Method and device for detecting data
By dividing and characteristic extraction of point cloud data collected by Lidar sensors, combined with the abnormal identification of machine learning models, it solves the problem of difficulty identifying forgery attacks in existing technologies, and improving the security of transportation.
Patent Information
- Application Number
- CN202280096215.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2022-05-18
- Publication Date
- 2025-01-28
AI Technical Summary
Existing lidar sensors have vulnerabilities in detecting data, making it difficult to effectively identify forged attacks, resulting in vehicles that may make incorrect security decisions.
By obtaining frame data, divide into multiple blocks, and extract the feature values of each block. Use the machine learning model to determine whether each block is abnormal based on the changes in the feature value, and then identify forgery attacks.
Improve the security of transportation, effectively detect the forgery attacks in point cloud data, and reduce the occurrence of error decisions.
Smart Images
Figure CN119384611A_ABST
Abstract
Description
Technical Field
[0001] Embodiments of the present application relate to the field of information technology, and more specifically, to a method and device for detecting data. Background Art
[0002] In modern vehicles, LiDAR (light detection and ranging) sensors play a vital role in autonomous driving technology, measuring various information about other vehicles and objects. LiDAR is a technology that uses infrared (IR) pulse signals to measure the distance, speed, direction, and other characteristics of a target between the LiDAR sensor and the object. In addition, when a vehicle equipped with a LiDAR sensor is driving, the LiDAR sensor identifies surrounding vehicles to ensure the safety of the vehicle. However, LiDAR sensors also have vulnerabilities that can be exploited physically due to these technical characteristics.
[0003] Various anomaly detection methods for LiDAR sensors have been proposed. For example, one anomaly detection method is to compare the information of the previous frame data with the information of the current frame data. Due to the large difference in information between two consecutive frame data, the anomaly detection method has limitations. Summary of the invention
[0004] Embodiments of the present application provide a method and device for detecting data, which can improve the safety of transportation vehicles.
[0005] In one possible design, the vehicle may include one or more different types of vehicles or movable objects that operate or move on land (e.g., highways, roads, railways), water (e.g., waterways, rivers, oceans), or space. For example, the vehicle may include a vehicle, a bicycle, a motorcycle, a train, a subway, an airplane, a ship, an aircraft, a robot, or other types of vehicles or movable objects.
[0006] According to a first aspect, an embodiment of the present application provides a method for detecting data, the method comprising: acquiring frame data, wherein the frame data comprises a snapshot of point cloud data within a preset time length; dividing the frame data into a plurality of blocks; extracting feature values of each of the plurality of blocks, wherein the feature values represent statistical characteristics of points in each block; and determining whether each block is abnormal or normal based on changes in the feature values of the plurality of blocks.
[0007] The above technical solution provides a method for detecting abnormal data generated by a forgery attack on point cloud data collected by a lidar sensor. The method can detect anomalies, such as inserting fake objects in point cloud data or deleting real objects in point cloud data. By detecting abnormal data, the method can help vehicles make safety decisions.
[0008] The embodiment of the present application analyzes the point cloud into blocks and provides a reasonable area where forgery occurs.
[0009] In one possible design, the frame data includes N point clouds, where N is greater than or equal to 1 and is a positive integer.
[0010] In one possible design, determining whether each block is abnormal or normal based on the change in the characteristic values of the multiple blocks includes: inputting the change in the characteristic value into a machine learning model to obtain a prediction result indicating whether each block is abnormal or normal, wherein the machine learning model is trained by training samples, and the training samples include the difference in characteristic values between abnormal blocks and normal blocks.
[0011] The above technical solution provides a machine learning model that learns the statistical value block time series difference between normal state and abnormal state. The machine learning model also determines whether each block is abnormal or normal, which helps to improve the intelligence of the vehicle and also helps to improve the accuracy of abnormal block determination.
[0012] In a possible design, the characteristic value includes one or more of the following: an average value of coordinate values, a standard deviation of coordinate values, an average value of intensity, and a standard deviation of intensity.
