An intelligent driving system, method, device and medium

Abstract

The application discloses an intelligent driving system, method, device and medium, and relates to the technical field of intelligent driving, and comprises a data acquisition module, a data processing module and a vehicle control module.The data acquisition module comprises a camera and an ultrasonic radar, and is used for collecting surrounding environment data of a target vehicle.The data processing module calculates the surrounding environment data through a preset lightweight AI algorithm model to obtain target markers and drivable areas.The preset lightweight AI algorithm model is an algorithm model obtained by integrating a deep learning model for parking and driving.The vehicle control module is used for executing parking and driving actions according to the target markers and the drivable areas.In the application, the simplified system architecture solves the problems of complex vehicle architecture development and excessively high cost caused by sensor accumulation, solves the problem of low data utilization rate, and can meet the functional scene demand of L2+ at the same time; and the lightweight AI algorithm model improves the data processing efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent driving, and particularly relates to an intelligent driving system, method, device and medium. BACKGROUND

[0002] In recent years, intelligent driving technology has developed rapidly and been widely applied, and has gradually become one of the core elements of automobile intelligence. However, the accumulation of sensors leads to complex vehicle architecture development and high development costs, and also brings economic burdens to consumers.

[0003] In addition, the data collected by a large number of sensors will produce data redundancy, so that the data utilization rate is not high. Therefore, how to improve the data utilization rate and the processing efficiency of data is also a problem that needs to be continuously explored by those skilled in the art.

[0004] In summary, how to simplify the vehicle architecture and reduce the development cost, and improve the data utilization rate and the processing efficiency of data, is a problem that needs to be solved in the field. SUMMARY

[0005] Therefore, the purpose of the present application is to provide an intelligent driving system, method, device and medium. The vehicle architecture development complexity can be simplified, the development cost can be reduced, and the perception efficiency of data can be improved. The specific scheme is as follows:

[0006] In a first aspect, the present application discloses an intelligent driving system, comprising: a data acquisition module, a data processing module, and a vehicle control module.

[0007] The data acquisition module comprises a camera and an ultrasonic radar, and is configured to acquire surrounding environment data of a target vehicle through the camera and the ultrasonic radar.

[0008] The data processing module is configured to calculate the surrounding environment data through a preset lightweight AI algorithm model to identify target markers of the surrounding environment of the target vehicle and a drivable area of the target vehicle. The preset lightweight AI algorithm model is an algorithm model obtained by integrating a parking deep learning model and a driving deep learning model.

[0009] The vehicle control module is configured to perform parking and driving actions according to the target markers and the drivable area to realize intelligent auxiliary driving.

[0010] Optionally, the camera comprises fisheye cameras arranged on the front side, the rear side, the left side and the right side of the vehicle body of the target vehicle, and a front-view camera arranged on the front side of the roof of the target vehicle and a rear-view camera arranged on the rear side of the roof of the target vehicle.

[0011] Optionally, the data processing module is configured to:

[0012] The first surrounding environment data collected by the fisheye camera, the front-view camera, and the ultrasonic radar are calculated by using a preset lightweight AI algorithm model to identify first target markers of a surrounding environment of the target vehicle and a first drivable area of the target vehicle.

[0013] Correspondingly, the vehicle control module is configured to perform a parking action according to the first target markers and the first drivable area to realize intelligent auxiliary parking.

[0014] Optionally, the data processing module is configured to:

[0015] The second surrounding environment data collected by the fisheye camera, the rear-view camera, and the ultrasonic radar are calculated by using a preset lightweight AI algorithm model to identify second target markers of a surrounding environment of the target vehicle and a second drivable area of the target vehicle.

[0016] Correspondingly, the vehicle control module is configured to perform a driving action according to the second target markers and the second drivable area to realize intelligent auxiliary driving.

[0017] Optionally, the preset lightweight AI algorithm model includes multiple neck networks and head networks, and for different target marker identification tasks and drivable area identification tasks, the preset lightweight AI algorithm model respectively uses independent neck networks and head networks.

