Hierarchical feedback self-learning method for vehicle artificial intelligence

Abstract

The invention belongs to the technical field of artificial intelligence, and discloses a hierarchical feedback self-learning method for vehicle artificial intelligence, which comprises the following steps: acquiring a video stream collected by a camera of a vehicle in a vehicle driving process; the method comprises the following steps: extracting feature information of multiple dimensions of each video frame in a video stream through a vehicle artificial intelligence perception network model; calculating first difference information between the first feature information and the second feature information; generating motion space information based on the first difference information and the motion information of the vehicle, and predicting feature information of the next video frame based on the motion space information; determining a hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information and the extracted feature information; and performing hierarchical feedback self-learning on the vehicle artificial intelligence perception network model based on a hierarchical feedback online self-learning strategy. According to the invention, hierarchical feedback self-learning of the artificial intelligence perception network model is realized, and the perception accuracy of the model is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of artificial intelligence, and in particular relates to a hierarchical feedback self-learning method for vehicle artificial intelligence. Background Art

[0002] In recent years, with the rapid development of artificial intelligence, artificial intelligence has gradually been applied to vehicles. Specifically, the driving data of the vehicle can be collected offline, and the collected driving data can be used as training data for model training to obtain a pre-trained model. During the driving process of the vehicle, the prediction results are obtained through the pre-trained model, and the vehicle is controlled accordingly based on the prediction results.

[0003] From the above description, it can be seen that in the related technology, a large amount of vehicle driving data is required to conduct offline training of the pre-trained model, resulting in a high training cost of the pre-trained model; and during the driving process of the vehicle, if a driving scenario not covered by the pre-trained model is encountered, the pre-trained model cannot make accurate predictions, resulting in a low prediction accuracy of the pre-trained model, thereby posing a safety hazard to the vehicle. Summary of the invention

[0004] The purpose of the present invention is to realize the hierarchical feedback self-learning of the vehicle artificial intelligence perception network model. The perception accuracy of the vehicle artificial intelligence perception network model after online training is higher, which improves the safety of the vehicle. In addition, online training does not require special collection of training data, which reduces the training cost of the vehicle artificial intelligence perception network model. Online hierarchical and collaborative learning realizes the self-adaptation and generalization capabilities of vehicle intelligent perception and intelligent decision-making. This hierarchical feedback self-learning vehicle artificial intelligence technology is suitable for safe vehicle intelligent driving.

[0005] In a first aspect, an embodiment of the present invention provides a hierarchical feedback self-learning method for vehicle artificial intelligence, the method comprising:

[0006] During the driving of the vehicle, obtaining a video stream captured by a camera of the vehicle;

[0007] The video stream is input into the vehicle artificial intelligence perception network model to be trained online, and the following steps are performed through the vehicle artificial intelligence perception network model:

[0008] Extracting feature information of multiple dimensions of each video frame in the video stream;

[0009] For feature information of each dimension of the current video frame, first difference information between first feature information and second feature information is calculated; the first feature information is feature information of the dimension of the current video frame, and the second feature information is feature information of the dimension of a video frame next to the current video frame; the current video frame is any video frame of the video stream;

[0010] generating motion space information of the vehicle based on the first differential information and the motion information of the vehicle, and predicting feature information of the dimension of the next video frame based on the motion space information;

[0011] Determining a hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame;

[0012] Based on the hierarchical feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to hierarchical feedback self-learning to obtain the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

[0013] Optionally, the vehicle artificial intelligence perception network model includes a two-dimensional feature information extraction sub-model, a three-dimensional feature information extraction sub-model and a four-dimensional feature information extraction sub-model;

[0014] The extracting feature information of multiple dimensions of each video frame in the video stream includes:

[0015] Extracting two-dimensional feature information of each video frame in the video stream through a two-dimensional feature information extraction sub-model, wherein the two-dimensional feature information is used to characterize the plane feature information of the video frame;

[0016] Extracting three-dimensional feature information of each video frame in the video stream by using a three-dimensional feature information extraction sub-model, wherein the three-dimensional feature information is used to characterize spatial feature information of the video frame;

[0017] The four-dimensional feature information of each video frame in the video stream is extracted through a four-dimensional feature information extraction sub-model, wherein the four-dimensional feature information is used to characterize the spatial feature information and the temporal feature information of the video frame.

[0018] Optionally, determining the hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame includes:

[0019] Determine first parameter adjustment gradient information corresponding to the two-dimensional feature information extraction sub-model based on second difference information between the predicted two-dimensional feature information of the next video frame and the extracted two-dimensional feature information of the next video frame, and construct a two-dimensional feature feedback network based on the first parameter adjustment gradient information;

[0020] Determining second parameter adjustment gradient information corresponding to the 3D feature information extraction sub-model based on second difference information between the predicted 3D feature information of the next video frame and the extracted 3D feature information of the next video frame, and constructing a 3D feature feedback network based on the second parameter adjustment gradient information;

[0021] Determine third parameter adjustment gradient information corresponding to the four-dimensional feature information extraction submodel based on second difference information between the predicted four-dimensional feature information of the next video frame and the extracted four-dimensional feature information of the next video frame, and construct a four-dimensional feature feedback network based on the third parameter adjustment gradient information;

[0022] Based on the three-dimensional feature feedback network and the four-dimensional feature feedback network, construct an inter-layer feedback network;

[0023] Based on the two-dimensional feature feedback network, the three-dimensional feature feedback network, the four-dimensional feature feedback network and the inter-layer feedback network, a layered feedback online autonomous learning strategy is determined.

[0024] Optionally, determining a hierarchical feedback online autonomous learning strategy based on the two-dimensional feature feedback network, the three-dimensional feature feedback network, the four-dimensional feature feedback network and the inter-layer feedback network includes:

[0025] Determining first parameter adjustment information of the two-dimensional feature information extraction sub-model based on the two-dimensional feature feedback network and the inter-layer feedback network;

[0026] Determining second parameter adjustment information of the three-dimensional feature information extraction sub-model based on the three-dimensional feature feedback network and the inter-layer feedback network;

[0027] The third parameter adjustment information of the four-dimensional feature information extraction sub-model is determined based on the four-dimensional feature feedback network.

