surrounding environment corresponding to any parking task collected in a historical period. The training data set further includes multiple token planning data sets, and each parking task corresponds to a token planning data set. Then, any parking command text and the second image corresponding to any parking task can be input into an initial parking planning model to obtain a third parking planning result corresponding to any moment. Finally, according to the difference between the third parking planning result and the second parking planning result corresponding to each parking task at each moment, the initial parking planning model is corrected until the initial parking planning model meets the preset available conditions, and the initial parking planning model is determined to be a trained parking planning model. Then, the parking command text and the first image can be input into the pre-constructed parking planning model to obtain a first parking planning result of the vehicle. The first parking planning result includes multiple parking trajectory points and the vehicle movement characteristics corresponding to each parking trajectory point. Finally, based on the first parking planning result, the vehicle is controlled to park. Thus, the image and text information can be combined, and the characteristics of the image-text model can be used for training. It can indeed maintain the consistency of model learning and training, and can effectively convert the continuous numerical information of the trajectory points into discrete representations that the model can understand and process, so that the model can better learn the semantic relationships and motion planning characteristics between the trajectory points. By discretizing the trajectory point coordinates into tokens representing motion planning, the model can better understand and learn the relationships between the trajectory points during the training process, and can also better combine the image information and the parking text instructions, thereby improving the performance and generalization ability of the model in the automatic parking task. It meets the current requirements of deep learning models for discretized and symbolic representations, and is beneficial for the model to learn more robust and general features. It can better utilize the advantages of the image-text model and help improve the model's understanding and processing ability for the automatic parking task.

[0110] Figure 4 is a schematic flowchart of a parking control method shown in the fourth embodiment of the present application. As Figure 4 shown, the method includes:

[0111] S401: Determine the parking command text and obtain a first image of the environment where the vehicle is located.

[0112] S402: Obtain a training data set, which includes multiple parking text sets and multiple parking driving image sets.

[0113] Among them, each parking text set includes multiple parking command texts, and each parking text set and each parking driving image set correspond to a parking task.

[0114] S403: Based on any parking task, obtain a second image corresponding to any moment from the corresponding parking driving image set, obtain any parking command text from the corresponding parking text set, and obtain a second parking planning result corresponding to the any moment from the corresponding token planning data set.

[0115] Among them, each parking driving image set includes images of the vehicle's surrounding environment corresponding to any parking task collected during a historical period. The training data set also includes multiple token planning data sets, and each parking task corresponds to a token planning data set.

[0116] S404: Input the any parking command text and the second image corresponding to the any parking task into the initial parking planning model to obtain a third parking planning result corresponding to the any moment.

[0117] It should be noted that the specific implementation manners of steps S401 - S404 can refer to the above embodiments and will not be elaborated here.

[0118] S405, correct the initial parking planning model according to the difference between each first trajectory point and the corresponding second trajectory point.

[0119] Among them, the third parking planning result includes each first trajectory point corresponding to any moment and the first motion planning token corresponding to each first trajectory point.

[0120] Among them, the first trajectory point can be the predicted trajectory point of the initial parking planning model, and the first motion planning token can be the predicted planning token corresponding to the predicted trajectory point.

[0121] Among them, the second parking planning result includes each second trajectory point corresponding to any moment and the second motion planning token corresponding to each second trajectory point.

[0122] Among them, the second trajectory point can be the real reference trajectory point, and the first motion planning token can be the real reference planning token.

[0123] It should be noted that each first trajectory point has a corresponding second trajectory point, and each first motion planning token has a corresponding second motion planning token.

[0124] Optionally, the first coordinates corresponding to each first trajectory point and the second coordinates corresponding to each second trajectory point can be determined first, and then the initial parking planning model can be corrected according to the differences between the first coordinates and the second coordinates.

[0125] For example, the coordinates of the first trajectory point are (x1, y1), and the coordinates of the second trajectory point are (x2, y2), and then the differences (△x, △y) between the first trajectory point and the second trajectory point can be determined. Then, the initial parking planning model can be corrected according to the evaluation results of the coordinate differences. For example, the performance of the model can be improved by adjusting the model parameters, increasing the training data, or improving the algorithm. Further, the corrected parking planning model can be used to input the coordinates of the first trajectory point into the model again to generate a corrected parking planning result. Through this iterative process, the parking planning model can be gradually optimized to better adapt to the actual situation.

