Video-assisted positioning method and device based on artificial intelligence, equipment and medium

Abstract

The invention discloses a video-assisted positioning method and device based on artificial intelligence, equipment and a medium, and relates to the technical field of vehicle positioning, and the method comprises the steps: collecting road video information and correction wheel speed meter information; and inputting the video road information and the correction wheel speed meter information into a deep learning positioning calculation model to obtain vision-assisted positioning information, the deep learning positioning calculation model comprises a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long-short-term memory network and a double-layer long-short-term memory network; judging whether satellite positioning fails or not or whether satellite positioning deviates according to satellite positioning information and the visual auxiliary positioning information; and when satellite positioning failure or satellite positioning deviation is judged, outputting the vision-assisted positioning information. The problem of inaccurate positioning of a single satellite can be solved.

Description

Technical Field

[0001] The present application relates to the field of vehicle positioning technology, and in particular to a video-assisted positioning method, device, equipment and medium based on artificial intelligence. Background Art

[0002] With the rapid development of global satellite positioning technology, satellite navigation systems represented by Beidou and GPS have achieved significant improvements in positioning accuracy. However, although satellite positioning technology can theoretically provide extremely high positioning accuracy, the satellite positioning equipment of the vehicle may be interfered with by various factors during driving, resulting in deviation of vehicle satellite positioning or even loss of positioning.

[0003] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0004] The main purpose of this application is to provide a video-assisted positioning method, device, equipment and medium based on artificial intelligence, aiming to solve the technical problem of vehicle satellite positioning deviation or even loss of positioning caused by interference from various factors to the satellite positioning equipment during vehicle driving.

[0005] To achieve the above objectives, the present application proposes a video-assisted positioning method based on artificial intelligence, and the video-assisted positioning method based on artificial intelligence includes:

[0006] Collect road video information and correct wheel speedometer information;

[0007] Inputting the video road information and the corrected wheel speedometer information into a deep learning positioning calculation model to obtain visual auxiliary positioning information, wherein the deep learning positioning calculation model includes a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a double-layer long short-term memory network;

[0008] Determining whether satellite positioning is invalid or deviated according to the satellite positioning information and the visual auxiliary positioning information;

[0009] When it is determined that the satellite positioning fails or the satellite positioning deviates, the visual auxiliary positioning information is output.

[0010] In one embodiment, before the step of inputting the video road information and the corrected wheel speedometer information into the deep learning positioning calculation model to obtain the visual auxiliary positioning information, the step further includes:

[0011] Collecting initial training data in different scenarios, wherein the initial training data includes target satellite positioning information, target wheel speed meter information, and target road video information;

[0012] Performing elimination processing on the initial training data to obtain first training data;

[0013] Performing anomaly location and addition processing on the first training data to obtain second training data;

[0014] Performing completion processing on the second training data to obtain target training data;

[0015] An initial deep learning positioning calculation model is trained based on the target training data to obtain a deep learning positioning calculation model.

[0016] In one embodiment, the step of inputting the video road information and the corrected wheel speedometer information into a deep learning positioning calculation model to obtain visually assisted positioning information includes:

[0017] Inputting the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model, and extracting a current frame feature map and a next frame feature map of a current frame and a next frame in the video road information through the double-layer convolutional neural network;

[0018] The current frame feature map and the next frame feature map are superimposed into a target feature map through the first connection layer, and the target feature map is converted into a graph feature vector through the convolutional neural network and the pooling layer;

[0019] Extracting the speed feature vectors of the current frame wheel speed meter speed and the next frame wheel speed meter speed in the corrected wheel speed meter information through the long short-term memory network;

[0020] Connecting the graph feature vector and the speed feature vector into a target feature vector through the second connection layer;

[0021] The visual auxiliary positioning information is output according to the target feature vector through the double-layer long short-term memory network.

[0022] In one embodiment, the step of outputting visual auxiliary positioning information according to the target feature vector through the double-layer long short-term memory network includes:

[0023] Outputting a positioning offset according to the target feature vector through the double-layer long short-term memory network;

[0024] Based on the positioning offset and the visual positioning reference point, visually assisted positioning information is determined.