[0028] Optionally, the layered feedback-based online autonomous learning strategy performs layered feedback self-learning on the vehicle artificial intelligence perception network model to obtain the vehicle artificial intelligence perception network model after online layered feedback self-learning, including:

[0029] Performing hierarchical feedback self-learning on the two-dimensional feature information extraction sub-model through the first parameter adjustment information;

[0030] Performing hierarchical feedback self-learning on the three-dimensional feature information extraction sub-model through the second parameter adjustment information;

[0031] The four-dimensional feature information extraction sub-model is subjected to hierarchical feedback self-learning through the third parameter adjustment information.

[0032] In a second aspect, an embodiment of the present invention provides a vehicle control method based on vehicle artificial intelligence, the method comprising:

[0033] Acquire a video stream collected by a camera of the vehicle during driving of the vehicle;

[0034] Inputting the video stream into the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning, so as to output the perception result through the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning;

[0035] The vehicle is controlled accordingly based on the perception result.

[0036] In a third aspect, an embodiment of the present invention provides a hierarchical feedback self-learning device for vehicle artificial intelligence, the device comprising:

[0037] A video stream acquisition module, used to acquire the video stream collected by the camera of the vehicle during the driving process of the vehicle;

[0038] The video stream input module is used to input the video stream into the vehicle artificial intelligence perception network model to be trained online, and perform the following steps through the vehicle artificial intelligence perception network model:

[0039] Extracting feature information of multiple dimensions of each video frame in the video stream;

[0040] For feature information of each dimension of the current video frame, first difference information between first feature information and second feature information is calculated; the first feature information is feature information of the dimension of the video frame, and the second feature information is feature information of the dimension of a next video frame of the video frame; the current video frame is any video frame of the video stream;

[0041] generating motion space information of the vehicle based on the first differential information and the motion information of the vehicle, and predicting feature information of the dimension of the next video frame based on the motion space information;

[0042] Determining a hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame;

[0043] Based on the hierarchical feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to hierarchical feedback self-learning to obtain the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

[0044] In a fourth aspect, an embodiment of the present invention provides a vehicle control device based on vehicle artificial intelligence, the device comprising:

[0045] A video stream acquisition module is used to acquire the video stream collected by the camera of the vehicle during the driving process of the vehicle;

[0046] A model online hierarchical feedback self-learning module, used to input the video stream into the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning, so as to output a perception result through the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning;

[0047] A vehicle control module is used to control the vehicle accordingly based on the perception result.

[0048] In a fifth aspect, an embodiment of the present invention provides an electronic device, including:

[0049] at least one processor;

[0050] a memory for storing the at least one processor-executable instruction;

[0051] The at least one processor is configured to execute the instructions to implement the method described in the first aspect or the second aspect.

[0052] In a sixth aspect, an embodiment of the present invention provides a computer-readable storage medium. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method described in the first aspect or the second aspect.

[0053] In a seventh aspect, an embodiment of the present invention provides a computer program product, including a computer program, which, when executed by a processor, implements the method described in either the first aspect or the second aspect.

[0054] The technical solution provided by the embodiment of the present invention is as follows: during the driving process of the vehicle, a video stream captured by the camera of the vehicle is obtained; the video stream is input into the vehicle artificial intelligence perception network model to be trained online, and the following steps are performed through the vehicle artificial intelligence perception network model: feature information of multiple dimensions of each video frame in the video stream is extracted; for the feature information of each dimension of the current video frame, the first difference information between the first feature information and the second feature information is calculated; based on the first difference information and the motion information of the vehicle, the motion space information of the vehicle is generated, and the feature information of the dimension of the next video frame is predicted based on the motion space information; based on the second difference information between the feature information of the predicted dimension of the next video frame and the feature information of the extracted dimension of the next video frame, a hierarchical feedback online autonomous learning strategy is determined; based on the hierarchical feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to hierarchical feedback self-learning to obtain the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

[0055] It can be seen that in the embodiment of the present invention, during the movement of the vehicle, the hierarchical feedback online autonomous learning strategy can be determined in real time, and the vehicle artificial intelligence perception network model can be self-learned by hierarchical feedback. The perception accuracy of the vehicle artificial intelligence perception network model after online training is higher, which improves the safety of the vehicle. In addition, online training does not require special collection of training data, which reduces the training cost of the vehicle artificial intelligence perception network model. Online hierarchical and collaborative learning realizes the adaptive and generalization capabilities of vehicle intelligent perception and intelligent decision-making. This hierarchical feedback self-learning vehicle artificial intelligence technology is suitable for safe vehicle intelligent driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A flowchart of a hierarchical feedback self-learning method for vehicle artificial intelligence provided by an embodiment of the present invention;

[0057] Figure 2 for Figure 1 A flowchart of a specific implementation method of S160;

[0058] Figure 3 A schematic diagram of a process of hierarchical feedback self-learning of vehicle artificial intelligence provided by an embodiment of the present invention;

[0059] Figure 4 A schematic diagram of another process of hierarchical feedback self-learning of vehicle artificial intelligence provided by an embodiment of the present invention;

[0060] Figure 5 A flow chart of a vehicle control method based on vehicle artificial intelligence provided by an embodiment of the present invention;

[0061] Figure 6 A schematic diagram of the structure of a hierarchical feedback self-learning device for vehicle artificial intelligence provided by an embodiment of the present invention;

[0062] Figure 7 A vehicle control device based on vehicle artificial intelligence provided by an embodiment of the present invention;

[0063] Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The present invention will be described in detail below through examples.

[0065] In recent years, with the rapid development of artificial intelligence, artificial intelligence has gradually been applied to vehicles. However, the artificial intelligence of existing technologies does not have the ability to learn online or update in real time. As a pre-trained model, it can only be continuously iterated based on pre-training, which means that its knowledge and capabilities are trained based on data before a specific point in time. In addition, during use, generative artificial intelligence cannot actively learn or memorize new information, nor can it update its knowledge base in real time based on user interaction. This means that if a scene not covered by pre-training is encountered during driving, the safety of the vehicle will be greatly challenged.

