A vehicle driving planning method, device, equipment and storage medium

Abstract

The application discloses a vehicle driving planning method, device and equipment and a storage medium, and comprises the following steps: determining a to-be-adjusted running state of a target obstacle by running an information prediction model according to obstacle environment information of the target obstacle around a target vehicle at a to-be-detected running time; determining to-be-adjusted running information of the target obstacle and running information occurrence probability corresponding to the to-be-adjusted running information by running the information prediction model according to the to-be-adjusted running state; and determining target running information of the target obstacle and target driving planning information of the target vehicle by running the information prediction model according to the to-be-adjusted running information, the running information occurrence probability and to-be-adjusted driving planning information of the target vehicle. The accuracy of the target running information can be improved, and thus the reliability of the target driving planning information of the target vehicle obtained can be improved.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the field of computers, and in particular to a vehicle driving planning method, device, equipment and storage medium. BACKGROUND

[0002] In an L4 level automatic driving system, scenarios that may affect the driving trajectory of the ego vehicle need to be dealt with, such as other vehicles cutting in, vehicle conflicts at intersections, and pedestrians crossing the road. The traditional method of making decisions on the running trajectory of the vehicle based on the instantaneous state has poor adaptability to the development and change of the environment around the vehicle, and when dealing with the above scenarios, it is easy to produce delay, thus leading to collision risk. In order to enable the vehicle to reasonably respond to environmental changes, the decision and planning module on the vehicle needs to be provided with the trend of environmental changes in a future period of time, i.e. the possible running information of the obstacles around the vehicle in a future period of time, so it is necessary to predict the running information of the obstacles around the vehicle. Obstacle running information prediction mainly includes two categories: rule-based prediction method and learning algorithm-based prediction method. The rule-based prediction method is a method of establishing a behavior rule library according to formal rules, traffic regulations and driving common sense, etc., and predicting the running information of the obstacles according to the behavior rule library. The learning algorithm-based prediction method is a method of establishing a behavior rule library by machine learning, combining the obstacle environmental information and the behavior information of the obstacles with the behavior rule library for behavior matching by using a machine learning algorithm to predict the running information of the obstacles. However, the existing obstacle running information prediction method only considers the dynamic change of the traffic conditions of the obstacles and the ego vehicle in time, ignores the mutual dependence of the obstacles and the ego vehicle in space, and is difficult to introduce the interaction between the obstacles and the ego vehicle into the vehicle driving planning system. Moreover, insufficient scene depth traversal of the traffic scene leads to low accuracy of the running information of the obstacles determined by the vehicle driving planning system. Therefore, how to improve the accuracy of the prediction result of the running information of the obstacles and thus improve the accuracy of the vehicle driving planning of the ego vehicle is a problem to be solved. SUMMARY

[0003] The present application provides a vehicle driving planning method, device, equipment and storage medium, which can improve the accuracy of the prediction result of the running information of the obstacles and thus improve the accuracy of the vehicle driving planning of the ego vehicle.

[0004] By means of the running information prediction model, the running state of the target obstacle to be adjusted is determined according to the obstacle environmental information of the target obstacle around the target vehicle in the to-be-detected running time;

[0005] By means of the running information prediction model, the running information of the target obstacle to be adjusted is determined according to the running state to be adjusted, and the running information occurrence probability corresponding to the running information to be adjusted is determined.

[0006] The running information prediction model is used to determine, according to the to-be-adjusted running information, the running information occurrence probability, and the to-be-adjusted driving planning information of the target vehicle, the target running information of the target obstacle, and the target driving planning information of the target vehicle.

[0007] According to another aspect of the present application, a vehicle driving planning device is provided, which comprises:

[0008] A running state determination module is configured to determine, by a running information prediction model, a to-be-adjusted running state of a target obstacle around a target vehicle according to obstacle environment information of the target obstacle at a to-be-detected running time.

[0009] A running information determination module is configured to determine, by the running information prediction model, to-be-adjusted running information of the target obstacle and a running information occurrence probability corresponding to the to-be-adjusted running information according to the to-be-adjusted running state.

