Lane line distance calculation method and system for performance test of automatic driving system

Abstract

The invention relates to the technical field of automatic driving system performance testing, and discloses a lane line distance calculation method and system for automatic driving system performance testing, and the method comprises the steps: employing inertial navigation to obtain corresponding latitude and longitude data along a lane line needed by testing as lane line data; establishing a vehicle body coordinate system and acquiring coordinates of preset feature points on the vehicle in the vehicle body coordinate system; establishing a lane line coordinate system, converting the lane line data into coordinate points in the lane line coordinate system, and decomposing the coordinate points into a plurality of fitting curve segments; coordinates of the feature points in the vehicle body coordinate system are converted into a lane line coordinate system, lane line data target fitting curve segments corresponding to the feature points are determined, and the position relation between the feature points and lane lines is judged; and determining a distance calculation mode according to the position relationship, and calculating the distance from the feature point to the target fitting curve segment as the distance from the feature point to the lane line. Lane line related working condition test data with high accuracy and good real-time performance are provided, ADAS related working condition tests are well supported, and ADAS test reliability is effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of performance testing of autonomous driving systems, and particularly to a lane line distance calculation method and system for performance testing of autonomous driving systems. Background Art

[0002] Performance testing of autonomous driving systems is mainly used to verify the stability and response ability of autonomous driving systems in specific environments, aiming to verify the real-time processing ability and decision-making ability of autonomous driving algorithms in the face of various complex situations through actual test data, and to optimize the algorithms with test results to improve the safety and reliability of autonomous driving.

[0003] In the performance testing of autonomous driving systems, the distance from the wheels to the lane line is often used as the data input for some regulatory tests. This distance information not only directly affects the lateral control accuracy of autonomous driving vehicles, but also relates to the overall safety and reliability of the system. Specifically, by accurately measuring the distance between the wheels and the lane line, the autonomous driving system can adjust the steering angle in real time to ensure that the vehicle always stays on the optimal driving path, avoiding deviating from the lane or colliding with other traffic participants; in addition, this distance information is also the basis for realizing advanced driving assistance functions such as automatic lane change and overtaking. Therefore, calculating the distance between the wheels and the lane line is one of the key factors to ensure that the vehicle can safely and accurately stay within the lane, and improving the accuracy of calculating the distance between the wheels and the lane line has also become the focus of current research.

[0004] Existing methods for calculating the distance between the wheels and the lane lines mainly include camera image processing and lidar data processing technologies. The scheme based on camera image processing calculates the distance between the wheels and the lane lines by identifying the lane lines in the image. This method relies on computer vision algorithms to analyze the road image and estimates the vertical distance from the center point of the wheels to the nearest lane line in combination with the vehicle's own positioning information. However, since the camera captures two-dimensional images and lacks depth information, there are large errors in practical applications. At the same time, the camera itself has problems such as lens distortion and installation angle deviation, and is significantly affected by changes in lighting conditions. For example, strong light, shadows, rain, snow, or haze weather will cause the image quality to decline, resulting in lane line recognition failure and large distance calculation errors. In addition, during the process of converting the image coordinate system to the world coordinate system, the calibration of the camera needs to be completed in advance. Once the installation position of the camera changes slightly, large errors will be introduced. For multi-lane line scenarios, if the contrast between the lane lines and the surrounding environment is low, the lane line type is blurred, or the edges are missing, etc., it will also seriously affect the recognition accuracy. If deep learning methods are used to recognize lane lines, although the robustness is improved to a certain extent, it has high requirements for hardware performance and requires a large amount of labeled data for training, resulting in a significant increase in cost and complexity. On the other hand, the method based on lidar constructs a three-dimensional environment model by emitting laser beams and receiving reflected signals, so as to accurately locate the position of the lane lines and the relative position of the vehicle. Although its accuracy is high, it also has obvious disadvantages: the equipment cost is expensive, and it is sensitive to the reflectivity of the object surface. In some cases, it cannot provide sufficient point cloud density to accurately define the lane boundary. Therefore, to obtain accurate data, high requirements are imposed on the lane lines. Summary of the Invention

[0005] The present invention aims to provide a lane line distance calculation method and system for the performance test of an autonomous driving system, which can provide test data related to lane lines with high accuracy and good real-time performance, well support the test of related working conditions of ADAS, and effectively ensure the reliability of ADAS testing.