[0066] From the above description, it can be seen that the training cost of the pre-trained model in the prior art is huge. It is necessary to prepare a large amount of offline data for pre-training, and then put the trained model into use. In addition, due to the extremely complex internal structure of the model, the training process cannot be analyzed, and the reliability of the results still needs to be considered. In addition, during driving, if a scene not covered by pre-training is encountered, the prediction accuracy of the pre-trained model is low, which will cause incorrect control of the vehicle and will bring serious consequences. Therefore, the reliability and safety of the application of artificial intelligence in vehicles is a huge challenge.

[0067] In order to solve the above technical problems, an embodiment of the present invention provides a vehicle artificial intelligence capable of online learning. Compared with the existing artificial intelligence technology, the present invention reduces the complexity of the model through online feedback learning technology, and can analyze the internal structure of the model to a certain extent, so that its safety and reliability meet driving requirements. This will not only promote the development of artificial intelligence theory, but also promote the industrial application of intelligent technology in the field of intelligent driving.

[0068] Specifically, the embodiment of the present invention introduces a multi-level dynamic feedback optimization network to improve the adaptability and reliability of the vehicle artificial intelligence perception network model to unknown environments. The present invention performs real-time optimization through dynamic feedback at the local level, and at the same time, the synergy between local optimization and local functions realizes feedback at the global level. The real-time feedback learning mechanism is the core to ensure the stability and robustness of the model. At the same time, the dynamic feedback and adjustment mechanism are the most effective measures for the model to shield noise and adapt to the environment, and are applied to the verification of vehicle active safety control. And the online learning technology of the present invention can significantly reduce the amount of data required for model training, thereby reducing the training cost of the model. Online hierarchical and collaborative learning realizes the adaptability and generalization capabilities of vehicle intelligent perception and intelligent decision-making. This hierarchical feedback self-learning vehicle artificial intelligence technology is suitable for safe vehicle intelligent driving.

[0069] The specific implementation method of the layered feedback self-learning method for vehicle artificial intelligence provided by an embodiment of the present invention will be described in detail below.

[0070] like Figure 1FIG. 1 is a flowchart of a hierarchical feedback self-learning method for vehicle artificial intelligence provided by an embodiment of the present invention. The method may include the following steps:

[0071] S110, while the vehicle is traveling, obtaining a video stream captured by a camera of the vehicle.

[0072] Specifically, the embodiment of the present invention trains the vehicle artificial intelligence perception network model through online learning technology, so the continuous video stream is the learning and perception basis of the vehicle artificial intelligence perception network model. During the driving process of the vehicle, the vehicle's camera will collect video streams in real time, so the video streams collected by the vehicle's camera can be obtained.

[0073] S120, input the video stream into the vehicle artificial intelligence perception network model to be trained online, and execute the following steps S130 to S170 through the vehicle artificial intelligence perception network model.

[0074] Among them, before online training of the vehicle artificial intelligence perception network model, the vehicle artificial intelligence perception network model can be pre-trained, and the pre-training process is an offline training process. In order to improve the perception accuracy of the vehicle artificial intelligence perception network model obtained by pre-training, in an embodiment of the present invention, during the driving process of the vehicle, the video stream collected by the vehicle's camera is input into the pre-trained vehicle artificial intelligence perception network model to perform online training on the pre-trained vehicle artificial intelligence perception network model. Because the vehicle will encounter various new scenes during driving, by online training of the vehicle artificial intelligence perception network model, the vehicle artificial intelligence perception network model can be adapted to various new scenes, thereby improving the perception accuracy of the vehicle artificial intelligence perception network model. In addition, online training does not require the special collection of a large amount of training data, which can reduce the training cost of the vehicle artificial intelligence perception network model.

[0075] After the vehicle artificial intelligence perception network model to be trained online receives the video stream, it will perform hierarchical feedback self-learning through steps S130 to S160.

[0076] S130, extracting feature information of multiple dimensions of each video frame in the video stream.

[0077] Specifically, after receiving the video stream, the vehicle artificial intelligence perception network model can extract feature information of multiple dimensions such as two-dimensional feature information, three-dimensional feature information, and four-dimensional feature information of each video frame in the video stream. Among them, the two-dimensional feature information can be the plane feature information of the video frame, for example, it can be feature information in the two directions of the X-axis and the Y-axis. The three-dimensional feature information can be the spatial feature information of the video frame, for example, it can be feature information in the three directions of the X-axis, the Y-axis, and the Z-axis. The four-dimensional feature information can be the spatial feature information and the temporal feature information of the video frame.

[0078] In one embodiment, the vehicle artificial intelligence perception network model may include a two-dimensional feature information extraction sub-model, a three-dimensional feature information extraction sub-model, and a four-dimensional feature information extraction sub-model;

[0079] At this time, S130, extracting feature information of multiple dimensions of each video frame in the video stream, may include the following three steps, namely, step 1 to step 3:

[0080] Step 1: extracting two-dimensional feature information of each video frame in the video stream through a two-dimensional feature information extraction sub-model, wherein the two-dimensional feature information is used to characterize the planar feature information of the video frame.

[0081] Step 2: extracting the three-dimensional feature information of each video frame in the video stream through the three-dimensional feature information extraction sub-model, wherein the three-dimensional feature information is used to characterize the spatial feature information of the video frame.

[0082] Step 3: extracting four-dimensional feature information of each video frame in the video stream through a four-dimensional feature information extraction sub-model, wherein the four-dimensional feature information is used to characterize the spatial feature information and temporal feature information of the video frame.

[0083] Among them, the two-dimensional feature information extraction sub-model, the three-dimensional feature information extraction sub-model and the four-dimensional feature information extraction sub-model are pre-trained by traditional back-propagation training methods respectively.

[0084] S140: For feature information of each dimension of the current video frame, calculate first difference information between the first feature information and the second feature information.

[0085] The first feature information is feature information of a dimension of a current video frame, and the second feature information is feature information of a dimension of a next video frame of the current video frame; the current video frame is any video frame of a video stream.

[0086] Specifically, since the video stream is a plurality of continuous video frames, for any video frame among the plurality of video frames, step S130 can extract feature information of multiple dimensions of the video, for example, two-dimensional feature information, three-dimensional feature information, and four-dimensional feature information of the video frame can be extracted.

