Detection method, electronic equipment, medium, processor, sensing device and vehicle

Abstract

The invention discloses a sensing data consistency detection method, electronic equipment, a storage medium, a processor, a sensing device and a vehicle, the sensing data consistency detection method is used for the processor of the sensing device, and the method comprises the following steps: obtaining a verification calculation value of each pre-configured pixel point data, the pre-configured pixel point data comprises sensing point cloud data and a verification pre-configured value; and sending the sensing point cloud data when the verification calculation value of each piece of pre-configured pixel point data and the corresponding verification pre-configured value meet a sensing data consistency condition. According to the method, verification can be carried out by taking pixel point data as a unit, the integrity of point-by-point data verification is realized, and the reliability of point cloud data is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of vehicles, and in particular to a method for detecting consistency of sensor data, as well as an electronic device, a computer-readable storage medium, a processor, a sensor device and a vehicle. Background Art

[0002] In existing technologies, the verification code for point cloud data from sensor systems is generated by the sensor processor's SOC, typically using CRC32. If the sensor processor's memory experiences data consistency issues or anomalies, there's no guarantee that the data sent externally via the protocol is consistent with the sensor data. Furthermore, the CRC32 verification algorithm is complex and time-consuming, and most use a single frame of data as the minimum verification unit, effectively determining the integrity of only one frame. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, one object of the present invention is to provide a method for detecting consistency of sensor data, which can verify the consistency of pixel data in units, confirm the integrity of each point, and enhance the reliability of point cloud data.

[0004] A second object of the present invention is to provide an electronic device.

[0005] A third object of the present invention is to provide a computer-readable storage medium.

[0006] A fourth objective of the present invention is to provide a processor.

[0007] A fifth object of the present invention is to provide a sensing device.

[0008] A sixth object of the present invention is to provide a vehicle.

[0009] In order to achieve the above-mentioned purpose, an embodiment of the first aspect of the present invention proposes a method for detecting consistency of sensor data, which is used for a processor of a sensor device. The method includes: obtaining a verification calculation value of each preconfigured pixel point data, the preconfigured pixel point data including sensor point cloud data and a verification preconfiguration value; when the verification calculation value of each preconfigured pixel point data and the corresponding verification preconfiguration value meet the sensor data consistency condition, sending the sensor point cloud data.

[0010] The sensor data consistency detection method of an embodiment of the present invention performs verification point by point in units of pixels. When the verification calculation value and the verification preconfigured value of each preconfigured pixel point data meet the sensor data consistency conditions, that is, the verification is passed, the sensor point cloud data is sent, thereby ensuring the integrity and validity of each pixel point data and improving the reliability of the transmitted sensor data.

[0011] In some embodiments, the pre-configured pixel point data also includes a frame count value of the data frame where the sensor point cloud data is located, and the method also includes: before obtaining and configuring the verification calculation value of the pixel point data, determining that the frame calculation value of the data frame where the sensor point cloud data is located is inconsistent with the count value of the current frame data.

[0012] In some embodiments, the verification calculation value is obtained based on the sensing point cloud data of the preconfigured pixel point data and the frame count value.

[0013] In some embodiments, the verification calculation value is obtained by performing an XOR operation on the current verification value and each data group of the preconfigured pixel point data, wherein the data group includes the sensor point cloud data and / or the frame count value.

[0014] In some embodiments, the bit width of the check calculation value is set based on the computational efficiency of the data block length of the pre-configured pixel data.

[0015] In some embodiments, the sensed data consistency condition includes the verification calculated value matching the verification pre-configured value.

[0016] In some embodiments, the preconfigured pixel data is obtained by configuring the verification preconfigured value and the frame count value based on the original pixel data of the sensor.

[0017] In some embodiments, before obtaining the verification calculation value of each preconfigured pixel point data, the method further includes: obtaining multiple preconfigured pixel point data stored in the cache space, each of the preconfigured pixel point data is any preconfigured pixel point data in the cache data block.

[0018] In some embodiments, the method further includes: executing an exception handling strategy when the verification calculation value and the verification pre-configured value meet a sensor data exception handling trigger condition.

[0019] In some embodiments, the sensor data abnormality processing trigger condition includes that the verification calculation value of at least one of the pre-configured pixel point data acquired this time does not match the verification pre-configuration value and / or the number of abnormal pre-configured pixel point data reaches a pixel number threshold, wherein the abnormal pre-configured pixel point data is the pre-configured pixel point data whose verification calculation value does not match the verification pre-configuration value.

[0020] In some embodiments, the exception handling strategy includes refreshing the cache space of the pre-configured pixel point data to obtain new pre-configured pixel point data.

[0021] In some embodiments, the pixel point number threshold is determined based on the probability of multiple preconfigured pixel point data being abnormal; the probability of multiple preconfigured pixel point data being abnormal is determined based on the probability of a single preconfigured pixel point data verification error and the number of preconfigured pixel points.

[0022] In order to achieve the above-mentioned purpose, the second aspect of the present invention proposes an electronic device, comprising: at least one processor; a memory communicatively connected to the at least one processor; the memory storing a computer program executable by the at least one processor, and the at least one processor implementing the sensor data consistency detection method when executing the computer program.