[0126] S406. Correct the initial parking planning model according to the differences between each first motion planning token and the corresponding second motion planning token.

[0127] Among them, the first motion planning token contains multiple first vehicle traveling characteristics corresponding to the vehicle at the first trajectory point.

[0128] Among them, the second motion planning token contains multiple second vehicle traveling characteristics corresponding to the vehicle at the second trajectory point.

[0129] Optionally, multiple first vehicle traveling characteristics corresponding to each trajectory point and second vehicle traveling characteristics corresponding to each first vehicle traveling characteristic can be determined, and then the initial parking planning model can be corrected according to the differences between each first vehicle traveling characteristic and the corresponding second vehicle traveling characteristic.

[0130] Optionally, by comparing the differences between each first vehicle traveling characteristic and the corresponding second vehicle traveling characteristic, the accuracy and adaptability of the initial parking planning result can be evaluated by comparing the differences between these characteristics. For example, the differences in position deviation, speed difference, steering angle, etc. can be compared. Further, according to the differences between each first vehicle traveling characteristic and the corresponding second vehicle traveling characteristic, the initial parking planning model can be corrected accordingly. For example, the path planning algorithm can be adjusted, the control strategy can be optimized, or the environment perception module can be updated to improve the accuracy and robustness of the parking planning. Finally, the corrected parking planning model can be used to generate a corrected parking planning result again, and it can be verified and evaluated.

[0131] S407: Input the parking command text and the first image into a pre - constructed parking planning model to obtain the first parking planning result of the vehicle. Among them, the first parking planning result includes multiple parking trajectory points and the vehicle movement characteristics corresponding to each parking trajectory point.

[0132] S408: Control the vehicle to park based on the first parking planning result.

[0133] It should be noted that the specific implementation manners of steps S407 - S408 can refer to the above - mentioned embodiments and will not be elaborated here.

[0134] It should be noted that in the embodiments of the present disclosure, the motion planning problem in the automatic parking task is transformed into a method of learning and generating an image - text model. The coordinates of the trajectory points are discretized, and then the discretized representation is used as the learning target of the model. The semantic information can be learned and generated by using the image - text model. The input of the model includes an image and a parking text instruction, and the output is a discretized token representing the motion planning. In this way, the model can learn how to generate a suitable motion planning according to the image and the text instruction. During the training process, the training set data is converted into a format suitable for the model to learn and train, where the input is an image and a parking text instruction, and the output is a discretized token representing the motion planning. In this way, the model can predict the corresponding motion planning according to the input image and text instruction. The motion planning problem of automatic parking is transformed into learning and generating discretized tokens to realize the learning and training of the model.

[0135] In an embodiment of the present disclosure, first, a parking command text is determined, and a first image of the environment where the vehicle is located is obtained. Then, a training data set is obtained. The training data set includes multiple parking text sets and multiple parking driving image sets. After that, based on any parking task, a second image corresponding to any moment is obtained from the corresponding parking driving image set, a parking command text corresponding to any parking task is obtained from the corresponding parking text set, and a second parking planning result corresponding to the any moment is obtained from the corresponding token planning data set. Then, the parking command text and the second image corresponding to the any parking task are input into an initial parking planning model to obtain a third parking planning result corresponding to the any moment. After that, the initial parking planning model is corrected according to the difference between each first trajectory point and the corresponding second trajectory point. Then, the initial parking planning model is corrected according to the difference between each first motion planning token and the corresponding second motion planning token. Then, the parking command text and the first image are input into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle. The first parking planning result includes multiple parking trajectory points and the vehicle traveling characteristics corresponding to each parking trajectory point. Finally, based on the first parking planning result, the vehicle is controlled to park. Thus, the traveling characteristics of the actual vehicle can be used to continuously correct and optimize the parking planning model, so that it can better adapt to the parking task requirements in different scenarios, and gradually optimize the parking planning model to better adapt to the actual situation.

