Parking space pre-judgment method and device and storage medium

Abstract

The invention provides a parking space pre-judgment method and device and a storage medium. The method comprises the following steps: acquiring current original data of a traffic participant; determining current track information of the traffic participants according to the current original data; the current track information represents the position change track of the traffic participant at the current moment; inputting the current original data and the current track information into a pre-judgment model so as to enable the pre-judgment model to output the state of the parking space; the pre-judgment model is obtained through training on the basis of historical original data and historical track information corresponding to the historical original data; displaying the state of the parking space; the prediction model is obtained by determining the historical track information of each traffic participant, and the feature learning ability of the prediction model and the rules contained in the historical track information of the traffic participants are combined with the current original data and the current track information, so that the accurate prediction of the parking space release condition is realized. And the intelligent level and the overall parking efficiency of the automatic parking system are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a parking space prediction method, device and storage medium. Background Art

[0002] Given the current shortage of parking spaces, vehicles take longer to find available spaces. Even with automated parking systems, overall parking efficiency is limited. Given the current imbalance between supply and demand for urban parking resources, improving parking efficiency faces multiple technical bottlenecks.

[0003] Related technologies usually determine whether a parking space is occupied based on real-time sensor data, or identify the current occupancy status of a parking space based on simple image recognition methods. This method can only reflect the current static situation and is difficult to meet the actual usage needs of increasingly complex and changing parking lots. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention are proposed to provide a parking space prediction method, device, and storage medium that overcome the above problems or at least partially solve the above problems.

[0005] According to a first aspect of the present invention, a parking space prediction method is provided, the method comprising: Obtaining current raw data of traffic participants; the traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; Determining current trajectory information of the traffic participant based on the current raw data; the current trajectory information represents a position change trajectory of the traffic participant at the current moment; Inputting the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the state of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; Displays the status of the parking space.

[0006] Optionally, the current original data includes: current image data and current distance data; the parking space includes a parking space sensor; The obtaining of current raw data of traffic participants includes: Acquire the current image data collected by the camera of the parking lot; The parking space sensor is used to acquire the current distance data between the traffic participant and the parking space.

[0007] Optionally, determining the current trajectory information of the traffic participant according to the current original data includes: Identifying the position coordinates of each traffic participant in the current image data; Constructing movement data of the traffic participants according to the position coordinates of each traffic participant within a preset time range; Current trajectory information corresponding to the traffic participant is determined according to the current distance data and the movement data.

[0008] Optionally, the movement data includes at least: movement speed, movement acceleration and movement vector; The determining the current trajectory information corresponding to the traffic participant according to the current distance data and the movement data includes: Determining a moving speed and a moving vector corresponding to the traffic participant according to the position coordinates; determining a movement acceleration corresponding to the traffic participant based on the movement speed; The movement data is corrected according to the current distance data to obtain current trajectory information corresponding to the traffic participant.

[0009] Optionally, the prediction model is trained in the following manner: Extracting historical image features of the historical original data; Determining a prediction result of the traffic participant leaving the parking space based on the historical image features and the historical trajectory information corresponding to the historical original data; The prediction model is trained based on the historical image features, the historical trajectory information and the prediction results.

[0010] Optionally, the training of the prediction model based on the historical image features, the historical trajectory information, and the prediction result includes: Marking the prediction result and determining corresponding marking data; Dividing the historical image features, the historical trajectory information, and the labeled data according to a preset division ratio to obtain a training set and a test set; The prediction model is trained based on the training set and the test set.

[0011] Optionally, the prediction model includes: a convolutional neural network and a gated recurrent neural network; Inputting the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the state of the parking space includes: Inputting the current raw data into the convolutional neural network to extract image features; Obtaining a fused feature vector according to the image features and the current trajectory information; The fused feature vector is input into the gated recurrent neural network to determine the state of the parking space.

[0012] Optionally, the status of the parking space includes: a first parking space status and a second parking space status; Inputting the fused feature vector into the gated recurrent neural network to determine the state of the parking space includes: Inputting the fused feature vector into the gated recurrent neural network to determine a corresponding probability value; When the probability value is greater than or equal to a probability threshold, determining the state of the parking space as a first parking space state; When the probability value is less than a probability threshold, the state of the parking space is determined to be a second parking space state.