[0087] For any video frame in a video stream, the video frame can be used as the current video frame, and the difference between the two-dimensional feature information of the video frame and the two-dimensional feature information of the next video frame of the video frame can be calculated, which is recorded as the first difference information. Similarly, for the three-dimensional feature information of any video frame in a video stream, the difference between the three-dimensional feature information of the video frame and the three-dimensional feature information of the next video frame of the video frame can be calculated, which is also recorded as the first difference information. For the four-dimensional feature information of any video frame in a video stream, the difference between the four-dimensional feature information of the video frame and the four-dimensional feature information of the next video frame of the video frame can be calculated, which is recorded as the first difference information.

[0088] For example, assuming that the feature information is the position of the vehicle in a video frame, after obtaining the vehicle position in any video frame, the differential information between the vehicle position in the video frame and the vehicle position in the next video frame can be calculated to obtain the position change of the vehicle in different video frames.

[0089] S150, generating motion space information of the vehicle based on the first differential information and the motion information of the vehicle, and predicting feature information of the dimension of the next video frame based on the motion space information.

[0090] Since the first differential information is a scalar, in order to more accurately predict the feature information of the next video frame, the motion space information of the vehicle can be generated based on the first differential information and the motion information of the vehicle. The motion space information is a vector, and the feature information of the next video frame can be accurately predicted through the motion space information. For example, the first differential information is the difference between the vehicle position of any video frame and the vehicle position of the next video frame, and the difference is a scalar; the motion information of the vehicle can be the vehicle's moving speed, and the motion space information of the vehicle is generated based on the first differential information and the motion information of the vehicle. The generated motion space information is a vector. The feature information of the next video frame can be predicted through the motion space information.

[0091] Specifically, if the first differential information is differential information between the two-dimensional feature information of a video frame and the two-dimensional feature information of the next video frame of the video frame, then the two-dimensional feature information of the next video frame can be predicted based on the motion space information generated by the first differential information and the motion information of the vehicle. If the first differential information is differential information between the three-dimensional feature information of a video frame and the three-dimensional feature information of the next video frame of the video frame, then the three-dimensional feature information of the next video frame can be predicted based on the motion space information generated by the first differential information and the motion information of the vehicle. If the first differential information is differential information between the four-dimensional feature information of a video frame and the four-dimensional feature information of the next video frame of the video frame, then the four-dimensional feature information of the next video frame can be predicted based on the motion space information generated by the first differential information and the motion information of the vehicle.

[0092] S160, determining a hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame.

[0093] For the sake of clarity in describing the solution, the next video frame may be referred to as the video frame corresponding to time t, which may be any time in the continuous video stream. As can be seen from the above description, the two-dimensional feature information, three-dimensional feature information and four-dimensional feature information of the video frame at time t may be extracted through step S130, and the two-dimensional feature information, three-dimensional feature information and four-dimensional feature information of the video frame at time t may be predicted through step S140.

[0094] For the video frame at time t, the difference information between the extracted two-dimensional feature information and the predicted two-dimensional feature information can be calculated to obtain the prediction error of the two-dimensional feature extraction sub-model, and a two-dimensional feature feedback network can be constructed based on the prediction error. The difference information between the extracted three-dimensional feature information and the predicted three-dimensional feature information can be calculated to obtain the prediction error of the three-dimensional feature extraction sub-model, and a three-dimensional feature feedback network can be constructed based on the prediction error. The difference information between the extracted four-dimensional feature information and the predicted four-dimensional feature information can also be calculated to obtain the prediction error of the four-dimensional feature extraction sub-model, and a four-dimensional feedback network can be constructed based on the prediction error. And an inter-layer feedback network can be constructed through a three-dimensional feedback network and a four-dimensional feedback network.

[0095] Then, the parameter adjustment information of the two-dimensional feature sub-model is determined according to the two-dimensional feature feedback network and the inter-layer feedback network; the parameter adjustment information of the three-dimensional feature sub-model is determined according to the three-dimensional feature feedback network and the inter-layer feedback network; and the parameter adjustment information of the four-dimensional feature sub-model can also be determined according to the four-dimensional feature feedback network.

[0096] In order to make the description of the solution clear and complete, the specific implementation of S160 will be described in detail in the following examples.

[0097] S170, based on the layered feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to layered feedback self-learning to obtain the vehicle artificial intelligence perception network model after online layered feedback self-learning.

[0098] Specifically, after obtaining the hierarchical feedback online autonomous learning strategy of the vehicle artificial intelligence perception network model, the vehicle artificial intelligence perception network model can be subjected to hierarchical feedback learning according to the hierarchical feedback online autonomous learning strategy. This process is the model parameter adjustment process of the vehicle artificial intelligence perception network model, and the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning is obtained. Assuming that the two-dimensional feature extraction submodel in the vehicle artificial intelligence perception network model needs to adjust three parameters a, b and c, the hierarchical feedback online autonomous learning strategy is: a remains unchanged, b increases, and c decreases. Then, based on the hierarchical feedback online autonomous learning strategy, a can be kept unchanged, b can be increased, and c can be decreased. For the three-dimensional feature extraction submodel and the four-dimensional feature extraction submodel in the vehicle artificial intelligence perception network model, their model parameters can be adjusted accordingly according to their corresponding hierarchical feedback online autonomous learning strategies. The perception accuracy of the vehicle artificial intelligence perception network model after online training is higher, and online training does not require special collection of training data, which reduces the training cost of the vehicle artificial intelligence perception network model.