[0013] In one possible design, the method also includes: prompting a user that the point cloud data is abnormal when a first block is abnormal, wherein the multiple blocks include the first block; and / or executing safety instructions when a second block is abnormal, wherein the multiple blocks include the second block.
[0014] When the multiple blocks include an abnormal block, a prompt may be given to the user and / or a safety instruction may be executed, so that the user can understand the abnormal data in a timely manner, which helps to improve the driving safety of the user.
[0015] In one possible design, the method is executed by a device located in a cloud server.
[0016] Since the computing power of cloud servers is stronger than that of vehicles, the abnormal data detection rate can be improved.
[0017] In one possible design, acquiring frame data includes: receiving frame data of a vehicle; the method further includes: sending an indication to the vehicle, wherein the indication is used to indicate whether each block is abnormal.
[0018] According to the second aspect, an embodiment of the present application provides a device for detecting data, comprising: an acquisition module, used to acquire frame data, wherein the frame data includes a snapshot of point cloud data within a preset time length; a division module, used to divide the frame data into multiple blocks; an extraction module, used to extract feature values of each of the multiple blocks, wherein the feature values represent statistical characteristics of points in each block; and a determination module, used to determine whether each block is abnormal or normal based on changes in the feature values of the multiple blocks.
[0019] In one possible design, the determination module is specifically used to input the change of the characteristic value into a machine learning model to obtain a prediction result indicating whether each block is abnormal, wherein the machine learning model is trained by training samples, and the training samples include the characteristic value differences of abnormal blocks and the characteristic value changes of normal blocks.
[0020] In a possible design, the characteristic value includes one or more of the following: an average value of coordinate values, a standard deviation of coordinate values, an average value of intensity, and a standard deviation of intensity.
[0021] In one possible design, the device also includes: a prompt module, used to prompt a user that the point cloud data is abnormal when a first block is abnormal, wherein the multiple blocks include the first block; and / or an execution module, used to execute safety instructions when a second block is abnormal, wherein the multiple blocks include the second block.
[0022] In one possible design, the device is located in a cloud server.
[0023] In one possible design, the acquisition module is specifically used to receive frame data of a vehicle; the device also includes: a sending module, used to send an indication to the vehicle, wherein the indication is used to indicate whether each block is abnormal.
[0024] According to a third aspect, an embodiment of the present application provides a vehicle having the function of implementing the method of the first aspect. The function can be implemented by hardware or by hardware executing corresponding software. The hardware executing the corresponding software includes one or more modules corresponding to the function.
[0025] In one possible design, the transportation vehicle is a vehicle.
[0026] According to a fourth aspect, a device is provided, comprising a processor and a memory. The processor is connected to the memory. The memory is used to store instructions, and the processor is used to execute the instructions. When the processor executes the instructions stored in the memory, the processor executes the method in the first aspect or any possible implementation of the first aspect.
[0027] According to the fifth aspect, a chip system is provided, wherein the chip system includes a memory and a processor, the memory is used to store a computer program, and the processor is used to call the computer program from the memory and run the computer program, so that the server where the chip system is located executes the method in the first aspect or any possible implementation of the first aspect.
[0028] According to a sixth aspect, a computer program product is provided. When the computer program product is run on a device, the device is enabled to execute the method in the first aspect or any possible implementation manner of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] Figure 1 is a schematic block diagram of a vehicle.
[0030] Figure 2 It is a schematic block diagram of the vehicle network topology.
[0031] Figure 3 It is a structural diagram of a miniature small outline package (MSOP).
[0032] Figure 4 is a schematic block diagram of an in-vehicle network topology including an attacker.
[0033] Figure 5 is a schematic block diagram of an in-vehicle network topology including an attacker.
[0034] Figure 6 It is a schematic block diagram of the system.
[0035] Figure 7 A point cloud is shown.
[0036] Figure 8 It shows that the point cloud is divided into 5 blocks.
[0037] Fig. 9 A graphical user interface (GUI) is shown.
[0038] Fig.10 It is a schematic block diagram of the system.
