Vehicle path planning method and device, vehicle and storage medium
By extracting multi-view features and generating bird's-eye view BEV features, detecting the range of safety enclosure box, autonomous vehicles can identify and avoid adjacent lane vehicles, solving the collision risk problem during emergency lateral intrusion of obstacles and improving the safety of path planning and intelligence.
Patent Information
- Application Number
- CN202411064514.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-08
AI Technical Summary
When an autonomous vehicle encounters an emergency lateral intrusion of obstacles, it is difficult to make accurate predictions, resulting in collision risks and reducing the safety and intelligence of path planning.
By acquiring the multi-view image data of the vehicle, using the preset neural network model to extract multi-view features, generate bird's-eye view BEV features, detect whether the vehicle is in the range of the safety enclosure box, and plan according to the driving information of the vehicles in adjacent lanes and the range of the safety enclosure box. The vehicle's driving path to ensure safe distance and avoidance.
It improves the safety and intelligence of autonomous driving path planning, reduces collision risks, and ensures advance avoidance when planning bicycle paths.
Smart Images

Figure CN118913305A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of automobile control technology, and in particular to a vehicle path planning method, device, vehicle and storage medium. Background Art
[0002] A typical scenario for autonomous vehicles is parallel driving or overtaking. When there are vehicles driving at a low speed or preparing to use the adjacent lane, the traditional collision detection based on obstacle trajectory prediction is poor because the motion model of the target object may not conform to the setting of the constant speed or constant curvature model. A feasible solution is to set a safe distance for the obstacle, and the path of the self-driving car should keep the distance from the obstacle greater than the safe distance. However, the setting of the safe distance is often written in the software configuration file and is not a variable parameter, so it may not meet the actual scene safety driving requirements in some extreme scenarios.
[0003] In the related technology, it is possible to predict the future driving trajectory of an obstacle through its cloud-based speed and direction, calculate the collision position and collision time between the autonomous driving vehicle and the obstacle based on the future driving trajectory of the autonomous driving vehicle and the future motion trajectory of the obstacle, calculate the avoidance start time of the autonomous driving vehicle based on the collision position and collision time, and construct the expected trajectory of the vehicle through preliminary planning, and perform collision detection based on the received dynamic obstacle predicted trajectory and the expected trajectory of the vehicle to avoid dynamic obstacles.
[0004] However, the relevant technology is difficult to make accurate predictions when encountering emergency lateral intrusion of obstacles, which can easily lead to collision risks, reduce the safety of path planning, and has low intelligence and practicality, which urgently needs to be improved. Summary of the invention
[0005] The present application provides a vehicle path planning method, device, vehicle and storage medium to solve the problems of the related technology that when encountering an emergency lateral intrusion of an obstacle, it is difficult to make an accurate prediction, which easily leads to the risk of collision, reduces the safety of path planning, and has low intelligence and practicality.
[0006] The first aspect of the present application provides a vehicle path planning method, comprising the following steps: acquiring multi-perspective image data of the vehicle, and extracting multi-perspective features of the vehicle from the multi-perspective image data using a preset neural network model; obtaining a bird's-eye view BEV feature based on the multi-perspective feature of the vehicle, and detecting whether the vehicle is in a preset safety bounding box range based on the BEV feature; if the vehicle is in the preset safety bounding box range, determining that the vehicle is in a dangerous driving state, and performing a dangerous driving reminder action, otherwise determining that the vehicle is in a safe driving state, and identifying driving information of vehicles in adjacent lanes, and planning the vehicle's driving path based on the driving information and the preset safety bounding box range to control the vehicle to travel along the driving path.
[0007] Optionally, in one embodiment of the present application, the identifying driving information of the vehicle in the adjacent lane and planning the driving path of the vehicle according to the driving information and the preset safety bounding box range includes: detecting the surrounding environment of the vehicle and extracting at least one driving information of the position, speed and direction of the vehicle in the adjacent lane from the surrounding environment; based on the at least one driving information and the preset safety bounding box range, planning the driving path of the vehicle using a preset path planning algorithm so that the vehicle and the vehicle in the adjacent lane meet a preset safety distance condition, and controlling the vehicle to travel along the driving path within the preset safety distance condition.