[0099] The technical solution provided by the embodiment of the present invention is as follows: during the driving process of the vehicle, a video stream captured by the camera of the vehicle is obtained; the video stream is input into the vehicle artificial intelligence perception network model to be trained online, and the following steps are performed through the vehicle artificial intelligence perception network model: feature information of multiple dimensions of each video frame in the video stream is extracted; for the feature information of each dimension of the current video frame, the first difference information between the first feature information and the second feature information is calculated; based on the first difference information and the motion information of the vehicle, the motion space information of the vehicle is generated, and the feature information of the dimension of the next video frame is predicted based on the motion space information; based on the second difference information between the feature information of the predicted dimension of the next video frame and the feature information of the extracted dimension of the next video frame, a hierarchical feedback online autonomous learning strategy is determined; based on the hierarchical feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to hierarchical feedback self-learning to obtain the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

[0100] It can be seen that in the embodiment of the present invention, during the movement of the vehicle, the hierarchical feedback online autonomous learning strategy can be determined in real time, and the vehicle artificial intelligence perception network model can be self-learned by hierarchical feedback. The perception accuracy of the vehicle artificial intelligence perception network model after online training is higher, which improves the safety of the vehicle. In addition, online training does not require special collection of training data, which reduces the training cost of the vehicle artificial intelligence perception network model. Online hierarchical and collaborative learning realizes the adaptive and generalization capabilities of vehicle intelligent perception and intelligent decision-making. This hierarchical feedback self-learning vehicle artificial intelligence technology is suitable for safe vehicle intelligent driving.

[0101] In order to make the description of the solution clear and complete, the specific implementation of S160 will be described in detail in the following examples.

[0102] S160, based on the second difference information between the feature information of the predicted dimension of the next video frame and the feature information of the extracted dimension of the next video frame, determine a hierarchical feedback online autonomous learning strategy, such as Figure 2 As shown, the following steps may be included:

[0103] S161, based on the second difference information between the predicted two-dimensional feature information of the next video frame and the extracted two-dimensional feature information of the next video frame, determine the first parameter adjustment gradient information corresponding to the two-dimensional feature information extraction sub-model, and construct a two-dimensional feature feedback network based on the first parameter adjustment gradient information.

[0104] Specifically, after predicting the two-dimensional feature information of the next video frame, the predicted two-dimensional feature information of the next video frame can be compared with the extracted real two-dimensional feature information of the next video frame to obtain second differential information between the two, and the parameter adjustment gradient information corresponding to the two-dimensional feature information extraction sub-model can be determined based on the second differential information. In order to clearly describe the scheme, the parameter adjustment gradient information can be referred to as the first parameter adjustment gradient information, and a two-dimensional feature feedback network can be constructed based on the first parameter adjustment gradient information. The two-dimensional feature feedback network can feed back the first parameter adjustment gradient information, i.e., the parameter adjustment information, to the two-dimensional feature information extraction sub-model, so that the two-dimensional feature information extraction sub-model can adjust the model parameters based on the parameter adjustment information.

[0105] S162, based on the second difference information between the predicted three-dimensional feature information of the next video frame and the extracted three-dimensional feature information of the next video frame, determine the second parameter adjustment gradient information corresponding to the three-dimensional feature information extraction sub-model, and construct a three-dimensional feature feedback network based on the second parameter adjustment gradient information.

[0106] Specifically, after predicting the three-dimensional feature information of the next video frame, the predicted three-dimensional feature information of the next video frame can be compared with the extracted real three-dimensional feature information of the next video frame to obtain second differential information between the two, and the parameter adjustment gradient information corresponding to the three-dimensional feature information extraction sub-model can be determined based on the second differential information. In order to clearly describe the scheme, the parameter adjustment gradient information can be called second parameter adjustment gradient information, and a three-dimensional feature feedback network can be constructed based on the second parameter adjustment gradient information. The three-dimensional feature feedback network can feed back the second parameter adjustment gradient information, i.e., the parameter adjustment information, to the three-dimensional feature information extraction sub-model, so that the three-dimensional feature information extraction sub-model can adjust the model parameters based on the parameter adjustment information.

[0107] S163, based on the second difference information between the predicted four-dimensional feature information of the next video frame and the extracted four-dimensional feature information of the next video frame, determine the third parameter adjustment gradient information corresponding to the four-dimensional feature information extraction sub-model, and construct a four-dimensional feature feedback network based on the third parameter adjustment gradient information.

[0108] Specifically, after predicting the four-dimensional feature information of the next video frame, the predicted four-dimensional feature information of the next video frame can be compared with the extracted real four-dimensional feature information of the next video frame to obtain the second differential information between the two, and the parameter adjustment gradient information corresponding to the four-dimensional feature information extraction sub-model is determined based on the second differential information. In order to clearly describe the scheme, the parameter adjustment gradient information can be called the third parameter adjustment gradient information, and a four-dimensional feature feedback network is constructed based on the third parameter adjustment gradient information. The four-dimensional feature feedback network can feed back the third parameter adjustment gradient information, i.e., the parameter adjustment information, to the four-dimensional feature information extraction sub-model, so that the four-dimensional feature information extraction sub-model can adjust the model parameters based on the parameter adjustment information.

[0109] Below Figure 3 The embodiment of the present invention is described with reference to FIG. Figure 3 The image space in is the image contained in each video frame in the video stream captured by the vehicle's camera, and the video input is to input the video stream into the vehicle's artificial intelligence perception network model. Figure 3 The feature extraction in can be the two-dimensional feature information extracted by the two-dimensional feature information extraction submodel, the three-dimensional feature information extracted by the three-dimensional feature information extraction submodel, or the four-dimensional feature information extracted by the four-dimensional feature information extraction submodel. The feature space corresponding to time t-1 can be two-dimensional feature information, three-dimensional feature information or four-dimensional feature information, and the motion information is vehicle motion information.

[0110] The differential space may be the first differential information, specifically, the differential information between the two-dimensional feature information of the video frame at time t-1 and the two-dimensional feature information of the video frame at time t, or the differential information between the three-dimensional feature information of the video frame at time t-1 and the three-dimensional feature information of the video frame at time t, or the differential information between the four-dimensional feature information of the video frame at time t-1 and the four-dimensional feature information of the video frame at time t. The motion space may be the vehicle motion space information at time t-1 generated based on the first differential information and the vehicle motion information.

[0111] The feature space corresponding to the prediction time t may be the predicted two-dimensional feature information, three-dimensional feature information, or four-dimensional feature information. For example, if the first differential information is the differential information between the two-dimensional feature information of the video frame at time t-1 and the two-dimensional feature information of the video frame at time t, the feature space corresponding to the prediction time t may be the predicted two-dimensional feature information. The feature space corresponding to time t may be the extracted real two-dimensional feature information, real three-dimensional feature information, or real four-dimensional feature information.