[0025] In one embodiment, the steps of collecting road video information and correcting wheel speedometer information include:

[0026] Collect road video information and wheel speed meter information;

[0027] Determine the deviation wheel speedometer speed and satellite positioning speed of the vehicle in the satellite positioning stable area;

[0028] Determine a wheel speed calculation offset according to the deviation wheel speed meter speed and the satellite positioning speed;

[0029] Determining a wheel speed meter calibration coefficient based on the wheel speed calculated offset;

[0030] Corrected wheel speedometer information is determined according to the wheel speedometer calibration coefficient and the wheel speedometer information.

[0031] In one embodiment, after the step of determining whether satellite positioning is invalid or whether satellite positioning is deviated according to the satellite positioning information and the visual auxiliary positioning information, the step further includes:

[0032] When it is determined that the satellite positioning is not failed and not offset, the visual positioning reference point and the wheel speed calculation offset are updated to determine the visual assisted positioning information under satellite positioning failure or satellite positioning offset based on the updated visual positioning reference point and the wheel speed calculation offset.

[0033] In one embodiment, the step of determining satellite positioning failure or satellite positioning deviation according to the satellite positioning information and the visual auxiliary positioning information includes:

[0034] Determining whether satellite positioning is invalid according to the reception status of the satellite positioning information;

[0035] Or, determine the deviation of the satellite positioning information and the visual auxiliary positioning information, and judge whether the satellite positioning is deviated according to the deviation.

[0036] In addition, to achieve the above-mentioned purpose, the present application also proposes a video-assisted positioning device based on artificial intelligence, and the video-assisted positioning device based on artificial intelligence includes:

[0037] A collection module, used to collect road video information and correct wheel speed meter information;

[0038] An input module, used to input the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model to obtain visual auxiliary positioning information, wherein the deep learning positioning calculation model includes a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a double-layer long short-term memory network;

[0039] A judgment module, used to judge whether the satellite positioning is invalid or the satellite positioning is deviated according to the satellite positioning information and the visual auxiliary positioning information;

[0040] The output module is used to output the visual auxiliary positioning information when it is determined that the satellite positioning fails or the satellite positioning deviates.

[0041] In addition, to achieve the above-mentioned objectives, the present application also proposes an artificial intelligence-based video-assisted positioning device, the device comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the artificial intelligence-based video-assisted positioning method as described above.

[0042] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the video-assisted positioning method based on artificial intelligence as described above are implemented.

[0043] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the artificial intelligence-based video-assisted positioning method as described above.

[0044] One or more technical solutions proposed in this application have at least the following technical effects:

[0045] The video-assisted positioning method, device, equipment and medium based on artificial intelligence proposed in the present application collect road video information and correct wheel speed meter information; input the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model to obtain visually assisted positioning information, wherein the deep learning positioning calculation model includes a two-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network and a two-layer long short-term memory network; judge whether the satellite positioning fails or the satellite positioning deviates according to the satellite positioning information and the visually assisted positioning information; when it is determined that the satellite positioning fails or the satellite positioning deviates, the visually assisted positioning information is output, which solves the technical problem that the satellite positioning of the vehicle deviates or even disappears due to interference from various factors on the satellite positioning device during the driving of the vehicle. Compared with the prior art, the present application can output visually assisted positioning information according to the road video information and the corrected wheel speed meter information in the deep learning positioning calculation model deployed on the vehicle when the vehicle satellite positioning deviates or even disappears, which can avoid affecting the vehicle driving due to the deviation of the vehicle satellite positioning or even the disappearance of the positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0048] Figure 1 A flowchart of the first embodiment of the video-assisted positioning method based on artificial intelligence of the present application is provided;

[0049] Figure 2 A neural network diagram provided for the first embodiment of the video-assisted positioning method based on artificial intelligence of the present application;

[0050] Figure 3 A video-assisted positioning application flow chart provided for the first embodiment of the video-assisted positioning method based on artificial intelligence of the present application;

[0051] Figure 4 A flowchart of the second embodiment of the video-assisted positioning method based on artificial intelligence of the present application is provided;

[0052] Figure 5 This is a schematic diagram of the module structure of the video-assisted positioning device based on artificial intelligence in an embodiment of the present application;

[0053] Figure 6 This is a schematic diagram of the device structure of the hardware operating environment involved in the artificial intelligence-based video-assisted positioning method in the embodiment of the present application.