[0039] Fig.11 is a flow chart of a method for detecting data.
[0040] Fig.12 is a schematic block diagram of a device for detecting data.
[0041] Fig.13 is a schematic block diagram of a device for detecting data. DETAILED DESCRIPTION
[0042] The technical solution in this application is described below in conjunction with the accompanying drawings.
[0043] Figure 1 is a schematic block diagram of a vehicle 100 .
[0044] like Figure 1 As shown, the vehicle 100 may include a sensing system 110, a display device 120, and a computing platform 130, wherein the sensing system 110 may include a variety of sensors for sensing environmental information around the vehicle 100. For example, the sensing system 110 may include a positioning system, which may be a global positioning system (GPS), a Beidou system, or other positioning systems. The sensing system 110 also includes an inertial measurement unit (IMU), a laser radar sensor, a millimeter wave radar, an ultrasonic radar, or an imaging device.
[0045] Part or all of the functions of the vehicle 100 may be controlled by a computing platform 130. The computing platform 130 may include processors 131 to 13n (n is a positive integer). A processor is a circuit with signal processing capabilities.
[0046] In some embodiments of the present application, the processor may be a central processing unit (CPU), a microprocessor, a graphics processing unit (GPU), a digital signal processor (DSP), or other circuits having the capability of reading and executing instructions.
[0047] In some embodiments of the present application, the processor can implement certain functions through the logical relationship of a fixed or reconfigurable hardware circuit. For example, the processor is a hardware circuit implemented by a programmable logic device (PLD) such as an application-specific integrated circuit (ASIC) or a field programmable gate array (FPGA). In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the configuration of the hardware circuit can be understood as the process of the processor loading instructions to implement the functions of some or all of the above-mentioned units. In addition, the processor can also be a hardware circuit designed for artificial intelligence (AI), which can be understood as an ASIC, such as a neural network processing unit (NPU), a tensor processing unit (TPU), a deep learning processing unit (DPU), etc. The computing platform 130 can also include a memory for storing instructions, and some or all of the processors 131 to 13n can call instructions in the memory and execute these instructions to implement corresponding functions.
[0048] LiDAR sensors act as the eyes of driverless vehicles, providing them with a 360-degree field of view. Autonomous vehicles can use LiDAR sensors for obstacle detection and avoidance to safely navigate in an environment.
[0049] Figure 2 is a schematic block diagram of an in-vehicle network topology 200 including a lidar sensor.
[0050] like Figure 2 As shown, the vehicle network topology 200 includes four nodes: a lidar sensor 210 , a vehicle network switch 220 , a pre-processing module 230 and a perception module 240 .
[0051] In some embodiments of the present application, the functions of the preprocessing module 230 can be implemented by a CPU and an AI chip. The CPU can segment the collected point cloud data, extract features, and obtain the time series change features of multiple blocks. Finally, the time series change features are input into the AI chip, and the AI chip outputs the result of whether each block is abnormal.
[0052] The lidar sensor 210 generates raw sensor data and sends the raw sensor data to the preprocessing module 230 through the vehicle network switch 220. The raw sensor data consists of R and α values, which represent the distance between a point and the lidar sensor and the azimuth of the point, respectively. The vertical angle (ω) and the horizontal angle offset (δ) are predefined constant values set by the user.
[0053] Figure 3 This is a diagram of the structure of a miniature small outline package (MSOP), which is a type of laser radar data package. Figure 3 As shown, an attacker can easily change the original information by clicking on the address value.
[0054] Figure 4 is a schematic block diagram of an in-vehicle network topology including an attacker.
[0055] like Figure 4 As shown, if the attacker is located between the vehicle network switch 220 and the lidar sensor 210, the attacker can easily modify the point. It can be assumed that the attacker is performing a man in the middle attack (MITM) through address resolution protocol (ARP) spoofing or network eavesdropping, and can physically access the network cable through the attacker's device.