[0006] The basic solution provided by the present invention is: a lane line distance calculation method for the performance test of an autonomous driving system, the method comprising: S1, obtaining corresponding longitude and latitude data along the lane line required for the test by using an inertial navigation system as lane line data; S2, establishing a vehicle body coordinate system and obtaining the coordinates of preset feature points on the vehicle in the vehicle body coordinate system; S3, establishing a lane line coordinate system, converting the lane line data into coordinate points in the lane line coordinate system, and performing sampling and piecewise fitting to decompose the lane line data into multiple piecewise fitting curve segments; S4. Convert the coordinates of the feature points in the vehicle body coordinate system to the lane line coordinate system, determine the target fitting curve segment of the lane line corresponding to the feature points, and judge the positional relationship between the feature points and the lane line; S5. Determine the distance calculation method according to the positional relationship, and calculate the distance from the feature point to the target fitting curve segment as the distance from the feature point to the lane line.

[0007] The present invention also provides a lane line distance calculation system for the performance test of an autonomous driving system to execute the above-mentioned lane line distance calculation method for the performance test of an autonomous driving system. The system includes: A lane line data acquisition module, configured to obtain corresponding longitude and latitude data along the lane line required for the test by using inertial navigation as lane line data; A coordinate system module, configured to establish a vehicle body coordinate system and obtain the coordinates of preset feature points on the vehicle in the vehicle body coordinate system; and also configured to establish a lane line coordinate system, convert the lane line data into coordinate points in the lane line coordinate system, and perform sampling and piecewise fitting to decompose the lane line data into multiple fitting curve segments; A positional relationship judgment module, configured to convert the coordinates of the feature points in the vehicle body coordinate system to the lane line coordinate system, determine the target fitting curve segment of the lane line corresponding to the feature points, and judge the positional relationship between the feature points and the lane line; A distance calculation module, configured to determine the distance calculation method according to the positional relationship to calculate the distance from the feature point to the target fitting curve segment as the distance from the feature point to the lane line; An offset speed calculation module, configured to calculate the offset speed by using the central difference method and smoothing processing according to the distance data obtained in real time.

[0008] The working principle and advantages of the present invention are as follows: Compared with the prior art, the present solution is mainly aimed at the conventional test of the ADAS test procedure for fixed sections in a closed site in the performance test of an autonomous driving system, optimizes the calculation method of the distance from vehicle feature points to the lane line and the deviation speed, provides test data related to lane line conditions with high accuracy and good real-time performance, well supports the ADAS-related condition tests, effectively guarantees the reliability of ADAS tests, and at the same time avoids existing problems while ensuring high accuracy of the calculated distance.

[0009] The equipment adopted in this solution is simple. The high-precision positioning inertial navigation system is a standard configuration for autonomous driving vehicles, and there is no need to additionally add a large number of high-cost sensor devices, so the overall cost is relatively low. Since the test is carried out in a closed site, it is less affected by the external environment, and weather factors such as light, rain and fog have little impact on the system, avoiding the errors caused by calibration deviation of vision devices such as cameras, significantly reducing the sources of uncertainty in the calculation process, and making the calculation results more stable and reliable. Through actual test verification, the maximum error between the calculated distance from the wheel to the lane line and the measured value is controlled within 2 cm, with a relatively high accuracy level. At the same time, this method simplifies the data processing process, improves the operation efficiency, the system response delay can reach the millisecond level, the algorithm has low requirements for hardware performance, and greatly improves the operation efficiency on the premise of ensuring the calculation accuracy. Considering that the ADAS-related working condition tests are usually carried out on fixed sections in a closed test site, the positions of the lane lines used are fixed and the range is limited, and the lane line information can be reused for a long time after being collected once, further reducing the difficulty of real-time calculation and providing stable, efficient and sustainable basic data support for various ADAS function tests. Brief Description of the Drawings

[0010] Figure 1 is a schematic flow chart of the lane line distance calculation method for autonomous driving system performance testing provided by an embodiment of the present invention; Figure 2 is a schematic diagram of the establishment of the vehicle body coordinate system and the setting of feature points provided by an embodiment of the present invention; Figure 3 is a schematic diagram of the positional relationship between the vehicle feature points and the lane line provided by an embodiment of the present invention; Figure 4 is a schematic structural diagram of the lane line distance calculation system for autonomous driving system performance testing provided by an embodiment of the present invention. Detailed Description of the Embodiment