[0013] According to a second aspect of the present invention, a parking space prediction device is provided, the device comprising: A data acquisition module is used to acquire current raw data of traffic participants; the traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; A current trajectory information determination module is used to determine the current trajectory information of the traffic participant based on the current raw data; the current trajectory information represents the position change trajectory of the traffic participant at the current moment; a prediction module, configured to input the current raw data and the current trajectory information into a prediction model, so that the prediction model outputs the state of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; The display module is used to display the status of the parking space.

[0014] According to a third aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned parking space prediction method are implemented.

[0015] The embodiments of the present invention include the following advantages: In the present invention, by obtaining the current original data of traffic participants; traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; the current trajectory information of the traffic participants is determined based on the current original data; the current trajectory information represents the position change trajectory of the traffic participants at the current moment; the current original data and the current trajectory information are input into a prediction model so that the prediction model outputs the status of the parking space; the prediction model is trained based on historical original data and historical trajectory information corresponding to the historical original data; the status of the parking space is displayed; the prediction model is obtained by determining the historical trajectory information of each traffic participant, and the feature learning ability of the prediction model and the rules contained in the historical trajectory information of the traffic participants are combined with the current original data and the current trajectory information to achieve accurate prediction of the parking space release situation, thereby improving the intelligence level of the automatic parking system and the overall parking efficiency.

[0016] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0018] Figure 1 This is a flowchart of a parking space prediction method embodiment of the present invention; Figure 2 This is a schematic diagram of a parking lot data collection layout provided by an embodiment of the present invention; Figure 3 This is a diagram of the architecture of the prediction model provided by an embodiment of the present invention; Figure 4 It is a structural block diagram of an embodiment of a parking space prediction device of the present invention. DETAILED DESCRIPTION

[0019] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0020] The terms "first," "second," and the like in the specification and claims of the present invention are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present invention can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects. For example, the first object can be one or more. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0021] The following describes in detail a parking space prediction method, device, and storage medium provided by the embodiments of the present invention through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0022] Given the current shortage of parking spaces, vehicles take longer to find available spaces. Even with automated parking systems, overall parking efficiency is limited. Given the current imbalance between supply and demand for urban parking resources, improving parking efficiency faces multiple technical bottlenecks.

[0023] Related technologies typically determine whether a parking space is occupied based on real-time sensor data. For example, when a vehicle approaches a parking space, ultrasonic sensors or cameras are used to detect the presence of objects within the space. If an obstacle is detected, the space is considered unavailable; otherwise, it is considered available. This method only provides current parking space occupancy information and cannot predict future parking space releases. Relying solely on real-time sensor data only reflects the current static situation and lacks consideration of dynamic factors such as whether a vehicle is about to leave the space or whether other traffic participants, such as pedestrians, will affect parking space usage.

[0024] In real-world parking lots, it's common for a vehicle to start moving out of a parking space. Due to a lack of anticipation for this impending change, other vehicles seeking to park miss the space, resulting in inefficient parking and wasted parking resources. Furthermore, in complex scenarios with multiple vehicles interacting and pedestrians frequently passing through, real-time sensor data struggles to comprehensively assess the impact of various traffic participant behavior trends on parking space release, making it difficult to provide effective guidance to the automated parking system in advance.

[0025] Therefore, in order to solve the problem in related technologies that the future release of parking spaces cannot be predicted, and there is a lack of in-depth mining of the historical trajectories of traffic participants and a lack of consideration of dynamic factors such as whether a vehicle is about to leave the parking space and whether other traffic participants such as pedestrians will affect the use of the parking space, resulting in an inability to accurately grasp the timing of parking space release, the present invention provides a parking space prediction method, which obtains current raw data of traffic participants; traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; determines the current trajectory information of the traffic participants based on the current raw data; the current trajectory information represents the position change trajectory of the traffic participants at the current moment; inputs the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the status of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; displays the status of the parking space; obtains a prediction model by determining the historical trajectory information of each traffic participant, and utilizes the feature learning ability of the prediction model and the laws contained in the historical trajectory data of the traffic participants in combination with the current raw data and the current trajectory information to achieve accurate prediction of the parking space release situation, thereby improving the intelligence level and overall parking efficiency of the automatic parking system.