[0023] According to the electronic device of the embodiment of the present invention, the sensor data consistency detection method of the above embodiment is executed by at least one processor, and a verification pre-configuration value is pre-configured for each pixel point data. Data verification can be performed in units of pixel point data to achieve point-by-point verification of data integrity, thereby greatly improving the reliability of point cloud data.

[0024] In order to achieve the above-mentioned object, a third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the sensor data consistency detection method when the computer program is executed.

[0025] The computer-readable storage medium of an embodiment of the present invention can implement the sensor data consistency detection method of the above embodiment by executing the computer program thereon, and can perform data verification in units of pixel point data to achieve point-by-point verification of data integrity, thereby greatly improving the reliability of point cloud data.

[0026] In order to achieve the above-mentioned purpose, the fourth aspect embodiment of the present invention proposes a processor, including: a field programmable gate array, used to obtain raw sensor pixel data, and configure the raw sensor pixel data to verify preconfigured values and frame count values to obtain preconfigured pixel data; a SOC chip, used to execute the sensor data consistency detection method.

[0027] According to the processor of an embodiment of the present invention, a field programmable gate array is used to pre-configure verification values for each pixel point data. The SOC chip can perform data verification in units of pixel point data to achieve point-by-point verification of data integrity, greatly improving the reliability of point cloud data.

[0028] In some embodiments, the processor further includes: a memory connected to the field programmable gate array, for storing the preconfigured pixel data.

[0029] In some embodiments, the SOC chip includes a cache memory, which is connected to the internal memory via a data bus to obtain the pre-configured pixel data.

[0030] In order to achieve the above-mentioned object, a fifth embodiment of the present invention proposes a sensing device, characterized in that it includes a sensor and the processor, and the sensor is connected to the processor.

[0031] According to the sensing device of the embodiment of the present invention, by adopting the processor of the above embodiment, the pre-configured verification value can be pre-configured for each pixel point data through the field programmable gate array. The SOC chip can perform data verification in units of pixel point data to achieve point-by-point verification of data integrity, thereby greatly improving the reliability of point cloud data.

[0032] In some embodiments, the sensor comprises a lidar sensor.

[0033] In order to achieve the above-mentioned purpose, a sixth aspect of the present invention proposes a vehicle, characterized in that the vehicle includes the electronic device; or the vehicle includes the processor; or the vehicle includes the sensing device.

[0034] According to the vehicle of the embodiment of the present invention, based on the pre-configured verification value configured for each pixel point data, data verification can be performed in units of pixel point data to achieve point-by-point verification of data integrity, greatly improving the reliability of point cloud data.

[0035] In some embodiments, the vehicle further includes: a controller connected to the SOC chip of the processor to obtain the sensor data after consistency detection processing.

[0036] In some embodiments, the controller includes a domain controller or a vehicle controller.

[0037] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments with reference to the following drawings, in which: Figure 1 is a schematic diagram of a framework of a processor of a sensing device according to an embodiment of the present invention; Figure 2 is a flow chart of a method for detecting consistency of sensor data according to one embodiment of the present invention; Figure 3is a schematic diagram of a point cloud data interval format definition according to one embodiment of the present invention; Figure 4 is a schematic diagram of a point cloud image with cache consistency anomaly according to an embodiment of the present invention; Figure 5 is a schematic diagram of point cloud data segmentation and collection according to one embodiment of the present invention; Figure 6 Schematic diagram of point cloud image target reliability detection according to one embodiment of the present invention Figure 7 is a schematic diagram of an interval format definition in a point cloud data packet according to an embodiment of the present invention; Figure 8 is a block diagram of an electronic device according to one embodiment of the present invention; Figure 9 is a block diagram of a processor according to one embodiment of the present invention; Figure 10 is a flowchart of a processor verification process according to one embodiment of the present invention; Figure 11 is a block diagram of a sensing device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The embodiments of the present invention will be described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. The embodiments of the present invention will be described in detail below.

[0040] In related technologies, CRC32 verification is generally used to detect the consistency of sensor data. The verification code is generated by the SOC end. When the data in the memory has a catch consistency problem or the memory, such as DDR data, is abnormal, it cannot be guaranteed that the data sent to the outside through the protocol is consistent with the sensor data. Due to the performance limitations of existing embedded systems, the CRC32 verification algorithm is complex and time-consuming, and it is impossible to achieve real-time verification in units of pixels. In addition, the processor sends the sensor data to the domain controller, and the sensor data is transmitted in frames, that is, only the integrity of one frame of data can be judged. In particular, when the target decision-making judgment in the autonomous driving algorithm requires the use of highly reliable regional pixel data, the integrity of a single pixel cannot be verified separately.

[0041] Figure 1 FIG. 1 is a schematic diagram of a processor framework of a sensing device according to an embodiment of the present invention. Figure 1 As shown, it includes an FPGA module, a DDR module and a SOC chip. The cache module of the SOC chip is connected to the DDR module through a data bus. The cache module is a buffer memory and the DDR module can be a memory.

