CN118082815A - Obstacle avoidance processing method and device for vehicle, vehicle and storage medium - Google Patents

Abstract

The application relates to an obstacle avoidance processing method and device for a vehicle, the vehicle and a storage medium. The method comprises the following steps: receiving a driving scene image of a vehicle in a driving direction, wherein the driving scene image at least comprises a self-vehicle lane; classifying the road condition information on the own vehicle lane according to a preset classification model to obtain corresponding driving decision information; wherein, road conditions information includes: the vehicle lane has an irregular obstacle, the vehicle lane has a preset obstacle or the vehicle lane has no obstacle. The scheme provided by the application can efficiently generate driving decision information, reduce the response time of the system, enhance the reliability of the obstacle avoidance strategy and reduce the probability of missed triggering and false triggering.

Description

Vehicle obstacle avoidance processing method, device, vehicle and storage medium

Technical Field

The present application relates to the field of autonomous driving technology, and in particular to a vehicle obstacle avoidance method, device, vehicle and storage medium.

Background technique

When a vehicle with an autonomous driving function is driving, if there is an obstacle in front of it, the vehicle needs to be able to identify the obstacle and make driving decisions to avoid it. In related technologies, especially for static obstacles, such as roadblocks placed on the lane, the generation of vehicle driving decisions requires the coordinated processing of the perception module, fusion module, and planning module. Among them, the perception module identifies obstacles and lane lines, the fusion module confirms the position of the obstacle in the lane by reconstructing the three-dimensional map, and the planning module selects the corresponding obstacle avoidance strategy according to the location and degree of danger of the obstacle.

However, the technical solution of the "perception-fusion-planning" multi-module collaborative operation needs to process information from different sources, making the implementation of the entire technical solution complicated and the system latency high. In addition, the three-dimensional reconstruction of distant obstacles has the problem of inaccurate construction, which makes it difficult to ensure the reliability and completeness of the obstacle avoidance strategy.

Summary of the invention

In order to solve or partially solve the problems existing in the related technology, the present application provides a vehicle obstacle avoidance processing method, device, vehicle and storage medium, which can efficiently generate driving decision information, reduce system response time, enhance the reliability of obstacle avoidance strategy and reduce the probability of missed triggering and false triggering.

The first aspect of the present application provides a vehicle obstacle avoidance processing method, comprising:

Receiving a driving scene image of the vehicle in a driving direction, wherein the driving scene image at least includes a lane of the vehicle;

The road condition information on the own vehicle lane is classified according to a preset classification model to obtain corresponding driving decision information; wherein the road condition information includes: the existence of unconventional obstacles in the own vehicle lane, the existence of preset obstacles in the own vehicle lane, or the absence of obstacles in the own vehicle lane.

In some embodiments, the driving decision information includes changing lanes, braking, or maintaining driving, wherein:

If the road condition information indicates that there is an unconventional obstacle in the lane of the vehicle, the driving decision information is braking;

If the road condition information indicates that there is a preset obstacle in the lane of the vehicle, then the driving decision information is changing lanes;

If the road condition information indicates that there is no obstacle in the vehicle lane, the driving decision information is to keep driving.

In some embodiments, the preset obstacle is a static obstacle;

If the road condition information indicates that there is a preset obstacle in the lane of the vehicle, then the driving decision information is lane change, including:

If the road condition information indicates that there is a preset obstacle in the vehicle's lane, the driving decision information is to change lanes to the left or to the right according to the placement of the preset obstacle.

In some implementations, the training method of the preset classification model includes:

Constructing training data based on the collected training images and the corresponding decision labels; wherein the corresponding decision labels are annotated according to the road condition information in each frame of the training image;

The training data is used to train the preset classification model to obtain a trained preset classification model.

In some implementations, the training images include: real scene images and/or simulated images;

The real scene image is a scene image containing preset obstacles or conventional obstacles; and the simulation image is a scene image containing unconventional obstacles.

In some implementations, the driving scene image also includes adjacent lanes;

If the road condition information indicates that there is an obstacle in an adjacent lane, the driving decision information includes an intrusion on the left side of the lane or an intrusion on the right side of the lane.

In some implementations, the method further includes: sending the driving decision information to a preset planning module for processing to output path planning information.