[0026] Reference Figure 1 , shows a flowchart of a parking space prediction method embodiment of the present invention, which may specifically include the following steps: Step 101: obtaining current raw data of traffic participants; the traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; In this embodiment, after determining the driving destination based on the user navigation data, parking lot information near the driving destination can be obtained, and then the current original data of traffic participants in the parking lot near the driving destination can be obtained. The current original data includes data used to reflect the characteristics of traffic participants and the dynamic relationship between parking spaces and traffic participants.

[0027] Since the above data can reflect the dynamic relationship between traffic participants and parking spaces, by obtaining the current original data of traffic participants in the parking lot, it is possible to predict the upcoming changes based on the dynamic relationship when other vehicles in the parking lot are starting to prepare to leave the parking space, or when the owners of other vehicles are walking towards the parked space. This avoids the problem in related technologies that only rely on real-time sensor data detection to reflect the current static situation, and lacks consideration of dynamic factors such as whether the vehicle is about to leave the parking space, and whether other traffic participants such as pedestrians will affect the use of the parking space.

[0028] Specifically, traffic participants include at least vehicles and pedestrians within a preset range of the parking lot where the parking space is located. Since there may be animals in the parking lot, which will also affect the use of the parking space, traffic participants can also include animals within the preset range of the parking lot.

[0029] Step 102: determining the current trajectory information of the traffic participant based on the current raw data; the current trajectory information represents the position change trajectory of the traffic participant at the current moment; In this embodiment, current raw data can be collected by cameras in a parking lot near the destination and / or ultrasonic sensors installed near parking spaces, characteristic data of the traffic participants can be extracted from the current raw data, and the position change trajectory of the corresponding traffic participants can be identified and tracked based on the characteristic data, thereby determining the trajectory information corresponding to each traffic participant.

[0030] Optionally, after obtaining the current raw data, the current raw data may be preprocessed, and the images in the current raw data may be subjected to denoising, normalization, cropping, etc. Data extraction is performed based on the preprocessed current raw data to determine the trajectory information of each traffic participant.

[0031] Denoising can reduce image quality issues caused by factors such as lighting changes and interference from camera electronic components, ensuring the clear and discernible features of traffic participants. Normalizing image pixel values to the range [0, 1] and standardizing image brightness and contrast can reduce the impact of image differences collected under different lighting conditions on subsequent analysis. The original image is cropped according to the location and size of the parking space, retaining only the parking space and the surrounding area within a certain range. The cropped image is then scaled to 256×256 pixels to facilitate subsequent model processing.

[0032] Step 103: Input the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the state of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; After obtaining the current trajectory information corresponding to the current raw data, the current raw data and the corresponding current trajectory information are input into the prediction model. The prediction model is used to predict the parking space release situation and output the predicted parking space status. Since the trajectory information corresponding to the current raw data identifies and tracks the position change trajectory of the corresponding traffic participant based on the feature data, it reflects the short-term historical trajectory of the traffic participant and can more accurately reflect the immediate intentions of the traffic participant near the parking space, such as: driving trajectory changes and speed fluctuations within a few minutes, providing a new perspective for parking space release prediction.

[0033] In practical applications, a prediction model can be trained based on historical raw data and the historical trajectory information corresponding to the historical raw data. A spatiotemporal correlation model is obtained by training with historical raw data and the historical trajectory information corresponding to the historical raw data. In terms of time, the time series characteristics of traffic participants' behaviors are analyzed based on short-term historical trajectory information. This spatiotemporal correlation model can more accurately predict the time point and possibility of parking space release, and has higher accuracy than traditional models that only consider single moment or local spatial information.

[0034] Step 104: Display the status of the parking space.

[0035] After outputting the status of the parking space according to the prediction model, information about possible vacant parking spaces in parking lots near the driving destination can be displayed to the user, so that the user can select the destination parking lot and navigate to the vacant parking spaces in the destination parking lot. Accurate parking space release prediction helps to allocate the parking spaces that are about to become vacant to the vehicles waiting for parking in a timely manner, avoiding the idle waiting period of parking spaces due to untimely information, reducing the idle time of parking spaces, and through accurate prediction, the vehicle can prepare to drive to the parking spaces that are about to be released in advance, avoiding unnecessary waiting and path reversal, significantly optimizing the entire parking process, and reducing parking time and energy consumption.