[0042] The FPGA module connects to the sensor through the data receiving module to obtain raw sensor data, such as lidar point cloud data. The SOC chip sends this data to a controller, such as an onboard domain controller, via the Ethernet port using a sensor data transmission protocol, such as a lidar point cloud protocol.

[0043] To solve the above problems, a first embodiment of the present invention provides a method for detecting consistency of sensing data. In the embodiment, the method can be applied to a processor of a sensing device, such as the above processor architecture or other processor architectures.

[0044] Reference below Figure 2-Figure 7 A method for detecting consistency of sensing data according to an embodiment of the present invention is described.

[0045] Figure 2 FIG. 1 is a flow chart of a method for detecting consistency of sensing data according to an embodiment of the present invention. Figure 2 As shown, the method includes steps S1 and S2.

[0046] S1, obtaining the verification calculation value of each pre-configured pixel point data.

[0047] S2: When the verification calculation value of each pre-configured pixel point data and the corresponding verification pre-configured value meet the sensor data consistency condition, the sensor point cloud data is sent.

[0048] Among them, each pre-configured pixel point data includes sensor point cloud data and verification pre-configured values.

[0049] The pre-configured pixel data may be pixel data obtained by configuring the verification algorithm parameters for the original sensor point cloud data. Figure 1 As shown, raw sensor data is obtained, and verification algorithm parameters are adapted according to the data bit width of the memory, such as DDR, and the platform architecture of the processor. In an embodiment, the verification algorithm parameters here may include but are not limited to verification pre-configured values, that is, a configured verification value.

[0050] Among them, the sensing point cloud data may include point coordinates, reflection intensity, etc.

[0051] Specifically, each preconfigured pixel data point is preconfigured with a check value, i.e., a check preconfigured value. Furthermore, a check calculation value for each preconfigured pixel data point can be obtained based on a preset algorithm or formula, i.e., a check value is calculated. When both the check calculation value and the check preconfigured value satisfy the sensor data consistency condition, the check passes, and the sensor point cloud data of the preconfigured pixel data obtained this time is transmitted. In this way, the integrity of each data point in the transmitted data frame can be determined through point-by-point verification.

[0052] The sensor data consistency condition may be set based on data reliability requirements or other verification conditions.

[0053] The sensor data consistency detection method of an embodiment of the present invention performs verification point by point in units of pixels. When the verification calculation value and the verification preconfigured value of each preconfigured pixel point data meet the sensor data consistency conditions, that is, the verification is passed, the sensor point cloud data is sent, thereby ensuring the integrity and validity of each pixel point data and improving the reliability of the transmitted sensor data.

[0054] In an embodiment, the sensor data consistency condition may include a match between a verification calculated value and a verification pre-configured value, that is, if the verification passes, the sensor point cloud data of this verification may be sent.

[0055] Furthermore, in some embodiments, the pre-configured pixel data also includes a frame count value of the data frame in which the sensor point cloud data is located. The frame count value is the number of the received frame of data. For example, the frame count value may be 2 bytes in size and start at 1. The frame count value may increase by 1 after each frame of point cloud data is updated.

[0056] In some embodiments, the preconfigured pixel data is obtained by configuring the verification preconfigured value and the frame count value based on the original pixel data of the sensor.

[0057] For example, taking laser radar point cloud data as an example, during the collection and transmission cycle of each laser radar point cloud data, the sensor data consistency detection method of an embodiment of the present invention is run.

[0058] like Figure 1 As shown in the figure, in the FPGA, the data receiving module drives the lidar sensor and obtains point cloud data; the point cloud data is added to the current frame's Counter, i.e., the frame count value, through the transmission module. The count value can be 2 bytes and the starting value can be 1. After each frame of point cloud data is updated, the frame count value is increased by 1 until the maximum value, for example, 65535, is cleared and the count is restarted. The corresponding 2-byte check pre-configured value is generated according to the check value calculation method, and the above two values, i.e., the frame count value and the check pre-configured value, are attached to the corresponding position of each pixel data. The DMA (Direct Memory Access) module is called in real time to transfer the data to the DDR memory.

[0059] Table 1 Definition of each pixel data block

[0060] Figure 3is a schematic diagram of a point cloud data interval format definition, namely, a PCD (Point Cloud Data) interval format definition according to an embodiment of the present invention. Figure 3 As shown in the figure, taking the lidar sensor as an example, the PCD interval is the measurement value part of the lidar. The figure shows a frame of data of a 96-line lidar (which can be understood as a column of 96 pixels in a two-dimensional plane). Each pixel in the figure is composed of the internal elements of each pixel defined by the data format in Table 1.

[0061] For example, in some embodiments, as shown in Table 1 above, the byte length of the data block (Data Block) of each pixel is 14 bytes, where each pixel consists of 10 bytes of data (from Data 0 to Data 4), a 2-byte counter value (Counter) and a 2-byte check value (Check value), and the entire PCD interval is 1344 bytes in total.