[0018] Optionally, the intelligent driving system further includes:

[0019] A data preprocessing module is configured to perform data desensitization preprocessing, data labeling processing, and data checking processing on the surrounding environment data of the target vehicle, and to automatically manage the processed surrounding environment data.

[0020] Optionally, the intelligent driving system further includes:

[0021] A visual abstraction module is configured to realize decoupling between an application layer and a bottom layer of the surrounding environment data based on a first preset middleware.

[0022] A running scheduling environment module is configured to realize efficient communication and efficient data sharing of zero-copy between homogeneous cores, heterogeneous cores, and different operating systems based on a second preset middleware, and to standardize and encapsulate data types.

[0023] In a second aspect, the present application discloses an intelligent driving method, including:

[0024] Collecting surrounding environment data of the target vehicle through a camera and ultrasonic radar;

[0025] Calculating the surrounding environment data through a preset lightweight AI algorithm model to identify target markers of the surrounding environment of the target vehicle and a drivable area of the target vehicle; wherein the preset lightweight AI algorithm model is an algorithm model obtained by integrating a deep learning model for parking and a deep learning model for driving;

[0026] Performing parking and driving actions according to the target markers and the drivable area to realize intelligent auxiliary driving.

[0027] In a third aspect, the present application discloses an electronic device, comprising:

[0028] A memory for saving a computer program;

[0029] A processor for executing the computer program to realize the intelligent driving method disclosed above.

[0030] In a fourth aspect, the present application discloses a computer readable storage medium for saving a computer program; wherein the computer program is executed by a processor to realize the intelligent driving method disclosed above.

[0031] It can be seen that the application provides an intelligent driving system, comprising a data acquisition module, a data processing module and a vehicle control module; the data acquisition module comprises a camera and an ultrasonic radar and is configured to acquire surrounding environment data of a target vehicle through the camera and the ultrasonic radar; the data processing module is configured to calculate the surrounding environment data through a preset lightweight AI algorithm model to identify target markers of the surrounding environment of the target vehicle and a drivable area of the target vehicle; wherein the preset lightweight AI algorithm model is an algorithm model obtained by integrating a parking deep learning model and a driving deep learning model; and the vehicle control module is configured to perform parking and driving actions according to the target markers and the drivable area to realize intelligent auxiliary driving. It can be seen that, in the application, only the camera and the ultrasonic radar are used as sensors to acquire the surrounding environment data of the target vehicle, compared with the traditional use of various complex types of radar sensors, the intelligent driving system architecture can be simplified, thereby solving the problems of complex vehicle architecture development and high cost caused by sensor accumulation, and solving the problem of low data utilization rate, while meeting the functional scene demand of L2+ automatic parking and driving; on the other hand, the preset lightweight AI algorithm model is obtained by integrating the parking deep learning model and the driving deep learning model, thereby improving the processing efficiency of the environment data collected by the sensors in the parking deep learning task and the driving deep learning task. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of the provided drawings.

[0033] Figure 1 It is a structural schematic diagram of an intelligent driving system disclosed by the present application.

[0034] Figure 2 It is a layout structural schematic diagram of a camera and an ultrasonic radar disclosed by the present application.

[0035] Figure 3 It is a structural schematic diagram of a specific intelligent driving disclosed by the present application.

[0036] Figure 4 It is a flow chart of an algorithm architecture design scheme based on vision disclosed by the present application.

[0037] Figure 5A flowchart for processing the input of 4-way language camera, 1-way front camera and 1-way rear camera based on a preset lightweight AI algorithm model is disclosed in the present application.

[0038] Figure 6 A specific intelligent driving structure diagram is disclosed in the present application.

[0039] Figure 7 A structure diagram for communication between different systems and different kernels is disclosed in the present application.

[0040] Figure 8 A flowchart for development of the intermediate layer of a visual abstraction module is disclosed in the present application.

[0041] Figure 9 A specific intelligent system architecture diagram is disclosed in the present application.

[0042] Figure 10 A flowchart for an intelligent driving method is disclosed in the present application.