[0010] A driving planning information determination module is configured to determine, by the running information prediction model, target running information of the target obstacle and target driving planning information of the target vehicle according to the to-be-adjusted running information, the running information occurrence probability, and the to-be-adjusted driving planning information of the target vehicle.

[0011] According to another aspect of the present application, an electronic device is provided, which comprises:

[0012] at least one processor; and

[0013] a memory connected with the at least one processor; wherein

[0014] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the vehicle driving planning method according to any one of the embodiments of the present application.

[0015] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for enabling a processor to execute the vehicle driving planning method according to any one of the embodiments of the present application.

[0016] The technical solution of the embodiment of the present invention uses an operation information prediction model to determine the target obstacle's to-be-adjusted operation state based on the obstacle environment information of the target obstacle around the target vehicle at the time of operation to be detected; uses the operation information prediction model to determine the target obstacle's to-be-adjusted operation information and the probability of occurrence of the operation information corresponding to the to-be-adjusted operation information based on the to-be-adjusted operation state; and uses the operation information prediction model to determine the target obstacle's target operation information and the target driving plan information of the target vehicle based on the to-be-adjusted operation information, the probability of occurrence of the operation information, and the target vehicle's to-be-adjusted driving plan information. This solution solves the problem that when predicting obstacle operation information and planning the vehicle's driving based on the predicted obstacle operation information, only the temporal dynamic changes in the traffic conditions of the obstacle and the vehicle are considered, while the spatial interdependence between the obstacle and the vehicle is ignored. This leads to errors in the prediction of obstacle operation information and unreliable vehicle driving plans. The operation state to be adjusted of the target obstacle is determined through an operation information prediction model, and then the operation information to be adjusted and the probability of occurrence of the operation information to be adjusted are determined based on the operation state to be adjusted. The target operation information of the target obstacle and the target driving planning information of the target vehicle are determined based on the analysis of the operation information to be adjusted, the probability of occurrence of the operation information to be adjusted, and the input driving planning information to be adjusted. When predicting the target operation information of the target obstacle, the spatial interdependence between the target obstacle and the target vehicle can be taken into account, thereby improving the accuracy of the target operation information and thus improving the reliability of the acquired target driving planning information of the target vehicle.

[0017] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A flowchart of a vehicle driving planning method provided in Example 1 of the present invention;

[0020] Figure 2 A flowchart of a vehicle driving planning method provided in the second embodiment of the present invention;

[0021] Figure 3A schematic structural diagram of a vehicle driving planning device provided in Embodiment 3 of the present invention;

[0022] Figure 4 This is a structural diagram of an electronic device provided in Example 4 of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "candidate" and "target" and the like in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this way are interchangeable where appropriate so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "etc." and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or apparatus.

[0025] Example 1

[0026] Figure 1 A flowchart of a vehicle driving planning method is provided for the first embodiment of the present invention. This embodiment is applicable to determining the operation information of target obstacles around a vehicle and the target driving planning information of the vehicle. The method can be executed by a vehicle driving planning device, which can be implemented in the form of hardware and / or software, and can be configured in an electronic device. Figure 1 As shown, the method includes:

[0027] S110 , determining the operating state to be adjusted of the target obstacle according to obstacle environment information of the target obstacle around the target vehicle at the operating time to be detected through the operating information prediction model.

[0028] The operation information prediction model includes a convolutional neural network and a long short-term memory network. The convolutional neural network can be a MobilenetV3 model, and the long short-term memory network is LSTM. The operating time to be tested can be set according to actual needs. For example, the operating time to be tested can be the first 8 seconds of the current operating time of the target vehicle. Target obstacles refer to obstacles around the target vehicle that may affect the target vehicle's driving trajectory. Target obstacles can be stationary or moving objects. Obstacle environment information refers to the environment around the target vehicle, including obstacles. The target obstacle's operating state to be adjusted refers to the initial operating state information of the target obstacle obtained through the operation information prediction model. There may be a certain error between the operating state to be adjusted and the actual operating state of the target obstacle, so further adjustments are required in subsequent steps.