[0036] This embodiment obtains current raw data of traffic participants; traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; current trajectory information of the traffic participants is determined based on the current raw data; the current trajectory information represents the position change trajectory of the traffic participants at the current moment; the current raw data and the current trajectory information are input into a prediction model so that the prediction model outputs the status of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; the status of the parking space is displayed; the prediction model is obtained by determining the historical trajectory information of each traffic participant, and the feature learning ability of the prediction model and the rules contained in the historical trajectory data of the traffic participants are combined with the current raw data and the current trajectory information to achieve accurate prediction of the parking space release situation, thereby improving the intelligence level of the automatic parking system and the overall parking efficiency.

[0037] In one embodiment of the present invention, the current original data includes: current image data and current distance data; the parking space includes a parking space sensor; The obtaining of current raw data of traffic participants includes: Acquire the current image data collected by the camera of the parking lot; The parking space sensor is used to acquire the current distance data between the traffic participant and the parking space.

[0038] In this embodiment, a camera is deployed in the parking lot, and the camera collects current image data at a preset frame rate (for example, 10 frames per second). Based on the current image data, information such as appearance features, position features, and posture features of traffic participants in the parking area of the parking lot can be extracted; wherein, the appearance features may include at least one appearance information of vehicle type, color, clothing, and type; the position features may include at least one position information of coordinates in the parking space and the relative position of the vehicle; the posture features may include at least one posture information of the traffic participant's direction and body movements.

[0039] In practical applications, ultrasonic sensors can be installed near parking spaces in parking lots. The ultrasonic sensors collect current distance data at regular intervals (such as 0.2 seconds) to obtain the current distance relationship between traffic participants and parking spaces, and assist in judging changes in their behavior.

[0040] Reference Figure 2 , shows a schematic diagram of the parking lot data collection layout provided by an embodiment of the present invention, which may specifically include the following contents: In a parking lot, current image data can be collected by deployed cameras 1, 2, and 3; current distance data can be collected by ultrasonic sensors-parking space 1, 2, and 3, and the current trajectory information of traffic participants can be determined based on the collected current image data and current distance data. It should be emphasized that the number of cameras and ultrasonic sensors can be set according to the actual needs of the parking lot, and the present invention does not impose any restrictions on the number of cameras and ultrasonic sensors.

[0041] In this embodiment, by fusing the visual data collected by the camera with the ultrasonic data collected by the sensor, the accuracy of the current trajectory information of the traffic participants can be improved, thereby improving the accuracy of the parking space status prediction.

[0042] In one embodiment of the present invention, determining the current trajectory information of the traffic participant based on the current raw data includes: Identifying the position coordinates of each traffic participant in the current image data; Constructing movement data of the traffic participants according to the position coordinates of each traffic participant within a preset time range; Current trajectory information corresponding to the traffic participant is determined according to the current distance data and the movement data.

[0043] After obtaining the current image data and the current distance data, the convolutional neural network image analysis technology is used to process the current image data of continuous image frames to obtain the characteristic data of each traffic participant in the current image data. Based on the characteristic data, the position coordinates of the traffic participants in the parking space are identified and tracked. Then, the movement data and the current distance data are constructed according to the position coordinates to obtain the corresponding current trajectory information. For example, the position coordinates of the traffic participants are recorded once every 1 second, and the movement data of the traffic participants are determined according to the position coordinates, so as to construct the historical trajectory of each traffic participant over a period of time.

[0044] Specifically, after constructing the movement data of traffic participants based on the position coordinates of each traffic participant within a preset time range, the local movement data can obtain the first trajectory information of each traffic participant, and then the first trajectory information is corrected in combination with the current distance data to obtain the second trajectory information.

[0045] For example, after extracting features from the current image data captured by the camera, the movement data of each traffic participant determined based on the extracted features may contain certain errors due to factors such as weather and equipment aging. The resulting movement data may not accurately determine whether a pedestrian approaching a parked vehicle in the parking lot is heading for the vehicle's left front door (also known as the main driver's door). When the pedestrian reaches the left rear door, the vehicle is not about to leave the parking space. Because the distance information collected by the ultrasonic sensor installed on the parking space represents the distance between the traffic participant and the parking space, the distance information collected when the traffic participant is standing at the left front door and the left rear door is different, which can assist in determining whether the pedestrian is heading for the vehicle's left front door. Therefore, the first trajectory information obtained from the movement data can be corrected by combining the distance information collected by the ultrasonic sensor installed on the parking space to obtain the second trajectory information, thereby improving the data accuracy of the trajectory information, enabling accurate prediction of parking space release, and enhancing the intelligence level and overall parking efficiency of the automatic parking system.