A second aspect of the present application provides a vehicle obstacle avoidance processing device, comprising:

An image receiving module, used for receiving a driving scene image of the vehicle in the driving direction, wherein the driving scene image at least includes a lane of the vehicle;

The classification decision module is used to classify the road condition information on the vehicle lane according to a preset classification model to obtain corresponding driving decision information; wherein the road condition information includes: the presence of unconventional obstacles in the vehicle lane, the presence of preset obstacles in the vehicle lane, or the absence of obstacles in the vehicle lane.

A third aspect of the present application provides a vehicle, comprising:

Processor; and

The memory stores executable codes thereon, and when the executable codes are executed by the processor, the processor is caused to execute the method as described above.

A fourth aspect of the present application provides a computer-readable storage medium having executable code stored thereon. When the executable code is executed by a processor of a vehicle, the processor is caused to execute the method as described above.

The technical solution provided by this application may have the following beneficial effects:

The vehicle obstacle avoidance processing method of the present application classifies the road condition information in the driving direction of the vehicle's own lane based on the real-time collected driving scene images, and outputs the corresponding driving decision information quickly and accurately without omissions for different types of road condition information, especially the presence of preset obstacles, so that the autonomous driving vehicle can avoid obstacles based on a shorter response time, improve processing efficiency, and ensure safe driving.

It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and other objects, features and advantages of the present application will become more apparent by describing in more detail exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

FIG1 is a schematic flow chart of a vehicle obstacle avoidance processing method according to an embodiment of the present application;

FIG2 is a training image of a vehicle lane with a preset obstacle facing the left front shown in an embodiment of the present application;

FIG3 is a training image of a vehicle lane with a preset obstacle facing the right front, shown in an embodiment of the present application;

FIG4 is another schematic flow chart of a vehicle obstacle avoidance method according to an embodiment of the present application;

FIG5 is a training image showing an unconventional obstacle in the vehicle lane shown in an embodiment of the present application;

FIG6 is a schematic diagram of the structure of an obstacle avoidance processing device for a vehicle according to an embodiment of the present application;

FIG. 7 is a schematic diagram of the structure of a vehicle according to an embodiment of the present application.

Detailed ways

The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and/or" used in this article refers to and includes any or all possible combinations of one or more associated listed items.

It should be understood that although the terms "first", "second", "third", etc. may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

In related technologies, when a vehicle is performing autonomous driving, based on the traditional rule-based system, it needs to be processed step by step through different functional modules to trigger the obstacle avoidance strategy. With the increase of links, it is difficult to ensure that the obstacle avoidance strategy is triggered for all scenes that require obstacle avoidance. To obtain a very accurate decision result, the fusion of multiple sensor data is required, and there is still a situation of missed triggering for obstacles at a long distance.

In view of the above problems, an embodiment of the present application provides a vehicle obstacle avoidance processing method, which can quickly, accurately and seamlessly trigger the vehicle's obstacle avoidance strategy in its own lane.

The technical solution of the embodiments of the present application is described in detail below with reference to the accompanying drawings.

FIG. 1 is a flow chart of a vehicle obstacle avoidance method according to an embodiment of the present application.

Referring to FIG. 1 , a vehicle obstacle avoidance processing method shown in an embodiment of the present application includes:

S110, receiving a driving scene image of the vehicle in a driving direction, where the driving scene image at least includes a lane of the vehicle.

During normal driving of the vehicle, the camera installed on the vehicle can capture the image in front of the driving direction in real time, and obtain a real-time driving scene image. It can be understood that the lane in which the vehicle is driving is the own vehicle lane, and the driving scene image can include the own vehicle lane in the driving direction.

For example, the vehicle is traveling forward along the own lane, and the driving scene image includes a real scene image in front of the own lane, so that the road condition information in front of the own lane can be displayed at a corresponding position in the image.

In the present application, the effect of what the human eye sees is what it gets can be simulated based on the driving scene image. Therefore, compared with sensors such as radar that are limited by sensing distance, obstacles farther away from the vehicle can be quickly identified based on the driving scene image without being restricted by distance, allowing the system to perform obstacle avoidance planning faster.

S120, classifying the road condition information on the own vehicle lane according to a preset classification model to obtain corresponding driving decision information; wherein the road condition information includes: the existence of unconventional obstacles in the own vehicle lane, the existence of preset obstacles in the own vehicle lane, or the absence of obstacles in the own vehicle lane.