[0043] Figure 11 An electronic device structure diagram is disclosed in the present application. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0045] Intelligent driving technology has developed rapidly and is widely used, and has gradually become one of the core elements of automobile intelligence. The accompanying problem is that the accumulation of sensors leads to complex vehicle architecture development and high development costs, and also brings economic burden to consumers. In addition, how to improve the perception efficiency of data collected by sensors is also a problem that needs to be continuously explored by those skilled in the art.

[0046] Therefore, an intelligent driving scheme is proposed in the embodiments of the present application, which can simplify the vehicle architecture and reduce the development cost, and improve the perception efficiency of data.

[0047] The embodiments of the present application disclose an intelligent driving system, as shown in Figure 1 The system comprises:

[0048] The data acquisition module 11 comprises a camera and an ultrasonic radar, and is configured to acquire surrounding environment data of a target vehicle through the camera and the ultrasonic radar.

[0049] In this embodiment, the camera 111 includes a first preset number of fisheye cameras, a second preset number of front-view cameras, and a third preset number of rear-view cameras. Specifically, the surrounding environment data of the target vehicle is collected by the first preset number of fisheye cameras, the second preset number of front-view cameras, the third preset number of rear-view cameras, and the fourth preset number of ultrasonic radars 112. In a specific implementation, the target vehicle in this embodiment includes four fisheye cameras, one front-view camera, one rear-view camera, and twelve ultrasonic radars. For example, the four fisheye cameras are arranged on the front side, the rear side, the left side, and the right side of the vehicle body of the target vehicle, one front-view camera is arranged on the front side of the roof of the target vehicle, and one rear-view camera is arranged on the rear side of the roof of the target vehicle. See Figure 2

[0050] The data processing module 12 is configured to calculate the surrounding environment data by using a preset lightweight AI algorithm model to identify target markers in the surrounding environment of the target vehicle and drivable areas of the target vehicle. The preset lightweight AI algorithm model is an algorithm model obtained by integrating a parking deep learning model and a driving deep learning model.

[0051] In this embodiment, the preset lightweight AI algorithm model is pre-deployed in a system on chip (SOC) of the target vehicle, and the preset lightweight AI algorithm model is an algorithm model obtained by integrating a parking deep learning model and a driving deep learning model. It should be noted that the parking deep learning model includes a stop line detection, a target object detection (obstacle detection), a drivable area detection, the driving deep learning model includes a lane line detection, a target object detection, a drivable area identification, and a traffic sign identification. In this embodiment, the parking deep learning task and the driving deep learning task are integrated to be line detection, target object detection, and drivable area detection, and traffic sign identification. This can not only detect obstacles, parking spaces, and drivable areas during low-speed parking, but also can detect obstacles, lane lines, and drivable areas during high-speed driving to realize the functions of NOA (Navigate on Autopilot).

[0052] ​In a specific embodiment of low-speed memory parking, a front-view camera, a fisheye camera and a visual perception scheme are adopted, and multiple information fusion based on visual semantic segmentation, visual vSLAM (visual simultaneous localization and mapping), high-precision odometry, etc. is adopted to realize centimeter-level accurate positioning in the memory valet parking process, ensure the memory parking mapping efficiency and the memory parking engine reference recommendation. Specifically, a preset lightweight AI algorithm model is used to calculate the first surrounding environment data collected by the fisheye camera, the front-view camera and the ultrasonic radar to identify the first target mark of the surrounding environment of the target vehicle and the first drivable area of the target vehicle. It can be understood that the first target mark includes a stop line mark and a target object mark.

[0053] In a specific embodiment of driving, considering that there are problems such as blind area, unstable transverse distance and lane position, and non-recognition of stationary objects in millimeter wave radar lateral perception, the blind area of the fisheye camera basically does not exist, and the perception ability can realize more than 50m perception with the improvement of pixels and computing power. At the same time, cooperating with the wide-angle rear-view camera, the NoA navigation auxiliary driving function based on pure vision is realized, and automatic lane changing is supported. Specifically, a preset lightweight AI algorithm model is used to calculate the second surrounding environment data collected by the fisheye camera, the rear-view camera and the ultrasonic radar to identify the second target mark of the surrounding environment of the target vehicle and the second drivable area of the target vehicle. It can be understood that the second target mark includes a lane line mark, an obstacle mark and a traffic sign mark.