[0054] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0055] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0056] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0057] The main solution of the embodiment of the present application is: collecting road video information and correcting wheel speedometer information; inputting the video road information and the corrected wheel speedometer information into a deep learning positioning calculation model to obtain visually assisted positioning information, wherein the deep learning positioning calculation model includes a two-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network and a two-layer long short-term memory network; judging whether the satellite positioning is invalid or the satellite positioning is deviated according to the satellite positioning information and the visually assisted positioning information; when it is judged that the satellite positioning is invalid or the satellite positioning is deviated, outputting the visually assisted positioning information.

[0058] It can be seen from the above embodiments that the present application obtains visually assisted positioning information by collecting road video information and correcting wheel speedometer information; inputting the video road information and the corrected wheel speedometer information into a deep learning positioning calculation model, wherein the deep learning positioning calculation model includes a two-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a two-layer long short-term memory network; judging whether the satellite positioning fails or deviates according to the satellite positioning information and the visually assisted positioning information; when it is determined that the satellite positioning fails or the satellite positioning deviates, the visually assisted positioning information is output, which solves the technical problem that the satellite positioning of the vehicle deviates or even disappears due to interference from various factors to the satellite positioning device during the driving process of the vehicle. Compared with the prior art, the present application can output visually assisted positioning information according to the road video information and the corrected wheel speedometer information in the deep learning positioning calculation model deployed on the vehicle when the vehicle satellite positioning deviates or even disappears, which can avoid affecting the vehicle driving due to the deviation of the vehicle satellite positioning or even the disappearance of the positioning.

[0059] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, an artificial intelligence-based video-assisted positioning device, etc. The following takes an artificial intelligence-based video-assisted positioning device as an example to illustrate this embodiment and the following embodiments.

[0060] Based on this, the embodiment of the present application provides a video-assisted positioning method based on artificial intelligence, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the video-assisted positioning method based on artificial intelligence in this application.

[0061] In this embodiment, the video-assisted positioning method based on artificial intelligence includes steps S10 to S40:

[0062] Step S10, collecting road video information and correcting wheel speed meter information;

[0063] It should be noted that the road video information can be collected in real time by the vehicle's onboard camera, and the corrected wheel speed information can also be determined by a device that measures the tire rotation speed and tire diameter to calculate the vehicle speed (i.e., a wheel speed meter).

[0064] In a feasible implementation manner, the steps of collecting road video information and correcting wheel speedometer information include: collecting road video information and wheel speedometer information; determining the deviated wheel speedometer speed and the satellite positioning speed of the vehicle in the satellite positioning stable area; determining the wheel speed calculation offset according to the deviated wheel speedometer speed and the satellite positioning speed; determining the wheel speedometer calibration coefficient according to the wheel speed calculation offset; and determining the corrected wheel speedometer information according to the wheel speedometer calibration coefficient and the wheel speedometer information.

[0065] It is understandable that the wheel speed meter will be affected by factors such as tire wear, tire pressure, and vehicle load. The wheel speed meter speed output by the wheel speed meter may deviate from the actual speed. Therefore, the wheel speed meter speed obtained by the wheel speed meter needs to be corrected.