[0008] Optionally, in one embodiment of the present application, generating BEV features based on multi-view features of the vehicle includes: based on the Transformer algorithm architecture, using self-attention and cross-attention mechanisms to process the multi-view features of the vehicle to generate the BEV features.
[0009] Optionally, in one embodiment of the present application, before detecting whether the vehicle is within a preset safety bounding box range according to the BEV feature, it also includes: training the preset neural network model using a training data set to obtain a trained neural network model; iteratively updating the trained neural network model until a preset iterative update stop condition is met, stopping the iteration, and obtaining the safety bounding box range.
[0010] Optionally, in one embodiment of the present application, after planning the driving path of the vehicle according to the driving information and the preset safety bounding box range, it also includes: judging whether the vehicle meets preset safety driving conditions according to the driving path; if the vehicle meets the preset safety driving conditions, controlling the vehicle to travel along the driving path, otherwise activating the path replanning mode of the vehicle; planning a new driving path based on the path replanning mode, so that the vehicle travels along the new driving path.
[0011] The second aspect of the present application provides a vehicle path planning device, including: an acquisition module, used to acquire multi-perspective image data of the vehicle, and use a preset neural network model to extract multi-perspective features of the vehicle from the multi-perspective image data; a detection module, used to obtain a bird's-eye view BEV feature based on the multi-perspective feature of the vehicle, and detect whether the vehicle is in a preset safety bounding box range based on the BEV feature; a planning module, used to determine that the vehicle is in a dangerous driving state when the vehicle is in the preset safety bounding box range, and perform a dangerous driving reminder action, otherwise it is in a safe driving state, and identifies the driving information of vehicles in adjacent lanes, and plans the driving path of the vehicle according to the driving information and the preset safety bounding box range to control the vehicle to travel along the driving path.
[0012] Optionally, in one embodiment of the present application, the planning module includes: a detection unit, used to detect the surrounding environment of the vehicle, and extract at least one driving information of the position, speed and direction of the vehicle in the adjacent lane from the surrounding environment; a planning unit, used to plan the driving path of the vehicle based on the at least one driving information and the preset safety bounding box range, using a preset path planning algorithm, so that the vehicle and the vehicle in the adjacent lane meet a preset safety distance condition, and control the vehicle to travel along the driving path within the preset safety distance condition.
[0013] Optionally, in one embodiment of the present application, the detection module includes: a processing unit, configured to process the multi-view features of the vehicle based on a Transformer algorithm architecture using self-attention and cross-attention mechanisms to generate the BEV features.
[0014] Optionally, in one embodiment of the present application, it also includes: a training module, used to train the preset neural network model using a training data set to obtain a trained neural network model before detecting whether the vehicle is in a preset safety bounding box range based on the BEV characteristics; an iteration module, used to iteratively update the trained neural network model until a preset iterative update stop condition is met, and then stop the iteration to obtain the safety bounding box range.
[0015] Optionally, in one embodiment of the present application, it also includes: a judgment module, which is used to plan the driving path of the vehicle according to the driving information and the preset safety bounding box range, and then judge whether the vehicle meets the preset safe driving conditions according to the driving path; a control module, which is used to control the vehicle to travel along the driving path when the vehicle meets the preset safe driving conditions, otherwise activate the path replanning mode of the vehicle; and a path planning module, which is used to plan a new driving path based on the path replanning mode so that the vehicle travels along the new driving path.
[0016] The third aspect of the present application provides a vehicle, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the vehicle path planning method as described in the above embodiment.
[0017] The fourth aspect of the present application provides a computer-readable storage medium, which stores a computer program that, when executed by a processor, implements the above-mentioned vehicle path planning method.
[0018] The embodiment of the present application can identify the driving information of vehicles in adjacent lanes, and plan the driving path of the vehicle according to the driving information and the range of the safety bounding box to control the vehicle to travel along the driving path, ensure early avoidance when planning the path of the vehicle, and improve the safety of automatic driving path planning, with high intelligence and reduced collision risk. Thus, it solves the problems of the related technology that it is difficult to make accurate predictions when encountering an emergency lateral intrusion of obstacles, which easily leads to collision risks, reduces the safety of path planning, and has low intelligence and practicality.