[0046] Optionally, vehicles can be tracked through license plate recognition and / or vehicle feature matching; pedestrians can be tracked based on body posture and / or appearance; and other traffic participants can be tracked based on their appearance and movement patterns. Current solutions are prone to misjudgment or failure when faced with complex parking environments (e.g., irregular layouts, a mix of various vehicle types, temporary obstacles, etc.) and diverse traffic participant behaviors (including novice drivers' unique parking habits and emergency stops). This embodiment leverages the characteristics of different traffic participants to deeply analyze the various features and dynamic behavior patterns in each participant's short-term trajectory information, better adapting to these complex scenarios and maintaining stable and reliable parking space release predictions.

[0047] In one embodiment of the present invention, the movement data includes at least: movement speed, movement acceleration and movement vector; The determining the current trajectory information corresponding to the traffic participant according to the current distance data and the movement data includes: Determining a moving speed and a moving vector corresponding to the traffic participant according to the position coordinates; determining a movement acceleration corresponding to the traffic participant based on the movement speed; The movement data is corrected according to the current distance data to obtain current trajectory information corresponding to the traffic participant.

[0048] In this embodiment, after feature extraction is performed on the current image data collected by the camera, the position coordinates of each traffic participant can be determined based on different features. The moving speed and moving vector of the traffic participant in the corresponding time period can be calculated based on the position coordinates, and then the corresponding acceleration can be calculated based on the moving speed. The moving speed, moving acceleration and moving vector of the traffic participant are corrected based on the current distance data to obtain the current trajectory information corresponding to the traffic participant.

[0049] For example, after extracting features from the current image data captured by a camera, the movement data of each traffic participant determined based on the extracted features may contain certain errors due to factors such as weather and equipment aging. The resulting movement data may not accurately determine whether a pedestrian approaching a parked vehicle in a parking lot is heading for the vehicle's left front door (also known as the main driver's door). When the pedestrian reaches the left rear door, the vehicle is not about to leave the parking space. Because the distance information collected by the ultrasonic sensor installed in the parking space represents the distance relationship between the traffic participant and the parking space, the distance information collected when the traffic participant is standing at the left front door and the left rear door is different, which can assist in determining whether the pedestrian is heading for the vehicle's left front door. If the trajectory information obtained based on the traffic participant's position coordinates, movement speed, movement vector, and corresponding acceleration points to the vehicle's left rear door, and the current distance data collected by the sensor is for the vehicle's left front door, the traffic participant's movement speed, movement acceleration, and movement vector can be corrected based on the current distance data to obtain accurate current trajectory information.

[0050] Specifically, at time t1, the traffic participant is at position P1 (x1, y1), and at time t2, the traffic participant is at position P2 (x2, y2). The moving speed v of the traffic participant is: ; Among them, x1 and y1 are the x-axis and y-axis data of the traffic participant at time t1; x2 and y2 are the x-axis and y-axis data of the traffic participant at time t2; v is the moving speed of the traffic participant.

[0051] After obtaining the position coordinates at time t1 and time t2, the corresponding movement vector can be obtained. : ; After obtaining the moving speed, the acceleration is also calculated based on the two adjacent moving speed values and the corresponding time interval. If the speed at time t3 is v1 and the speed at time t4 is v2, then the acceleration a is: ; Based on movement speed, movement acceleration, and movement vector, the movement trajectory of traffic participants over a short period of time can be obtained. This movement trajectory reflects the time series characteristics of traffic participants' behavior. Current distance data can reflect the spatial information between traffic participants and parking spaces in the local area surrounding the parking space and the layout information of the entire parking lot. Therefore, by correcting the movement data based on the current distance data, the resulting current trajectory information combines temporal and spatial data, enabling more accurate predictions of the timing and likelihood of parking space release. This approach offers higher accuracy than traditional models that only consider single-moment or local spatial information. Furthermore, while traditional parking space release determination methods rely heavily on long-term statistical data, this embodiment focuses on the short-term trajectory information of traffic participants, capturing more timely and targeted behavioral information. This short-term trajectory information can more accurately reflect the immediate intentions of traffic participants near parking spaces, such as changes in driving trajectory and speed fluctuations over a period of several minutes, providing a new perspective for predicting parking space release.