[0066] It should be noted that the satellite positioning stable area refers to the area where the satellite positioning signal is better during the vehicle's driving process. When the vehicle is driving in the satellite positioning stable area, the wheel speed calculation offset can be calculated based on the wheel speed meter speed and satellite positioning speed at this time, and then the wheel speed meter calibration coefficient is calculated based on this deviation (i.e., the wheel speed calculation offset). When the vehicle is driving on a section where the positioning signal fails, the currently acquired wheel speed meter speed can be multiplied by the deviation coefficient to obtain more accurate vehicle speed information (i.e., corrected wheel speed meter information).

[0067] Step S20, inputting the video road information and the corrected wheel speedometer information into a deep learning positioning calculation model to obtain visual auxiliary positioning information, wherein the deep learning positioning calculation model includes a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a double-layer long short-term memory network;

[0068] It should be noted that the deep learning positioning calculation model is a trained model, which is deployed on the vehicle terminal and can output accurate positioning information in real time; the deep learning positioning calculation model is an end-to-end deep learning positioning calculation model structure.

[0069] In the specific implementation, Figure 2As shown in the figure, the deep learning positioning calculation model includes a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network (CNN), a pooling layer, a long short-term memory network (LSTM) and a double-layer long short-term memory network. Among them, the FlowNetCorr structure commonly used in the optical flow method is used to extract the image features of the front and back frame images. This structure imitates the standard matching process, first extracting features from the two front and back images respectively, and then merging and calculating the correlation between the features. At the same time, in order to increase the feature extraction ability of the network, the convolution layer is replaced with a C2f network structure with a residual structure (that is, a double-layer convolutional neural network). The wheel speed meter speed corresponding to the two video frames is connected to the feature vector of the picture through a layer of LSTM structure, and the input speed modal information can further improve the accuracy of the positioning calculation.

[0070] In a feasible implementation manner, before the step of inputting the video road information and the corrected wheel speedometer information into the deep learning positioning calculation model to obtain the visual assisted positioning information, the step also includes: collecting initial training data under different scenarios, wherein the initial training data includes target satellite positioning information, target wheel speedometer information and target road video information; performing elimination processing on the initial training data to obtain first training data; performing abnormal positioning addition processing on the first training data to obtain second training data; performing completion processing on the second training data to obtain target training data; and training the initial deep learning positioning calculation model based on the target training data to obtain a deep learning positioning calculation model.

[0071] It should be noted that different scenarios refer to scenarios with different weather, road conditions, speeds, etc. When collecting initial training data, it is necessary to cover data in different scenarios to ensure the output accuracy of the deep learning positioning calculation model obtained through the final training.

[0072] It should be noted that the initial deep learning positioning calculation model refers to an untrained model; after obtaining the target training data, the target training data needs to be divided into a training set and a validation set, and then the initial deep learning calculation model is trained based on the training set and the validation set.

[0073] In the specific implementation, the vehicle-mounted camera first collects road video information, corrected wheel speedometer information, and satellite positioning information. The collected data covers various weather, road conditions, speed and other scenarios. Secondly, the data is processed. In sections where positioning deviation or positioning failure occurs, the positioning is matched to the map based on speed and map. The positioning and speed are interpolated according to the video frame rate to obtain the positioning and speed information corresponding to each video frame in the target road video information.

[0074] In the specific implementation, the data processing method for the initial training data includes the following parts:

[0075] The first is abnormal information processing (i.e., the initial training data is eliminated to obtain the first training data), and some missing or damaged data or data that has been parked for a long time in the initial training data is eliminated; the second is to identify and extract abnormal positioning information (i.e., the second training data is obtained after the first training data is processed for abnormal positioning). Through the analysis of the continuity and stability of positioning, the comparison of the vehicle speed of the satellite positioning information and the actual vehicle speed, etc., the sections where positioning deviation or positioning failure occurs are found, and the proportion of abnormal positioning data in the data set is increased in the first training data; the third is to process the data (i.e., the target training data is obtained after the second training data is completed). If the visualization finds that the trajectory of the section has indeed deviated or is missing, it is necessary to model the trajectory based on the road information and speed information of the map, complete the positioning information of the section where positioning failure or deviation occurs, and finally convert the satellite positioning coordinates into plane coordinates and normalize the data.