[0011] The following is a more detailed description through specific embodiments: The embodiment is basically as shown in the attached Figure 1 figures: The lane line distance calculation method for autonomous driving system performance testing provided by this embodiment includes: S1. Using the inertial navigation to obtain the corresponding longitude and latitude data along the lane line required for the test as the lane line data; S2. Establishing a vehicle body coordinate system and obtaining the coordinates of preset feature points on the vehicle in the vehicle body coordinate system; S3. Establishing a lane line coordinate system, converting the lane line data into coordinate points in the lane line coordinate system, and performing sampling and piecewise fitting to decompose the lane line data into multiple piecewise fitting curve segments; S4, converting the coordinates of the feature point in the vehicle body coordinate system into the lane line coordinate system, determining the target fitting curve segment of the lane line corresponding to the feature point, and determining the positional relationship between the feature point and the lane line; S5, determining a distance calculation method based on the positional relationship, and calculating the distance from the feature point to the target fitting curve segment as the distance from the feature point to the lane line; S6, calculating the offset velocity based on the distance data obtained in real time by using the central difference method and smoothing processing.

[0012] Specifically: In S1, in order to calculate the real-time distance from the wheel to the lane line, it is necessary to first collect the lane line point set data and store it in a file.

[0013] In this embodiment, lane line data is obtained by dotting with a cart. The process is as follows: (1) Install a high-precision positioning inertial navigation antenna at the center of the cart; install a laser device vertically downward at the center of the cart, so that the center point of the inertial navigation antenna and the laser emission direction are on the same vertical line. A laser pen can be installed below the inertial navigation antenna to facilitate the laser to aim at the mark on the ground for marking.

[0014] (2) After the inertial navigation positioning is calibrated, confirm that the inertial navigation positioning status meets the data collection requirements, that is, the RTK solution status of the inertial navigation reaches the fixed solution state and can achieve centimeter-level measurement; set a fixed frequency (for example, 10 Hz) to collect longitude and latitude data.

[0015] (3) Push the cart at a constant speed. During the pushing process, the laser needs to be strictly aligned with the inner edge contour of the lane line, and the inertial navigation data is continuously collected as the lane line data corresponding to the lane line.

[0016] (4) After completing the lane line collection, clean up the data, convert the longitude and latitude data collected by the inertial navigation into geodetic coordinates, draw a curve in the coordinate system, remove the data points that obviously do not conform to the corresponding lane line trajectory, and save it as a lane line data file to be used. In order to conveniently display the lane line position on the map, the file format is kml format. For the removal of abnormal points, the saved KML file can be loaded and opened using map software, and by observing its distribution in the map, the points that are obviously inconsistent with the distribution pattern of the previous and next points can be removed; or, the longitude and latitude points of the lane line file can be converted into coordinate values of the rectangular coordinate system, and a curve can be drawn for these points using drawing software. In theory, the line connecting these points should be a relatively smooth curve. Points with obvious burrs or steps in the line are removed, that is, the abnormal points collected during the collection process due to the unstable state of the inertial navigation positioning are removed to ensure the accuracy of the lane line data.

[0017] It should be noted that in this embodiment, a single lane is used, including left and right lane markings. The data collection process is the same: the left and right lane marking data are acquired and processed using the aforementioned cart-dotting method, and the corresponding left and right lane marking data files are saved for future use. Since ADAS operating condition testing is generally conducted on fixed sections of road in closed test sites, the lane markings used are also fixed and have a limited distance between them. Therefore, once collected in this manner, it can be used long-term, providing sustainable basic data for ADAS testing.

[0018] In S2, the vehicle body coordinate system is established and feature points are set.