[0136] Figure 5 is a block diagram of a parking control device shown according to the present application, as Figure 5 shown, the parking control device 500 includes:

[0137] A determination module 510, configured to determine a parking command text and obtain a first image of the environment where the vehicle is located;

[0138] An input module 520, configured to input the parking command text and the first image into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle, where the first parking planning result includes multiple parking trajectory points and the vehicle traveling characteristics corresponding to each parking trajectory point;

[0139] A control module 530, configured to control the vehicle to park based on the first parking planning result.

[0140] Optionally, the determination module is specifically configured to:

[0141] In response to receiving a parking instruction for the first time, determine the parking command text corresponding to the parking instruction;

[0142] Collect an image of the environment where the vehicle is located at the first time as the first image,

[0143] Alternatively, capture an image of the vehicle's environment at a second time as the first image, where the first time is a historical time of the second time.

[0144] Alternatively, capture an image of the vehicle's environment at a third time as the first image, where the third time includes the first time, the second time, and the time interval between the first time and the second time.

[0145] Optionally, the input module further includes:

[0146] An acquisition unit for acquiring a training data set, which includes a plurality of parking text sets and a plurality of parking driving image sets. Each of the parking text sets includes a plurality of parking command texts, and each parking text set and each parking driving image set correspond to a parking task.

[0147] A training unit for training an initial parking planning model based on the training data set until the initial parking planning model meets a preset available condition, and determining that the initial parking planning model is a trained parking planning model.

[0148] Optionally, each of the parking driving image sets includes an image of the vehicle's surrounding environment corresponding to any parking task collected in a historical period. The training data set further includes a plurality of token planning data sets, and each parking task corresponds to a token planning data set. The training unit includes:

[0149] A first acquisition subunit for, based on any parking task, acquiring a second image corresponding to any moment from the corresponding parking driving image set, acquiring any parking command text from the corresponding parking text set, and acquiring a second parking planning result corresponding to any moment from the corresponding token planning data set.

[0150] A second acquisition subunit for inputting the any parking command text and the second image corresponding to the any parking task into the initial parking planning model to obtain a third parking planning result corresponding to any moment.

[0151] A correction subunit for correcting the initial parking planning model according to the difference between the third parking planning result and the second parking planning result corresponding to each parking task at each moment until the initial parking planning model meets a preset available condition.

[0152] Optionally, the third parking planning result includes each first trajectory point corresponding to any moment and a first motion planning token corresponding to each first trajectory point.

[0153] The second parking planning result includes each second trajectory point corresponding to any moment and a second motion planning token corresponding to each second trajectory point;

[0154] The correction subunit includes:

[0155] A first correction subunit for correcting the initial parking planning model according to the difference between each first trajectory point and the corresponding second trajectory point;

[0156] A second correction subunit for correcting the initial parking planning model according to the difference between each motion planning token and the corresponding second motion planning token.

[0157] Optionally, the first motion planning token includes multiple first vehicle traveling features corresponding to the vehicle at the first trajectory point;

[0158] The second motion planning token includes multiple second vehicle traveling features corresponding to the vehicle at the second trajectory point;

[0159] The second correction subunit is specifically configured to:

[0160] Determine multiple first vehicle traveling features corresponding to each trajectory point and the second vehicle traveling features corresponding to each first vehicle traveling feature;

[0161] Correct the initial parking planning model according to the difference between each first vehicle traveling feature and the corresponding second vehicle traveling feature.

[0162] Optionally, the first correction subunit is specifically configured to:

[0163] Determine a first coordinate corresponding to each first trajectory point and a second coordinate corresponding to each second trajectory point;

[0164] Correct the initial parking planning model according to the difference between the first coordinate and the second coordinate.

[0165] Optionally, the initial parking planning model is a Path Language and Image Model PaLI-X.

