Freight vehicle loading and unloading point judgment method, device and equipment and storage medium

Abstract

The invention discloses a loading and unloading point judgment method, device and equipment of a freight vehicle and a storage medium. Comprising the following steps: extracting statistical characteristics of surrounding environment of a vehicle target parking point and generating statistical characteristic vectors; acquiring a satellite image map within a preset range around the target parking point of the vehicle, and constructing a satellite image map superposed with a vehicle movement track; inputting the satellite image map superposed with the vehicle motion trail into a pre-trained image feature coding model to generate an image feature vector; splicing the statistical feature vector and the image feature vector to generate a fused feature vector; and inputting the fusion feature vector into a pre-trained discrimination model to obtain a discrimination result about whether the target stop point is a loading and unloading point. According to the method, the trajectory data of the freight vehicle stop point, the surrounding satellite images and the statistical characteristics of the freight vehicle stop point are fused, and deep semantic information of different types of data is fully utilized, so that the model can judge the loading and unloading behaviors from multiple dimensions, and the judgment accuracy is remarkably improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent transportation technology, and more specifically, to a method, device, equipment, and storage medium for determining a loading and unloading point of a freight vehicle. Background Art

[0002] With the rapid development of the logistics industry, efficient management and monitoring of freight vehicles has become a key component in improving logistics efficiency. Accurately determining whether vehicles are loading or unloading at stops based on trajectory data during the freight process is crucial for optimizing logistics scheduling, reducing operating costs, and improving transportation efficiency. However, traditional methods for determining loading and unloading points rely primarily on single data sources, such as vehicle stop times or location information. These methods have numerous limitations and cannot accurately reflect loading and unloading behavior. Summary of the Invention

[0003] The embodiments of the present application provide a method, device, equipment and storage medium for determining the loading and unloading points of freight vehicles, so as to at least solve the technical problem in the related art that it is difficult to accurately determine the loading and unloading points of freight vehicles.

[0004] According to one aspect of an embodiment of the present application, a method for determining a loading and unloading point of a freight vehicle is provided, comprising:

[0005] Extract the statistical features of the environment around the target vehicle stop and generate a statistical feature vector;

[0006] Obtain satellite images within a preset range around the target vehicle stop and construct a satellite image with the vehicle's trajectory superimposed on it;

[0007] Inputting the satellite image with the superimposed vehicle motion trajectory into a pre-trained image feature encoding model to generate an image feature vector;

[0008] splicing the statistical feature vector and the image feature vector to generate a fused feature vector;

[0009] The fused feature vector is input into a pre-trained discriminant model to obtain a discrimination result of whether the target stop point is a loading and unloading point.

[0010] In one embodiment, extracting statistical features of the environment surrounding the target vehicle stop point and generating a statistical feature vector includes:

[0011] Obtaining one or more statistical data of POI type, number of vehicles, distribution of vehicle stop time intervals, distance to the nearest road, grade of the nearest road, and distance to previous and subsequent stops of the vehicle within a preset range of the target stop point;

[0012] The statistical feature vector is generated based on the statistical data.

[0013] In one embodiment, obtaining a satellite image within a preset range around a target vehicle stop and constructing a satellite image with a superimposed vehicle motion trajectory includes:

[0014] Obtain the vehicle's motion trajectory within a preset time range before and after the target stop point;

[0015] Convert the latitude and longitude of the target stop point into satellite image tile coordinates, and use this as the center to search and obtain the tile and the surrounding tiles within a preset range to obtain a satellite image map;

[0016] The movement trajectory within a preset time range before and after the docking is superimposed on the satellite image.

[0017] In one embodiment, the fused feature vector is input into a pre-trained discriminant model to obtain a discriminant result of whether the target stop point is a loading or unloading point, including:

[0018] Inputting the fused feature vector into a pre-trained discriminant model to obtain the probability of whether the target stop point is a loading and unloading point;

[0019] When the probability is greater than or equal to a preset threshold, the target stop point is determined to be a loading and unloading point.

[0020] In one embodiment, before inputting the satellite image with the superimposed vehicle motion trajectory into the pre-trained image feature coding model, the method further includes:

[0021] Automatically generate a training dataset containing positive and negative samples based on vehicle driving data;

[0022] Superimpose the trajectories in the positive and negative samples with the satellite image to obtain a superimposed satellite image;

[0023] Inputting the superimposed satellite image into a neural network for training;

[0024] After the training is completed, the last fully connected layer of the model is removed to obtain the image feature encoding model.