[0054] In addition, in the present embodiment, the backbone (network skeleton) of the preset lightweight AI algorithm model mainly adopts the backbone structure of yolov5s, the neck (neck network) mainly adopts the structure of bi-FPN, the line detection, target object detection and drivable area detection adopt independent neck layers respectively; the head (head network) mainly adopts a traditional convolution structure, and different tasks adopt independent head layers, and the three main tasks include: identifying obstacles, identifying parking spaces, identifying lane lines and identifying drivable areas. That is, the preset lightweight AI algorithm model includes multiple neck networks and head networks, and for different target mark identification tasks and drivable area identification tasks, the preset lightweight AI algorithm model adopts independent neck networks and head networks. In this way, multiple tasks have separate neck layers and head layers, and have the ability of multi-channel information fusion, so that the overall detection accuracy is improved compared with single task, and the effective extraction of image features is also improved.

[0055] Further, the intelligent driving system described in the present application further comprises a data preprocessing module 14, as shown inFigure 3 As shown, for data desensitization preprocessing, data labeling processing and data checking processing of the surrounding environment data of the target vehicle, and for automatic management of the processed surrounding environment data. In this way, the application can train the algorithm on the cloud platform based on effective visual data, so as to realize the optimization of algorithm maturity.

[0056] Figure 4 A flowchart of a visual-based algorithm architecture design scheme disclosed in the present application is shown in Figure 2 As shown, the present application first corrects the distortion of the original image through the data preprocessing module and performs downsampling processing, further reconstructs the column chart (CLV) and the bird's-eye view (BEV) of the preprocessed image, then marks the target and detects the free space based on the column chart and the bird's-eye view, and fuses the lane line and object tracking results based on the target marking and the visual signal processing results of the free space detection, so as to assist driving during driving. On the other hand, the parking line based on the target marking and the visual signal processing results of the free space detection are fused to select the parking space during parking.

[0057] Figure 5 A flowchart for processing the input of 4-way language camera, 1-way front-view camera and 1-way rear-view camera based on a preset lightweight AI algorithm model is disclosed in the present application. Exemplarily, 4xYUV-TV Image represents the input image data of 4-way language camera, YUV422, HDR-LFM, 1920x1536 represents the data format of the image data. In the present application, the image correction rendering module, image encoding processor, image processor, optical flow algorithm module, visual processing model and convolutional neural network algorithm module are used to process the input images of 4-way language camera, 1-way front-view camera and 1-way rear-view camera and the intermediate generated images. It should be noted that 4-way fisheye camera can effectively replace panoramic camera to complete front / rear and lateral target recognition, lane line recognition and drivable area recognition. The front-view camera mainly completes the recognition of forward long-distance targets and the recognition of traffic signs / signals. The rear-view camera mainly completes the recognition of forward long-distance targets. In this way, the multi-camera target is pre-fused, and the target recognition efficiency is improved. At the same time, the preset lightweight AI algorithm model is a unified model for target detection, lane line detection, parking space detection and drivable area detection, and combines the advantages of CNN (convolutional neural network) for bottom feature extraction and the advantages of Transformer (a kind of neural network) for global feature extraction. In this way, the perception accuracy and efficiency are improved.

[0058] The vehicle control module 13 performs parking and driving actions according to the target mark and the drivable area to realize intelligent auxiliary driving.

[0059] In this embodiment, after the target mark and the drivable area are determined, parking and driving actions are performed to realize intelligent auxiliary driving. Specifically, after the first target mark of the target vehicle surrounding environment and the first drivable area of the target vehicle are determined, parking actions are performed according to the first target mark and the first drivable area to realize intelligent auxiliary parking; after the second target mark of the target vehicle surrounding environment and the second drivable area of the target vehicle are determined, driving actions are performed according to the second target mark and the second drivable area to realize intelligent auxiliary driving.