[0019] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:
[0021] Figure 1 A flowchart of a vehicle path planning method provided according to an embodiment of the present application;
[0022] Figure 2 A schematic diagram of a vehicle path planning method according to an embodiment of the present application;
[0023] Figure 3 A schematic diagram of path planning of a vehicle path planning method according to an embodiment of the present application;
[0024] Figure 4 A schematic diagram of the structure of a vehicle path planning device provided according to an embodiment of the present application;
[0025] Figure 5 It is a schematic diagram of the structure of a vehicle provided according to an embodiment of the present application. DETAILED DESCRIPTION
[0026] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.
[0027] The path planning method, device, vehicle and storage medium of the vehicle of the embodiment of the present application are described below with reference to the accompanying drawings. In view of the problem that the related technologies mentioned in the above background technology are difficult to make accurate predictions when encountering an emergency lateral intrusion of obstacles, which easily leads to collision risks, reduces the safety of path planning, and has low intelligence and practicality, the present application provides a vehicle path planning method, in which the driving information of vehicles in adjacent lanes can be identified, and the driving path of the vehicle can be planned according to the driving information and the range of the safety bounding box to control the vehicle to travel along the driving path, ensure early avoidance during the self-vehicle path planning, improve the safety of automatic driving path planning, have high intelligence, and reduce the risk of collision. As a result, the problem that the related technologies are difficult to make accurate predictions when encountering an emergency lateral intrusion of obstacles, which easily leads to collision risks, reduces the safety of path planning, and has low intelligence and practicality is solved.
[0028] Specifically, Figure 1 A schematic flow chart of a vehicle path planning method provided in an embodiment of the present application.
[0029] like Figure 1 As shown, the path planning method of the vehicle includes the following steps:
[0030] In step S101, multi-view image data of a vehicle is acquired, and multi-view features of the vehicle are extracted from the multi-view image data using a preset neural network model.
[0031] It can be understood that the multi-view image data in the embodiments of the present application can be obtained by a camera installed on the vehicle.
[0032] In the actual implementation process, the embodiment of the present application can prepare sample data, which can come from a public data set or data collected by an autonomous vehicle. For the scenario where a vehicle in an adjacent lane borrows the lane where the vehicle is located, the n frames of image data before borrowing the lane are used as input, and the bounding box of the activity space of the vehicle in the adjacent lane in the future t frames is used as output to construct a data set. The embodiment of the present application can capture multi-view image data through a camera, and use a preset neural network model to extract the multi-view features of the vehicle from the multi-view image data.
[0033] The embodiments of the present application can utilize a preset neural network model to extract multi-view features of a vehicle from multi-view image data, thereby enhancing detection accuracy, improving driving safety, and facilitating safer obstacle avoidance and path planning.
[0034] In step S102 , a bird's-eye view (BEV) feature is obtained according to the multi-view feature of the vehicle, and it is detected according to the BEV feature whether the vehicle is within a preset safety bounding box range.
[0035] It is understandable that when driving in the city, you will often encounter parallel or overtaking scenes. For example, if your car is driving in the second lane on the right, and there is a vehicle driving at a low speed in the first lane on the right, when the car encounters an obstacle in front and needs to detour, it is very likely to borrow the lane where the car is located. This application believes that the vehicle in the adjacent lane borrowing the lane to drive may be a sudden behavior, which is difficult to model and predict through a simple geometric model of the speed and heading angle of the vehicle in the adjacent lane. However, in some actual traffic scenarios, it is often possible to predict through some environmental conditions, such as the lane line is in a dotted line, which is more likely to encounter vehicles in the adjacent lane borrowing the lane where the car is located. For example, if the vehicle in the adjacent lane encounters a low-speed obstacle (pedestrians, two-wheeled vehicles, etc.) in front, there is a high probability that the vehicle in the adjacent lane will borrow the lane where the car is located. For example, there is a merging intersection in front of the first lane on the right, and there is also a high probability that the vehicle in the adjacent lane will borrow the lane where the car is located. These environmental factors and road features can be identified through Bird's-Eye-View (BEV) feature extraction.
[0036] During the actual implementation process, the embodiment of the present application can obtain the bird's-eye view BEV features based on the multi-perspective features of the vehicle, and detect whether the vehicle is within a preset safety bounding box range based on the BEV features, so as to avoid it in advance when planning the self-driving path based on the judgment results, thereby improving the safety of autonomous driving path planning.