[0019] like Figure 2 As shown, dual antennas are installed one in front of the other on the horizontal centerline of the roof. The front antenna is installed on the vertical centerline of the roof near the upper edge of the front windshield, and the rear antenna is installed on the vertical centerline of the roof above the rear seats. The rear antenna installation position is used as the origin of the vehicle body coordinate system, with the front direction of the vehicle body as the x-axis and the left direction of the vehicle body as the y-axis to establish the vehicle body coordinate system. Figure 2 The six points shown are: 1 is the center edge of the front bumper; 2 is the center edge of the rear bumper; 3 is the outer edge of the left front wheel; 4 is the outer edge of the left rear wheel; 5 is the outer edge of the right front wheel; and 6 is the outer edge of the right rear wheel. According to the requirements of ADAS test regulations, the preset feature points include the left front wheel point and the right front wheel point, and points 3 and 5 are taken as the feature points for the test.

[0020] Measure and record the coordinates of the feature points in the vehicle body coordinate system. Specifically, use a surveying pole for dot measurement. The operation is as follows: After the vehicle body coordinate system is established and the inertial navigation positioning accurately meets the measurement requirements, remove the rear antenna and install it on the top of the surveying pole. The surveying pole is perpendicular to the ground and placed at positions 3 and 5 in the figure, respectively, which are the midpoints of the outer edges of the left and right front wheels. The surveying pole is also equipped with a universal circular level. Slightly adjust the surveying pole so that the bubble in the circular level moves to the center of the circular level. Then, dot the position data of the point and record it. The coordinate value of the preset feature point in the vehicle body coordinate system can be obtained by conversion.

[0021] If the lane line is the left lane line, the corresponding left front wheel point is the feature point, and the distance from the left front wheel point to the left lane line is calculated, which represents the distance from the left front wheel to the left lane line; if the lane line is the right lane line, the corresponding right front wheel point is the feature point, and the distance from the right front wheel point to the right lane line is calculated, which represents the distance from the right front wheel to the left lane line.

[0022] In S3, the lane line coordinate system is established, sampled and segmented.

[0023] Before starting the ADAS test, data configuration is performed. The data file of the left lane line corresponding to the left lane line is configured with the coordinates of the left front wheel point, and the data file of the right lane line corresponding to the right lane line is configured with the coordinates of the right front wheel point to form a configuration file. This configuration file will be sent to the algorithm calculation model before calculation, and the left and right lane line coordinate systems will be created and sampled and fitted separately. When real-time position-related information based on the vehicle body coordinate system is obtained from devices such as inertial navigation installed on the vehicle body and the real-time data is given to the algorithm calculation model, the distances from the left front wheel feature point to the left lane line and from the right front wheel feature point to the right lane line can be calculated respectively.

[0024] From the perspective of the distance calculation model, the lane line point set data corresponding to the lane line can be fitted into a curve for calculation. In order to fit a more realistic lane line and reduce the calculation amount at the same time, the lane line is segmented and fitted. Since the lane line data is collected by a trolley, it is impossible to ensure uniform speed during the pushing process, and there may even be pauses. Therefore, there will be problems such as uneven distribution of data points and even multiple points overlapping in some areas. At the same time, the collected lane data is too dense, affecting the calculation efficiency. So, before performing segmented fitting, the data points need to be sampled. Since the collected points are unevenly distributed, sampling cannot be performed according to a fixed number of points. Therefore, sampling is performed according to a fixed distance.

[0025] Before sampling, coordinate conversion needs to be performed on the data points because the collected data points are in longitude and latitude and cannot be used directly. They need to be converted into the coordinates of the local coordinate system to be used. The prerequisite for coordinate conversion is to establish a lane line coordinate system and ensure that the established lane line coordinates are reasonable. If they are not reasonable, there may be a situation where multiple points are repeated within a certain interval azimuth, resulting in no solution for subsequent mathematical calculations. Therefore, two points, the starting point and the ending point of the lane line, are selected to calculate the heading angle. With the starting point as the origin, the calculated heading angle is used to establish a rectangular coordinate system, and thus the lane coordinate system is obtained. This can ensure that the points of the lane line are relatively evenly distributed in the coordinate system.

[0026] The coordinate points of the lane line data are converted to the coordinates in the lane line coordinate system and sampled in the form of distance segments: Starting from the starting point of the lane line, find the data points whose distance is greater than or equal to the preset distance (such as 0.1 meter) from front to back as the ending point. In order to reduce the data points and obtain more accurate lane line data points at the same time, the lane line data collected in this distance segment is averaged, and one point is used to replace all the data points in this distance segment. It is equivalent to sampling the lane line data at intervals of the preset distance, and using the average value of the data in this distance segment as the lane line data of this distance segment. Taking the ending point of the previous distance segment as the starting point, continue to find the data points whose distance is greater than or equal to the preset distance (such as 0.1 meter) from back to front as the ending point for sampling, and so on until all the data points are sampled.