[0025] In one embodiment, a training data set containing positive and negative samples is automatically generated based on vehicle driving data, including:

[0026] Obtain the driving trajectories of multiple vehicles and identify the stop points in the trajectories;

[0027] According to the vehicle's stop points, the vehicle's driving trajectory is divided into segments containing the starting and ending points, and the stop points outside the starting and ending areas in the trajectory segments are used as negative samples;

[0028] Perform grid clustering on the vehicle stops to obtain suspected loading and unloading areas; then mark the suspected loading and unloading areas obtained by clustering with fences;

[0029] Perform point-to-surface calculations on the stop points and the marked fences, and the stop points within the fences are identified as positive samples;

[0030] A training data set is obtained based on the negative samples and the positive samples.

[0031] In one embodiment, before inputting the fused feature vector into the pre-trained discriminant model, the method further includes:

[0032] Training the discriminant model based on a machine learning model;

[0033] The machine learning model is an XGBoost, logistic regression or random forest model.

[0034] According to another aspect of the embodiments of the present application, a device for determining a loading and unloading point of a freight vehicle is provided, comprising:

[0035] Statistical feature extraction module, used to extract statistical features of the environment around the target vehicle stop and generate statistical feature vectors;

[0036] The satellite image extraction module is used to obtain satellite images within a preset range around the vehicle's target stop point and construct a satellite image with the vehicle's motion trajectory superimposed on it;

[0037] An image feature extraction module is used to input the satellite image with the superimposed vehicle motion trajectory into a pre-trained image feature encoding model to generate an image feature vector;

[0038] A feature fusion module, configured to concatenate the statistical feature vector and the image feature vector to generate a fused feature vector;

[0039] The discrimination module is used to input the fused feature vector into a pre-trained discrimination model to obtain a discrimination result of whether the target stop point is a loading and unloading point.

[0040] According to another aspect of an embodiment of the present application, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-mentioned method for determining a loading and unloading point of a freight vehicle through the above-mentioned computer program.

[0041] According to another aspect of the embodiment of the present application, a computer-readable storage medium is further provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for determining the loading and unloading point of a freight vehicle when running.

[0042] The technical solutions provided by the embodiments of the present application may have the following beneficial effects:

[0043] This application accurately determines the loading and unloading behavior of freight vehicles by fusing multi-source data. First, statistical features of the environment surrounding the stop are extracted and vectors are generated. This process can quantify the static environmental information of the stop and provide basic data support for subsequent analysis. Second, satellite imagery is acquired and overlaid with the vehicle's motion trajectory. This operation combines the vehicle's dynamic behavior with the surrounding geographical environment, allowing the model to understand the vehicle's docking behavior from a macro perspective. Third, image feature vectors are generated using a pre-trained image feature encoding model. This step leverages the powerful capabilities of deep learning to extract the rich information contained in the satellite imagery. The statistical feature vectors are concatenated with the image feature vectors to form a fused feature vector, achieving deep integration of multimodal data and providing a more comprehensive input for the model. Finally, the discrimination result is obtained using a pre-trained discriminant model. This process uses the efficient discrimination capabilities of machine learning to accurately determine whether the stop is a loading and unloading point. The entire process not only fully utilizes the advantages of multiple data sources, but also improves the efficiency and accuracy of data processing and analysis through deep learning and machine learning technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0045] Figure 1 is a flow chart of a method for determining a loading and unloading point of a freight vehicle according to an embodiment of the present application;

[0046] Figure 2 A satellite image with a superimposed motion trajectory according to an embodiment of the present application;

[0047] Figure 3 is a schematic diagram of trajectory segmentation according to an embodiment of the present application;

[0048] Figure 4 is a schematic diagram of a loading and unloading area fence according to an embodiment of the present application;

[0049] Figure 5 This is a schematic diagram of docking at a loading and unloading point according to an embodiment of the present application;

[0050] Figure 6 This is a schematic diagram of a non-loading and unloading point docking according to an embodiment of the present application;

[0051] Figure 7 This is a schematic diagram of a method for determining a loading and unloading point of a freight vehicle according to an embodiment of the present application;

[0052] Figure 8 This is a schematic diagram of a method for determining a loading and unloading point of a freight vehicle according to an embodiment of the present application;

[0053] Figure 9 This is a schematic diagram of a device for determining a loading and unloading point of a freight vehicle according to an embodiment of the present application;

[0054] Figure 10 It is a schematic structural diagram of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0056] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0057] This application relates to the field of trajectory data mining. Its purpose is to identify whether a vehicle is loading or unloading at a target stop by integrating three different modal data types: the trajectory of a freight vehicle before and after a stop (spatiotemporal sequence), satellite imagery (images) around the stop, and the statistical characteristics (vectors) of the stop itself, using a machine learning model.