[0029] Specifically, the convolutional neural network in the operation information prediction model determines the target obstacle's operating motion and trajectory during the time period to be detected based on the obstacle environment information surrounding the target vehicle. These motions include left turns, right turns, left lane changes, right lane changes, straight driving, and U-turns. The target obstacle's operating motion and trajectory during the time period to be detected are used as the target obstacle's operating state to be adjusted.

[0030] Exemplarily, the operating state to be adjusted of the target obstacle may be determined through the following sub-steps:

[0031] S1101, determining a target position relationship graph of the target vehicle and the target obstacle based on the vehicle position of the target vehicle during the time of operation to be detected and the obstacle position of the target obstacle during the time of operation to be detected, and determining a target position relationship vector based on the target position relationship graph;

[0032] The target position relationship graph refers to an image that can represent the position relationship between the target vehicle and the target obstacle. The target position relationship vector can be a position relationship vector corresponding to the first second of the detection time.

[0033] Specifically, a target position relationship diagram of the target vehicle and the target obstacle is drawn based on the vehicle position and obstacle position of the target vehicle during the time to be detected, as well as the obstacle position of the target obstacle during the time to be detected. One target position relationship diagram can be drawn every second. The position relationship diagram includes a vehicle geometric image representing the target vehicle and an obstacle geometric image representing the target obstacle. There is a connecting line between the vehicle geometric image and the obstacle geometric image, and the connecting line is provided with a distance value representing the target vehicle and the target obstacle. The target position relationship diagram for the first second of the time to be detected is read, and the target position relationship diagram for the first second of the time to be detected is processed into a feature vector in the range of [0,1]. The feature vector obtained according to the target position relationship diagram is used as the target position relationship vector.

[0034] S1102 : Determine target feature data of the target obstacle based on obstacle environment information and target position relationship vector of the target obstacle around the target vehicle during the running time to be detected.

[0035] Specifically, the obstacle environment information of the target obstacle around the target vehicle during the detection run time is rendered as a raster image, processed into a feature vector in the range [0, 1], and the feature vector obtained from the raster image is used as the obstacle environment vector. The obstacle environment vector and the target position relationship vector are used as the target feature data of the target obstacle.

[0036] S1103: Determine the operating state of the target obstacle to be adjusted according to the target characteristic data through the operating information prediction model.

[0037] The operation state to be adjusted refers to the initial acquisition of the target obstacle.

[0038] Specifically, the target feature data is used as input data of the operation information prediction model, and the convolutional neural network in the operation information prediction model determines the operation state of the target obstacle to be adjusted based on the input target feature data.

[0039] It can be understood that by determining the target feature data of the target obstacle based on the obstacle environment information and the target position relationship vector that can characterize the position relationship between the target vehicle and the target obstacle, the position relationship between the target vehicle and the target obstacle can be taken into account when determining the target feature data, thereby improving the reliability of the target feature data.

[0040] S120 , determining the operation information to be adjusted of the target obstacle and the probability of occurrence of the operation information corresponding to the operation information to be adjusted according to the operation state to be adjusted using the operation information prediction model.

[0041] The "operation information to be adjusted" refers to the initially acquired target obstacle's motion and trajectory. There may be some discrepancy between the "operation information to be adjusted" and the actual obstacle's motion information. Therefore, further adjustments are required to obtain more accurate target obstacle motion information. The "operation information occurrence probability" refers to the probability that the target obstacle's actual motion matches the "operation information to be adjusted."

[0042] Specifically, through the convolutional neural network in the operation information prediction model, at least one operation information to be adjusted of the target obstacle is determined according to the operation state to be adjusted, and the operation information occurrence probability corresponding to each operation information to be adjusted is obtained.

[0043] For example, five frames of obstacle environment information can be rendered as a raster image with a time step of 0.2 seconds. Using the operational information prediction model, the five frames of target obstacle operational states to be adjusted can be output based on the raster images rendered from the obstacle environment information. Based on these five frames of target obstacle operational states to be adjusted, the operational information to be adjusted for each target obstacle in each state to be adjusted, as well as the probability of occurrence of the operational information corresponding to the operational information to be adjusted, can be determined using the operational information prediction model.