[0027] Since the sampled data is arranged at equal intervals, it can be directly segmented and fitted in the number of sampling points, that is, segmented according to a fixed number of points. The number of points selected should not be too large, as it will affect curve fitting, nor too small, as it will result in too many segments, both of which will lead to ineffective fitting. Select 30 points for segmented fitting, that is, segmented fitting at an interval of 3 meters. For the convenience of calculation, the least squares method is selected for polynomial fitting, and the order of the polynomial is set to 2. Thus, the lane line data is decomposed into multiple segmented fitting curve segments, and the fitting equation corresponding to each fitting curve segment is obtained at the same time. The coefficients of the fitting equation are obtained by fitting all the points on the corresponding curve segment, which is related to the curvature of the curve segment.

[0028] S4. Determine the positional relationship between the vehicle feature point and the lane line.

[0029] Convert the coordinates of the feature point in the vehicle body coordinate system to the lane line coordinate system. In the lane line coordinate system, for the segmented fitting curve segments, connect the starting point and the ending point of each fitting curve segment to form a straight line segment corresponding to the fitting curve segment and number it. It can be numbered from the starting point to the ending point of the lane line as 1, 2, 3... N, that is, the first straight line segment, the second straight line segment, the third straight line segment... the last straight line segment. The number represents the corresponding relationship between the straight line segment and its corresponding fitting curve segment.

[0030] From the first straight line segment to the last straight line segment, calculate and analyze the positional relationship between the foot of the perpendicular from the feature point to the straight line segment and the straight line segment in a preset manner to determine whether the feature point enters the lane line range or the target fitting curve segment of the lane line corresponding to the feature point.

[0031] Analyze the positional relationship between the foot of the perpendicular from the feature point to the straight line segment and the straight line segment by determining whether the foot of the perpendicular from the feature point to the straight line segment is on the straight line segment. Use the following formula: Assume the vehicle feature point A, the starting point of the line segment is B, the ending point is C, and the coordinates of the point are x and y. Then: r = ((A.x - B.x) * (C.x - B.x) + (A.y - B.y) * (C.y - B.y)) / pow(len, 2) Among them, r is the value representing whether the foot of the perpendicular is on the line segment; A.x and A.y are the x and y axis coordinate values of A respectively; B.x and B.y are the x and y axis coordinate values of B respectively; C.x and C.y are the x and y axis coordinate values of C respectively; pow(len, 2) is the square of len; len is the distance between points B and C.

[0032] From the front view Figure 3 Define the left and right as left and right; when r <= 0, the foot of the perpendicular is at point B, indicating that point A is on the left side of the line segment BC, asFigure 3 as shown in (a); when r >= 1, the foot of the perpendicular is at point C, indicating that point A is on the right side of line segment BC, as Figure 3 shown in (c); when 0 < r < 1, the foot of the perpendicular is on the line segment, indicating that point A is within the range of line segment BC, as Figure 3 shown in (b).

[0033] Regarding the judgment of the positional relationship between the feature point and the lane line: 1) First, judge the positional relationship between the feature point and the first straight line segment: Calculate the r of the feature point and the first straight line segment. If it is judged that the feature point is on the left side of the first straight line segment, then the feature point is outside the lane line range. The distance calculated in S5 is the distance from the feature point to the starting point of the first line segment, and it is calculated according to the distance between two coordinate points; 2) If it is judged that the feature point is on the right side of the first straight line segment, then judge the positional relationship between the feature point and the last straight line segment: Calculate the r of the feature point and the last straight line segment. If it is judged that the feature point is on the right side of the last straight line segment, then the feature point is outside the lane line range. The distance calculated in S5 is the distance from the feature point to the end point of the last line segment, and it is calculated according to the distance between two coordinate points; 3) If it is judged that the feature point is on the left side of the last straight line segment, then the feature point is within the lane line range. Next, start judging the positional relationship between the feature point and the corresponding straight line segment from the second straight line segment: Calculate the r of the feature point and the straight line segment until it is judged that the foot of the perpendicular from the feature point to the straight line segment is on the straight line segment, then stop the calculation. Take this straight line segment as the target straight line segment, and select the target fitting curve segment corresponding to the target straight line segment to perform mathematical solution using the fitting equation to obtain the coordinates of the foot of the perpendicular point. The calculated distance is the distance from the feature point to the foot of the perpendicular point, and it is calculated according to the distance between two coordinate points.