[0058] The following is combined with Figure 1 The method for determining the loading and unloading points of freight vehicles in the embodiment of the present application is described in detail. Figure 1 As shown, the method mainly includes the following steps:

[0059] S101 extracts statistical features of the surrounding environment of the target vehicle stop point and generates a statistical feature vector.

[0060] In one embodiment, the vehicle's stop points are first acquired. Potential loading and unloading stops that meet the requirements are calculated from the trajectory data. First, the stop points are calculated based on a given stationary standard, taking into account factors such as trajectory point speed and offset range to ultimately obtain the stop points. Then, a duration threshold is set based on business knowledge, and stops that meet the duration requirements are selected as potential loading and unloading stops. The core fields of these fields include vehicle ID, longitude, latitude, stop start time, stop end time, and stop duration. Stop points where loading and unloading activities are likely to occur are obtained.

[0061] Then, we traverse the selected stops and select them as target stops. We then find the k-minute trajectory of the corresponding vehicle before and after the stop based on the vehicle ID and the start and end time of the stop, and retain the longitude, latitude, time and other information of each track point.

[0062] Furthermore, one or more statistical data are obtained, including POI type (service area, logistics park, factory area, gas station), number of vehicles, distribution of vehicle stop time intervals, distance to the nearest road, grade of the nearest road, and distance to the previous and subsequent stops of the vehicle within a preset range of the target stop point; and a statistical feature vector is generated based on the statistical data.

[0063] In one implementation, this data is quantified, for example, by converting POI types into category codes and converting vehicle stop time interval distributions into time-frequency distributions. Ultimately, this quantized data is integrated into a feature vector Vs of length m, which is used as input for subsequent models. This statistical feature vector effectively reflects the surrounding environment of the stop point and vehicle behavior patterns, providing rich semantic information for the discriminant model.

[0064] S102 obtains a satellite image within a preset range around the target stop point of the vehicle, and constructs a satellite image with the vehicle's motion trajectory superimposed thereon.

[0065] Specifically, first, the motion trajectory of the vehicle within a preset time range before and after the vehicle stops at the target stop point is obtained. For example, the motion trajectory within 2 minutes before and after the vehicle stops is obtained.

[0066] The latitude and longitude of the target stop point are converted into satellite image tile coordinates, and with this tile as the center, the tiles within the preset range of the surrounding area are searched and obtained to obtain the satellite image map. The specific size of the satellite image map is not subject to application and is not specifically limited. It can be set according to actual conditions.

[0067] Furthermore, the movement trajectory within a preset time range before and after the stop is superimposed on the satellite image.

[0068] In order to generate a satellite image containing the vehicle's trajectory, we first extract the target stop point and the vehicle trajectory data within k minutes before and after the stop point. These trajectory data include the latitude and longitude coordinates of the vehicle at each time point. Then, we plot these trajectory points into a trajectory map. Then, we overlay this trajectory map with the generated satellite image tiles. The overlaid satellite image is as follows: Figure 2 As shown in the figure, it not only contains the geographical environment information around the stop, but also intuitively shows the dynamic behavior of the vehicle in the area, providing rich visual information for subsequent feature extraction and model discrimination.

[0069] The reason for choosing to fuse trajectory sequences and satellite images in advance rather than in the prediction stage is that if features are extracted from the two types of data separately, the relative position information in the feature encoding will be lost. The relationship between the stop point and the surrounding environment is an important basis for judging the nature of loading and unloading.

[0070] Therefore, this application chooses to express the unique position and morphological information of the spatiotemporal sequence in satellite images. Figure 2 As shown in the figure, observing the satellite image or trajectory alone cannot lead to the conclusion that this candidate stop is a loading and unloading point. However, after superimposing the data, the semantics of "the vehicle enters the factory area and stops" can be clearly obtained.