[0052] In one embodiment of the present invention, the prediction model is trained in the following manner: Extracting historical image features of the historical original data; Determining a prediction result of the traffic participant leaving the parking space based on the historical image features and the historical trajectory information corresponding to the historical original data; The prediction model is trained based on the historical image features, the historical trajectory information and the prediction results.

[0053] In this embodiment, the prediction model is trained based on historical original data, and features are extracted from the historical original data to obtain corresponding historical image features. Then, the prediction result of the traffic participant leaving the parking space is determined in combination with the historical trajectory information of the traffic participant corresponding to the historical image features, that is, the idle state of the parking space that has a dynamic relationship with the traffic participant. The prediction model is trained based on the historical image features, historical trajectory information and corresponding prediction results as the data set.

[0054] The prediction model, trained on spatiotemporal correlations of historical trajectory information, is spatiotemporally correlated. Spatially, it comprehensively considers the layout of the local area surrounding the parking space and the entire parking lot. Temporally, it analyzes the temporal characteristics of traffic participant behavior based on short-term historical trajectory information. This spatiotemporal correlation model can more accurately predict the timing and likelihood of parking space release, achieving higher precision than traditional models that only consider single-moment or spatially localized information.

[0055] In one embodiment of the present invention, the training of the prediction model based on the historical image features, the historical trajectory information, and the prediction result includes: Marking the prediction result and determining corresponding marking data; Dividing the historical image features, the historical trajectory information, and the labeled data according to a preset division ratio to obtain a training set and a test set; The prediction model is trained based on the training set and the test set.

[0056] During the model training process, the prediction results are first marked. If the traffic participant drives the vehicle out of the parking space after the observation period, the marked data is 1, indicating that the parking space is released; if the traffic participant still stays in the parking space, the marked data is 0, indicating that the parking space is not released; then the historical image features, historical trajectory information and marked data are divided according to the preset division ratio to obtain the training set and test set, and the prediction model is trained based on the training set and test set.

[0057] In practical applications, the 80 / 20 principle can be used to divide the data set into training and test sets, and the training parameters and evaluation indicators can be set to complete the training of the prediction model, effectively ensuring that the model learns the characteristic rules under sufficient training samples. At the same time, the generalization ability is verified through an independent test set to avoid overfitting.

[0058] Optionally, during the training of the prediction model, its update gate and reset gate mechanism can be used to flexibly capture short-term dynamic changes in the time series, learn the dynamic change patterns of traffic participants' behaviors in a short period of time, and thus predict whether they will leave the parking space.

[0059] In one embodiment of the present invention, the prediction model includes: a convolutional neural network and a gated recurrent neural network; Inputting the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the state of the parking space includes: Inputting the current raw data into the convolutional neural network to extract image features; Obtaining a fused feature vector according to the image features and the current trajectory information; The fused feature vector is input into the gated recurrent neural network to determine the state of the parking space.

[0060] In this embodiment, the prediction model includes a CNN (Convolutional Neural Network) component and a Gated Recurrent Unit (GRU) component, a type of gated recurrent neural network. A lightweight CNN is used to process the preprocessed current image data, extract spatial features, and reduce computing resource consumption. The GRU is used to analyze the dynamic changes in the current trajectory information of traffic participants over a short period of time.

[0061] Specifically, refer to Figure 3 , shows the architecture diagram of the prediction model provided by an embodiment of the present invention, which may specifically include the following contents: The convolutional neural network (CNN) consists of an input layer, two convolutional layers, two pooling layers, a fully connected layer, and an output layer. During the prediction phase, preprocessed image data is fed into the input layer. The convolutional layer uses a 3×3 convolution kernel to capture the local features of traffic participants. The pooling layer downsamples using max pooling. The fully connected layer integrates the convolutional and pooling features, and the output layer outputs image features related to the appearance and location of traffic participants.