[0034] Specifically, find the number of the target straight line segment. The straight line segment is composed of the first and last points of the corresponding lane line fitting curve segment. By finding the corresponding target fitting curve segment through the number, the fitting curve equation can be obtained. According to the known conditions, 3 equations can be listed for solution: 1) Let the fitting curve equation be f(x) = ax 2 + bx + c; Assume that the foot of the perpendicular point on the curve is (x0, f(x0)); Assume that the coordinates of the current feature point are (x1, f(x1)); Taking the derivative of the equation can obtain: f(x)' = 2ax + b; 2) The slope k of the straight line equation passing through x0 and x1 is: k = (f(x1) - f(x0)) / (x1 - x0); According to the relationship between the tangent line and the perpendicular line, it can be obtained that: k * f(x0)' = -1; 3) By combining and simplifying the equations, a cubic equation in one variable about x0 can be obtained: y = 2a 2*x0 3 +3ab*x0 2 +(2ac - 2a*f(x1) + b 2 +1)*x0 + bc - b*f(x1) - x1; Among them, x0 is the unknown, and the others are known. The Shengjin formula is used to solve the roots of the cubic equation of one variable. There may be multiple solutions for the solved roots. It is necessary to screen out the roots that meet the interval according to the interval position of the curve points. After solving x0, f(x0) can be solved according to the curve equation, and the coordinates of the foot point (x0, f(x0)) can be obtained. Then, according to the distance formula between two points, the distance between the feature point (x1, f(x1)) and the foot point (x0, f(x0)) is calculated, which is the distance from the vehicle feature point to the lane line, that is, the distance from the wheel to the lane line.

[0035] Since the lane line is in the form of a curve, when judging whether the foot of the feature point is on the line segment, there will be a situation where the feature point is on the right side of the current line segment and on the left side of the next line segment, or on the left side of the current line segment and on the right side of the previous line segment. In this case, the distance is taken as the average value of the distances from the feature point to the two curves as the result. Of course, these situations are very few, but by adopting the above method, the accuracy of the result can still be ensured.

[0036] S6. Calculate the offset speed according to the distance data obtained in real time by the central difference method and smoothing processing.

[0037] Through the above method, the real-time distance from the vehicle feature point to the lane line can be obtained in real time, and the offset speed can be further obtained according to the obtained distance. The offset speed is determined by calculating the change rate of the distance between the vehicle (feature point) and the lane line over time. The real-time data of the feature point obtained from the inertial navigation includes position information and time information. From this, the time stamp corresponding to each real-time data can be known. Using the central difference method, the data of adjacent 3 points are selected for differential calculation. It is observed that the data has very large spikes after differentiation and cannot well reflect the change trend of the data. Therefore, the differentiated data is low-pass filtered by the first order to smooth the data, and finally the smoothly changing offset speed data is obtained.

[0038] Specifically, the solution process is as follows: Step 1: Starting from the third calculated distance data, differentiate the distance data and record the time information (t0, t1, t2) and distances (d0, d1, d2) of these three points. Obtain the differential diff = (d2 - d0) / (t2 - t0); thereafter, after each calculated distance, take the relevant data of this distance and the previous two calculated distances for differentiation, that is, starting from the third obtained distance data, for each calculated distance, a differential value can be calculated simultaneously. Step 2: Perform low-pass filtering on the obtained differential data using the following formula: output = alpha * input + (1 - alpha) * prevOutput, where alpha = 0.05, input: is the input differential value, prevOutput: is the result of the previous filtering.

[0039] During the test, it is necessary to collect the data of the left and right lane lines and the coordinate data of the left and right feature points, and simultaneously calculate the distances and offset speeds of the vehicle's left and right feature points to the left and right lane lines.