[0037] It should be noted that the preset safety bounding box range can be set by those skilled in the art according to actual conditions and is not specifically limited here.
[0038] Optionally, in one embodiment of the present application, BEV features are generated based on multi-view features of the vehicle, including: based on the Transformer algorithm architecture, using self-attention and cross-attention mechanisms to process the multi-view features of the vehicle to generate BEV features.
[0039] It can be understood that the Transformer in the embodiment of the present application is a deep learning model.
[0040] As a possible way to achieve this, Figure 2 As shown, the embodiment of the present application can be based on the Transformer algorithm architecture, use the self-attention and cross-attention mechanism to process the multi-view features of the vehicle, generate BEV features, and the BEV features are connected to the linear layer. The final output result is a 4-dimensional vector (x, y, w, h), which means the center coordinates, width, and height of the safety bounding box that the vehicle needs to maintain a safe distance from the vehicle in the adjacent lane and must avoid, such as Figure 3 shown.
[0041] The embodiment of the present application can predict whether vehicles in adjacent lanes will borrow the lane based on the BEV+transformer algorithm, so as to consider the safety distance in the path planning process and reduce the risk of collision. The embodiment of the present application adopts a spatiotemporal cross-attention mechanism in the feature extraction process, and the data set does not need manual annotation. The neural network calculates the safety bounding box through self-supervised learning, provides constraints for path planning, and improves the safety of path planning.
[0042] Optionally, in one embodiment of the present application, before detecting whether the vehicle is within a preset safety bounding box range based on BEV features, it also includes: training a preset neural network model using a data set to obtain a trained neural network model; iteratively updating the trained neural network model until a preset iterative update stop condition is met, stopping the iteration, and obtaining a safety bounding box range.
[0043] It can be understood that the preset iterative update stopping condition in the embodiment of the present application can be a condition for stopping the iteration when the neural network model reaches the maximum number of iterations.
[0044] In the actual implementation process, the embodiment of the present application can use the training data set to train the preset neural network model, split the data set into a training set, a test set and a validation set, and enrich the data set with the data collected by the autonomous driving car to obtain the trained neural network model, and iteratively update the trained neural network model until a certain iterative update stop condition is met, then stop the iteration and obtain the safe bounding box range. In the scenario of parallel overtaking, when encountering obstacles in adjacent lanes, the embodiment of the present application calculates the above-mentioned safe bounding box parameters (x, y, w, h) through a neural network based on the multi-view image input of the current frame and the historical n-1 frames.
[0045] The embodiment of the present application can extract environmental features from multi-view image data based on the BEV+transformer architecture, calculate the safe bounding box range of the adjacent lane vehicle when it may borrow the lane through the neural network, and avoid it in advance when planning the path of the vehicle, which can improve the safety of autonomous driving path planning. The embodiment of the present application has better generalization and can have data-driven iteration of better path planning strategies.
[0046] It should be noted that the preset iterative update stop condition can be set by those skilled in the art according to actual conditions and is not specifically limited here.
[0047] In step S103, if the vehicle is within the preset safety bounding box range, the vehicle is determined to be in a dangerous driving state and a dangerous driving reminder action is performed; otherwise, the vehicle is in a safe driving state, and the driving information of vehicles in adjacent lanes is identified, and the vehicle's driving path is planned according to the driving information and the preset safety bounding box range to control the vehicle to travel along the driving path.
[0048] It is understandable that the dangerous driving reminder action in the embodiment of the present application can be a sound prompt and / or a display light flashing prompt when the vehicle invades the safety enclosure and is in a dangerous driving state.
[0049] In actual implementation, the embodiment of the present application can determine that the vehicle is in a dangerous driving state when the vehicle is within the range of the safety encircling box, and perform a dangerous driving reminder action, for example, the embodiment of the present application performs a sound prompt and / or a display light flashing prompt. The embodiment of the present application can determine that the vehicle is in a safe driving state when the vehicle is not within the range of the safety encircling box, and identify the driving information of the vehicle in the adjacent lane, and plan the vehicle's driving path according to the driving information and the range of the safety encircling box to control the vehicle to drive along the driving path.
[0050] The embodiment of the present application can identify the driving information of vehicles in adjacent lanes, and plan the vehicle's driving path according to the driving information and the range of the safety bounding box to control the vehicle to travel along the driving path and ensure early avoidance when planning the vehicle's path. This can improve the safety of autonomous driving path planning, has high intelligence, and reduces the risk of collision.
