Vehicle body sensor calibration method and calibration coordinate system positioning module used by same

Abstract

The invention provides a vehicle body sensor calibration method and a calibration coordinate system positioning module used by the same, and relates to the technical field of sensor calibration, and the method comprises the following steps: building a vehicle calibration coordinate system based on the calibration coordinate system positioning module; randomly dividing the target test area into a plurality of calibration groups, carrying out preliminary calibration on the parameters of the vehicle body sensor in each calibration group, carrying out secondary calibration after positioning and removing an abnormal target test area, and determining a plurality of groups of calibration parameters of the vehicle body sensor; and with the minimization of the measurement precision deviation and the maximization of the vehicle body sensor calibration parameter application universality as optimization targets, the optimal vehicle body sensor calibration parameter applied to each target test area is determined based on an optimization algorithm. The vehicle calibration coordinate system is quickly established through the calibration coordinate system positioning module, the convenience and efficiency of vehicle body sensor calibration are improved, the optimal vehicle body sensor calibration parameters applied to each target test area are found based on optimization operation, and the comprehensive measurement precision is ensured.

Description

A method for calibrating vehicle body sensors and a calibration coordinate system positioning module used therein. Technical Field

[0001] This invention relates to the field of sensor calibration technology, specifically to a vehicle body sensor calibration method and a calibration coordinate system positioning module used therein. Background Technology

[0002] With the booming development of the intelligent vehicle industry, the demand for autonomous driving in automobile R&D is increasing, and the algorithms for autonomous driving and automatic parking are evolving at an increasingly rapid pace. This means that during the R&D phase, it may be necessary to constantly change the solution or update the algorithm to meet new challenges. In this process, it is necessary to frequently perform basic calibration of vehicle sensors. If calibration is performed in a professional calibration room, it often requires application and waiting and is costly. Furthermore, the calibration results of conventional body sensors are unique and cannot fully adapt to the measurement needs of all locations in the external area. There are some locations where the fit between the calibrated body sensors and the actual location is poor, resulting in poor measurement accuracy at these locations.

[0003] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this invention is to provide a vehicle body sensor calibration method and a calibration coordinate system positioning module used therein, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for calibrating a vehicle body sensor includes the following steps:

[0007] S1, Based on the calibration coordinate system positioning module that can be detachably installed at the rear wheel hub of the vehicle, a vehicle calibration coordinate system with the rear axle of the vehicle as the Y-axis is established, and the X-axis is perpendicular to the Y-axis in the horizontal direction and passes through the center of the vehicle body;

[0008] S2, define the vehicle calibration area, and divide the vehicle calibration area into three calibration zones: near distance, medium distance and far distance, according to the order of the distance between the vehicle and the sensors from near to far. Each calibration zone is further divided into several target test areas.

[0009] S3. For each calibration zone, the target test area within it is randomly divided into several calibration groups. The preliminary calibration of the vehicle body sensor parameters is performed in each calibration group. Based on the elimination method, the target test area in the calibration group is divided into normal target test area or abnormal target test area. The normal target test areas in the same calibration group are summarized as a target test area set.

[0010] S4. When there is an abnormal target test area in the calibration group, the vehicle body sensor is calibrated a second time based on the target test area set to determine the vehicle body sensor calibration parameters applied to the target test area set. Otherwise, the preliminary calibration result is used as the vehicle body sensor calibration parameters for the target test area set.

[0011] S5. For each target test area, calculate the measurement accuracy deviation of each set of vehicle body sensor calibration parameters, and determine the optimal vehicle body sensor calibration parameters for each target test area based on the optimization algorithm with the optimization objectives of minimizing the measurement accuracy deviation and maximizing the universality of the application of vehicle body sensor calibration parameters.

[0012] Furthermore, the logic for dividing each calibration zone is as follows: a near-distance threshold and a far-distance threshold are preset, and the near-distance threshold is less than the far-distance threshold. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is not greater than the near-distance threshold is designated as the near-distance calibration zone. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is not less than the far-distance threshold is designated as the far-distance calibration zone. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is between the near-distance threshold and the far-distance threshold is designated as the medium-distance calibration zone.