[0060] It can be seen that the present application proposes an intelligent driving system, which comprises a data acquisition module, a data processing module, and a vehicle control module. The data acquisition module comprises a camera and an ultrasonic radar and is used to acquire surrounding environment data of a target vehicle through the camera and the ultrasonic radar. The data processing module is used to calculate the surrounding environment data through a preset lightweight AI algorithm model to identify target marks of the target vehicle surrounding environment and drivable areas of the target vehicle. The preset lightweight AI algorithm model is an algorithm model obtained by integrating a parking deep learning model and a driving deep learning model. The vehicle control module is used to perform parking and driving actions according to the target marks and the drivable areas to realize intelligent auxiliary driving. It can be seen that, in the present application, only the camera and the ultrasonic radar are used as sensors to acquire the surrounding environment data of the target vehicle. Compared with the traditional use of various complex radar sensors, the intelligent driving system architecture can be simplified. In this way, the problems of complex vehicle architecture development and high cost caused by sensor accumulation are solved, and the problem of low data utilization rate is solved. At the same time, the functional scene demand of L2+ automatic parking and driving can be met. On the other hand, the preset lightweight AI algorithm model is obtained by integrating the parking deep learning model and the driving deep learning model. In this way, the processing efficiency of the environment data collected by the sensors is improved in the parking deep learning task and the driving deep learning task.

[0061] This embodiment further illustrates and optimizes the intelligent driving system. As shown in Figure 6 The intelligent driving system specifically further comprises the following parts:

[0062] The visual abstraction module 15 is used to realize decoupling between the application layer and the bottom layer of the surrounding environment data based on a first preset middleware.

[0063] The runtime scheduling environment module 16 is used to achieve efficient communication and zero-copy efficient data sharing between homogeneous cores, heterogeneous cores, and different operating systems based on the second preset middleware; and is used to standardize and encapsulate the data types of the surrounding environment data.

[0064] It is understood that the visual abstraction module 15 is used to decouple the application layer from the underlying image data based on the first preset middleware. This ensures that the preset lightweight algorithm model is not strongly correlated with hardware. In other words, this application can be deployed seamlessly on any platform and operating system, from the lightest single-core MCU to complex Linux multi-core processors, and the software components can be seamlessly ported. On the other hand, since the second preset middleware in this application can standardize and encapsulate data types, it can guarantee the consistency of interface data types across any platform. Furthermore, the second preset middleware also has the following advantages: it enables this application to have the lowest CPU core memory usage, ensuring the lowest system latency, even when transmitting large data such as images; it guarantees absolute process security, with a data verification mechanism that reaches the ASIL-D functional safety level; and its fully self-developed compilation framework modularizes the software development process, ensuring an efficient software development process.

[0065] Figure 7 This diagram illustrates the architecture for communication between different systems and kernels, including process C (heterogeneous system), process B (homogeneous system), kernel A, kernel B, thread A, thread B, thread C, and instance A to instance F. It should be noted that... Figure 7 The lines in the diagram indicate that different cores need to communicate with each other.

[0066] Figure 8 For the development process of the intermediate layer of a visual abstraction module disclosed in this application, see [link to relevant documentation]. Figure 8 As shown, this specifically includes: development of the image abstraction layer, hardware-accelerated application development, SoC image application programming interface development, kernel algorithm development, DMA-enabled data exchange, and register-level algorithm development.

[0067] Figure 9For a specific intelligent system architecture disclosed in the present application, specifically including: User Functions, Services, Algorithms, Data Driven: data closed loop (i.e. data preprocessing module); Runtime Environment; Vision Abstraction Layer (ViAL); Operating Sys; Boot Loader; Board Support Packages; Hardware.

[0068] The embodiment of the present application discloses an intelligent driving method, see Figure 10 As shown in the figure, including:

[0069] Step S11: collecting surrounding environment data of the target vehicle through a camera and an ultrasonic radar.