[0062] When a frame of LiDAR point cloud data is transmitted, a data completion interrupt is triggered. The SOC chip's verification and detection module retrieves the point cloud data corresponding to the DDR address based on the interrupt signal and verifies the point cloud data using an algorithm. If the verification passes, the formatted output module is called to output the point cloud data through the network interface.

[0063] In some embodiments, before obtaining the verification calculation value of the pre-configured pixel point data, the method further includes: determining that the frame calculation value of the data frame where the sensing point cloud data is located is inconsistent with the count value of the current frame data.

[0064] Specifically, the frame count value of each pre-configured pixel point data is compared with the current frame calculation value to see if they are consistent. If they are inconsistent, the point-by-point verification data is triggered, that is, the calculation of the verification calculation value of the above pre-configured pixel point data and the comparison of the verification calculation value with the verification pre-configured value are triggered, that is, the verification of the pixel point data is performed one by one.

[0065] In some embodiments, when the frame count value of the pixel point data is consistent with the current frame count value, it is considered that there is no consistency problem with the pixel point data, and the point cloud data can be directly output.

[0066] In some embodiments, the verification calculation value is obtained based on the sensor point cloud data and the frame calculation value of the preconfigured pixel data. For example, the verification calculation value may be a function of the sensor point cloud data and the frame calculation value of each preconfigured pixel data, or the verification calculation value may be calculated using a preset formula that includes the sensor point cloud data and configuration parameters in the preconfigured pixel data.

[0067] Furthermore, in some embodiments, the verification calculation value is obtained by performing an XOR operation on the current verification value and each data group of pre-configured pixel point data, wherein the data group includes sensor point cloud data and / or frame count values.

[0068] For example, in some embodiments, the check calculation value can be obtained by the following formula: (Formula 1) (Formula 2) in, It can be 14 bytes, i.e., 1 pixel point cloud data block, as shown in Table 1, which can include 10 bytes of data, 2 bytes of frame count value and 16-bit check preconfigured value. for The data bytes in i are as follows Figure 3 The sequence number of the pixel data in .

[0069] In formula (2), To verify the configuration value, The initial value of may be a set value such as 0 or other values. Wherein, every two data bytes may be a data group.

[0070] Specifically, as in the above formulas (1) and (2), the right side of the equal sign is a group of two bytes of data. Divide it into 6 data groups (each group is 16 bits), and perform XOR operation on the current check calculation value and each data group. After the calculation is completed, the R on the left side of the equal sign is the 16-bit check calculation value.

[0071] Furthermore, based on the calculation process of the above checksum calculation value, the bit width of the checksum calculation value can be set based on the computational efficiency of the data block length of the pre-configured pixel data. That is, the bit width of R can also be adjusted to different bit widths such as 8bit, 32bit, 64bit, to adapt to different The highest computing efficiency of data block length under different hardware architecture platforms.

[0072] In some embodiments, when the calculated verification value and the pre-configured verification value meet the sensor data exception handling trigger condition, the exception handling strategy is executed. That is, when the sensor data consistency condition is not met, the sensor point cloud data is not transmitted. When the sensor data exception trigger condition is met, exception handling is performed. This ensures that the transmitted data is complete and reliable, and abnormal data is not transmitted, thereby improving the reliability of the sensor data.

[0073] In some embodiments, the exception handling strategy may include refreshing the cache space of the pre-configured pixel point data to obtain new pre-configured pixel point data, and then verifying the new pre-configured pixel point data according to the above process.

[0074] For example, Figure 1 As shown, when the verification test passes, the formatted output module can be called to output the point cloud data through the network interface. Otherwise, exception processing is performed, such as calling the exception processing module to refresh the point cloud data cache to ensure that the transmitted sensor data is reliable data.

[0075] In some embodiments, the trigger conditions for sensor data abnormality processing may include the verification calculation value not matching the verification preconfiguration value and / or the number of abnormal preconfigured pixel point data reaching a pixel point number threshold, wherein the abnormal preconfigured pixel point data is the preconfigured pixel point data whose verification calculation value does not match the verification preconfiguration value.

[0076] If the calculated verification value does not match the pre-configured verification value, the pre-configured data point is considered abnormal and the point cloud data will not be transmitted. Exception handling, such as refreshing the point cloud data cache, will be performed. Alternatively, if the calculated verification value does not match the pre-configured verification value, it is determined that the pre-configured pixel data is abnormal and the number of abnormal pre-configured pixel data reaches the pixel number threshold, an exception handling strategy will be implemented, such as clearing the current cache data and refreshing the point cloud data cache. Verification will then be performed according to the above verification process to ensure that the transmitted sensor data is reliable.

[0077] In point cloud data, it's often necessary to balance the false positive rate and the false negative rate. This can be achieved by setting an appropriate pixel count threshold. For example, the pixel count threshold is represented by k. If k is set too low, occasional single-point anomalies may trigger false positives for regional anomalies, resulting in a high false positive rate. However, if k is set too high, multi-point verification anomalies may be overlooked, and true regional anomalies may not be detected in a timely manner, resulting in a high false negative rate.