It is understood that there may be obstacles or no obstacles in front of the vehicle's own lane. Further, the road condition information includes the presence of unconventional obstacles in the own lane, the presence of preset obstacles in the own lane, or the absence of obstacles in the own lane. Among them, unconventional obstacles can be, for example, abnormal, rare or randomly appearing obstacles such as overturned vehicles, wheels, cartons, garbage, etc. that hinder the vehicle's travel. Preset obstacles refer to static obstacles, and static obstacles can be, for example, roadblocks, such as ice cream cones, fences, etc., which are only examples for illustration and not limitation.

In some specific implementations, if the road condition information indicates that there is an unconventional obstacle in the vehicle lane, the driving decision information is braking; if the road condition information indicates that there is a preset obstacle in the vehicle lane, the driving decision information is changing lanes; if the road condition information indicates that there is no obstacle in the vehicle lane, the driving decision information is to keep driving. It can be understood that for unconventional obstacles, that is, abnormal and rare obstacles, in order to ensure the safety of vehicle driving, the driving decision information of braking can be output. For preset obstacles, the preset classification model has been sufficiently trained during the training process and can be accurately identified, so that the driving decision information of changing lanes can be used for obstacle avoidance. When there is no obstacle in the vehicle lane, the vehicle can drive normally without obstacle avoidance.

In this step, based on whether there is an obstacle in front of the vehicle lane and the type of obstacle, the corresponding driving decision information is directly generated without introducing data processing links of other different functional modules, thereby improving data processing efficiency and triggering error rate, so that the system can control the vehicle in a more rapid and accurate manner based on the obtained driving decision information.

In some embodiments, the preset classification model may be a deep learning model based on a CNN network (convolutional neural network), which is a model obtained by pre-training with training data and capable of classifying input driving scene images.

From this example, it can be seen that the vehicle obstacle avoidance processing method of the present application classifies the road condition information in the driving direction of the vehicle's own lane based on the real-time collected driving scene images, and outputs the corresponding driving decision information quickly and accurately for different types of road condition information, so that the autonomous driving vehicle can avoid obstacles based on a shorter response time, improve processing efficiency, and ensure safe driving.

An embodiment of the present application further provides a method for training a preset classification model, including:

S210, constructing training data according to the collected training images and the corresponding decision labels; wherein the corresponding decision labels are annotated according to the road condition information in each frame of the training image.

In some embodiments, the training images include: real-life images and/or simulated images; wherein the real-life images are scene images containing preset obstacles or conventional obstacles; and the simulated images are scene images containing unconventional obstacles. That is to say, the real-life images are images captured through real shooting, and at least part of the real-life images contain scenes where preset obstacles or conventional obstacles exist in front of the vehicle's lane. A small part of the real-life images contain scenes where there are no obstacles in front of the vehicle's lane. By enriching the various scenes of the real-life images, the model can be more comprehensively trained to recognize different scenes and trigger them in a timely and accurate manner. Among them, conventional obstacles, for example, can be other normally traveling vehicles located in front of the current vehicle, and the models of other vehicles are not restricted, such as cars, trucks, bicycles, motorcycles, etc.

Furthermore, the amount of data for real-life images with unconventional obstacles is small, so there is not enough training data for training the model. Based on this, simulation images are designed to supplement and make up for the lack of rare scenes. In some embodiments, the simulation image can be a completely virtual designed image, or an image obtained by combining part of the real scene with part of the virtual elements, or an image obtained by combining different real-life images. For example, a real-life image containing a vehicle lane can be used as the background, and a virtually designed unconventional obstacle can be placed on the vehicle lane of the real-life image to form a frame of simulation image.

The training method of the present application, by including real-life images of conventional obstacles and simulated images of unconventional obstacles as training data, can improve the triggering rate of the trained preset classification model and avoid missed triggering, thereby timely generating driving decision information and improving the safety of autonomous driving.

In some implementations, corresponding decision labels are pre-set based on the road condition information in the training image, i.e., the obstacle position and obstacle type in the vehicle lane. The decision labels may include lane change, braking, or keeping driving. In other words, each decision label corresponds to a driving decision information. By associating the decision label with the road condition information of the training image, the model can learn the mapping relationship between the road condition information and the driving decision information during the training process.