[0112] By calculating the difference information between the predicted two-dimensional feature information and the real two-dimensional feature information (the above-mentioned second difference information), the first parameter adjustment gradient information corresponding to the two-dimensional feature information extraction sub-model can be determined, that is, Figure 3 In the gradient space shown, assuming that there are three model parameters to be adjusted, namely a, b and c, then the gradient space can be a unchanged, b increased, and c decreased. Of course, this is just an example, and the specific adjustment needs to be determined according to the actual situation. After obtaining the first parameter adjustment gradient information, a two-dimensional feature feedback network can be constructed according to the first parameter adjustment gradient information, and the two-dimensional feature feedback network can feed back the first parameter adjustment gradient information to the two-dimensional feature information extraction sub-model.

[0113] Similarly, by calculating the difference information (second difference information) between the predicted three-dimensional feature information and the real three-dimensional feature information, the second parameter adjustment gradient information corresponding to the three-dimensional feature information extraction sub-model can be determined, that is, Figure 3 In the gradient space shown, after obtaining the second parameter adjustment gradient information, a three-dimensional feature feedback network can be constructed according to the second parameter adjustment gradient information, and the three-dimensional feature feedback network can feed back the second parameter adjustment gradient information to the three-dimensional feature information extraction sub-model. By calculating the difference information (second difference information) between the predicted four-dimensional feature information and the real four-dimensional feature information, the third parameter adjustment gradient information corresponding to the four-dimensional feature information extraction sub-model can be determined, that is, Figure 3 In the gradient space shown, after obtaining the third parameter adjustment gradient information, a four-dimensional feature feedback network can be constructed according to the third parameter adjustment gradient information, and the four-dimensional feature feedback network can feed back the third parameter adjustment gradient information to the four-dimensional feature information extraction sub-model.

[0114] S164, constructing an inter-layer feedback network based on the three-dimensional feature feedback network and the four-dimensional feature feedback network.

[0115] Specifically, in the embodiment of the present invention, not only a two-dimensional feature feedback network, a three-dimensional feature feedback network and a four-dimensional feature feedback network can be constructed, but also an inter-layer feedback network can be constructed based on the three-dimensional feature feedback network and the four-dimensional feature feedback network.

[0116] S165, based on the two-dimensional feature feedback network, the three-dimensional feature feedback network, the four-dimensional feature feedback network and the inter-layer feedback network, determine the layered feedback online autonomous learning strategy.

[0117] As an implementation method of an embodiment of the present invention, S165, based on the two-dimensional feature feedback network, the three-dimensional feature feedback network, the four-dimensional feature feedback network and the inter-layer feedback network, determines the layered feedback online autonomous learning strategy, which may include the following three steps, namely step 1 to step 3:

[0118] Step 1: Determine first parameter adjustment information of a two-dimensional feature information extraction sub-model based on a two-dimensional feature feedback network and an inter-layer feedback network.

[0119] Step 2: Determine second parameter adjustment information of the three-dimensional feature information extraction sub-model based on the three-dimensional feature feedback network and the inter-layer feedback network.

[0120] Step 3: Determine third parameter adjustment information of the four-dimensional feature information extraction sub-model based on the four-dimensional feature feedback network.

[0121] Specifically, the two-dimensional feature feedback network and the inter-layer feedback network can collaboratively feed back parameter adjustment information to the two-dimensional feature information extraction sub-model. Therefore, the first parameter adjustment information of the two-dimensional feature information extraction sub-model can be determined based on the two-dimensional feature feedback network and the inter-layer feedback network. That is, the two-dimensional feature feedback network and the inter-layer feedback network can determine the first parameter adjustment information of the two-dimensional feature information extraction sub-model by weighted summation. Among them, the weights of the two-dimensional feature feedback network and the inter-layer feedback network can be determined according to actual conditions.

[0122] Similarly, the three-dimensional feature feedback network and the inter-layer feedback network can collaboratively feed back parameter adjustment information to the three-dimensional feature information extraction sub-model. Therefore, the second parameter adjustment information of the three-dimensional feature information extraction sub-model can be determined based on the three-dimensional feature feedback network and the inter-layer feedback network. That is, the three-dimensional feature feedback network and the inter-layer feedback network can determine the second parameter adjustment information of the three-dimensional feature information extraction sub-model by weighted summation. Among them, the weights of the three-dimensional feature feedback network and the inter-layer feedback network can also be determined according to actual conditions.

[0123] The four-dimensional feature feedback network feeds back parameter adjustment information to the four-dimensional feature information extraction sub-model. Therefore, the third parameter adjustment information of the four-dimensional feature information extraction sub-model can be determined based on the four-dimensional feature feedback network.

[0124] Below Figure 4 Take this as an example to illustrate: Figure 4The image space in is the image contained in each video frame in the video stream captured by the vehicle's camera, and the video input is to input the video stream into the vehicle's artificial intelligence perception network model. 2D network refers to a two-dimensional feature information extraction sub-model, 2D feature space refers to two-dimensional feature information, 2D-3D network refers to a three-dimensional feature information extraction sub-model, 3D feature space refers to three-dimensional feature information, 3D-4D network refers to a four-dimensional feature information extraction sub-model, and 4D feature space refers to four-dimensional feature information. The 2D feedback network is a two-dimensional feature feedback network, the 3D feedback network is a three-dimensional feature feedback network, and the 4D feedback network is a four-dimensional feature feedback network. An inter-layer feedback network can be constructed based on the 3D feedback network and the 4D feedback network. From Figure 4 It can be seen that the 2D feedback network and the inter-layer feedback network jointly feed back the online autonomous learning strategy, i.e., parameter adjustment information, to the 2D network. The 3D feedback network and the inter-layer feedback network jointly feed back the online autonomous learning strategy, i.e., parameter adjustment information, to the 2D-3D network. The 4D feedback network feeds back the online autonomous learning strategy, i.e., parameter adjustment information, to the 3D-4D network.