[0076] Step S30, judging whether the satellite positioning is invalid or deviated according to the satellite positioning information and the visual auxiliary positioning information;

[0077] In a feasible implementation, the step of determining satellite positioning failure or satellite positioning deviation based on the satellite positioning information and the visually assisted positioning information includes: determining whether satellite positioning has failed based on the reception status of the satellite positioning information; or, determining the deviation between the satellite positioning information and the visually assisted positioning information, and determining whether the satellite positioning has deviated based on the deviation.

[0078] It should be noted that when the satellite positioning disappears and the satellite positioning information cannot be received, it can be determined that the satellite positioning has failed. When the deviation between the satellite positioning information and the visual auxiliary positioning information is too large, it can be determined that the satellite positioning has deviated.

[0079] Step S40: when it is determined that the satellite positioning fails or the satellite positioning deviates, the visual auxiliary positioning information is output.

[0080] In a feasible implementation manner, the step of determining whether satellite positioning has failed or whether satellite positioning has deviated based on the satellite positioning information and the visual assisted positioning information also includes: when it is determined that satellite positioning has not failed and has not deviated, outputting satellite positioning information, and updating the visual positioning reference point and the wheel speed calculation offset, so as to determine the visual assisted positioning information under satellite positioning failure or satellite positioning deviation based on the updated visual positioning reference point and the wheel speed calculation offset.

[0081] It should be noted that when it is determined that the satellite positioning is not invalid and not offset, the satellite positioning information can be directly output, and the visual positioning reference point and the wheel speed calculation offset can be updated according to the satellite positioning information at this time.

[0082] In the specific implementation, Figure 3 As shown in the figure, the satellite positioning information and the visual auxiliary positioning information are obtained at the same time, and then it is determined whether the satellite positioning fails or whether the positioning deviation occurs. If the satellite positioning fails or deviates, the visual auxiliary positioning information is output. If the satellite positioning is normal, the reference point of the visual positioning is updated and the wheel speed meter speed offset is calculated and updated at the same time, and the satellite positioning information is output. According to this process, accurate positioning information is obtained, and then the process is repeated to continue the next positioning calculation.

[0083] This embodiment collects road video information and corrects the wheel speed meter information; inputs the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model to obtain visually assisted positioning information, wherein the deep learning positioning calculation model includes a two-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network and a two-layer long short-term memory network; determines whether the satellite positioning fails or deviates according to the satellite positioning information and the visually assisted positioning information; when it is determined that the satellite positioning fails or the satellite positioning deviates, the visually assisted positioning information is output, which solves the technical problem that the satellite positioning of the vehicle deviates or even disappears due to interference from various factors on the satellite positioning device during the driving process of the vehicle. Compared with the prior art, the present application can output visually assisted positioning information based on the road video information and the corrected wheel speed meter information in the deep learning positioning calculation model deployed on the vehicle when the vehicle satellite positioning deviates or even disappears, which can avoid affecting the vehicle driving due to the deviation of the vehicle satellite positioning or even the disappearance of the positioning.

[0084] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 4 , step S20 also includes steps S201 to S205:

[0085] Step S201, inputting the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model, and extracting a current frame feature map and a next frame feature map of a current frame and a next frame in the video road information through the double-layer convolutional neural network;

[0086] In a specific implementation, the current frame and the next frame need to be respectively extracted through two convolutional neural networks in a double-layer convolutional neural network to obtain the current feature map and the next frame feature map.

[0087] Step S202, superimposing the current frame feature map and the next frame feature map into a target feature map through the first connection layer, and converting the target feature map into a graph feature vector through the convolutional neural network and the pooling layer;

[0088] Step S203, extracting the speed feature vectors of the current frame wheel speed and the next frame wheel speed in the corrected wheel speed meter information through the long short-term memory network;

[0089] Step S204, connecting the graph feature vector and the speed feature vector into a target feature vector through the second connection layer;

[0090] Step S205: outputting visual auxiliary positioning information according to the target feature vector through the double-layer long short-term memory network.