[0071] S103 inputs the satellite image with the vehicle motion trajectory superimposed thereon into a pre-trained image feature encoding model to generate an image feature vector.

[0072] The satellite image with the superimposed vehicle motion trajectory is encoded through a trained image feature encoding model to finally generate a feature vector Vp of length n. The deep learning model used in this application includes but is not limited to CNN.

[0073] In one embodiment, before inputting the satellite image with the superimposed vehicle motion trajectory into the pre-trained image feature coding model, the method further includes training the image feature coding model.

[0074] Specifically, a training dataset containing positive and negative samples is automatically generated based on vehicle driving data. This application proposes a method for batch production of positive and negative samples, which can greatly reduce the cost of manual labeling.

[0075] In one embodiment, the driving trajectories of multiple vehicles are obtained and the stop points in the trajectories are identified. The vehicle driving trajectories are divided into segments containing the start and end points according to the vehicle stop points, and the stop points outside the start and end points in the trajectory segments are used as negative samples.

[0076] like Figure 3The figure shows the trajectory of a vehicle within a certain time range. The points in the figure represent the vehicle's intermediate stops. By calculating the long-distance deflection angle, the vehicle's trajectory can be divided into two segments: Chongqing-Haixi and Haixi-Lhasa. Loading and unloading only occur in these three cities. Points outside the starting and ending areas can be directly used as negative samples without additional labeling.

[0077] Furthermore, the vehicle stops are clustered in a grid-like manner to obtain suspected loading and unloading areas; the suspected loading and unloading areas obtained by clustering are marked with fences; the stop points and the marked fences are calculated based on point-to-surface calculations, and the stop points within the fences are identified as positive samples.

[0078] First, all the long-term stops of vehicles are clustered into grids to form clues of suspected loading and unloading areas, such as Figure 4 The arrows point to the area. The clues are then handed over to humans for fence annotation. The arrows in the image above indicate the annotated fence. After generating a sufficiently diverse set of fences (e.g., logistics parks, enterprises, orchards, farms, mining areas, docks, etc.), all vehicle stops are then compared against the annotated fences for point-to-surface calculations. Ultimately, the stops within the fences are considered positive samples.

[0079] Furthermore, a training data set is obtained based on negative samples and positive samples.

[0080] The trajectories from the positive and negative samples are superimposed on the satellite image to obtain a superimposed satellite image. This superimposed satellite image is then fed into a neural network for training. After training, the final fully connected layer of the model is removed to obtain an image feature encoding model.

[0081] S104 concatenates the statistical feature vector and the image feature vector to generate a fused feature vector.

[0082] In one embodiment, the generated m-dimensional statistical feature vector Vs is concatenated with the generated n-dimensional image feature vector Vp to generate an m+n-dimensional fusion feature vector Vf, ultimately achieving a fusion representation of multimodal data such as trajectory sequences, images, and structured vectors.

[0083] By fusing multimodal data, the model can understand the characteristics of docking points from multiple perspectives, thereby improving the accuracy of identifying loading and unloading behaviors. The fused feature vector contains rich information, allowing the model to better adapt to different scenarios and conditions, enhancing its generalization capabilities.

[0084] S105 inputs the fused feature vector into a pre-trained discriminant model to obtain a discrimination result of whether the target stop point is a loading and unloading point.

[0085] In one embodiment, the fused feature vector is input into a pre-trained discriminant model to obtain a discriminant result of whether the target stop is a loading or unloading point, including:

[0086] The fused feature vector is input into a pre-trained discriminant model to determine the probability of the target stop being a loading or unloading point. If the probability is greater than or equal to a preset threshold, the target stop is determined to be a loading or unloading point. The preset probability threshold can be set based on actual conditions.

[0087] The fused feature vector generated in the previous step is input into the pre-trained discriminant model. This vector contains the statistical features and image features of the vehicle stop and is the result of multimodal data fusion.

[0088] The discriminant model calculates and outputs a probability value based on the input fused feature vector. This probability value indicates the likelihood of loading or unloading at the target stop.

[0089] Based on the actual application scenario and requirements, a probability threshold is set. This threshold is a value between 0 and 1, which is used to determine when a stop is considered a loading or unloading point.