[0044] S130. Determine the target operation information of the target obstacle and the target driving plan information of the target vehicle through the operation information prediction model according to the operation information to be adjusted, the probability of occurrence of the operation information, and the driving plan information to be adjusted of the target vehicle.

[0045] The driving plan information to be adjusted may be manually input initial vehicle driving plan information. The vehicle driving plan information, namely the ego vehicle planning information, refers to the target vehicle's driving action and vehicle trajectory determined after planning the target vehicle's subsequent action and subsequent driving trajectory.

[0046] Specifically, through the long short-term memory network in the operation information prediction model, the target operation information of the target obstacle and the target driving plan information of the target vehicle are determined according to the operation information to be adjusted, the probability of occurrence of the operation information and the driving plan information to be adjusted of the target vehicle.

[0047] Exemplarily, a method for determining target operating information of a target obstacle and target driving planning information of a target vehicle may be: using the operating information to be adjusted and the probability of occurrence of the operating information as obstacle-related information, and performing a game interaction on the obstacle-related information and the target vehicle's driving planning information to be adjusted through an operating information prediction model; and determining the target operating information of the target obstacle and the target driving planning information of the target vehicle based on the game interaction results.

[0048] Specifically, the obstacle-related information and the probability of the operation being adjusted are used as the obstacle-related information. The obstacle-related information and the target vehicle's driving plan information to be adjusted are used as the LSTM in the operation information prediction model. In this case, the LSTM can serve as the encoder and decoder in the operation information prediction model. The LSTM uses a game interaction between the obstacle-related information and the target vehicle's driving plan information to be adjusted. Based on the game interaction results, the target obstacle's target operation information and the target vehicle's target driving plan information are determined.

[0049] The above method determines the target operation information of the target obstacle and the target driving plan information of the target vehicle based on the game interaction results between the obstacle-related information and the target vehicle's driving plan information to be adjusted, which can improve the reliability of the target operation information of the target obstacle and the target driving plan information of the target vehicle.

[0050] The technical solution provided in this embodiment uses an operation information prediction model to determine the target obstacle's to-be-adjusted operation state based on the obstacle environment information of the target obstacle around the target vehicle at the time of operation to be detected. The operation information prediction model also determines the target obstacle's to-be-adjusted operation information and the probability of occurrence of the operation information corresponding to the to-be-adjusted operation information based on the to-be-adjusted operation state. The operation information prediction model also determines the target obstacle's target operation information and the target vehicle's target driving plan information based on the to-be-adjusted operation information, the probability of occurrence of the operation information, and the target vehicle's to-be-adjusted driving plan information. This solution solves the problem that when predicting obstacle operation information and planning the ego vehicle's driving based on the predicted obstacle operation information, only the temporal dynamic changes in the traffic conditions of the obstacle and the ego vehicle are considered, while the spatial interdependence between the obstacle and the ego vehicle is ignored. This leads to errors in the prediction of obstacle operation information and unreliable driving plans for the ego vehicle. The operation state to be adjusted of the target obstacle is determined through an operation information prediction model, and then the operation information to be adjusted and the probability of occurrence of the operation information to be adjusted are determined based on the operation state to be adjusted. The target operation information of the target obstacle and the target driving planning information of the target vehicle are determined based on the analysis of the operation information to be adjusted, the probability of occurrence of the operation information to be adjusted, and the input driving planning information to be adjusted. When predicting the target operation information of the target obstacle, the spatial interdependence between the target obstacle and the target vehicle can be taken into account, thereby improving the accuracy of the target operation information and thus improving the reliability of the acquired target driving planning information of the target vehicle.

[0051] Example 2

[0052] Figure 2This is a flowchart of a vehicle driving planning method provided in Example 2 of the present invention. This embodiment is optimized based on the above embodiment and provides a preferred implementation method for training a dual-flow network model based on the target vehicle's historical driving planning information, historical location, and historical obstacle data of the vehicle's historical obstacles to determine the operation information prediction model. Specifically, Figure 2 As shown, the method includes:

[0053] S210 : Determine model training data for the dual-stream network model based on historical driving planning information and historical positions of the target vehicle and historical obstacle data of historical obstacles encountered by the vehicle.