[0062] The GRU gated recurrent neural network consists of an input layer, two gated recurrent neural network units, and an output layer. During the prediction phase, the image features output by the CNN convolutional neural network are fused with features such as speed, acceleration, and motion vector extracted from the current trajectory information to generate a fused feature vector. This fused feature vector is then fed into the input layer. The gated recurrent neural network unit then uses this fused feature vector to learn the dynamic changes in the current trajectory of traffic participants over a short period of time, and then outputs the predicted parking space status through the output layer.

[0063] The prediction model built based on the CNN convolutional neural network and the GRU gated recurrent neural network is able to predict the release of parking spaces through model training. It can accurately predict the release of parking spaces and improve the intelligence level and overall parking efficiency of the automatic parking system.

[0064] In one embodiment of the present invention, the status of the parking space includes: a first parking space status and a second parking space status; Inputting the fused feature vector into the gated recurrent neural network to determine the state of the parking space includes: Inputting the fused feature vector into the gated recurrent neural network to determine a corresponding probability value; When the probability value is greater than or equal to a probability threshold, determining the state of the parking space as a first parking space state; When the probability value is less than a probability threshold, the state of the parking space is determined to be a second parking space state.

[0065] In this embodiment, parking space states include a first parking space state and a second parking space state. The first parking space state indicates that the space is about to be released, while the second parking space state indicates that the space has not been released. After the current raw data and current trajectory information are input into the prediction model, the CNN component extracts image features from the current raw data and fuses them with real-time features such as movement speed, movement vector, and movement acceleration extracted from the current traffic participant's current trajectory information. The fused feature vector is input into the GRU component, which uses this fusion to predict whether the current traffic participant is about to leave the parking space. The prediction model outputs a probability value for the traffic participant's imminent departure. When the probability value is greater than or equal to a set probability threshold, the parking space state is determined to be the first parking space state, indicating that the space is about to be released. When the probability value is less than the set probability threshold, the parking space state is determined to be the second parking space state, indicating that the space has not been released. By effectively utilizing the current trajectory information of traffic participants over a short period of time to accurately predict parking space release, the system not only helps improve the efficiency and intelligence of the automated parking system, reduces vehicle waiting time, but also optimizes parking lot resource utilization, providing strong support for the development of intelligent transportation.

[0066] In an embodiment of the present invention, current raw data of traffic participants are obtained; traffic participants include vehicles and pedestrians within a preset range of a parking lot where a parking space is located; current trajectory information of the traffic participants is determined based on the current raw data; the current trajectory information represents the position change trajectory of the traffic participants at the current moment; the current raw data and the current trajectory information are input into a prediction model so that the prediction model outputs the status of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; the status of the parking space is displayed; the prediction model is obtained by determining the historical trajectory information of each traffic participant, and the feature learning ability of the prediction model and the rules contained in the historical trajectory data of the traffic participants are combined with the current raw data and the current trajectory information to achieve accurate prediction of the parking space release situation, thereby improving the intelligence level of the automatic parking system and the overall parking efficiency.

[0067] Reference Figure 4 , shows a structural block diagram of a parking space prediction device of the present application, which may specifically include the following modules: The data acquisition module 401 is used to acquire the current raw data of traffic participants; the traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; The current trajectory information determination module 402 is configured to determine the current trajectory information of the traffic participant based on the current raw data; the current trajectory information represents the position change trajectory of the traffic participant at the current moment; A prediction module 403 is configured to input the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the state of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; The display module 404 is used to display the status of the parking space.

[0068] In an embodiment of the present application, the parking space prediction device provided by the embodiment of the present application obtains the current original data of traffic participants; traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; the current trajectory information of the traffic participants is determined based on the current original data; the current trajectory information represents the position change trajectory of the traffic participants at the current moment; the current original data and the current trajectory information are input into the prediction model so that the prediction model outputs the status of the parking space; the prediction model is trained based on historical original data and historical trajectory information corresponding to the historical original data; the status of the parking space is displayed; the prediction model is obtained by determining the historical trajectory information of each traffic participant, and the feature learning ability of the prediction model and the rules contained in the historical trajectory data of the traffic participants are combined with the current original data and the current trajectory information to achieve accurate prediction of the parking space release situation, thereby improving the intelligence level of the automatic parking system and the overall parking efficiency.