[0166] In an embodiment of the present application, first, a parking command text is determined, and a first image of the vehicle's environment is obtained. Then, the parking command text and the first image are input into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle. The first parking planning result includes a plurality of parking trajectory points and the vehicle traveling characteristics corresponding to each parking trajectory point. Finally, based on the first parking planning result, the vehicle is controlled to park. Thus, an efficient and practical automatic parking motion planning strategy can be generated based on the parking planning model, which can handle more complex and detailed parking tasks, allow effective reasoning according to images and texts during operation, and achieve real-time control.

[0167] According to an embodiment of the present application, the present application also provides an electronic device, a readable storage medium, and a computer program product.

[0168] Figure 6 A block diagram of an exemplary electronic device suitable for implementing the embodiments of the present disclosure is shown. Figure 6 The illustrated electronic device 12 is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of the present disclosure.

[0169] As Figure 6 shown, the electronic device 12 is presented in the form of a general-purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 connecting different system components (including the system memory 28 and the processing unit 16).

[0170] The bus 18 represents one or more of several types of bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the multiple bus architectures. For example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnection (PCI) bus.

[0171] The electronic device 12 typically includes a variety of computer system-readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0172] The memory 28 can include computer system-readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 12 can further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 can be used for reading and writing on non-removable, non-volatile magnetic media ( Figure 6 not shown, commonly referred to as a "hard disk drive"). Although Figure 6 not shown in the figure, a disk drive for reading and writing on removable non-volatile disks (such as a "floppy disk") and an optical disk drive for reading and writing on removable non-volatile optical disks (such as compact disc read only memory (CD-ROM), digital video disc read only memory (DVD-ROM) or other optical media) can be provided. In these cases, each drive can be connected to the bus 18 through one or more data media interfaces. The memory 28 can include at least one program product having a set (such as at least one) of program modules that are configured to perform the functions of the embodiments of the present disclosure.

[0173] A program / utility 40 having a set (at least one) of program modules 42 can be stored, for example, in the memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include the implementation of a network environment. The program modules 42 generally perform the functions and / or methods in the embodiments described in the present disclosure.

[0174] The electronic device 12 can also communicate with one or more external devices 14 (such as a keyboard, a pointing device, a display 24, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 12, and / or communicate with any device that enables the electronic device 12 to communicate with one or more other computing devices (such as a network card, a modem, etc.). Such communication can be carried out through the input / output (I / O) interface 22. Moreover, the electronic device 12 can also communicate with one or more networks (such as a Local Area Network (LAN), a Wide Area Network (WAN), and / or a public network, such as the Internet) through the network adapter 20. As shown in the figure, the network adapter 20 communicates with other modules of the electronic device 12 through the bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in combination with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0175] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the methods mentioned in the foregoing embodiments.

[0176] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0177] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of the features. In the description of the present disclosure, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0178] Any process or method description represented in a flowchart or otherwise described herein can be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logic function or process. The scope of the preferred embodiments of the present disclosure includes additional implementations where functions may be executed not in the order shown or discussed, including in a substantially simultaneous manner according to the involved functions or in a reverse order, which should be understood by those skilled in the technical field to which the embodiments of the present disclosure pertain.

[0179] The logic and / or steps represented in a flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing a logical function, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with such instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0180] It should be understood that various parts of the present disclosure can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one of the following techniques known in the art or a combination thereof can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0181] Those of ordinary skill in the art can understand that all or part of the steps carried out in the methods of the above embodiments can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0182] In addition, in each of the various embodiments of the present disclosure, the functional units can be integrated into one processing module, or each unit can exist physically alone, or two or more units can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0183] The storage medium mentioned above can be a read-only memory, a magnetic disk, an optical disk, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present disclosure.

Claims

1. A parking control method, characterized in that, Including: Determine a parking command text and obtain a first image of the environment where the vehicle is located; Input the parking command text and the first image into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle, where the first parking planning result includes a plurality of parking trajectory points and the vehicle traveling characteristics corresponding to each parking trajectory point; Based on the first parking planning result, control the vehicle to park.