[0056] If the attacker exists Figure 5 If the position shown in Figure 1 is the same as the position shown in Figure 2, object insertion can be performed by modifying the original sensor data with the value of the object point owned by the attacker. Figure 5 As shown, an object (vehicle) that does not exist in actual situation will appear as an object (vehicle) on the perception module 240.
[0057] The embodiment of the present application provides a method and related equipment for detecting data. The technical solution can identify point cloud modification attacks to prevent vehicles from making wrong decisions.
[0058] Figure 6 6 is a schematic block diagram of a system 600. The system 600 includes a partitioning module 610, a feature extraction module 620, a time series management module 630, a machine learning model 640, and an attack detection module 650.
[0059] In some embodiments of the present application, the system 600 is included in the above-mentioned pre-processing module 230.
[0060] System 600 detects abnormal changes in the statistical values of points in a series of point clouds. In order to calculate the statistical values of the points, the system divides the point cloud into blocks of equal size. When an attack occurs, the statistical values of the points in the attacked blocks will change abnormally over time.
[0061] The segmentation module 610 receives point cloud data from a laser radar sensor. The laser radar sensor scans the direction of objects around the car and the distance to the objects around the car to obtain laser radar data, and the laser radar data is formatted to obtain point cloud data. In addition, the laser radar data includes the laser pulse reflection intensity of the object.
[0062] Figure 7 The point cloud is shown. The segmentation module 610 divides the point cloud into blocks. There are countless points in a point cloud (usually 10,000 to 100,000 points). Therefore, it is more efficient to process the point cloud after it is divided into blocks.
[0063] Figure 8 It shows that the point cloud is divided into 5 blocks.
[0064] Figure 8 For illustration purposes only, the point cloud is shown divided into 5 blocks. The size of each block depends on the developer. A reasonable size for each block is 2m×2m×1m, where m represents meters. The raw data of the lidar, which consists of thousands of laser points from the lidar, is called point cloud data. A frame is nothing more than a snapshot of the point cloud data within a preset time length. As the vehicle moves, the point cloud data changes, and so does the data in the frame. Each frame of data can be divided into a total of 3600 blocks.
[0065] The feature extraction module 620 extracts features of points in each block. The features of points in each block are values that can represent statistical characteristics of the points.
[0066] In some embodiments of the present application, the characteristic value of each block includes an average value of coordinate values or a standard deviation of coordinate values. The average value of coordinate values or the standard deviation of coordinate values shows the distribution of points in each block.
[0067] In some embodiments of the present application, the characteristic value of each block includes an average value of intensity or a standard deviation of intensity. The average value of intensity or the standard deviation of intensity shows the number of various object surfaces in each block.
[0068] The time series management module 630 generates a time series of block features within the length of a certain time window (e.g., 1 second). The time series contains the feature values of each block in the most recent N point clouds. The entire raw data of the lidar is called point cloud data, which consists of thousands of laser points. Each time the lidar scans its surroundings, the point cloud data is continuously updated (e.g., once every T milliseconds). A snapshot of the point cloud data within a preset time length is called frame data. Each frame data can be divided into multiple blocks of the same size (3D). Then, the data in each block is monitored over a period of time. If any anomaly is detected within the block, it is considered abnormal data. Continuous feature values can represent changes in the distribution and characteristics of points. If there is an attack, the sudden appearance (or disappearance) of an object will significantly change the feature values of the points in the relevant blocks. Therefore, the time series-based method helps to detect anomalies in point clouds.
[0069] The machine learning model 640 analyzes the time series of features of the blocks and automatically classifies each block into a normal class or an attack class. The machine learning model 640 learns the normal state and the abnormal state from a pre-collected data set (the data set may contain normal and attacked sensor data, or may contain only normal sensor data). After training, the machine learning model 640 can determine whether each block in the point cloud is attacked.
[0070] In some embodiments of the present application, the machine learning model 640 is included in an advanced driving assistant system (ADAS).
[0071] The attack detection module 650 receives the inference results from the machine learning model 640. If there is an attacked block, the module will issue an alarm to the autonomous driving system, the driver, or the system manager.