[0040] As Figure 4 shown, to execute the above lane line distance calculation method for autonomous driving system performance testing, this solution also provides a lane line distance calculation system for autonomous driving system performance testing. The system includes: A lane line data acquisition module, which is used to obtain the corresponding longitude and latitude data along the lane line required for the test using inertial navigation as the lane line data; A coordinate system module, which is used to establish a vehicle body coordinate system and obtain the coordinates of the preset feature points on the vehicle in the vehicle body coordinate system; it is also used to establish a lane line coordinate system, convert the lane line data into coordinate points in the lane line coordinate system, and perform sampling and piecewise fitting to decompose the lane line data into multiple piecewise fitting curve segments; A position relationship judgment module, which is used to convert the coordinates of the feature points in the vehicle body coordinate system into the lane line coordinate system, determine the target fitting curve segment of the lane line corresponding to the feature points, and judge the position relationship between the feature points and the lane line; A distance calculation module, which is used to determine the distance calculation method according to the position relationship to calculate the distance from the feature point to the target fitting curve segment as the distance from the feature point to the lane line; An offset speed calculation module, which is used to calculate the offset speed by the central difference method and smoothing processing according to the real-time obtained distance data.

[0041] It can be understood that the above system can fully execute the above method and achieve the same effect, which will not be elaborated here.

[0042] The lane line distance calculation method and system for autonomous driving system performance testing provided in this embodiment are mainly aimed at the conventional testing of the ADAS test regulations for fixed sections of closed sites in autonomous driving system performance testing. The calculation methods for the distance and deviation speed from vehicle feature points to lane lines are optimized to provide lane line-related working condition test data with high accuracy and good real-time performance, which well supports the ADAS-related working condition testing, effectively guarantees the reliability of ADAS testing, and at the same time avoids existing problems while ensuring high calculation distance accuracy. The equipment adopted in this solution is simple. The high-precision positioning inertial navigation system is a standard configuration for autonomous driving vehicles, without the need to additionally add a large number of high-cost sensor devices, and the overall cost is relatively low. Since the test is carried out in a closed site, it is less affected by the external environment, and weather factors such as light, rain, and fog have less impact on the system, avoiding the errors caused by calibration deviations of vision devices such as cameras, significantly reducing the sources of uncertainty in the calculation process, and making the calculation results more stable and reliable. Through actual test verification, the maximum error between the calculated distance from the wheel to the lane line and the measured value is controlled within 2 cm, with a relatively high accuracy level. At the same time, this method simplifies the data processing flow, improves the operation efficiency, the system response delay can reach the millisecond level, the algorithm has low requirements for hardware performance, and greatly improves the operation efficiency on the premise of ensuring calculation accuracy. Considering that ADAS-related working condition testing is usually carried out on fixed sections in a closed test site, the positions of the lane lines used are fixed and the range is limited, and the lane line information can be reused for a long time after being collected once, further reducing the difficulty of real-time calculation and providing stable, efficient, and sustainable callable basic data support for various ADAS function tests.

[0043] The above are only embodiments of the present invention. Specific structures and common knowledge such as characteristics that are well known in the art are not described in detail here. Those of ordinary skill in the art know all the common general technical knowledge in the technical field to which the invention belongs before the application date or priority date, can know all the existing technologies in this field, and have the ability to apply the conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, combine their own abilities to improve and implement this solution. Some typical well-known structures or well-known methods should not become obstacles for those of ordinary skill in the art to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can still be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent.

Claims

1. Lane line distance calculation method for autonomous driving system performance testing, characterized in that S1. Use inertial navigation to obtain corresponding longitude and latitude data along the lane line required for testing as lane line data. S2. Establish a vehicle body coordinate system and obtain the coordinates of preset feature points on the vehicle in the vehicle body coordinate system. S3. Establish a lane line coordinate system, convert the lane line data into coordinate points in the lane line coordinate system, and perform sampling and piecewise fitting to decompose the lane line data into multiple piecewise fitting curve segments. S4. Convert the coordinates of the feature points in the vehicle body coordinate system into the lane line coordinate system, determine the target fitting curve segment of the lane line corresponding to the feature points, and judge the positional relationship between the feature points and the lane line. S5. Determine the distance calculation method according to the positional relationship to calculate the distance from the feature point to the target fitting curve segment as the distance from the feature point to the lane line.

2. The lane line distance calculation method for the performance test of the autonomous driving system according to claim 1, wherein It further includes S6. Calculate the offset speed by central difference method and smoothing processing based on the distance data obtained in real time.