2. The method according to claim 1, characterized in that, The determining the parking command text and obtaining the first image of the environment where the vehicle is located includes: In response to receiving a parking instruction at a first time, determine the parking command text corresponding to the parking instruction; Collect an image of the environment where the vehicle is located at the first time as the first image, Or, collect an image of the environment where the vehicle is located at a second time as the first image, the first time being the historical time of the second time, Or, collect an image of the environment where the vehicle is located at a third time as the first image, the third time including the first time, the second time, and the time interval between the first time and the second time.

3. The method according to claim 1, characterized in that, Before inputting the parking command text and the first image into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle, it further includes: Obtain a training data set, where the training data set includes a plurality of parking text sets and a plurality of parking driving image sets, Wherein, each parking text set includes a plurality of parking command texts, and each parking text set and each parking driving image set correspond to a parking task; Based on the training data set, train an initial parking planning model until the initial parking planning model meets a preset available condition, and determine the initial parking planning model as a trained parking planning model.

4. The method according to claim 3, wherein Each parking driving image set includes an image of the vehicle's surrounding environment corresponding to any parking task collected in a historical period, and the training data set further includes a plurality of token planning data sets, and each parking task corresponds to a token planning data set, The training the initial parking planning model based on the training data set until the initial parking planning model meets a preset available condition includes: Based on any parking task, obtain a second image corresponding to any moment from the corresponding parking driving image set, obtain any parking command text from the corresponding parking text set, and obtain a second parking planning result corresponding to the any moment from the corresponding token planning data set; Input the any parking command text and the second image corresponding to the any parking task into the initial parking planning model to obtain a third parking planning result corresponding to the any moment; According to the difference between the third parking planning result and the second parking planning result corresponding to each parking task at each moment, correct the initial parking planning model until the initial parking planning model meets a preset available condition.

5. The method according to claim 3, wherein Wherein, The third parking planning result includes each first trajectory point corresponding to any moment and a first motion planning token corresponding to each first trajectory point; The second parking planning result includes each second trajectory point corresponding to any moment and a second motion planning token corresponding to each second trajectory point; The modification of the initial parking planning model according to the difference between the third parking planning result and the second parking planning result corresponding to each parking task at each moment includes: Modifying the initial parking planning model according to the difference between each first trajectory point and the corresponding second trajectory point; Modifying the initial parking planning model according to the difference between each motion planning token and the corresponding second motion planning token.

6. The method according to claim 5, wherein Wherein, The first motion planning token includes a plurality of first vehicle traveling characteristics corresponding to the vehicle at the first trajectory point; The second motion planning token includes a plurality of second vehicle traveling characteristics corresponding to the vehicle at the second trajectory point; The modification of the initial parking planning model according to the difference between each motion planning token and the corresponding second motion planning token includes: Determining a plurality of first vehicle traveling characteristics corresponding to each trajectory point and the second vehicle traveling characteristics corresponding to each first vehicle traveling characteristic; Modifying the initial parking planning model according to the difference between each first vehicle traveling characteristic and the corresponding second vehicle traveling characteristic.

7. The method according to claim 5, wherein The modification of the initial parking planning model according to the difference between each first trajectory point and the corresponding second trajectory point includes: Determining a first coordinate corresponding to each first trajectory point and a second coordinate corresponding to each second trajectory point; Modifying the initial parking planning model according to the difference between the first coordinate and the second coordinate.

8. The method according to claim 3, characterized in that, The initial parking planning model is the Path Language and Image Model PaLI-X.

9. A parking control device, characterized in that, Including: A determination module for determining a parking command text and acquiring a first image of the vehicle's environment; An input module for inputting the parking command text and the first image into a pre-constructed parking planning model to obtain a first parking planning result of the vehicle, wherein the first parking planning result includes a plurality of parking trajectory points and vehicle traveling characteristics corresponding to each parking trajectory point; A control module for controlling the vehicle to park based on the first parking planning result.

10. An electronic device, characterized in that, Including: A processor, And a memory communicatively connected to the processor, the memory storing computer execution instructions; The processor executes the computer execution instructions stored in the memory to implement the method according to any one of claims 1-7.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer execution instructions, and when the computer execution instructions are executed by a processor, they are used to implement the method according to any one of claims 1-7.

Images