[0051] It should be noted that the preset safety bounding box range can be set by those skilled in the art according to actual conditions and is not specifically limited here.
[0052] Optionally, in one embodiment of the present application, driving information of vehicles in adjacent lanes is identified, and a driving path of the vehicle is planned based on the driving information and a preset safety bounding box range, including: detecting the surrounding environment of the vehicle, and extracting at least one driving information of the position, speed and direction of the vehicle in the adjacent lane from the surrounding environment; based on at least one driving information and a preset safety bounding box range, planning the driving path of the vehicle using a preset path planning algorithm so that the vehicle and the vehicle in the adjacent lane meet a preset safety distance condition, and controlling the vehicle to travel along the driving path within the preset safety distance condition.
[0053] It can be understood that the preset safety distance condition in the embodiment of the present application can be that the distance between the current vehicle and the vehicle in the adjacent lane satisfies the safety distance condition.
[0054] Specifically, the embodiments of the present application can detect the surrounding environment of the vehicle and extract driving information such as the position, speed and direction of vehicles in adjacent lanes from the surrounding environment. When the vehicle is doing path planning, in the parallel and overtaking stages, the path planning algorithm based on RRT or A*, based on the position of the target object in the adjacent lane identified, re-plans the path of the vehicle according to the range of the safety bounding box, so that the vehicle and the vehicles in the adjacent lane meet a certain safety distance condition, and controls the vehicle to travel along the driving path within a certain safety distance condition, ensuring that the vehicle will not intrude into the safety bounding box when traveling on the planned path, thereby improving the safety of path planning.
[0055] It should be noted that the preset safety bounding box range and the preset safety distance condition can be set by those skilled in the art according to actual conditions and are not specifically limited here.
[0056] Optionally, in one embodiment of the present application, after planning the vehicle's driving path according to the driving information and the preset safety bounding box range, it also includes: judging whether the vehicle meets the preset safe driving conditions based on the driving path; if the vehicle meets the preset safe driving conditions, controlling the vehicle to drive along the driving path, otherwise activating the vehicle's path replanning mode; planning a new driving path based on the path replanning mode, so that the vehicle drives along the new driving path.
[0057] It can be understood that the preset safe driving conditions in the embodiment of the present application can be driving conditions that maintain a safe state when the vehicle is driving; the path replanning mode in the embodiment of the present application can be a mode for planning a new path for the vehicle.
[0058] In the actual implementation process, the embodiment of the present application can determine whether the vehicle meets certain safe driving conditions based on the driving path, and control the vehicle to drive along the driving path if the vehicle meets certain safe driving conditions. The embodiment of the present application can activate the vehicle's path replanning mode when the planned path can no longer ensure that the vehicle drives in the current lane, plan a new driving path based on the path replanning mode, and trigger a left lane change or deceleration and parking scene to make the vehicle drive along the new driving path.
[0059] The embodiment of the present application can plan a new driving path based on the path replanning mode, so that the vehicle can travel along the new driving path, thereby avoiding obstacles or dangerous areas, improving the operating efficiency and safety of the vehicle, and improving the passenger experience.
[0060] It should be noted that the preset safe driving conditions can be set by those skilled in the art according to actual conditions and are not specifically limited here.
[0061] According to the vehicle path planning method proposed in the embodiment of the present application, the driving information of the vehicles in the adjacent lanes can be identified, and the vehicle's driving path can be planned according to the driving information and the range of the safety bounding box to control the vehicle to travel along the driving path, and ensure early avoidance when planning the vehicle's own path. This can improve the safety of automatic driving path planning, have high intelligence, and reduce the risk of collision. This solves the problem that the related technology is difficult to make accurate predictions when encountering an emergency lateral intrusion of obstacles, which easily leads to collision risks, reduces the safety of path planning, and has low intelligence and practicality.
[0062] Next, a path planning device for a vehicle proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.
[0063] Figure 4 It is a structural schematic diagram of a path planning device for a vehicle according to an embodiment of the present application.
[0064] like Figure 4 As shown, the vehicle path planning device 10 includes: an acquisition module 100, a detection module 200 and a planning module 300.