[0091] It should be noted that the next frame refers to the frame of the current frame at the next moment.

[0092] In the specific implementation, the current frame and the next frame are first obtained through the video road information, and then the RGB of the current frame is converted into t And the next frame RGB t+1 The feature map of the current frame and the feature map of the next frame are extracted respectively through two convolutional neural networks (i.e., two-layer convolutional neural networks), and the two feature maps are superimposed together through the first connection layer to obtain the target feature map, and then the target feature map is converted into a graph feature vector through a convolutional neural network and a pooling layer. At the same time, the real-time speed information collected in the time period of the current frame and the next frame is connected to the graph feature vector after passing through a long short-term memory network LSTM, and finally a two-layer LSTM network structure is used to obtain the offset of the plane coordinates at the current time t+1 relative to the time t.

[0093] In a feasible implementation, the step of outputting visually assisted positioning information according to the target feature vector through the double-layer long short-term memory network includes: outputting a positioning offset according to the target feature vector through the double-layer long short-term memory network; and determining visually assisted positioning information based on the positioning offset and a visual positioning reference point.

[0094] It should be noted that since the final output of the deep learning positioning calculation model is a two-dimensional vector, which represents the offset of the plane coordinates (x, y) of the current video frame relative to the previous video frame (i.e., the positioning offset), it is necessary to infer the current coordinates (i.e., visually assisted positioning information) based on the positioning offset and the initial position (i.e., the visual positioning reference point).

[0095] In this embodiment, the video road information and the corrected wheel speed meter information are input into a deep learning positioning calculation model, and the current frame feature map and the next frame feature map of the current frame and the next frame in the video road information are extracted through the double-layer convolutional neural network; the current frame feature map and the next frame feature map are superimposed into a target feature map through the first connection layer, and the target feature map is converted into a graph feature vector through the convolutional neural network and the pooling layer; the speed feature vector of the current frame wheel speed meter speed and the next frame wheel speed meter speed in the corrected wheel speed meter information is extracted through the long short-term memory network; the graph feature vector and the speed feature vector are connected into a target feature vector through the second connection layer; and the visual auxiliary positioning information is output according to the target feature vector through the double-layer long short-term memory network. Through the above method, the FlowNetCorr structure commonly used in the optical flow method is used to extract the features of the previous and next frames. The feature extraction of the two previous and next image frames is performed by imitating the standard matching process. The correlation between the features is calculated and the convolution layer is replaced with a C2f network structure with a residual structure to increase the feature extraction ability of the network. The wheel speed meter speeds corresponding to the two videos are connected to the image feature vector through a layer of LSTM structure. The input speed modal feature can further improve the accuracy of the positioning calculation. The image features are extracted through the CNN convolutional neural network and the long-term time series information is recorded through the LSTM long short-term memory network, which can effectively reduce the accumulation of positioning errors.

[0096] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the artificial intelligence-based video-assisted positioning method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0097] This application also provides a video-assisted positioning device based on artificial intelligence, please refer to Figure 5 , the video-assisted positioning device based on artificial intelligence includes:

[0098] The acquisition module 10 is used to acquire road video information and correct wheel speed meter information;

[0099] An input module 20, used to input the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model to obtain visual auxiliary positioning information, wherein the deep learning positioning calculation model includes a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a double-layer long short-term memory network;

[0100] A judgment module 30, used to judge whether the satellite positioning is invalid or the satellite positioning is deviated according to the satellite positioning information and the visual auxiliary positioning information;

[0101] The output module 40 is used to output the visual auxiliary positioning information when it is determined that the satellite positioning fails or the satellite positioning deviates.