[0013] Furthermore, for any calibration group, the method for preliminary calibration of the vehicle body sensor parameters is as follows: a target is randomly placed in each target test area in the calibration group, the coordinate measurement values ​​of the center of each target are obtained based on the vehicle body sensor, and the vehicle body sensor parameters are preliminarily calibrated by combining the actual coordinate values ​​of the center of each target.

[0014] Furthermore, the logic for dividing the target test area in the calibration group into normal target test area or abnormal target test area based on the elimination method is as follows: Based on the coordinate measurement values ​​of the center of each target in the calibration group output by the vehicle body sensor after preliminary calibration, and combined with the true coordinate values, the measurement deviation of each target is determined. The target corresponding to the maximum measurement deviation is taken as the evaluation target. The other target test areas in the calibration group, except for the target test area where the evaluation target is located, are taken as normal target test areas. Based on the relative reduction of the average measurement deviation of the elimination method compared with the average measurement deviation, it is determined whether the target test area where the evaluation target is located is an abnormal target test area.

[0015] Furthermore, for any target, the calculation logic for its measurement deviation is as follows: extract the coordinate measurement value output by the vehicle body sensor after the target center has been preliminarily calibrated, and calculate the Euclidean distance between it and the true coordinate value of the target center, which is used as the measurement deviation of the target.

[0016] The calculation logic for the mean measurement deviation is as follows: the measurement errors of all targets within the same calibration group are averaged to obtain the mean measurement error.

[0017] The calculation logic for the average measurement deviation of the first row is as follows: the measurement errors of other targets in the same calibration group, excluding the evaluation target, are averaged to obtain the average measurement error of the first row.

[0018] The calculation logic for the relative reduction is as follows: calculate the difference between the mean measurement deviation and the mean measurement deviation of the first-ranked unit, and use the ratio of this difference to the mean measurement deviation as the relative reduction.

[0019] Furthermore, the logic for determining abnormal target test areas is as follows: a relative reduction threshold is preset. If the relative reduction is greater than the relative reduction threshold, the target test area where the evaluation target is located is regarded as an abnormal target test area; otherwise, the target test area where the evaluation target is located is regarded as a normal target test area.

[0020] Furthermore, the mathematical expression for the measurement accuracy deviation is as follows:

[0021]

[0022] In the formula, Δd i,j When applying the j-th set of vehicle body sensor calibration parameters to the target in the i-th target test area, the Euclidean distance Δd between the measured value and the true value of the target center coordinates output by the vehicle body sensor is... max =max{Δd i,j |i∈[1,n],j∈[1,m]} represents the maximum Euclidean distance, Δd min =min{Δd i,j |i∈[1,n],j∈[1,m]} represents the minimum Euclidean distance, i is the index of the target test area within the vehicle calibration area, n is the number of target test areas within the vehicle calibration area, j is the group index of the vehicle body sensor calibration parameters, m is the total number of groups of vehicle body sensor calibration parameters, and ε i,j The measurement accuracy deviation when applying the calibration parameters of the j-th group of vehicle body sensors to the i-th target test area.

[0023] Furthermore, a genetic algorithm is used to determine the optimal vehicle body sensor calibration parameters for each target test area. The specific steps are as follows:

[0024] 1) Randomly generate multiple sets of calibration parameter configuration schemes as individuals in the initial population. Each individual corresponds to a pairing combination of calibration parameters for each target test area and each vehicle body sensor, represented as: A k ={a i,k |i∈[1,n]},A k Let a represent the k-th individual. i,k This represents the group of vehicle body sensor calibration parameters paired with the i-th target test area in the k-th individual, where k is the index of the individual in the population, and a i,k=1,2,3,…,m respectively represent that in the k-th individual, the i-th target test area is paired with the first group of vehicle body sensor calibration parameters, the second group of vehicle body sensor calibration parameters, the third group of vehicle body sensor calibration parameters, …, the m-th group of vehicle body sensor calibration parameters;

[0025] 2) Based on the optimization objective of minimizing measurement accuracy deviation, the first fitness value of each individual is calculated, and its mathematical expression is as follows:

[0026]

[0027] In the formula, Fit1 k Let be the first fitness value of the k-th individual. To apply the scheme of the kth individual, the mean measurement accuracy deviation of all target test areas, f k To apply the scheme of the kth individual, the percentage of target test areas with substandard measurement accuracy is determined, where f0 is a preset percentage threshold and γ is the logistic function. The slope is preset, and ω1 and ω2 are preset weight coefficients. The specific values ​​of the two are determined by the analytic hierarchy process, and ω1+ω2=1.