[0078] In an embodiment, when verifying a specific area of point cloud data (which can be determined by a domain controller), it is necessary to reasonably set a threshold for the number of points at which multiple pre-configured pixel point data are simultaneously abnormal, i.e., a pixel point number threshold, based on the number of pixel point data and the probability of single-point verification abnormality, to effectively determine whether the area is abnormal.

[0079] In some embodiments, the pixel point number threshold is determined based on the probability of multiple pre-configured pixel point data being abnormal (hereinafter referred to as the probability of multiple-point abnormalities); the probability of multiple pre-configured pixel point data being abnormal is determined based on the probability of a single pre-configured pixel point data verification error (hereinafter referred to as the probability of a single-point abnormality) and the number of pre-configured pixel points.

[0080] The following describes how to determine the pixel number threshold.

[0081] The probability of the same checksum value for different point cloud data can be understood as the probability of a checksum error. The following is the probability that two random 16-bit B (data groups) will get the same 16-bit R (checksum calculation value) calculated by this checksum method: ; For n 16-bit random data groups B, the 16-bit R calculated according to the above formula (2) is the same. The premise of the following probability calculation formula (3) is that the different calculation values of each operation check in formula (1) and formula (2) are independent events: ;Formula (3) When the number n of data groups B is small relative to the total possible value d (65536 for 16-bit), the approximate calculation formula is: ;Formula (4) Let e be approximately equal to 2.718, and A is divided into 6 groups, so n = 6. Substituting this into the formula we can get the result.

[0082] The P probability value obtained by the above calculation is the probability of single-point calibration error.

[0083] Furthermore, when verifying a specific area of point cloud data, it is necessary to reasonably set a threshold for the number of simultaneously abnormal pixels based on the number of points and the probability of a single-point verification anomaly to effectively determine whether the area is abnormal. In some embodiments, the specific area may be the area corresponding to the pre-configured pixel data obtained this time, i.e., the pixel data being verified this time. The specific area or quantity can be set or determined based on other vehicle parameters.

[0084] For example, set the following parameter values: N (the total number of pixel data points) means that a specific area contains N pixels.

[0085] P (probability of abnormality of single-point calibration) is calculated as P = 0.00027 according to formula (4).

[0086] k (pixel number threshold) When k points are detected as abnormal in a specific area, the area is judged to be abnormal.

[0087] In some embodiments, the probability of multi-point verification anomaly is calculated based on the assumption that pixel anomalies are independent events. The probability of multiple pixel data anomalies, i.e., the law of multi-point verification anomalies, can be expressed as follows: ; Among them, is the number of combinations of selecting k pixels from N pixels, that is ; Among them, is the probability that k pixels are abnormal at the same time.

[0088] is the probability that the remaining Nk pixels are normal.

[0089] In summary, the probability formula for checking anomalies for at least k pixels can be expressed as follows: ;Formula (5) Assume that the total number of pixels in a specific area to be verified is N = 1000. According to the single point abnormality probability P = 0x00027, The probability of at least k pixels checking for anomalies can be calculated according to formula (5). The calculation results for different k values are shown in Table 2 below.

[0090] Table 2: Corresponding values of different k value

[0091] Among them, in the pixel anomaly detection in a specific area, The significance of k is to quantify the randomness of multi-point anomalies. The lower the value, the less rare they are, and the closer they are to non-random events. For example, as shown in Table 2 above, for an area with N=1000, k=3 means that at least three pixel data verification anomalies must be detected before anomaly processing is triggered. This is suitable for low false alarm requirements (probability below 0.27%), while k=5 is suitable for high reliability requirements.

[0092] In the embodiment of the present invention, the consistency check of the embodiment of the present invention can be applied to detection in different application scenarios.

[0093] In some embodiments, the method of the embodiment of the present invention can be applied to the cache consistency anomaly detection process, such as Figure 4 The figure shows a schematic diagram of a point cloud image with cache consistency anomaly according to one embodiment of the present invention, wherein the framed area may be a specific area to be verified. Specifically, the following steps may be included: Step 1: Data chunking.

[0094] Figure 5 FIG. 1 is a schematic diagram of point cloud data segmentation and collection according to an embodiment of the present invention. Figure 5 As shown, each column of data is divided into blocks according to the Cache Line Size, that is, each data block is a Cache Line Size. Among them, each column of data can refer to the amount of data transferred between the memory and the cache once, for example Figure 3 1344Bytes in.

[0095] It can be understood that the data in the cache is pre-configured pixel data, that is, the original sensor pixel data is configured on the FPGA side to obtain the pre-configured pixel data and stored in the memory such as DDR.

[0096] Step 2: Collect data.

[0097] Use a random sampling strategy, such as Figure 5 For the pixel shown in , one pixel point data is randomly selected from each data block for verification.

[0098] That is, before obtaining the check calculation value of each preconfigured pixel data, a plurality of preconfigured pixel data stored in the cache space is obtained, and each preconfigured pixel data is any preconfigured pixel data in the cache data block.

[0099] Step 3: Verify the data.

[0100] The following verification process is performed on each pixel data selected in the above sampling strategy: Step 3.1, compare frame count values. Compare the frame count value of each data point to see if it is consistent with the current frame count value. If they are inconsistent, trigger point-by-point data verification.