[0065] Specifically, the acquisition module 100 is used to acquire multi-view image data of the vehicle and extract multi-view features of the vehicle from the multi-view image data using a preset neural network model.
[0066] The detection module 200 is used to obtain a bird's-eye view BEV feature according to the multi-view feature of the vehicle, and detect whether the vehicle is within a preset safety bounding box range according to the BEV feature.
[0067] The planning module 300 is used to determine that the vehicle is in a dangerous driving state and perform a dangerous driving reminder action when the vehicle is in a preset safety bounding box range. Otherwise, it is in a safe driving state, and identifies the driving information of vehicles in adjacent lanes, and plans the vehicle's driving path according to the driving information and the preset safety bounding box range to control the vehicle to travel along the driving path.
[0068] Optionally, in one embodiment of the present application, the planning module 300 includes: a detection unit and a planning unit.
[0069] The detection unit is used to detect the surrounding environment of the vehicle and extract at least one driving information of the position, speed and direction of the vehicle in the adjacent lane from the surrounding environment.
[0070] The planning unit is used to plan the vehicle's driving path based on at least one driving information and a preset safety bounding box range using a preset path planning algorithm so that the vehicle and the vehicles in the adjacent lanes meet the preset safety distance conditions, and control the vehicle to travel along the driving path within the preset safety distance conditions.
[0071] Optionally, in one embodiment of the present application, the detection module 200 includes: a processing unit.
[0072] Among them, the processing unit is used to process the multi-view features of the vehicle based on the Transformer algorithm architecture, using self-attention and cross-attention mechanisms to generate BEV features.
[0073] Optionally, in one embodiment of the present application, the vehicle path planning device 10 further includes: a training module and an iteration module.
[0074] Among them, the training module is used to train a preset neural network model using a training data set before detecting whether the vehicle is within a preset safety bounding box range according to the BEV characteristics to obtain a trained neural network model.
[0075] The iteration module is used to iteratively update the trained neural network model until the preset iterative update stop condition is met, then stop the iteration and obtain the safe bounding box range.
[0076] Optionally, in one embodiment of the present application, the vehicle path planning device 10 further includes: a judgment module, a control module and a path planning module.
[0077] The judgment module is used to judge whether the vehicle meets the preset safe driving conditions according to the driving path after planning the vehicle's driving path according to the driving information and the preset safety bounding box range.
[0078] The control module is used to control the vehicle to travel along the driving path when the vehicle meets the preset safe driving conditions, otherwise the vehicle's path replanning mode is activated.
[0079] The path planning module is used to plan a new driving path based on the path replanning mode so that the vehicle can travel along the new driving path.
[0080] It should be noted that the above explanation of the vehicle path planning method embodiment is also applicable to the vehicle path planning device of this embodiment, and will not be repeated here.
[0081] According to the vehicle path planning device proposed in the embodiment of the present application, the driving information of the vehicles in the adjacent lanes can be identified, and the vehicle's driving path can be planned according to the driving information and the range of the safety bounding box to control the vehicle to travel along the driving path, and ensure early avoidance when planning the vehicle's own path. This can improve the safety of automatic driving path planning, have high intelligence, and reduce the risk of collision. This solves the problem that the related technology is difficult to make accurate predictions when encountering an emergency lateral intrusion of obstacles, which easily leads to collision risks, reduces the safety of path planning, and has low intelligence and practicality.
[0082] Figure 5 A schematic diagram of the structure of a vehicle provided in an embodiment of the present application. The vehicle may include:
[0083] A memory 501 , a processor 502 , and a computer program stored in the memory 501 and executable on the processor 502 .
[0084] When the processor 502 executes the program, the vehicle path planning method provided in the above embodiment is implemented.
[0085] Furthermore, the vehicle also includes:
[0086] The communication interface 503 is used for communication between the memory 501 and the processor 502 .
[0087] The memory 501 is used to store computer programs that can be executed on the processor 502 .
[0088] The memory 501 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.
[0089] If the memory 501, the processor 502 and the communication interface 503 are implemented independently, the communication interface 503, the memory 501 and the processor 502 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.
[0090] Optionally, in a specific implementation, if the memory 501, the processor 502 and the communication interface 503 are integrated on a chip, the memory 501, the processor 502 and the communication interface 503 can communicate with each other through an internal interface.