[0125] Based on the above embodiment, in one implementation, S170, the layered feedback online autonomous learning strategy is based on which the vehicle artificial intelligence perception network model is subjected to layered feedback self-learning to obtain the vehicle artificial intelligence perception network model after online layered feedback self-learning, which may include the following steps:

[0126] The model parameters of the two-dimensional feature information extraction sub-model are subjected to hierarchical feedback self-learning through the first parameter adjustment information.

[0127] The model parameters of the three-dimensional feature information extraction sub-model are subjected to hierarchical feedback self-learning through the second parameter adjustment information.

[0128] The model parameters of the four-dimensional feature information extraction sub-model are subjected to hierarchical feedback self-learning through the third parameter adjustment information.

[0129] From the above description, it can be known that the first parameter adjustment information of the two-dimensional feature information extraction sub-model can be obtained. Therefore, the two-dimensional feature information extraction sub-model can be subjected to hierarchical feedback self-learning through the first parameter adjustment information, that is, the model parameters of the two-dimensional feature information extraction sub-model can be adjusted. Similarly, the three-dimensional feature information extraction sub-model can be subjected to hierarchical feedback self-learning through the second parameter adjustment information, that is, the model parameters of the three-dimensional feature information extraction sub-model can be adjusted, and the four-dimensional feature information extraction sub-model can be subjected to hierarchical feedback self-learning through the third parameter adjustment information, that is, the model parameters of the four-dimensional feature information extraction sub-model can be adjusted.

[0130] It can be seen that in the embodiment of the present invention, during the movement of the vehicle, the hierarchical feedback online autonomous learning strategy can be determined in real time, and the vehicle artificial intelligence perception network model can be self-learned by hierarchical feedback. The perception accuracy of the vehicle artificial intelligence perception network model after online training is higher, which improves the safety of the vehicle. In addition, online training does not require special collection of training data, which reduces the training cost of the vehicle artificial intelligence perception network model. Online hierarchical and collaborative learning realizes the adaptive and generalization capabilities of vehicle intelligent perception and intelligent decision-making. This hierarchical feedback self-learning vehicle artificial intelligence technology is suitable for safe vehicle intelligent driving.

[0131] The embodiment of the present invention also provides a vehicle control method based on vehicle artificial intelligence, such as Figure 5 As shown, the method includes:

[0132] S510, obtaining a video stream collected by a camera of the vehicle during the driving process of the vehicle.

[0133] S520, inputting the video stream into the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning, so as to output the perception result through the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

[0134] S530: Control the vehicle accordingly based on the perception result.

[0135] Specifically, in practical applications, during the driving process of the vehicle, the video stream collected by the vehicle's camera is input into the online trained vehicle artificial intelligence perception network model obtained through the above embodiment. After receiving the video stream, the online trained vehicle artificial intelligence perception network model extracts the two-dimensional feature information, three-dimensional feature information and four-dimensional feature information of each video frame in the video stream, and outputs a prediction structure based on the four-dimensional feature information, and then controls the vehicle accordingly according to the prediction result. For example, if the prediction result is that there is an obstacle in front of the vehicle, then the vehicle can be controlled to avoid the obstacle. It can be understood that there are multiple prediction results and multiple ways to control the vehicle, which are not listed here one by one.

[0136] It can be seen that in the embodiment of the present invention, during the movement of the vehicle, the vehicle artificial intelligence perception network model can be trained online in real time, and the model parameters of the vehicle artificial intelligence perception network model can be adjusted, thereby realizing online feedback and autonomous learning of the vehicle artificial intelligence perception network model. The prediction accuracy of the vehicle artificial intelligence perception network model after online training is higher, which improves the safety of the vehicle, and online training does not require special collection of training data, which reduces the training cost of the vehicle artificial intelligence perception network model. Online hierarchical and collaborative learning realizes the adaptive and generalization capabilities of vehicle intelligent perception and intelligent decision-making. This hierarchical feedback self-learning vehicle artificial intelligence technology is suitable for safe vehicle intelligent driving.

[0137] The embodiment of the present invention also provides a hierarchical feedback self-learning device 60 for vehicle artificial intelligence, such as Figure 6 As shown, the device comprises:

[0138] The video stream acquisition module 610 is used to acquire the video stream collected by the camera of the vehicle during the driving process of the vehicle;

[0139] The video stream input module 620 is used to input the video stream into the vehicle artificial intelligence perception network model to be trained online, and perform the following steps through the vehicle artificial intelligence perception network model:

[0140] Extracting feature information of multiple dimensions of each video frame in the video stream;

[0141] For feature information of each dimension of the current video frame, first difference information between first feature information and second feature information is calculated; the first feature information is feature information of the dimension of the video frame, and the second feature information is feature information of the dimension of a next video frame of the video frame; the current video frame is any video frame of the video stream;

[0142] generating motion space information of the vehicle based on the first differential information and the motion information of the vehicle, and predicting feature information of the dimension of the next video frame based on the motion space information;

[0143] Determining a hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame;

[0144] Based on the hierarchical feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to hierarchical feedback self-learning to obtain the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

[0145] The embodiment of the present invention further provides a vehicle control device 70 based on vehicle artificial intelligence, such as Figure 7 As shown, the device comprises:

[0146] The video stream acquisition module 710 is used to acquire the video stream collected by the camera of the vehicle during the driving process of the vehicle;

[0147] A model online hierarchical feedback self-learning module 720 is used to input the video stream into the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning, so as to output a perception result through the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning;

[0148] The vehicle control module 730 is used to control the vehicle accordingly based on the perception result.

[0149] In a fifth aspect, an embodiment of the present invention provides an electronic device 800, such as Figure 8 As shown, including:

[0150] at least one processor 801;

[0151] a memory 802 for storing the at least one processor executable instruction;

[0152] Wherein, the at least one processor is configured to execute the instructions to implement the above method.

[0153] In a sixth aspect, an embodiment of the present invention provides a computer-readable storage medium. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the above method.

[0154] In a seventh aspect, an embodiment of the present invention provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0155] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and intent of the present invention.