[0101] Step 3.2: Verify data point by point. For each pixel data, obtain the verification calculation value according to formula (2). Compare the verification calculation value with the pre-configured verification pre-configured value to see if they match. If they do not match, record an exception. If the exception value reaches k (the pixel number threshold), trigger the exception handling strategy.

[0102] In other embodiments, the method of the embodiment of the present invention can also be applied to the scenario of reliability detection of point cloud data in DDR.

[0103] Specifically, based on the scenario where the autonomous driving algorithm needs to rely on highly reliable data to make decisions, the target tracking module in the autonomous driving algorithm needs to judge (e.g. Figure 4 When determining whether each pixel of the vehicle in the frame is highly reliable, we can first use a cache line-sized point cloud data block (such as Figure 6 The dotted box in the middle is used to perform random quick verification, and the pixel data in a specific area is verified point by point (as shown in the dotted box in the middle Figure 6 The reliability of the pixel data scanned above is determined based on the corresponding k value in Table 2. The smaller the k value shown in Table 2, the higher the reliability of the pixel. It can be used as a trust metric to ensure the credibility of data and used in driving decisions to improve safety.

[0104] By implementing the above method, the reliability of point cloud data can be monitored in real time. When abnormalities in point cloud data are found, the autonomous driving system can process them in a timely manner, which can reduce the probability of misjudgment of the target by the perception layer of the autonomous driving algorithm to a certain extent, thereby improving the overall reliability and robustness of the autonomous driving system.

[0105] For each frame of point cloud data, if the accuracy of its output result cannot be guaranteed by the verification value, the point cloud data will not be used as the input of subsequent modules. This avoids the use of erroneous point cloud information that affects the accuracy of obstacle perception, improves the reliability of the autonomous driving process, and eliminates safety hazards.

[0106] Figure 7 This is a schematic diagram of the interval format definition in the point cloud data packet according to one embodiment of the present invention. The specific process of calculating the verification value can be as follows: Capture the data fragments of the PCD packet of the lidar through Wireshark (such as Figure 7 As shown), the following is the detailed algorithm flow for obtaining the checksum calculation value in Xilinx's ZynqMP platform SOC (Cortex-A53 architecture little endian mode): The data block identifier of the pre-configured pixel data is: ; The checksum calculation value is expressed as: ; Finally, we get: R = 0xa2ae.

[0107] Furthermore, the verification calculation value of each pre-configured pixel point data in the area to be verified is compared with the verification pre-configured value. If the two are not the same, an exception is recorded. When the abnormal value reaches the threshold k, the exception processing is triggered. If the verification passes, the point cloud data is sent.

[0108] In short, the sensor data consistency detection method of the present invention implements a point-by-point verification method for sensor data, such as LiDAR data, via the FPGA. This method then verifies pixel consistency in real time on the SOC, eliminating point cloud data inconsistencies under specific operating conditions and enhancing the reliability of point cloud data. This method addresses the inability of traditional single-frame verification methods to verify point-by-point integrity, enabling highly reliable point cloud data, such as LiDAR data, to be output in a variety of complex and demanding scenarios, thereby improving the environmental adaptability of the autonomous driving system's back-end algorithms.

[0109] Based on the sensor data consistency detection method of the above embodiment, a second embodiment of the present invention provides an electronic device.

[0110] Figure 8is a block diagram of an electronic device according to an embodiment of the present invention.

[0111] like Figure 8 As shown, the electronic device 10 includes at least one processor 11 and a memory 12 in communication with the at least one processor 11 .

[0112] The memory 12 stores a computer program that can be executed by at least one processor 11 . When the at least one processor 11 executes the computer program, the sensor data consistency detection method of the above embodiment is implemented.

[0113] According to the electronic device 10 of the embodiment of the present invention, the sensor data consistency detection method of the above embodiment is executed by at least one processor 11, and a verification pre-configuration value is pre-configured for each pixel point data. Data verification can be performed in units of pixel point data to achieve point-by-point verification of data integrity, thereby greatly improving the reliability of point cloud data.

[0114] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the sensor data consistency detection method of the above embodiment can be implemented.

[0115] The computer-readable storage medium of an embodiment of the present invention can implement the sensor data consistency detection method of the above embodiment by executing the computer program thereon, and can perform data verification in units of pixel point data to achieve point-by-point verification of data integrity, thereby greatly improving the reliability of point cloud data.

[0116] A fourth embodiment of the present invention provides a processor.

[0117] Figure 9 is a block diagram of a processor according to an embodiment of the present invention, such as Figure 9 As shown, the processor 11 includes a field programmable gate array (FPGA) 101 and a SOC chip 102. That is, the processor has a multi-core heterogeneous architecture.

[0118] Among them, the field programmable gate array 101 is used to obtain raw sensor pixel data, and configure the raw sensor pixel data to check preconfigured values and frame count values to obtain preconfigured pixel data; the SOC chip 102 is used to execute the sensor data consistency detection method of the above embodiment.