[0072] Fig. 9 A graphical user interface (GUI) is shown.
[0073] like Fig. 9 As shown, when the vehicle is in an autonomous driving state, if the system 600 determines that the point cloud is abnormal, a prompt message "The data collected by the lidar is abnormal, please take over the vehicle" can be displayed on the display screen.
[0074] In some embodiments of the present application, the user may be prompted by changing the color of the ambient light.
[0075] In some embodiments of the present application, the system 600 may be located in a server.
[0076] Fig.10 is a schematic block diagram of system 1000 .
[0077] like Fig.10 As shown, the system 1000 includes a vehicle 1010 and a server 1020. The vehicle 1010 includes a laser radar sensor, and the server 1020 includes the above-mentioned system 600. The vehicle 1010 can send the point cloud data collected by the laser radar sensor to the server 1020, and the system 600 in the server 1020 can determine whether each block is abnormal or normal by analyzing the point cloud data. The server 1020 can send the result of whether each block is abnormal to the vehicle 1010. Then, the vehicle 1010 can perform corresponding operations according to the result.
[0078] The proposed system 600 can identify point cloud modification attacks, especially removing or adding objects in the driving direction of the autonomous vehicle, so as to prevent the perception module from making wrong decisions.
[0079] Furthermore, the system 600 can detect anomalies caused by cyber attacks, and the system 600 does not require data from other sources.
[0080] From a practical point of view, the system 600 works on tiled areas rather than analyzing the entire point cloud. This is faster than doing a point-by-point analysis in the scene.
[0081] Fig.11 1 is a flow chart of a method 1100 for detecting data. The method may be performed by a device. For example, the device is the computing platform, vehicle, or server described above.
[0082] 1101: The device obtains frame data. The frame data includes a snapshot of the point cloud data within a preset time period.
[0083] Optionally, the frame data includes N point clouds, where N is greater than or equal to 1 and is a positive integer.
[0084] For example, the preset duration is 1 second.
[0085] Optionally, the device obtains frame data, including: the device obtains frame data from a lidar sensor.
[0086] 1102: The device divides the frame data into multiple blocks.
[0087] Optionally, the device divides the frame data into a plurality of blocks, including: the device divides the frame data into a plurality of blocks of preset sizes.
[0088] For example, the preset size is 2m×2m×1m.
[0089] 1103: The device extracts a feature value of each block in the plurality of blocks. The feature value represents a statistical characteristic of points in each block.
[0090] Optionally, the characteristic value includes one or more of the following: an average value of coordinate values, a standard deviation of coordinate values, an average value of intensity, and a standard deviation of intensity.
[0091] For example, the mean or standard deviation of the coordinate values shows the distribution of the points in the block. The mean or standard deviation of the intensity shows the number of various object surfaces in the block.
[0092] 1104: The device determines whether each block is abnormal or normal according to changes in characteristic values of the multiple blocks.
[0093] Optionally, determining whether each block is abnormal or normal based on changes in characteristic values of multiple blocks includes: the device inputs the changes in characteristic values into a machine learning model to obtain a prediction result indicating whether each block is abnormal or normal, wherein the machine learning model is trained by training samples, and the training samples include characteristic value differences between abnormal blocks and normal blocks.
[0094] Optionally, if the changes in the characteristic values of the plurality of blocks match the changes in the characteristic values of the normal blocks, the plurality of blocks are normal blocks. If the changes in the characteristic values of the plurality of blocks match the changes in the characteristic values between the abnormal blocks and the normal blocks, the plurality of blocks include the abnormal blocks.
[0095] Fig.12 FIG. 1 is a schematic block diagram of a device 1200 for detecting data provided by an embodiment of the present application. Fig.12 As shown, the device 1200 includes: an acquisition module 1210 , a division module 1220 , an extraction module 1230 and a determination module 1240 .
[0096] The acquisition module 1210 is used to acquire frame data, wherein the frame data includes a snapshot of the point cloud data within a preset time length.
[0097] The division module 1220 is used to divide the frame data into multiple blocks.