[0070] Step S12: calculating the surrounding environment data through a preset lightweight AI algorithm model to identify target markers of the surrounding environment of the target vehicle and a drivable area of the target vehicle; wherein the preset lightweight AI algorithm model is an algorithm model obtained by integrating a deep learning model for parking and a deep learning model for driving.

[0071] Step S13: performing parking and driving actions according to the target markers and the drivable area to realize intelligent auxiliary driving.

[0072] It can be seen that the application provides an intelligent driving method, which comprises the following steps: collecting surrounding environment data of a target vehicle through a camera and an ultrasonic radar; calculating the surrounding environment data through a preset lightweight AI algorithm model to identify target marks of the surrounding environment of the target vehicle and a drivable area of the target vehicle; wherein the preset lightweight AI algorithm model is an algorithm model obtained by integrating a deep learning model for parking and a deep learning model for driving; and performing parking and driving actions according to the target marks and the drivable area to realize intelligent auxiliary driving. It can be seen that, in the application, only the camera and the ultrasonic radar are used as sensors to collect the surrounding environment data of the target vehicle, compared with the traditional use of various complex types of radar sensors, the intelligent driving system architecture can be simplified, so that the problems of complex vehicle architecture development and high cost caused by sensor accumulation are solved, the problem of low data utilization rate is solved, and the functional scene demand of L2+ for automatic parking and driving is met; on the other hand, the preset lightweight AI algorithm model is obtained by integrating the deep learning model for parking and the deep learning model for driving, so that the processing efficiency of the environment data collected by the sensor is improved in the deep learning task for parking and the deep learning task for driving.

[0073] Further, the application embodiment further provides an electronic device. Figure 11 The electronic device 20 structure diagram shown in the figure is not considered as any limitation on the use range of the application.

[0074] Figure 11 The electronic device 20 structure diagram provided by the application embodiment. The electronic device 20 can specifically comprise at least one processor 21, at least one memory 22, a display screen 23, an input / output interface 24, a communication interface 25, a power supply 26 and a communication bus 27. The memory 22 is used to store a computer program, the computer program is loaded and executed by the processor 21 to realize the related steps in the intelligent driving method disclosed in any of the preceding embodiments. In addition, the electronic device 20 in the embodiment can be an electronic computer.

[0075] In the embodiment, the power supply 26 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 25 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol followed by the communication interface 25 can be any communication protocol applicable to the technical solution of the application, which is not limited here; the input / output interface 24 is used to obtain external input data or output data to the outside world, and the specific interface type can be selected according to the specific application needs, which is not limited here.

[0076] Furthermore, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk, or optical disk, etc. The resources stored thereon may include computer programs 221, and the storage method may be temporary storage or permanent storage. The computer programs 221 may include, in addition to computer programs capable of performing the intelligent driving method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, computer programs capable of performing other specific tasks.

[0077] Furthermore, embodiments of this application also disclose a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned intelligent driving method.

[0078] For the specific steps of this method, please refer to the relevant content disclosed in the foregoing embodiments, which will not be repeated here.

[0079] The various embodiments in this application are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. For the same or similar parts between the various embodiments, refer to each other. As for the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and relevant parts can be referred to in the method section.

[0080] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0081] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0082] Finally, it needs to be pointed out that in this article, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "includes a" does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0083] The above provides a detailed description of the intelligent driving method, device, equipment and storage medium provided by the present application. The principles and implementation modes of the present application are described by applying specific examples. The above example is only used to help understand the method and its core idea of the present application; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed; in view of the above, the content of the specification should not be understood as a limitation of the present application.