[0119] According to the processor 11 of the embodiment of the present invention, the field programmable gate array 101 pre-configures the verification pre-configuration value for each pixel point data, and the SOC chip 102 can perform data verification in units of pixel point data to achieve point-by-point verification of data integrity, thereby greatly improving the reliability of point cloud data.

[0120] like Figure 1 As shown, the processor 11 may also include a memory, such as DDR, connected to the field programmable gate array 101 for storing preconfigured pixel data. Specifically, the FPGA receives raw sensor data, such as lidar data, verifies preconfigured values for each point cloud data, and then transfers it to the memory for storage.

[0121] like Figure 1 As shown, the SOC chip 101 includes a cache memory such as Cache, and the cache memory is connected to the memory through a data bus to obtain pre-configured pixel data.

[0122] In an embodiment, data can be transmitted between the cache memory and the memory with a set data size, and then the SOC chip 102 collects pre-configured pixel point data from the cache memory to perform data verification according to the sensor data consistency detection method of the above embodiment. If the verification is passed, the point cloud data is sent, otherwise an exception handling strategy is executed, such as refreshing the data in the cache memory, to ensure the reliability of the sent sensor point cloud data.

[0123] like Figure 10 FIG. 1 is a flowchart of a processor verification process according to one embodiment of the present invention, including: S01: a data receiving module receives data; S02: a transmission module transmits data; S03: a verification detection module verifies the data; S04: if verification passes, the formatted output module outputs point cloud data; S05: if verification fails, the exception handling module performs exception handling.

[0124] Specifically, refer to Figure 1 As shown, taking a LiDAR sensor as an example, S01, the data receiving module, can be implemented in an FPGA and is responsible for receiving the raw sensor point cloud data from the LiDAR sensor's MIPI interface, such as point coordinates and reflection intensity, and adapting the verification algorithm parameters, such as the verification pre-configured value and frame count value, based on the DDR data bit width and the CPU platform architecture, and providing the corresponding verification parameters to the transmission module.

[0125] S02, the transmission module can be implemented in FPGA, which receives radar point cloud data and formats it into mipi signal data. After calculating and adding the check value according to the corresponding check algorithm parameters, it transmits the point cloud data to the corresponding DDR address space on the SOC side through the DMA (Direct Memory Access) controller and triggers the data transmission completion interrupt.

[0126] S3, the verification and detection module can be implemented in the SOC code, triggered by the interruption of the detected data transmission completion, and obtains the point cloud data from the corresponding address, and verifies the consistency and credibility of the point cloud data according to different application scenario algorithms.

[0127] S4: The exception handling module can be triggered by the check detection module to refresh the cache and repair the corresponding DDR address data according to the exception address.

[0128] S5, the formatted output module can be triggered by the verification detection module, and formatted output high-quality point cloud data after consistency detection and exception processing.

[0129] Based on the processor of the above embodiment, a fifth embodiment of the present invention provides a sensing device.

[0130] like Figure 11 As shown, the sensing device 200 of the embodiment of the present invention includes a sensor 201 and the processor 11 of the above embodiment, and the processor 11 is connected to the sensor 201, as shown in FIG. Figure 1 In the embodiment, the FPGA in the processor 11 is connected to the sensor 201 to obtain the original sensing point cloud data.

[0131] In an embodiment, the sensor 200 may include a lidar sensor or other perception sensor.

[0132] According to the sensing device 200 of the embodiment of the present invention, by adopting the processor 11 of the above embodiment, the field programmable gate array 101 can be used to pre-configure the verification preconfiguration value for each pixel point data, and the SOC chip 102 can perform data verification in units of pixel point data to achieve point-by-point verification of data integrity, thereby greatly improving the reliability of point cloud data.

[0133] Based on the electronic device, processor and sensor device of the above embodiments, a sixth embodiment of the present invention proposes a vehicle.

[0134] The vehicle according to the embodiment of the present invention includes the electronic device according to the above embodiment, or includes the sensing device according to the above embodiment, or includes the processor according to the above embodiment.

[0135] Of course, the vehicle also includes other infrastructure systems and components, which are not limited here.

[0136] According to the vehicle of the embodiment of the present invention, based on the pre-configured verification value configured for each pixel point data, data verification can be performed in units of pixel point data to achieve point-by-point verification of data integrity, greatly improving the reliability of point cloud data.

[0137] The vehicle of the embodiment of the present invention also includes a controller, which is connected to the SOC chip of the processor to receive the sensor data after consistency detection processing, and then can perform control based on the sensor data, such as automatic driving control or other scene control.

[0138] In some embodiments, the controller may be a domain controller or a vehicle controller of the vehicle, which is not specifically limited here.

[0139] Vehicles in this embodiment of the present invention perform point-by-point data verification on a pixel-by-pixel basis, ensuring the integrity and validity of each pixel. This is particularly true when highly reliable regional pixel data is required for target decision-making within autonomous driving algorithms, enabling the integrity of individual pixels to be verified independently. This overcomes the inability of traditional single-frame verification methods to confirm integrity point by point, enabling sensors to output highly reliable point cloud data in a variety of complex and demanding scenarios, thereby improving the environmental adaptability of the autonomous driving system's back-end algorithms.