[0054] The dual-stream network model includes a convolutional neural network and a long short-term memory network; the historical obstacle data includes a historical environment dataset and a historical operating status dataset corresponding to the vehicle's historical obstacles.

[0055] Historical obstacles refer to obstacles around the target vehicle during its historical driving process. The historical environment dataset consists of historical obstacle environment information. The historical operation status dataset consists of historical obstacle operation status information. This historical obstacle operation status information includes the historical obstacle operation movements and operation trajectories.

[0056] Exemplarily, a method for determining model training data for a dual-stream network model may be: determining a historical position relationship between a target vehicle and the vehicle's historical obstacles based on the target vehicle's historical position and the vehicle's historical obstacle data; determining historical feature data of the vehicle's historical obstacles based on the historical position relationship and the historical obstacle data, and using the historical feature data and the target vehicle's historical driving planning information as model training data for the dual-stream network model.

[0057] Among them, the historical position relationship can be represented by a historical position relationship graph between the target vehicle and historical obstacles.

[0058] Specifically, based on the historical position of the target vehicle and the historical obstacle data of the vehicle's historical obstacles, a historical position relationship graph between the target vehicle and the vehicle's historical obstacles is determined, the historical position relationship graph is processed into a feature vector in the range of [0,1], and the feature vector obtained from the historical position relationship graph is used as the historical position relationship vector. A historical environment dataset is extracted from the historical obstacle data, and the historical environment image in the historical environment dataset is processed into a feature vector in the range of [0,1]. The feature vector obtained from the historical environment image is used as the historical environment vector. At the same time, the historical operating status is extracted from the historical obstacle data, the historical operating status is processed into a feature vector in the range of [0,1], and the feature vector obtained from the historical operating status is used as the historical operating status vector. The historical position relationship vector, historical environment vector, and historical operating status vector are used as model training data for the dual-stream network model.

[0059] The above scheme takes into account the positional relationship between the target vehicle and the vehicle's historical obstacles when determining the model training data of the dual-stream network model, which can improve the reliability of the target operation information of the target obstacle and the target driving planning information of the target vehicle determined by the trained dual-stream network model.

[0060] S220: Perform model training on the dual-stream network model based on the model training data to determine the operation information prediction model.

[0061] The operation information prediction model is used to determine the target operation information of the target obstacle corresponding to the target vehicle and the target driving plan information of the target vehicle based on the obstacle environment information of the target vehicle at the operation time to be detected and the driving plan information to be adjusted.

[0062] Specifically, the historical location relationship vector and the historical environment vector are used as sample training data in the model training data, and the historical operation state vector is used as sample supervision data in the model training data. The two-stream network model is supervised and trained based on the sample training data and sample supervision data to determine the operation information prediction model.

[0063] An exemplary method for determining the operation information prediction model may include: training the dual-stream network model using a gradient descent method based on model training data, and determining the obstacle prediction operation status during training. Based on the historical operation status dataset and the obstacle prediction operation status, the mean squared error of the dual-stream network model and the corresponding prediction success rate for the obstacle prediction operation status are determined. When the mean squared error and prediction success rate meet training completion criteria, the dual-stream network model training is determined to be complete, and the trained dual-stream network model is used as the operation information prediction model.

[0064] The above solution provides a method for training a dual-stream network model using model training data, which can improve the model accuracy of the operation information prediction model.

[0065] The technical solution of this embodiment determines the model training data of the dual-stream network model based on the historical driving plan information, historical position, and historical obstacle data of the vehicle's historical obstacles; based on the model training data, the dual-stream network model is model-trained to determine the operation information prediction model. It is possible to obtain an operation information prediction model that determines the target operation information of the target obstacle corresponding to the target vehicle and the target driving plan information of the target vehicle based on the obstacle environment information and the driving plan information to be adjusted of the target vehicle at the time of operation to be detected. This improves the efficiency of obtaining the target operation information and the target driving plan information. At the same time, the model training data of the dual-stream network model is determined based on the historical driving plan information, historical position, and historical obstacle data of the target vehicle's historical obstacles, fully considering the historical position relationship between the target vehicle and the vehicle's historical obstacles. Model training of the dual-stream network model based on the model training data can improve the reliability of the operation information prediction model.