[0098] The extraction module 1230 is used to extract a feature value of each block in the plurality of blocks, wherein the feature value represents a statistical characteristic of points in each block.
[0099] The determination module 1240 is used to determine whether each block is abnormal or normal according to changes in the characteristic values of the multiple blocks.
[0100] Optionally, the determination module 1240 is specifically used to input the change of the characteristic value into the machine learning model to obtain a prediction result indicating whether each block is abnormal, wherein the machine learning model is trained by training samples, and the training samples include the characteristic value differences between abnormal blocks and normal blocks.
[0101] Optionally, the characteristic value includes one or more of the following: an average value of coordinate values, a standard deviation of coordinate values, an average value of intensity, and a standard deviation of intensity.
[0102] Optionally, the device 1200 also includes: a prompt module, used to prompt a user of point cloud data abnormality when a first block is abnormal, wherein the multiple blocks include the first block; and / or an execution module, used to execute safety instructions when a second block is abnormal, wherein the multiple blocks include the second block.
[0103] Optionally, device 1200 is located in a cloud server.
[0104] Optionally, the acquisition module 1210 is specifically used to receive frame data of the vehicle.
[0105] The device 1200 further includes: a sending module, configured to send an indication to the vehicle, wherein the indication is used to indicate whether each block is abnormal.
[0106] like Fig.13 As shown, the device 1300 may include a transceiver 1301, a processor 1302, and a memory 1303. The memory 1303 may be used to store codes, instructions, etc. executed by the processor 1302.
[0107] In some embodiments, device 1300 may be a computing platform and may execute Fig.10 The above method is performed by the vehicle shown.
[0108] In some embodiments, device 1300 may be a vehicle and may perform the Fig.10 The above method is performed by the vehicle shown.
[0109] In some embodiments, device 1300 may be a server and may execute Fig.10 The above method is performed by the vehicle shown.
[0110] It should be understood that the processor 1302 can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by a hardware integrated logic circuit in the processor or by instructions in the form of software. The processor can be a general processor, a digital signal processor (digital signal processor, DSP), an application-specific integrated circuit (application-specific integrated circuit, ASIC), a field programmable gate array (field programmable gate array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, or discrete hardware components. The processor can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general processor can be a microprocessor, or the processor can be any traditional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly executed and completed by a hardware decoding processor, or executed and completed using a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in conjunction with the hardware in the processor.
[0111] It can be understood that the memory 1303 in the embodiment of the present application can be a volatile memory or a non-volatile memory, and can also include both a volatile memory and a non-volatile memory. The non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM may be used, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus random access memory (DR RAM).
[0112] It should be noted that the memories in the systems and methods described in this specification include but are not limited to these memories and any other suitable types of memories.
[0113] An embodiment of the present application further provides a system chip, wherein the system chip includes an input / output interface, at least one processor, at least one memory and a bus. The at least one memory is used to store instructions, and the at least one processor is used to call instructions in the at least one memory to perform the operations in the method in the above embodiment.
[0114] An embodiment of the present application further provides a computer storage medium, wherein the computer storage medium can store program instructions to execute any of the above methods.
[0115] Optionally, the storage medium may specifically be the memory 1303 .
[0116] Those skilled in the art will appreciate that, in conjunction with the examples described in the embodiments disclosed in this specification, each unit and algorithm step can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are performed by 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 of each specific application, but it should not be considered that the implementation is beyond the scope of this application.
[0117] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-mentioned systems, devices and units can refer to the corresponding processes in the above-mentioned method embodiments, which will not be described in detail here.
[0118] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the described device embodiments are only examples. For example, unit division is only a logical function division, and other division methods can be used in actual implementation. For example, multiple units or components can be merged or integrated into another system, or some features can be ignored or not performed. In addition, the mutual coupling or direct coupling or communication connection shown or described can be implemented through some interfaces. The indirect coupling or communication connection between devices or units can be implemented electronically, mechanically or in other ways.
[0119] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, may be located in one location, or may be distributed over multiple network units. Some or all of the units may be selected based on actual needs to achieve the purpose of the solution of the embodiment.