In some specific implementations, the decision labels for lane changes may also include changing lanes to the left and changing lanes to the right. When the training image contains a preset obstacle with a preset placement position, the label of the training image is set to change lanes to the left or change lanes to the right. As shown in FIG2, the preset obstacles are multiple static obstacles on the vehicle lane, such as multiple ice cream cones. If the multiple static obstacles are regularly placed toward the left front, it means that the vehicle needs to change lanes to the left, and the decision label of the training image is changing lanes to the left. Similarly, as shown in FIG3, when multiple static obstacles are regularly placed toward the right front, it means that the vehicle needs to change lanes to the right, and the decision label of the training image corresponds to changing lanes to the right. By introducing training images with preset obstacles, the obstacle avoidance performance of the model for static obstacles can be efficiently improved.

In some embodiments, the training image may also include a scene where an obstacle exists on an adjacent lane. If the adjacent lane is located on the left side of the own vehicle lane and there is an obstacle, the decision label of the training image is increased with intrusion on the left side; if the adjacent lane is located on the right side of the own vehicle lane and there is an obstacle, the decision label of the training image is increased with intrusion on the right side; if there are adjacent lanes on both the left and right sides of the own vehicle lane and there are obstacles on both sides, the decision label of the training image is increased with intrusion on the left side and intrusion on the right side at the same time. In other words, after first determining the corresponding position of the adjacent vehicle in the own vehicle lane, the road condition information on the adjacent lane is further determined. By marking the corresponding decision label according to the road condition information of the own vehicle lane, the corresponding decision label is also added according to the road condition information of the adjacent lane, so that the subsequently trained preset classification model can output richer driving decision information for reference to the driving scene image.

In this step, training data is formed by combining multiple frames of training images containing different road condition information and corresponding decision labels, so as to facilitate subsequent training of the preset classification model.

S220, using the training data to train a preset classification model to obtain a trained preset classification model.

In this step, the preset classification model can be a deep learning model based on a convolutional neural network. Based on the above training data as input data, through model iteration, a preset classification model that can output corresponding driving decision information according to different driving scene images can be trained.

The preset classification model of the present application can be trained through various real-life images and simulated images with different scenes to form a classification model that can output corresponding driving decision information based on the input driving scene images end-to-end, thereby efficiently improving the model performance and system performance, avoiding missed triggers and false triggers, and improving the accuracy and timeliness of vehicle obstacle avoidance.

Referring to FIG. 4 , an embodiment of the present application further provides a vehicle obstacle avoidance processing method, including:

S310, receiving a driving scene image of the vehicle in a driving direction, where the driving scene image includes a vehicle lane and/or an adjacent lane.

The driving scene image captured in real time during the driving process of the vehicle may include not only the own lane, but also the adjacent lanes located on the left and/or right side of the own lane. In other words, in addition to obstacles in front of the vehicle in the own lane, there may also be obstacles in front of the adjacent lanes.

S320, classifying the road condition information on the own vehicle lane and the adjacent lane according to a preset classification model to obtain corresponding classification results.

In this step, in addition to classifying the road condition information of the vehicle lane, the road condition information of the adjacent lanes can also be classified to obtain classification results of road condition information corresponding to different lanes. In some implementations, the road condition information corresponding to the vehicle lane is used to generate the main driving decision information, and the road condition information corresponding to the adjacent lane is used to generate the auxiliary driving decision information.

S330: If the road condition information indicates that there is an unconventional obstacle in the lane of the vehicle, the output driving decision information is braking.

Referring to FIG. 5 , in this step, if the preset classification model recognizes that the road condition information of the own vehicle lane indicates that there are unconventional obstacles in the own vehicle lane, such as overturned vehicles, wheels, cartons, garbage and other abnormal, rare or randomly appearing obstacles that hinder the vehicle's travel, the corresponding driving decision information braking is output.

S340: If the road condition information indicates that there is a preset obstacle in the vehicle lane, the output driving decision information is to change lanes to the left or to the right according to the placement of the preset obstacle.

Referring to Figures 2 and 3, in this step, the preset classification model recognizes that the road condition information of the vehicle lane is that there are preset obstacles in the vehicle lane, such as multiple roadblocks placed in sequence in front of the vehicle lane toward the left front, and the output driving decision information is to change lanes to the left; for example, multiple roadblocks are placed in sequence in front of the vehicle lane toward the right front, and the output driving decision information is to change lanes to the right. The preset classification model of the present application can efficiently output obstacle avoidance strategies for static obstacles.