[0028] 3) Based on the optimization objective of maximizing the universality of the vehicle body sensor calibration parameters, the second fitness value of each individual is calculated, and its mathematical expression is as follows:

[0029]

[0030] In the formula, Fit2 k Let n be the second fitness value of the k-th individual. k Q represents the total number of groups of vehicle body sensor calibration parameters that appear in the scheme of the k-th individual. k,q This represents the number of target test areas using the q-th group of vehicle body sensor calibration parameters within the scheme of the k-th individual, where q is the index of the vehicle body sensor calibration parameter group within the individual. k,max =max{Q k,q |q∈[1,n k ]} represents the maximum number of target test areas using each group of vehicle body sensor calibration parameters within the scheme of the k-th individual, T0 is the preset proportional threshold, and δ is the Logistic function. The preset slopes, ω3, ω4 and ω5 are preset weight coefficients, and the specific values ​​of the three are determined by the analytic hierarchy process, and ω3+ω4+ω5=1;

[0031] 5) Determine the weighted weights of the first fitness value and the second fitness value based on the analytic hierarchy process (AHP), and then multiply the first fitness value and the second fitness value of each individual by their respective weighted weights and sum them up to obtain the comprehensive fitness value of each individual.

[0032] 6) Select superior individuals from the population based on the comprehensive fitness value as parents, perform crossover and mutation operations on the parents to generate offspring, combine the parents and offspring to form a new population and recalculate the comprehensive fitness value of each individual, repeat the selection, crossover and mutation operations until the predetermined number of iterations threshold is reached, and determine the optimal vehicle body sensor calibration parameters for each target test area based on the individual with the largest comprehensive fitness value. When performing the mutation operation, perform mutation on the parent individuals based on the preset mutation probability. The mutation operation is to randomly mutate the gene values ​​in the parent individuals to integer values ​​between 1 and m.

[0033] Furthermore, the logic for determining whether the measurement accuracy of the target test area meets the standard is as follows: a distance threshold is preset. If, after applying the calibration parameters of the vehicle body sensor paired with the k-th individual, the Euclidean distance between the measured value of the target center coordinates and the true value of the coordinates in the i-th target test area is not less than the distance threshold, then it is determined that the measurement accuracy of the i-th target test area does not meet the standard when applying the scheme of the k-th individual; otherwise, it is determined that it meets the standard.

[0034] A calibration coordinate system positioning module is applied to the above-mentioned vehicle body sensor calibration method. The calibration coordinate system positioning module consists of two vehicle body positioning units installed one by one at the two rear wheel hubs of the vehicle. The vehicle body positioning unit includes an axle balance module and a laser ranging module.

[0035] The axle balance module includes a fixed plate, a bearing, and a balance shaft. The fixed plate is detachably fixed to the rear wheel hub of the vehicle by three positioning bolts. The bearing is fixed in the center hole of the fixed plate by a first snap ring, and the balance shaft is fixed in the inner ring of the bearing by a second snap ring.

[0036] The laser ranging module includes a laser rangefinder base fixed to the mounting plate with screws, several laser ranging heads fixed to the laser rangefinder base with screws, a wireless transmission module fixed to the laser rangefinder base with adhesive backing and communicating with the laser ranging heads, and a laser transmitter fixed to the laser rangefinder base with screws and emitting laser light.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] The vehicle body sensor calibration method and the calibration coordinate system positioning module used in this invention achieve the technical effect of quickly establishing a vehicle calibration coordinate system by detachably installing the calibration coordinate system positioning module at the rear wheel hub of the vehicle. This eliminates the need for calibration operations in a professional calibration room, greatly improving the convenience and efficiency of vehicle body sensor calibration. Furthermore, by performing preliminary calibration in different regions within the vehicle calibration area, and then performing a secondary calibration after eliminating abnormal target test areas based on the preliminary calibration results, the fit between the target test areas and the calibration results is improved. On this basis, an optimization operation is performed by integrating measurement accuracy requirements and application universality requirements to find the optimal vehicle body sensor calibration parameters for each target test area. The vehicle body sensor calibration parameters are then adaptively set for each target test area to ensure that each area can maintain good measurement accuracy. Attached Figure Description