[0102] The artificial intelligence-based video-assisted positioning device provided by the present application adopts the artificial intelligence-based video-assisted positioning method in the above-mentioned embodiment, which can solve the technical problem of vehicle satellite positioning deviation or even positioning disappearance caused by interference of various factors on the satellite positioning device of the vehicle during driving. Compared with the prior art, the beneficial effects of the artificial intelligence-based video-assisted positioning device provided by the present application are the same as the beneficial effects of the artificial intelligence-based video-assisted positioning method provided by the above-mentioned embodiment, and the other technical features of the artificial intelligence-based video-assisted positioning device are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.

[0103] The present application provides an artificial intelligence-based video-assisted positioning device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the artificial intelligence-based video-assisted positioning method in the above-mentioned embodiment one.

[0104] Reference below Figure 6 , which shows a schematic diagram of the structure of an artificial intelligence-based video-assisted positioning device suitable for implementing the embodiment of the present application. The artificial intelligence-based video-assisted positioning device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The AI-based video-assisted positioning device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present application.

[0105] like Figure 6As shown, the video-assisted positioning device based on artificial intelligence may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the video-assisted positioning device based on artificial intelligence are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 can allow the video-assisted positioning device based on artificial intelligence to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a video-assisted positioning device based on artificial intelligence with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have alternatively.

[0106] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0107] The artificial intelligence-based video-assisted positioning device provided by the present application adopts the artificial intelligence-based video-assisted positioning method in the above-mentioned embodiment, which can solve the technical problem that the satellite positioning of the vehicle deviates or even disappears due to interference from various factors to the satellite positioning device during the driving of the vehicle. Compared with the prior art, the beneficial effects of the artificial intelligence-based video-assisted positioning device provided by the present application are the same as the beneficial effects of the artificial intelligence-based video-assisted positioning method provided by the above-mentioned embodiment, and the other technical features of the artificial intelligence-based video-assisted positioning device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0108] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0109] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0110] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the artificial intelligence-based video-assisted positioning method in the above-mentioned embodiment.

[0111] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0112] The above-mentioned computer-readable storage medium may be included in the video-assisted positioning device based on artificial intelligence; or it may exist independently without being assembled into the video-assisted positioning device based on artificial intelligence.

[0113] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by a video-assisted positioning device based on artificial intelligence, the video-assisted positioning device based on artificial intelligence: collects road video information and corrects wheel speedometer information; inputs the video road information and the corrected wheel speedometer information into a deep learning positioning calculation model to obtain visually assisted positioning information, wherein the deep learning positioning calculation model includes a two-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a two-layer long short-term memory network; determines whether satellite positioning is invalid or whether satellite positioning is deviated based on satellite positioning information and the visually assisted positioning information; and outputs the visually assisted positioning information when it is determined that satellite positioning is invalid or satellite positioning is deviated.

[0114] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0115] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0116] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0117] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned video-assisted positioning method based on artificial intelligence, and can solve the technical problem of vehicle satellite positioning deviation or even positioning disappearance caused by interference from various factors to the satellite positioning device of the vehicle during driving. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the video-assisted positioning method based on artificial intelligence provided by the above-mentioned embodiment, and will not be repeated here.

[0118] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned artificial intelligence-based video-assisted positioning method.

[0119] The computer program product provided by the present application can solve the technical problem that the satellite positioning of the vehicle deviates or even disappears due to interference from various factors to the satellite positioning device of the vehicle during driving. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the video-assisted positioning method based on artificial intelligence provided by the above embodiment, and will not be elaborated here.

[0120] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A video-assisted positioning method based on artificial intelligence, characterized in that: The method comprises: Collect road video information and correct wheel speedometer information; Inputting the video road information and the corrected wheel speedometer information into a deep learning positioning calculation model to obtain visual auxiliary positioning information, wherein the deep learning positioning calculation model includes a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a double-layer long short-term memory network; Determining whether satellite positioning is invalid or deviates according to the satellite positioning information and the visual auxiliary positioning information; When it is determined that the satellite positioning fails or the satellite positioning deviates, the visual auxiliary positioning information is output.