It can be understood that in this step, if the road condition information indicates that there is a conventional obstacle in the vehicle lane, the output driving decision information is changing lanes.

S350: If the road condition information indicates that there is no obstacle in the vehicle lane, the output driving decision information is to keep driving.

Obviously, if the preset classification model recognizes that the road condition information of the ego vehicle lane is that there are no obstacles, the driving decision information is to keep driving, that is, the vehicle does not need to avoid obstacles and can continue to drive along the ego vehicle lane normally.

The above steps S330 to S350 are selectively executed according to the real road condition information corresponding to the driving scene image. It can be understood that if an obstacle is displayed on the lane of the vehicle in the driving scene image, the distance between the obstacle and the current vehicle can be ignored, and the corresponding driving decision information can be executed regardless of the actual distance.

S360, if the road condition information indicates that there is an obstacle in the adjacent lane, the output driving decision information includes an intrusion on the left side of the lane or an intrusion on the right side of the lane.

While identifying the road condition information of the own vehicle lane, the preset classification model can optionally also simultaneously identify the road condition information of the adjacent lane. For example, if there is an adjacent lane on the left side of the own vehicle lane and there is an obstacle in the adjacent lane on the left side, the preset classification model not only outputs the driving decision information corresponding to the own vehicle lane, but also synchronously outputs the driving decision information corresponding to the adjacent lane, such as intrusion on the left side of the lane; or, the preset classification model only outputs the driving decision information corresponding to the own vehicle lane.

This step S360 can be performed simultaneously with or after one of the above steps S330 to S350, or S360 can be performed selectively.

In some implementations, after the driving decision information is obtained, the driving decision information is sent to a preset planning module for processing to output path planning information.

It can be understood that the driving decision information can be output faster according to the preset classification model, and then can be sent to the preset planning module for path planning earlier, so that the vehicle can perform automatic driving according to the latest path planning information.

Compared with the traditional method of using multiple functional modules to process step by step to obtain driving decision information, the present application can directly generate driving decision information through a preset classification model end-to-end, shortening the processing flow of the entire system, and then shortening the response time of the system. It can also cope with longer detection distances, obtain obstacle avoidance strategies more quickly, and reserve more sufficient time and distance for the vehicle to complete path planning and lane change distance, thereby improving the safety performance of autonomous driving.

Corresponding to the aforementioned application function implementation method embodiment, the present application also provides a vehicle obstacle avoidance processing device, a vehicle and corresponding embodiments.

FIG. 6 is a schematic diagram of the structure of an obstacle avoidance device for a vehicle according to an embodiment of the present application.

Referring to FIG. 6 , an obstacle avoidance device for a vehicle shown in an embodiment of the present application includes an image receiving module 610 and a classification decision module 620 . Among them:

The image receiving module 610 is used to receive a driving scene image of the vehicle in the driving direction, and the driving scene image at least includes the vehicle lane.

The classification decision module 620 is used to classify the road condition information on the own vehicle lane according to the preset classification model to obtain corresponding driving decision information; wherein the road condition information includes: the existence of unconventional obstacles in the own vehicle lane, the existence of preset obstacles in the own vehicle lane, or the absence of obstacles in the own vehicle lane.

In a specific embodiment, the classification decision module 620 is used to determine that if the road condition information indicates that there is an unconventional obstacle in the vehicle lane, the driving decision information is braking; if the road condition information indicates that there is a preset obstacle in the vehicle lane, the driving decision information is changing lanes; if the road condition information indicates that there is no obstacle in the vehicle lane, the driving decision information is keeping driving.

In a specific implementation, the classification decision module 620 is used to determine the driving decision information to change lanes to the left or to the right according to the placement of the preset obstacle if the road condition information indicates that there is a preset obstacle in the vehicle lane.

In a specific implementation, the classification decision module 620 is used for, if the road condition information indicates that there is an obstacle in an adjacent lane, the driving decision information includes an intrusion on the left side of the lane or an intrusion on the right side of the lane.