[0066] Example 3

[0067] Figure 3 This is a schematic diagram of the structure of a vehicle driving planning device provided by the third embodiment of the present invention. This embodiment is applicable to the case of determining the operation information of target obstacles around the vehicle and the target driving planning information of the vehicle. Figure 3 As shown, the vehicle driving planning device includes: an operating state determination module 310, an operating information determination module 320 and a driving planning information determination module 330.

[0068] The operating state determination module 310 is configured to determine the operating state of the target obstacle to be adjusted based on the obstacle environment information of the target obstacle around the target vehicle at the operating time to be detected by using the operating information prediction model;

[0069] The operation information determination module 320 is configured to determine the operation information to be adjusted of the target obstacle and the probability of occurrence of the operation information corresponding to the operation information to be adjusted based on the operation state to be adjusted using the operation information prediction model;

[0070] The driving plan information determination module 330 is used to determine the target operating information of the target obstacle and the target driving plan information of the target vehicle through the operating information prediction model according to the operating information to be adjusted, the probability of occurrence of the operating information and the driving plan information to be adjusted of the target vehicle.

[0071] The technical solution provided in this embodiment uses an operation information prediction model to determine the target obstacle's to-be-adjusted operation state based on the obstacle environment information of the target obstacle around the target vehicle at the time of operation to be detected. The operation information prediction model also determines the target obstacle's to-be-adjusted operation information and the probability of occurrence of the operation information corresponding to the to-be-adjusted operation information based on the to-be-adjusted operation state. The operation information prediction model also determines the target obstacle's target operation information and the target vehicle's target driving plan information based on the to-be-adjusted operation information, the probability of occurrence of the operation information, and the target vehicle's to-be-adjusted driving plan information. This solution solves the problem that when predicting obstacle operation information and planning the ego vehicle's driving based on the predicted obstacle operation information, only the temporal dynamic changes in the traffic conditions of the obstacle and the ego vehicle are considered, while the spatial interdependence between the obstacle and the ego vehicle is ignored. This leads to errors in the prediction of obstacle operation information and unreliable driving plans for the ego vehicle. The operation state to be adjusted of the target obstacle is determined through an operation information prediction model, and then the operation information to be adjusted and the probability of occurrence of the operation information to be adjusted are determined based on the operation state to be adjusted. The target operation information of the target obstacle and the target driving planning information of the target vehicle are determined based on the analysis of the operation information to be adjusted, the probability of occurrence of the operation information to be adjusted, and the input driving planning information to be adjusted. When predicting the target operation information of the target obstacle, the spatial interdependence between the target obstacle and the target vehicle can be taken into account, thereby improving the accuracy of the target operation information and thus improving the reliability of the acquired target driving planning information of the target vehicle.

[0072] Exemplarily, the operating status determination module 310 is specifically configured to:

[0073] Determining a target position relationship graph of the target vehicle and the target obstacle based on the vehicle position of the target vehicle during the time of operation to be detected and the obstacle position of the target obstacle during the time of operation to be detected, and determining a target position relationship vector based on the target position relationship graph;

[0074] Determining target feature data of the target obstacle based on obstacle environment information and target position relationship vector of the target obstacle around the target vehicle during the running time to be detected;

[0075] The operation information prediction model is used to determine the operation status of the target obstacle to be adjusted based on the target characteristic data.

[0076] Exemplarily, the driving plan information determination module 330 is specifically configured to:

[0077] The operation information to be adjusted and the probability of occurrence of the operation information are used as obstacle-related information, and the obstacle-related information and the target vehicle's driving plan information to be adjusted are interactively analyzed through the operation information prediction model.

[0078] The target operation information of the target obstacle and the target driving planning information of the target vehicle are determined according to the game interaction results.