Claims

1. A hierarchical feedback self-learning method for vehicle artificial intelligence, characterized in that: The method comprises: During the driving of the vehicle, obtaining a video stream captured by a camera of the vehicle; The video stream is input into the vehicle artificial intelligence perception network model to be trained online, and the following steps are performed through the vehicle artificial intelligence perception network model: Extracting feature information of multiple dimensions of each video frame in the video stream; For feature information of each dimension of the current video frame, first difference information between first feature information and second feature information is calculated; the first feature information is feature information of the dimension of the current video frame, and the second feature information is feature information of the dimension of a video frame next to the current video frame; the current video frame is any video frame of the video stream; generating motion space information of the vehicle based on the first differential information and the motion information of the vehicle, and predicting feature information of the dimension of the next video frame based on the motion space information; Determining a hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame; Based on the hierarchical feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to hierarchical feedback self-learning to obtain the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

2. The method according to claim 1, characterized in that The vehicle artificial intelligence perception network model includes a two-dimensional feature information extraction sub-model, a three-dimensional feature information extraction sub-model and a four-dimensional feature information extraction sub-model; The extracting feature information of multiple dimensions of each video frame in the video stream includes: Extracting two-dimensional feature information of each video frame in the video stream through a two-dimensional feature information extraction sub-model, wherein the two-dimensional feature information is used to characterize the plane feature information of the video frame; Extracting three-dimensional feature information of each video frame in the video stream by using a three-dimensional feature information extraction sub-model, wherein the three-dimensional feature information is used to characterize spatial feature information of the video frame; The four-dimensional feature information of each video frame in the video stream is extracted through a four-dimensional feature information extraction sub-model, wherein the four-dimensional feature information is used to characterize the spatial feature information and the temporal feature information of the video frame.

3. The method according to claim 2, characterized in that The determining of the hierarchical feedback online autonomous learning strategy based on the second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame includes: Determine first parameter adjustment gradient information corresponding to the two-dimensional feature information extraction sub-model based on second difference information between the predicted two-dimensional feature information of the next video frame and the extracted two-dimensional feature information of the next video frame, and construct a two-dimensional feature feedback network based on the first parameter adjustment gradient information; Determining second parameter adjustment gradient information corresponding to the 3D feature information extraction sub-model based on second difference information between the predicted 3D feature information of the next video frame and the extracted 3D feature information of the next video frame, and constructing a 3D feature feedback network based on the second parameter adjustment gradient information; Determine third parameter adjustment gradient information corresponding to the four-dimensional feature information extraction submodel based on second difference information between the predicted four-dimensional feature information of the next video frame and the extracted four-dimensional feature information of the next video frame, and construct a four-dimensional feature feedback network based on the third parameter adjustment gradient information; Based on the three-dimensional feature feedback network and the four-dimensional feature feedback network, construct an inter-layer feedback network; Based on the two-dimensional feature feedback network, the three-dimensional feature feedback network, the four-dimensional feature feedback network and the inter-layer feedback network, a layered feedback online autonomous learning strategy is determined.

4. The method according to claim 3, characterized in that: The step of determining a hierarchical feedback online autonomous learning strategy based on the two-dimensional feature feedback network, the three-dimensional feature feedback network, the four-dimensional feature feedback network, and the inter-layer feedback network includes: Determining first parameter adjustment information of the two-dimensional feature information extraction sub-model based on the two-dimensional feature feedback network and the inter-layer feedback network; Determining second parameter adjustment information of the three-dimensional feature information extraction sub-model based on the three-dimensional feature feedback network and the inter-layer feedback network; The third parameter adjustment information of the four-dimensional feature information extraction sub-model is determined based on the four-dimensional feature feedback network.

5. The method according to claim 4, characterized in that The vehicle artificial intelligence perception network model is subjected to layered feedback self-learning based on the layered feedback online autonomous learning strategy to obtain the vehicle artificial intelligence perception network model after online layered feedback self-learning, including: Performing hierarchical feedback self-learning on the two-dimensional feature information extraction sub-model through the first parameter adjustment information; Performing hierarchical feedback self-learning on the three-dimensional feature information extraction sub-model through the second parameter adjustment information; The four-dimensional feature information extraction sub-model is subjected to hierarchical feedback self-learning through the third parameter adjustment information.

6. A vehicle control method based on vehicle artificial intelligence, characterized in that: The method comprises: Acquire a video stream collected by a camera of the vehicle during driving of the vehicle; Inputting the video stream into the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning as claimed in any one of claims 1 to 5, so as to output a perception result through the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning; The vehicle is controlled accordingly based on the perception result.

7. A hierarchical feedback self-learning device for vehicle artificial intelligence, characterized in that: The device comprises: A video stream acquisition module is used to acquire the video stream collected by the camera of the vehicle while the vehicle is traveling; The video stream input module is used to input the video stream into the vehicle artificial intelligence perception network model to be trained online, and perform the following steps through the vehicle artificial intelligence perception network model: Extracting feature information of multiple dimensions of each video frame in the video stream; For feature information of each dimension of the current video frame, first difference information between first feature information and second feature information is calculated; the first feature information is feature information of the dimension of the video frame, and the second feature information is feature information of the dimension of a next video frame of the video frame; the current video frame is any video frame of the video stream; generating motion space information of the vehicle based on the first differential information and the motion information of the vehicle, and predicting feature information of the dimension of the next video frame based on the motion space information; Determining a hierarchical feedback online autonomous learning strategy based on second difference information between the predicted feature information of the dimension of the next video frame and the extracted feature information of the dimension of the next video frame; Based on the hierarchical feedback online autonomous learning strategy, the vehicle artificial intelligence perception network model is subjected to hierarchical feedback self-learning to obtain the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning.

8. A vehicle control device based on vehicle artificial intelligence, characterized in that: The device comprises: A video stream acquisition module is used to acquire the video stream collected by the camera of the vehicle during the driving process of the vehicle; A model online hierarchical feedback self-learning module, used to input the video stream into the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning as claimed in any one of claims 1 to 5, so as to output a perception result through the vehicle artificial intelligence perception network model after online hierarchical feedback self-learning; A vehicle control module is used to control the vehicle accordingly based on the perception result.

9. An electronic device, characterized in that: include: at least one processor; a memory for storing the at least one processor-executable instruction; The at least one processor is configured to execute the instructions to implement the method according to any one of claims 1 to 5 or claim 6.

10. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method according to any one of claims 1 to 5 or claim 6.

Images