[0090] In an exemplary scenario, the threshold is 0.7. When the probability output by the discriminant model is greater than or equal to 0.7, the target stop is determined to be a loading and unloading point. The vehicle loads and unloads cargo when it stops at this point.

[0091] Before inputting the fused feature vector into the pre-trained discriminant model, the method further includes: training the discriminant model based on a machine learning model; the machine learning model is an XGBoost, logistic regression, or random forest model. This application does not make any specific restrictions.

[0092] like Figure 5 As shown in , this is an example of a stop that the discriminant model identifies as a loading and unloading point. Figure 6 The figure shows an example of a stop that is identified by the discriminant model as a non-loading or unloading point, but the actual stop semantics are roadside stops. This application can accurately identify whether a stop is a loading or unloading point.

[0093] In order to facilitate understanding of the method of the embodiment of the present application, the following Figure 7 Further explanation.

[0094] like Figure 7 As shown in the figure, this figure shows a flowchart for determining whether loading and unloading occurs at a freight vehicle stop. The flowchart is divided into several main steps:

[0095] Stop point calculation: First, the vehicle's stop point needs to be calculated, which is usually determined based on the vehicle's trajectory data by setting certain speed and time thresholds.

[0096] Stop trajectory extraction: Extract the trajectory information related to the stop point from the vehicle's trajectory data, which may include the trajectory within a certain period of time before and after the stop.

[0097] Satellite image matching: Obtain satellite images around the stop and possibly match them with trajectory data for subsequent image processing.

[0098] Stop feature calculation: Calculate the statistical features related to the stop points, such as POI type, number of vehicles, stop time distribution, etc., and generate statistical feature vectors.

[0099] Satellite image overlay trajectory: The extracted trajectory information is overlaid on the satellite image to generate a satellite image map containing the vehicle stop point identification and the front and rear trajectories.

[0100] Image feature coding: Use a pre-trained image feature coding model to process satellite images with superimposed trajectories, extract image features, and generate image feature vectors.

[0101] Feature fusion: Fuse the image feature vector with the statistical feature vector to form a comprehensive feature vector.

[0102] Model discrimination: The fused feature vector is input into the pre-trained discrimination model, and the model will output the discrimination result of whether the stop is a loading and unloading point.

[0103] This flowchart describes a multimodal data processing and fusion system that combines vehicle trajectories, satellite imagery, and statistical data to improve the accuracy of vehicle loading and unloading behavior.

[0104] like Figure 8 As shown in the figure, the data modality conversion process can ultimately convert various types of feature data related to freight vehicle stops, such as spatiotemporal sequences, images, and numerical vectors, into a unified feature vector, which is then handed over to the discriminant model for processing. The flowchart is divided into several main steps:

[0105] Multimodal feature extraction: Stop Trajectory - Spatiotemporal Sequence: Extracts spatiotemporal sequence features associated with stops from vehicle trajectory data. Satellite Imagery - Imagery: Acquires satellite imagery of vehicle stops as an image data source. Stop Feature Vector: Calculates statistical features associated with stops and generates feature vectors.

[0106] Image Feature Coding: Overlaying Track Satellite Imagery - Image: Extracted track information is overlaid onto satellite imagery to generate a satellite image that includes vehicle stop identification and the preceding and following tracks. Image Feature Coding - Vector: Image processing techniques (such as deep learning models) are used to perform feature encoding on the overlaid track satellite imagery to generate an image feature vector.

[0107] Feature concatenation: Concatenates the image feature vector and the statistical feature vector to form a unified feature vector. This unified feature vector integrates information from different modalities (spatiotemporal sequence, image, and statistical features) to more comprehensively describe the characteristics of the stop.

[0108] By integrating multiple types of data (trajectory data, satellite imagery, and statistical data) to extract rich features and ultimately generate a unified feature vector, this approach can improve the understanding of stop characteristics and provide more comprehensive information support for subsequent analysis and decision-making.

[0109] This solution significantly improves the accuracy and efficiency of identifying loading and unloading points for freight vehicles by integrating multimodal data such as stop trajectories, satellite imagery, and statistical features. First, the fusion of multimodal data enables the model to understand the characteristics of stop points from multiple dimensions, such as spatiotemporal sequences, images, and statistical information, thereby more accurately identifying loading and unloading behaviors. Second, by overlaying satellite imagery and trajectory data, the model can capture the dynamic interaction between vehicles and the environment, enhancing its understanding of stop behavior. In addition, the unified feature vector simplifies the data processing process and improves the efficiency of model training and prediction. Finally, by reducing reliance on manual annotation, this solution reduces costs while enhancing the model's generalization capabilities.