[0079] Exemplarily, the vehicle driving planning device further includes:

[0080] The model training data determination module is used to determine the model training data of the dual-stream network model based on the historical driving plan information and historical location of the target vehicle, as well as the historical obstacle data of the vehicle's historical obstacles. The dual-stream network model includes a convolutional neural network and a long short-term memory network. The historical obstacle data includes a historical environment dataset and a historical operating status dataset corresponding to the vehicle's historical obstacles.

[0081] The model training module is used to train the dual-stream network model based on the model training data and determine the operation information prediction model; the operation information prediction model is used to determine the target operation information of the target obstacle corresponding to the target vehicle and the target driving plan information of the target vehicle based on the obstacle environment information of the target vehicle at the time of operation to be detected and the driving plan information to be adjusted.

[0082] Exemplarily, the model training data determination module is specifically used to:

[0083] Determine the historical position relationship between the target vehicle and the historical obstacles of the vehicle based on the historical position of the target vehicle and the historical obstacle data of the historical obstacles of the vehicle;

[0084] According to the historical position relationship and historical obstacle data, the historical feature data of the vehicle's historical obstacles are determined, and the historical feature data and the historical driving planning information of the target vehicle are used as model training data for the dual-stream network model.

[0085] Exemplarily, the model training module is specifically used to:

[0086] Based on the model training data, the gradient descent method is used to train the dual-stream network model to determine the obstacle prediction operation status during the training process;

[0087] Based on the historical operating status dataset and the obstacle prediction operating status, determine the mean square error of the dual-stream network model and the prediction success rate corresponding to the obstacle prediction operating status;

[0088] When the mean square error and the prediction success rate meet the training completion conditions, the dual-stream network model training is determined to be completed, and the dual-stream network model after training is used as the operation information prediction model.

[0089] Exemplarily, the above-mentioned operation information prediction model includes a convolutional neural network and a long short-term memory network.

[0090] The vehicle driving planning device provided in this embodiment can be applied to the vehicle driving planning method provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0091] Example 4

[0092] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0093] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0094] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0095] The processor 11 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, and the like. The processor 11 performs various methods and processes described above, such as the vehicle travel planning method.

[0096] In some embodiments, the vehicle travel planning method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded onto and / or installed in the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded onto the RAM 13 and executed by the processor 11, one or more steps of the vehicle travel planning method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the vehicle travel planning method by any other suitable means, such as by means of firmware.

[0097] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (PLD), a computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0098] Computer programs used to implement the methods of the application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program, when executed by the processor of the machine, implements the functions / acts specified in the flowcharts and / or block diagrams. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, and partially on a machine or a remote machine or a server.

[0099] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0100] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0101] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0102] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0103] It should be understood that the various forms of flow shown above can be used to reorder, add or delete steps. For example, each step described in the present application can be executed in parallel, sequentially or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, which is not limited herein.

[0104] The above detailed description does not constitute a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A vehicle driving planning method, characterized in that: include: Determining the operating state to be adjusted of the target obstacle based on obstacle environment information of the target obstacle around the target vehicle at the operating time to be detected using an operating information prediction model; Determining, by the operation information prediction model, the operation information to be adjusted of the target obstacle and the probability of occurrence of the operation information corresponding to the operation information to be adjusted according to the operation state to be adjusted; Using the to-be-adjusted operating information and the probability of occurrence of the operating information as obstacle-related information, and performing a game interaction between the obstacle-related information and the to-be-adjusted driving plan information of the target vehicle through the operating information prediction model; Determining target operation information of the target obstacle and target driving planning information of the target vehicle according to the game interaction result; The method further comprises: Determining model training data for a dual-stream network model based on historical driving plan information and historical locations of the target vehicle, as well as historical obstacle data of historical obstacles encountered by the vehicle; the dual-stream network model includes a convolutional neural network and a long short-term memory network; the historical obstacle data includes a historical environment dataset and a historical operating status dataset corresponding to the historical obstacles encountered by the vehicle; Based on the model training data, the dual-stream network model is trained to determine an operation information prediction model; the operation information prediction model is used to determine the target operation information of the target obstacle corresponding to the target vehicle and the target driving plan information of the target vehicle based on the obstacle environment information of the target vehicle at the time of operation to be detected and the driving plan information to be adjusted.