From this example, it can be seen that the obstacle avoidance processing device of the vehicle of the present application can directly output driving decision information based on a preset classification model, obtain accurate and timely obstacle avoidance strategies, shorten the processing flow, reduce the response time of the system, and is suitable for longer detection distances, thereby improving the safety factor of the vehicle during automatic driving.

Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

FIG. 7 is a schematic diagram of the structure of a vehicle according to an embodiment of the present application.

7 , the vehicle 1000 includes a memory 1010 and a processor 1020 .

The processor 1020 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

The memory 1010 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. Among them, ROM can store static data or instructions required by the processor 1020 or other modules of the computer. The permanent storage device may be a readable and writable storage device. The permanent storage device may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In some other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at runtime. In addition, the memory 1010 may include any combination of computer-readable storage media, including various types of semiconductor storage chips (such as DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and/or optical disks may also be used. In some embodiments, the memory 1010 may include a readable and/or writable removable storage device, such as a laser disc (CD), a read-only digital versatile disc (such as a DVD-ROM, a double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (such as an SD card, a mini SD card, a Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not include carrier waves and transient electronic signals transmitted wirelessly or wired.

The memory 1010 stores executable codes, and when the executable codes are processed by the processor 1020 , the processor 1020 can execute part or all of the methods described above.

In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

Alternatively, the present application can also be implemented as a computer-readable storage medium (or non-temporary machine-readable storage medium or machine-readable storage medium) on which executable code (or computer program or computer instruction code) is stored. When the executable code (or computer program or computer instruction code) is executed by a processor of a vehicle (or a vehicle server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

The embodiments of the present application have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims (10)

1. A vehicle obstacle avoidance processing method, comprising:

receiving a driving scene image of a vehicle in a driving direction, wherein the driving scene image at least comprises a self-vehicle lane;

Classifying the road condition information on the self-vehicle lane according to a preset classification model to obtain corresponding driving decision information; wherein, the road condition information includes: the vehicle lane has an irregular obstacle, the vehicle lane has a preset obstacle or the vehicle lane has no obstacle.

2. The method of claim 1, wherein the driving decision information comprises lane change, braking, or holding, wherein:

If the road condition information is that an irregular obstacle exists in the self-vehicle lane, the driving decision information is braking;

If the road condition information is that a preset barrier exists in the self-vehicle lane, the driving decision information is lane changing;

And if the road condition information is that the own vehicle lane has no obstacle, the driving decision information is to keep running.

3. The method of claim 2, wherein the predetermined obstacle is a static obstacle;

if the road condition information is that a preset obstacle exists in the self-vehicle lane, the driving decision information is lane change, and the method comprises the following steps:

If the road condition information is that a preset obstacle exists in the self-vehicle lane, the driving decision information is left lane change or right lane change according to the placement position of the preset obstacle.

4. The method according to claim 1, wherein the training method of the preset classification model comprises:

constructing training data according to the acquired training images and the corresponding decision labels; marking a corresponding decision tag according to the road condition information in each frame of the training image;

And training the preset classification model by using the training data to obtain a trained preset classification model.

5. The method of claim 4, wherein the training image comprises: live-action images and/or simulated images;

The live-action image is a scene image containing a preset obstacle or a conventional obstacle; the simulated image is a scene image that includes non-conventional obstructions.

6. The method of claim 1, wherein the driving scene image further comprises adjacent lanes;

The method further comprises the steps of: if the road condition information is that an obstacle exists in the adjacent lane, the driving decision information comprises that the left side of the lane is invaded or the right side of the lane is invaded.

7. The method according to claim 1, wherein the method further comprises:

And sending the driving decision information to a preset planning module for processing so as to output path planning information.

8. An obstacle avoidance processing device for a vehicle, comprising:

the image receiving module is used for receiving a driving scene image of the vehicle in the driving direction, and the driving scene image at least comprises a self-vehicle lane;

The classification decision module is used for classifying the road condition information on the self-vehicle lane according to a preset classification model to obtain corresponding driving decision information; wherein, the road condition information includes: the vehicle lane has an irregular obstacle, the vehicle lane has a preset obstacle or the vehicle lane has no obstacle.

9. A vehicle, characterized by comprising:

A processor; and

A memory having executable code stored thereon, which when executed by the processor, causes the processor to perform the method of any of claims 1-7.

10. A computer readable storage medium having executable code stored thereon, which when executed by a processor of a vehicle causes the processor to perform the method of any of claims 1-7.