Claims

1. An intelligent driving system, characterized in that, include: Data acquisition module, data processing module, vehicle control module; The data acquisition module includes a camera and an ultrasonic radar, and is used to acquire environmental data of the target vehicle's surroundings through the camera and the ultrasonic radar. The data processing module is used to calculate the surrounding environment data using a preset lightweight AI algorithm model to identify target markers in the environment surrounding the target vehicle and the drivable area of ​​the target vehicle; wherein, the preset lightweight AI algorithm model is an algorithm model obtained by integrating a deep learning model for parking and a deep learning model for driving. The vehicle control module is used to perform parking and driving actions based on the target marker and the drivable area to achieve intelligent assisted driving; The deep learning model for parking includes parking line detection, object detection, and drivable area detection, while the deep learning model for driving includes lane line detection, object detection, drivable area recognition, and traffic sign recognition. Integrating the deep learning tasks for parking and driving can unify them into line detection, object detection, drivable area detection, and traffic sign recognition. It can detect obstacles, parking spaces, and drivable areas during low-speed parking, and can also achieve NOA (Noise, Arrival, and Assurance) function for assisted driving during high-speed driving by detecting obstacles, lane lines, and drivable areas.

2. The intelligent driving system according to claim 1, characterized in that, The camera includes: Fisheye cameras deployed on the front, rear, left, and right sides of the target vehicle body, a front-view camera deployed on the front of the roof of the target vehicle, and a rear-view camera deployed on the rear of the roof of the target vehicle.

3. The intelligent driving system according to claim 2, characterized in that, The data processing module is used for: The first surrounding environment data collected by the fisheye camera, the forward-looking camera and the ultrasonic radar are calculated by a preset lightweight AI algorithm model to identify the first target marker in the surrounding environment of the target vehicle and the first drivable area of ​​the target vehicle. Accordingly, the vehicle control module is used to perform parking actions based on the first target marker and the first drivable area to achieve intelligent assisted parking.

4. The intelligent driving system according to claim 2, characterized in that, The data processing module is used for: The second surrounding environment data collected by the fisheye camera, the rearview camera and the ultrasonic radar are calculated by a preset lightweight AI algorithm model to identify the second target marker in the surrounding environment of the target vehicle and the second drivable area of ​​the target vehicle. Accordingly, the vehicle control module is used to perform driving actions based on the second target marker and the second drivable area to achieve intelligent assisted driving.

5. The intelligent driving system according to claim 3 or 4, characterized in that, The preset lightweight AI algorithm model includes multiple neck networks and head networks, and for different target marker recognition tasks and drivable area recognition tasks, the preset lightweight AI algorithm model adopts independent neck networks and head networks respectively.

6. The intelligent driving system according to any one of claims 1 to 4, characterized in that, The intelligent driving system also includes: The data preprocessing module is used to perform data anonymization preprocessing, data labeling, and data inspection on the surrounding environmental data of the target vehicle, and to automatically manage the processed surrounding environmental data.

7. The intelligent driving system according to claim 6, characterized in that, The intelligent driving system also includes: The visual abstraction module is used to decouple the application layer from the underlying surrounding environment data based on the first preset middleware; The runtime scheduling environment module is used to achieve efficient communication and zero-copy data sharing between homogeneous cores, heterogeneous cores, and different operating systems based on the second preset middleware; and is used to standardize and encapsulate data types.

8. An intelligent driving method, characterized in that, include: Data on the surrounding environment of the target vehicle is collected using cameras and ultrasonic radar. The surrounding environment data is calculated using a preset lightweight AI algorithm model to identify target markers in the environment surrounding the target vehicle and the drivable area of ​​the target vehicle; wherein, the preset lightweight AI algorithm model is an algorithm model obtained by integrating a deep learning model for parking and a deep learning model for driving. Parking and driving actions are performed based on the target markers and the drivable area to achieve intelligent assisted driving; The deep learning model for parking includes parking line detection, object detection, and drivable area detection, while the deep learning model for driving includes lane line detection, object detection, drivable area recognition, and traffic sign recognition. Integrating the deep learning tasks for parking and driving can unify them into line detection, object detection, drivable area detection, and traffic sign recognition. It can detect obstacles, parking spaces, and drivable areas during low-speed parking, and can also achieve NOA (Noise, Arrival, and Assurance) function for assisted driving during high-speed driving by detecting obstacles, lane lines, and drivable areas.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the intelligent driving method as described in claim 8.

10. A computer-readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the intelligent driving method as described in claim 8.

Images