[0091] The processor 502 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.
[0092] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned vehicle path planning method is implemented.
[0093] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0094] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
[0095] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.
[0096] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.
[0097] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0098] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.
[0099] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.
[0100] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.
Claims
1. A vehicle path planning method, characterized in that: The following steps are involved: Acquire multi-view image data of a vehicle, and extract multi-view features of the vehicle from the multi-view image data using a preset neural network model; Obtaining a bird's-eye view (BEV) feature according to the multi-view feature of the vehicle, and detecting whether the vehicle is within a preset safety bounding box range according to the BEV feature; If the vehicle is within the preset safety bounding box range, the vehicle is determined to be in a dangerous driving state and a dangerous driving reminder action is performed; otherwise, the vehicle is in a safe driving state, and the driving information of vehicles in adjacent lanes is identified, and the driving path of the vehicle is planned according to the driving information and the preset safety bounding box range to control the vehicle to travel along the driving path.
2. The method according to claim 1, characterized in that The identifying driving information of the vehicle in the adjacent lane and planning the driving path of the vehicle according to the driving information and the preset safety bounding box range includes: Detecting the surrounding environment of the vehicle, and extracting at least one driving information of the position, speed and direction of the vehicle in the adjacent lane from the surrounding environment; Based on the at least one driving information and the preset safety bounding box range, a driving path of the vehicle is planned using a preset path planning algorithm so that the vehicle and the vehicle in the adjacent lane meet a preset safety distance condition, and the vehicle is controlled to travel along the driving path within the preset safety distance condition.
3. The method according to claim 1, characterized in that The generating of BEV features according to the multi-view features of the vehicle includes: Based on the Transformer algorithm architecture, the multi-view features of the vehicle are processed using self-attention and cross-attention mechanisms to generate the BEV features.
4. The method according to claim 1, characterized in that: Before detecting whether the vehicle is within a preset safety bounding box range according to the BEV feature, the method further includes: Using the data set to train the preset neural network model to obtain a trained neural network model; The trained neural network model is iteratively updated until a preset iterative update stopping condition is met, and then the iteration is stopped to obtain the safe bounding box range.
5. The method according to claim 1, characterized in that After planning the driving path of the vehicle according to the driving information and the preset safety bounding box range, the method further includes: Determining whether the vehicle meets preset safe driving conditions according to the driving path; If the vehicle meets the preset safe driving condition, controlling the vehicle to travel along the driving path, otherwise activating the path replanning mode of the vehicle; A new driving path is planned based on the path replanning mode, so that the vehicle travels along the new driving path.
6. A vehicle path planning device, characterized in that: include: An acquisition module, used to acquire multi-view image data of a vehicle, and extract multi-view features of the vehicle from the multi-view image data using a preset neural network model; A detection module, configured to obtain a bird's-eye view (BEV) feature according to the multi-view features of the vehicle, and detect whether the vehicle is within a preset safety bounding box range according to the BEV feature; The planning module is used to determine that the vehicle is in a dangerous driving state and execute a dangerous driving reminder action when the vehicle is within the preset safety bounding box range; otherwise, it is in a safe driving state, and identifies the driving information of vehicles in adjacent lanes, and plans the driving path of the vehicle according to the driving information and the preset safety bounding box range to control the vehicle to drive along the driving path.
7. The device according to claim 6, characterized in that The planning module includes: a detection unit, configured to detect the surrounding environment of the vehicle and extract at least one driving information of the position, speed and direction of the vehicle in the adjacent lane from the surrounding environment; A planning unit is used to plan the driving path of the vehicle based on the at least one driving information and the preset safety bounding box range using a preset path planning algorithm so that the vehicle and the vehicle in the adjacent lane meet a preset safety distance condition, and control the vehicle to travel along the driving path within the preset safety distance condition.
8. The device according to claim 6, characterized in that The detection module comprises: The processing unit is used to process the multi-view features of the vehicle based on the Transformer algorithm architecture by using self-attention and cross-attention mechanisms to generate the BEV features.
9. A vehicle, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the path planning method for a vehicle as described in any one of claims 1 to 5.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the path planning method for a vehicle as described in any one of claims 1 to 5.
Citation Information
Cited By
Vehicle avoidance method, device, equipment, storage medium and program product
CN120646012A