[0110] According to another aspect of the embodiment of the present application, there is also provided a device for determining a loading and unloading point of a freight vehicle for implementing the above-mentioned method for determining a loading and unloading point of a freight vehicle. Figure 9 As shown, the device includes:

[0111] Statistical feature extraction module 901, used to extract statistical features of the surrounding environment of the target vehicle stop point and generate a statistical feature vector;

[0112] Satellite image extraction module 902, used to obtain satellite images within a preset range around the target vehicle stop point and construct a satellite image with the vehicle's motion trajectory superimposed on it;

[0113] Image feature extraction module 903, used to input the satellite image with the vehicle motion trajectory superimposed thereon into a pre-trained image feature encoding model to generate an image feature vector;

[0114] A feature fusion module 904 is used to combine the statistical feature vector and the image feature vector to generate a fused feature vector;

[0115] The discrimination module 905 is used to input the fused feature vector into a pre-trained discrimination model to obtain a discrimination result of whether the target stop point is a loading and unloading point.

[0116] It should be noted that the above-mentioned embodiment of the device for determining the loading and unloading point of a freight vehicle, when executing the method for determining the loading and unloading point of a freight vehicle, only uses the division of the above-mentioned functional modules as an example. In actual application, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the above-mentioned embodiment of the device for determining the loading and unloading point of a freight vehicle and the embodiment of the method for determining the loading and unloading point of a freight vehicle are based on the same concept. The implementation process is detailed in the method embodiment and will not be repeated here.

[0117] According to another aspect of an embodiment of the present application, an electronic device corresponding to the method for determining a loading and unloading point of a freight vehicle provided in the aforementioned embodiment is also provided to execute the aforementioned method for determining a loading and unloading point of a freight vehicle.

[0118] Please refer to Figure 10 , which shows a schematic diagram of an electronic device provided by some embodiments of the present application. Figure 10 As shown, the electronic device includes: a processor 1000, a memory 1001, a bus 1002 and a communication interface 1003, and the processor 1000, the communication interface 1003 and the memory 1001 are connected via the bus 1002; the memory 1001 stores a computer program that can be run on the processor 1000, and when the processor 1000 runs the computer program, it executes the method for determining the loading and unloading point of a freight vehicle provided in any of the aforementioned embodiments of the present application.

[0119] The memory 1001 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. The system network element and at least one other network element are connected via at least one communication interface 1003 (which may be wired or wireless), and may utilize the Internet, a wide area network, a local area network, a metropolitan area network, or the like.

[0120] Bus 1002 may be an ISA bus, a PCI bus, or an EISA bus. Buses may be classified as address buses, data buses, and control buses. Memory 1001 is used to store programs, and processor 1000 executes the programs upon receiving execution instructions. The method for determining a loading and unloading point for a freight vehicle disclosed in any of the aforementioned embodiments of the present application may be applied to or implemented by processor 1000.

[0121] The processor 1000 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor 1000 or by software instructions. The above processor 1000 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of this application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory 1001 , and the processor 1000 reads the information in the memory 1001 and completes the steps of the above method in combination with its hardware.

[0122] The electronic device provided in the embodiment of the present application and the method for determining the loading and unloading point of a freight vehicle provided in the embodiment of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, operated or implemented therein.

[0123] According to another aspect of the embodiments of the present application, a computer-readable storage medium corresponding to the method for determining the loading and unloading points of freight vehicles provided in the aforementioned embodiments is also provided, on which a computer program (i.e., a program product) is stored. When the computer program is run by the processor, it will execute the method for determining the loading and unloading points of freight vehicles provided in any of the aforementioned embodiments.