2. The method according to claim 1, characterized in that Determining the operating state to be adjusted of the target obstacle based on obstacle environment information of the target obstacle around the target vehicle at the operating time to be detected by the operating information prediction model includes: Determining a target position relationship graph of the target vehicle and the target obstacle based on the vehicle position of the target vehicle during the time of operation to be detected and the obstacle position of the target obstacle during the time of operation to be detected, and determining a target position relationship vector based on the target position relationship graph; determining target feature data of the target obstacle based on obstacle environment information of the target obstacle around the target vehicle during the running time to be detected and the target position relationship vector; The operation state to be adjusted of the target obstacle is determined according to the target characteristic data through the operation information prediction model.

3. The method according to claim 1, characterized in that Based on the historical driving planning information, historical location, and historical obstacle data of the target vehicle, the model training data of the dual-stream network model is determined, including: Determining a historical position relationship between the target vehicle and the historical obstacle data of the vehicle according to the historical position of the target vehicle and the historical obstacle data of the vehicle; According to the historical position relationship and the historical obstacle data, historical feature data of the historical obstacle of the vehicle is determined, and the historical feature data and the historical driving planning information of the target vehicle are used as model training data of the dual-stream network model.

4. The method according to claim 1, wherein The dual-stream network model is trained according to the model training data to determine an operation information prediction model, including: Training the dual-stream network model using a gradient descent method based on the model training data, and determining an obstacle prediction operating state during the training process; Determining a mean square error of the dual-stream network model and a prediction success rate corresponding to the obstacle prediction operating state based on the historical operating state dataset and the obstacle prediction operating state; When the mean square error and the prediction success rate meet the training completion condition, it is determined that the training of the dual-stream network model is completed, and the dual-stream network model after the training is completed is used as the operation information prediction model.

5. The method according to any one of claims 1 to 2, characterized in that The operation information prediction model includes a convolutional neural network and a long short-term memory network.

6. A vehicle driving planning device, characterized in that: include: An operating state determination module is configured to determine the operating state to be adjusted of the target obstacle according to obstacle environment information of the target obstacle around the target vehicle at the operating time to be detected using an operating information prediction model; an operation information determination module, configured to determine, by using the operation information prediction model and according to the operation state to be adjusted, the operation information to be adjusted of the target obstacle and an occurrence probability of the operation information corresponding to the operation information to be adjusted; a driving plan information determination module, configured to determine, by means of the operation information prediction model, target operation information of the target obstacle and target driving plan information of the target vehicle according to the operation information to be adjusted, the probability of occurrence of the operation information, and the driving plan information to be adjusted of the target vehicle; The driving plan information determination module is specifically configured to: use the to-be-adjusted operating information and the probability of occurrence of the operating information as obstacle-related information, and perform a game interaction between the obstacle-related information and the to-be-adjusted driving plan information of the target vehicle through the operating information prediction model; and determine the target operating information of the target obstacle and the target driving plan information of the target vehicle based on the game interaction results; A model training data determination module is configured to determine model training data for a dual-stream network model based on historical driving plan information and historical locations of the target vehicle, as well as historical obstacle data of historical obstacles encountered by the vehicle; the dual-stream network model includes a convolutional neural network and a long short-term memory network; the historical obstacle data includes a historical environment dataset and a historical operating status dataset corresponding to the historical obstacles encountered by the vehicle; A model training module is used to perform model training on the dual-stream network model based on the model training data to determine an operation information prediction model; the operation information prediction model is used to determine the target operation information of the target obstacle corresponding to the target vehicle and the target driving plan information of the target vehicle based on the obstacle environment information of the target vehicle at the time of operation to be detected and the driving plan information to be adjusted.

7. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the vehicle driving planning method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the vehicle driving planning method according to any one of claims 1 to 5 when executed.

Images