[0069] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0070] The present invention also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the above-described parking space prediction method embodiment and achieves the same technical effects. To avoid repetition, the details are not described here. The computer-readable storage medium may be, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0071] An embodiment of the present application also provides a vehicle-mounted terminal, comprising: a processor, a memory, and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory, and when the computer program is executed by the processor, the various processes of the above-mentioned parking space prediction method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0072] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0073] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, apparatuses, or computer program products. Therefore, the embodiments of the present application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware. Furthermore, the embodiments of the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0074] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0075] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0077] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0078] Finally, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0079] It should also be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0080] The above is a detailed introduction to the parking space prediction method, device, and storage medium provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A parking space prediction method, characterized in that: The method comprises: Obtaining current raw data of traffic participants; the traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; Determining current trajectory information of the traffic participant based on the current raw data; the current trajectory information represents a position change trajectory of the traffic participant at the current moment; Inputting the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the state of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; Displays the status of the parking space.

2. The method according to claim 1, characterized in that The current original data includes: current image data and current distance data; the parking space includes a parking space sensor; The obtaining of current raw data of traffic participants includes: Acquire the current image data collected by the camera of the parking lot; The parking space sensor is used to acquire the current distance data between the traffic participant and the parking space.

3. The method according to claim 2, characterized in that The determining the current trajectory information of the traffic participant according to the current raw data includes: Identifying the position coordinates of each traffic participant in the current image data; Constructing movement data of the traffic participants according to the position coordinates of each traffic participant within a preset time range; Current trajectory information corresponding to the traffic participant is determined according to the current distance data and the movement data.

4. The method according to claim 3, characterized in that The movement data at least includes: movement speed, movement acceleration and movement vector; The determining the current trajectory information corresponding to the traffic participant according to the current distance data and the movement data includes: Determining a moving speed and a moving vector corresponding to the traffic participant according to the position coordinates; determining a movement acceleration corresponding to the traffic participant based on the movement speed; The movement data is corrected according to the current distance data to obtain current trajectory information corresponding to the traffic participant.

5. The method according to claim 1, wherein The prediction model is trained in the following way: Extracting historical image features of the historical original data; Determining a prediction result of the traffic participant leaving the parking space based on the historical image features and the historical trajectory information corresponding to the historical original data; The prediction model is trained based on the historical image features, the historical trajectory information and the prediction results.

6. The method according to claim 5, characterized in that The training of the prediction model based on the historical image features, the historical trajectory information, and the prediction result includes: Marking the prediction result and determining corresponding marking data; Dividing the historical image features, the historical trajectory information, and the labeled data according to a preset division ratio to obtain a training set and a test set; The prediction model is trained based on the training set and the test set.

7. The method according to claim 1, characterized in that The prediction model includes: a convolutional neural network and a gated recurrent neural network; Inputting the current raw data and the current trajectory information into a prediction model so that the prediction model outputs the state of the parking space includes: Inputting the current raw data into the convolutional neural network to extract image features; Obtaining a fused feature vector according to the image features and the current trajectory information; The fused feature vector is input into the gated recurrent neural network to determine the state of the parking space.

8. The method according to claim 7, characterized in that The parking space status includes: a first parking space status and a second parking space status; Inputting the fused feature vector into the gated recurrent neural network to determine the state of the parking space includes: Inputting the fused feature vector into the gated recurrent neural network to determine a corresponding probability value; When the probability value is greater than or equal to a probability threshold, determining the state of the parking space as a first parking space state; When the probability value is less than a probability threshold, the state of the parking space is determined to be a second parking space state.

9. A parking space prediction device, characterized in that: The device comprises: A data acquisition module is used to acquire current raw data of traffic participants; the traffic participants include vehicles and pedestrians within a preset range of the parking lot where the parking space is located; A current trajectory information determination module is used to determine the current trajectory information of the traffic participant based on the current raw data; the current trajectory information represents the position change trajectory of the traffic participant at the current moment; a prediction module, configured to input the current raw data and the current trajectory information into a prediction model, so that the prediction model outputs the state of the parking space; the prediction model is trained based on historical raw data and historical trajectory information corresponding to the historical raw data; The display module is used to display the status of the parking space.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the parking space prediction method according to any one of claims 1 to 8 is implemented.

Images