[0140] In the description of this specification, any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes additional implementations in which functions may be performed in a different order than shown or discussed, including performing functions in a substantially simultaneous manner or in a reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0141] The logic and / or steps represented in a flowchart or otherwise described herein, for example, can be considered a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0142] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the aforementioned embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any of the following technologies known in the art, or a combination thereof, may be used: a discrete logic circuit having logic gates for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gates, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0143] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0144] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0145] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and are not to be construed as limiting the present invention. Persons skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

[0146] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "illustrative embodiments," "example," "specific example," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

[0147] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A method for detecting consistency of sensor data, characterized in that: For use with a processor of a sensing device, the method comprising: Obtaining a verification calculation value for each preconfigured pixel point data, wherein the preconfigured pixel point data includes sensor point cloud data and a verification preconfigured value; When the verification calculation value of each of the preconfigured pixel point data and the corresponding verification preconfigured value meet the sensing data consistency condition, the sensing point cloud data is sent.

2. The method according to claim 1, characterized in that The pre-configured pixel point data also includes a frame count value of the data frame where the sensing point cloud data is located, and the method further includes: Before obtaining and configuring the verification calculation value of the pixel point data, it is determined that the frame calculation value of the data frame where the sensing point cloud data is located is inconsistent with the count value of the current frame data.

3. The method according to claim 1, characterized in that The verification calculation value is obtained based on the sensing point cloud data of the pre-configured pixel point data and the frame count value.

4. The method according to claim 3, characterized in that The verification calculation value is obtained by performing an exclusive OR operation on the current verification value and each data group of the pre-configured pixel point data, wherein the data group includes the sensing point cloud data and / or the frame count value.

5. The method according to claim 4, characterized in that The bit width of the check calculation value is set based on the computation efficiency of the data block length of the pre-configured pixel point data.

6. The method according to any one of claims 1 to 5, characterized in that The sensed data consistency condition includes that the verification calculated value matches the verification pre-configured value.

7. The method according to claim 2, characterized in that The preconfigured pixel data is obtained by configuring the verification preconfigured value and the frame count value based on the original pixel data of the sensor.

8. The method according to claim 1, characterized in that Before obtaining the verification calculation value of each pre-configured pixel point data, the method further includes: A plurality of preconfigured pixel point data stored in the cache space is obtained, where each of the preconfigured pixel point data is any preconfigured pixel point data in the cache data block.

9. The method according to claim 1, characterized in that The method further comprises: When the verification calculation value and the verification pre-configured value meet the sensor data abnormality processing trigger condition, the abnormality processing strategy is executed.

10. The method according to claim 9, characterized in that The sensor data abnormality processing trigger condition includes that the verification calculation value of at least one of the pre-configured pixel point data acquired this time does not match the verification pre-configuration value and / or the number of abnormal pre-configured pixel point data reaches a pixel point number threshold, wherein the abnormal pre-configured pixel point data is the pre-configured pixel point data whose verification calculation value does not match the verification pre-configuration value.

11. The method according to claim 10, characterized in that The exception handling strategy includes refreshing the cache space of the pre-configured pixel point data to obtain new pre-configured pixel point data.

12. The method according to claim 10, characterized in that The pixel number threshold is determined based on the probability of abnormal data of multiple pre-configured pixel points; The probability that the data of the plurality of pre-configured pixels are abnormal is determined based on the probability that a single pre-configured pixel data check is erroneous and the number of the pre-configured pixels.

13. An electronic device, characterized in that: include: at least one processor; a memory communicatively coupled to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, and when the at least one processor executes the computer program, the sensor data consistency detection method according to any one of claims 1 to 12 is implemented.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the method for detecting consistency of sensing data according to any one of claims 1 to 12 is implemented.

15. A processor, characterized in that: include: A field programmable gate array, configured to obtain raw sensor pixel data, and configure a checksum preconfigured value and a frame count value on the raw sensor pixel data to obtain preconfigured pixel data; A SOC chip, used to execute the sensor data consistency detection method according to any one of claims 1 to 12.

16. The processor according to claim 15, wherein: The processor further includes: A memory is connected to the field programmable gate array and is used to store the preconfigured pixel point data.

17. The processor according to claim 16, wherein: The SOC chip includes a cache memory, which is connected to the internal memory via a data bus to obtain the pre-configured pixel point data.

18. A sensing device, characterized in that: The device comprises a sensor and the processor according to any one of claims 15 to 17, wherein the sensor is connected to the processor.

19. The sensing device according to claim 18, characterized in that The sensor includes a lidar sensor.

20. A vehicle, characterized in that: The vehicle includes the electronic device according to claim 13; Alternatively, the vehicle comprises the processor according to any one of claims 15 to 17; Alternatively, the vehicle comprises the sensing device according to claim 18 or 19.

21. The sensing device according to claim 20, characterized in that The vehicle further comprises: A controller is connected to the SOC chip of the processor to obtain the sensor data after consistency detection processing.

22. The vehicle according to claim 21, characterized in that The controller includes a domain controller or a vehicle controller.

Images