[0120] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0121] When these functions are implemented in the form of software functional units and sold or used as independent products, these functions can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution in this application can be implemented in the form of a software product in essence or in the part that contributes to the prior art or in the part of the technical solution. The computer software product is stored in a storage medium, including several instructions to instruct a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The above-mentioned storage medium includes any medium that can store program code, such as a USB disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.
[0122] The above description is only a specific implementation of the present application and is not intended to limit the scope of protection of the present application. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should belong to the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.
Claims
1. A method for detecting data, characterized in that: include: Acquire frame data, wherein the frame data includes a snapshot of point cloud data within a preset time length; Dividing the frame data into a plurality of blocks; Extracting a feature value of each block in the plurality of blocks, wherein the feature value represents a statistical characteristic of points in each block; Whether each block is abnormal or normal is determined according to changes in the characteristic values of the plurality of blocks.
2. The method according to claim 1, characterized in that Determining whether each block is abnormal or normal according to changes in the characteristic values of the plurality of blocks comprises: The change in the feature value is input into a machine learning model to obtain a prediction result indicating whether each block is abnormal or normal, wherein the machine learning model is trained by training samples, and the training samples include the feature value differences between abnormal blocks and normal blocks.
3. The method according to claim 1 or 2, characterized in that: The characteristic values include one or more of the following: The average value of the coordinates; Standard deviation of coordinate values; The average value of intensity; The standard deviation of the intensity.
4. The method according to any one of claims 1 to 3, characterized in that Also includes: Prompting a user that the point cloud data is abnormal if a first block is abnormal, wherein the plurality of blocks include the first block; and / or The security instruction is executed if a second block is abnormal, wherein the plurality of blocks includes the second block.
5. The method according to any one of claims 1 to 4, characterized in that: The method is executed by a device located in a cloud server.
6. The method according to claim 5, characterized in that The acquiring frame data comprises: Receive frame data of the vehicle; The method further comprises: An indication is sent to the vehicle, wherein the indication is used to indicate whether each of the blocks is abnormal.
7. A device for detecting data, characterized in that: include: An acquisition module, used to acquire frame data, wherein the frame data includes a snapshot of point cloud data within a preset time length; A division module, used for dividing the frame data into a plurality of blocks; An extraction module, configured to extract a feature value of each of the plurality of blocks, wherein the feature value represents a statistical characteristic of points in each of the blocks; A determination module is used to determine whether each block is abnormal or normal according to changes in the characteristic values of the multiple blocks.
8. The device according to claim 7, characterized in that The determination module is specifically used to input the change of the characteristic value into a machine learning model to obtain a prediction result indicating whether each block is abnormal or normal, wherein the machine learning model is trained by training samples, and the training samples include the characteristic value differences between abnormal blocks and normal blocks.
9. The device according to claim 7 or 8, characterized in that The characteristic values include one or more of the following: The average value of the coordinates; Standard deviation of coordinate values; The average value of intensity; The standard deviation of the intensity.
10. The device according to any one of claims 7 to 9, characterized in that The device also includes: a prompting module, configured to prompt a user that the point cloud data is abnormal if a first block is abnormal, wherein the plurality of blocks include the first block; and / or An execution module is configured to execute a security instruction when a second block is abnormal, wherein the plurality of blocks includes the second block.
11. The device according to any one of claims 7 to 10, characterized in that The device is located in a cloud server.
12. The device according to claim 11, characterized in that The acquisition module is specifically used to receive frame data of the vehicle; The device also includes: A sending module is used to send an indication to the vehicle, wherein the indication is used to indicate whether each block is abnormal.
13. A device for detecting data, characterized in that: It comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to call the computer program from the memory and run the computer program, so that the computer where the chip is located executes the method according to any one of claims 1 to 6.
14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to perform the method according to any one of claims 1 to 6.
15. A computer program product, characterized in that When the computer program product is run on a computer, the computer is caused to execute the method according to any one of claims 1 to 6.