[0124] It should be noted that examples of computer-readable storage media may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0125] The computer-readable storage medium provided in the above-mentioned embodiments of the present application and the method for determining the loading and unloading points of freight vehicles provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0126] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] The above embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for determining a loading and unloading point of a freight vehicle, characterized in that: include: Extract the statistical features of the environment around the target vehicle stop and generate a statistical feature vector; Obtain satellite images within a preset range around the target vehicle stop and construct a satellite image with the vehicle's trajectory superimposed on it; Inputting the satellite image with the superimposed vehicle motion trajectory into a pre-trained image feature encoding model to generate an image feature vector; splicing the statistical feature vector and the image feature vector to generate a fused feature vector; The fused feature vector is input into a pre-trained discriminant model to obtain a discrimination result of whether the target stop point is a loading and unloading point.

2. The method according to claim 1, characterized in that Extract the statistical features of the surrounding environment of the vehicle target stop and generate a statistical feature vector, including: Obtaining one or more statistical data of POI type, number of vehicles, distribution of vehicle stop time intervals, distance to the nearest road, grade of the nearest road, and distance to previous and subsequent stops of the vehicle within a preset range of the target stop point; The statistical feature vector is generated based on the statistical data.

3. The method according to claim 1, characterized in that Obtain satellite imagery within a preset range around the target vehicle stop and construct a satellite imagery with the vehicle's trajectory superimposed on it, including: Obtain the vehicle's motion trajectory within a preset time range before and after the target stop point; Convert the latitude and longitude of the target stop point into satellite image tile coordinates, and use this as the center to search and obtain the tile and the surrounding tiles within a preset range to obtain a satellite image map; The movement trajectory within a preset time range before and after the docking is superimposed on the satellite image.

4. The method according to claim 1, wherein The fused feature vector is input into the pre-trained discriminant model to obtain a discriminant result of whether the target stop point is a loading and unloading point, including: Inputting the fused feature vector into a pre-trained discriminant model to obtain the probability of whether the target stop point is a loading and unloading point; When the probability is greater than or equal to a preset threshold, the target stop point is determined to be a loading and unloading point.

5. The method according to claim 1, wherein Before inputting the satellite image with the superimposed vehicle motion trajectory into the pre-trained image feature coding model, the method further includes: Automatically generate a training dataset containing positive and negative samples based on vehicle driving data; Superimpose the trajectories in the positive and negative samples with the satellite image to obtain a superimposed satellite image; Inputting the superimposed satellite image into a neural network for training; After the training is completed, the last fully connected layer of the model is removed to obtain the image feature encoding model.

6. The method according to claim 5, characterized in that Automatically generate a training dataset containing positive and negative samples based on vehicle driving data, including: Obtain the driving trajectories of multiple vehicles and identify the stop points in the trajectories; According to the vehicle's stop points, the vehicle's driving trajectory is divided into segments containing the starting and ending points, and the stop points outside the starting and ending areas in the trajectory segments are used as negative samples; Perform grid clustering on the vehicle stops to obtain suspected loading and unloading areas; then mark the suspected loading and unloading areas obtained by clustering with fences; Perform point-to-surface calculations on the stop points and the marked fences, and the stop points within the fences are identified as positive samples; A training data set is obtained based on the negative samples and the positive samples.

7. The method according to claim 1, characterized in that Before inputting the fused feature vector into the pre-trained discriminant model, the method further includes: Training the discriminant model based on a machine learning model; The machine learning model is an XGBoost, logistic regression or random forest model.

8. A device for determining a loading and unloading point of a freight vehicle, characterized in that: include: Statistical feature extraction module, used to extract statistical features of the environment around the target vehicle stop and generate statistical feature vectors; The satellite image extraction module is used to obtain satellite images within a preset range around the vehicle's target stop point and construct a satellite image with the vehicle's motion trajectory superimposed on it; An image feature extraction module is used to input the satellite image with the superimposed vehicle motion trajectory into a pre-trained image feature encoding model to generate an image feature vector; A feature fusion module, configured to concatenate the statistical feature vector and the image feature vector to generate a fused feature vector; The discrimination module is used to input the fused feature vector into a pre-trained discrimination model to obtain a discrimination result of whether the target stop point is a loading and unloading point.

9. An electronic device, characterized in that: The system comprises a processor and a memory storing program instructions, wherein the processor is configured to execute the method for determining a loading and unloading point of a freight vehicle according to any one of claims 1 to 7 when executing the program instructions.

10. A computer-readable medium, characterized in that Computer-readable instructions are stored thereon, and the computer-readable instructions are executed by a processor to implement a method for determining a loading and unloading point of a freight vehicle as described in any one of claims 1 to 7.

Images