3. The lane line distance calculation method for the performance test of the autonomous driving system according to claim 1, characterized in that, When the lane line is the left lane line and the corresponding left front wheel point is the feature point, calculate the distance from the left front wheel point to the left lane line; when the lane line is the right lane line and the corresponding right front wheel point is the feature point, calculate the distance from the right front wheel point to the right lane line.

4. The lane line distance calculation method for the performance test of the autonomous driving system according to claim 1, wherein In S1, install an inertial navigation antenna and a laser device at the center position of the trolley; the laser emission direction vertically downward of the laser device and the center point of the inertial navigation antenna are on the same vertical line; push the trolley at a constant speed, and during the process, the laser is aligned with the inner edge contour of the lane line, and continuously collect longitude and latitude data through inertial navigation, and form lane line data after cleaning the data.

5. The lane line distance calculation method for the performance test of the automatic driving system according to claim 1, wherein In S2, install two antennas one after the other on the front and rear horizontal center lines of the roof, use the installation position of the rear antenna as the origin of the vehicle body coordinate system, the forward direction of the vehicle body as the x-axis, and the left side direction of the vehicle body as the y-axis to establish the vehicle body coordinate system.

6. The lane line distance calculation method for the performance test of the autonomous driving system according to claim 1, wherein, In S3, select the starting point and ending point of the lane line to calculate the heading angle, use the starting point as the origin, and establish a rectangular coordinate system with the calculated heading angle as the lane line coordinate system of the lane line. Perform sampling in the form of distance segments and perform piecewise fitting in the form of the number of sampling points.

7. The lane line distance calculation method for the performance test of the autonomous driving system according to claim 1, characterized in that In S4, in the lane line coordinate system, connect the starting point and the ending point of each fitting curve segment of the lane line respectively to form a straight line segment corresponding to the fitting curve segment; from the first straight line segment to the last straight line segment, calculate and analyze the positional relationship between the feature point and the foot of the perpendicular of the straight line segment and the straight line segment in a preset manner, and judge whether the feature point enters the lane line range or the target fitting curve segment of the lane line corresponding to the feature point.

8. The lane line distance calculation method for the performance test of the autonomous driving system according to claim 7, wherein When the feature point is on the left side of the first straight line segment, the feature point is outside the lane line range, and the calculated distance is the distance from the feature point to the starting point of the first line segment; when the feature point is on the right side of the last straight line segment, the feature point is outside the lane line range, and the calculated distance is the distance from the feature point to the ending point of the last straight line segment; otherwise, the feature point is within the lane line range.

9. The lane line distance calculation method for the performance test of the autonomous driving system according to claim 7, wherein When the foot of the perpendicular of the feature point and the straight line segment is on the straight line segment, use the straight line segment as the target straight line segment, select the fitting equation of the target fitting curve segment corresponding to the target straight line segment for mathematical solution to obtain the coordinates of the foot of the perpendicular point, and the calculated distance is the distance from the feature point to the foot of the perpendicular point.

10. A lane line distance calculation system for the performance test of an autonomous driving system, characterized in that, Implement the lane line distance calculation method for the performance test of the autonomous driving system according to any one of claims 1-9, wherein the system comprises: A lane line data acquisition module, configured to obtain corresponding longitude and latitude data along the lane line required for the test by using inertial navigation as lane line data; A coordinate system module, configured to establish a vehicle body coordinate system and obtain the coordinates of preset feature points on the vehicle in the vehicle body coordinate system; and further configured to establish a lane line coordinate system, convert the lane line data into coordinate points in the lane line coordinate system, and perform sampling and piecewise fitting to decompose the lane line data into multiple piecewise fitting curve segments; A position relationship judgment module, configured to convert the coordinates of the feature points in the vehicle body coordinate system into the lane line coordinate system, determine the target fitting curve segment of the lane line corresponding to the feature points, and judge the position relationship between the feature points and the lane line; A distance calculation module, configured to determine a distance calculation method according to the position relationship to calculate the distance from the feature points to the target fitting curve segment as the distance from the feature points to the lane line; An offset speed calculation module, configured to calculate the offset speed by using the central difference method and smoothing processing based on the distance data obtained in real time.

Images