2. The method according to claim 1, characterized in that Before the step of inputting the video road information and the corrected wheel speedometer information into the deep learning positioning calculation model to obtain the visual auxiliary positioning information, the step further includes: Collecting initial training data in different scenarios, wherein the initial training data includes target satellite positioning information, target wheel speed meter information, and target road video information; Performing elimination processing on the initial training data to obtain first training data; Performing anomaly location and addition processing on the first training data to obtain second training data; Performing completion processing on the second training data to obtain target training data; An initial deep learning positioning calculation model is trained based on the target training data to obtain a deep learning positioning calculation model.

3. The method according to claim 1, characterized in that The step of inputting the video road information and the corrected wheel speedometer information into the deep learning positioning calculation model to obtain the visual auxiliary positioning information comprises: Inputting the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model, and extracting a current frame feature map and a next frame feature map of a current frame and a next frame in the video road information through the double-layer convolutional neural network; The current frame feature map and the next frame feature map are superimposed into a target feature map through the first connection layer, and the target feature map is converted into a graph feature vector through the convolutional neural network and the pooling layer; Extracting the speed feature vectors of the current frame wheel speed meter speed and the next frame wheel speed meter speed in the corrected wheel speed meter information through the long short-term memory network; Connecting the graph feature vector and the speed feature vector into a target feature vector through the second connection layer; The visual auxiliary positioning information is output according to the target feature vector through the double-layer long short-term memory network.

4. The method according to claim 3, characterized in that The step of outputting visual auxiliary positioning information according to the target feature vector through the double-layer long short-term memory network includes: Outputting a positioning offset according to the target feature vector through the double-layer long short-term memory network; Based on the positioning offset and the visual positioning reference point, visually assisted positioning information is determined.

5. The method according to claim 1, characterized in that The steps of collecting road video information and correcting wheel speedometer information include: Collect road video information and wheel speed meter information; Determine the deviation wheel speedometer speed and satellite positioning speed of the vehicle in the satellite positioning stable area; Determine a wheel speed calculation offset according to the deviation wheel speed meter speed and the satellite positioning speed; Determining a wheel speed meter calibration coefficient based on the wheel speed calculated offset; Corrected wheel speedometer information is determined according to the wheel speedometer calibration coefficient and the wheel speedometer information.

6. The method according to claim 1, characterized in that After the step of judging whether satellite positioning is invalid or whether satellite positioning is deviated according to the satellite positioning information and the visual auxiliary positioning information, the following step further comprises: When it is determined that the satellite positioning is not failed and not offset, the satellite positioning information is output, and the visual positioning reference point and the wheel speed calculation offset are updated to determine the visual assisted positioning information in the case of satellite positioning failure or satellite positioning offset based on the updated visual positioning reference point and the wheel speed calculation offset.

7. The method according to claim 1, characterized in that The step of judging satellite positioning failure or satellite positioning deviation according to the satellite positioning information and the visual auxiliary positioning information comprises: Determining whether satellite positioning is invalid according to the reception status of the satellite positioning information; Or, determine the deviation of the satellite positioning information and the visual auxiliary positioning information, and judge whether the satellite positioning is deviated according to the deviation.

8. A video-assisted positioning device based on artificial intelligence, characterized in that: The device comprises: A collection module, used to collect road video information and correct wheel speed meter information; An input module, used to input the video road information and the corrected wheel speed meter information into a deep learning positioning calculation model to obtain visual auxiliary positioning information, wherein the deep learning positioning calculation model includes a double-layer convolutional neural network, a first connection layer, a second connection layer, a convolutional neural network, a pooling layer, a long short-term memory network, and a double-layer long short-term memory network; A judgment module, used to judge whether the satellite positioning is invalid or the satellite positioning is deviated according to the satellite positioning information and the visual auxiliary positioning information; The output module is used to output the visual auxiliary positioning information when it is determined that the satellite positioning fails or the satellite positioning deviates.

9. A video-assisted positioning device based on artificial intelligence, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the artificial intelligence-based video-assisted positioning method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the artificial intelligence-based video-assisted positioning method as described in any one of claims 1 to 7 are implemented.

Images