[0039] Figure 1 is a schematic diagram of the overall method flow of the present invention;

[0040] Figure 2 is a schematic diagram of the curve of the average number of iterations and the overall fitness value in this scheme. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0042] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0043] Example:

[0044] Please refer to Figures 1-2. This invention provides a method for calibrating a vehicle body sensor, comprising the following steps:

[0045] S1, Based on the calibration coordinate system positioning module that can be detachably installed at the rear wheel hub of the vehicle, a vehicle calibration coordinate system with the rear axle of the vehicle as the Y-axis is established, and the X-axis is perpendicular to the Y-axis in the horizontal direction and passes through the center of the vehicle body;

[0046] The calibration coordinate system positioning module consists of two body positioning units installed one by one at the two rear wheel hubs of the vehicle. The body positioning unit includes an axle balance module and a laser ranging module. The detachable calibration coordinate system positioning module can quickly locate the vehicle and determine the vehicle calibration coordinate system without having to send the vehicle to a specific calibration room, which greatly improves the calibration efficiency of the body sensors.

[0047] The axle balance module includes a fixed plate, a bearing, and a balance shaft. The fixed plate is detachably fixed to the rear wheel hub of the vehicle by three positioning bolts. Specifically, the fixed plate has three slots for the positioning bolts to pass through. After removing the three bolts on the rear wheel, the three positioning bolts are passed through the rear wheel hub and the fixed plate in sequence, and the fixed plate is fixed to the rear wheel hub of the vehicle with nuts. The bearing is fixed in the center hole of the fixed plate by a first retaining ring, and the balance shaft is fixed in the inner ring of the bearing by a second retaining ring.

[0048] The laser ranging module includes a laser rangefinder base fixed to the fixing plate by screws, several laser ranging heads fixed to the laser rangefinder base by screws, a wireless transmission module fixed to the laser rangefinder base by adhesive and communicating with the laser ranging heads, and a laser transmitter fixed to the laser rangefinder base by screws and emitting laser light. Preferably, there are 4 laser ranging heads, which are evenly installed in a ring on the laser rangefinder base.

[0049] The method for establishing the vehicle calibration coordinate system is as follows:

[0050] 1) After the vehicle has come to a complete stop, remove the three lug nuts on the rear wheel and install the calibration coordinate system positioning module at the rear wheel hub.

[0051] 2) Fix the wire harness of the plumb bob to the balance shaft and adjust the installation position of the fixing plate in real time until the plumb bob stops, indicating that the shaft balance module has been placed horizontally.

[0052] 3) Place a baffle directly in front of each of the two rear wheel hubs of the vehicle. Emmit a laser beam through a laser emitter and measure the distance between the laser ranging module and the baffle using a laser ranging head. Adjust the positions of the two baffles until the distances from the center points of the two baffles to the two laser ranging modules are the same. Connect the center points of the two baffles to construct a baseline segment parallel to the Y-axis. Take the midpoint of the baseline segment as a reference point and draw a straight line through the reference point and parallel to the Y-axis as the X-axis passing through the center point of the vehicle. This will construct the vehicle calibration coordinate system.

[0053] The positive directions of the X-axis and Y-axis can be set according to the actual situation. For example, the direction from the rear axle to the center of the vehicle can be taken as the positive direction of the X-axis, and the direction from the left rear wheel to the right rear wheel can be taken as the positive direction of the Y-axis. There are no restrictions here.

[0054] S2, define the vehicle calibration area, and divide the vehicle calibration area into three calibration zones: near distance, medium distance and far distance, according to the order of the distance between the vehicle and the sensors from near to far. Each calibration zone is further divided into several target test areas.

[0055] It should be noted that the specific vehicle calibration area is set by the staff according to the actual situation. For example, for a vehicle sensor such as a lidar that detects the position of obstacles around the vehicle, the area around the vehicle can be 1 meter to 50 meters as the vehicle calibration area. The size of the target test area can be set to 1 meter × 1 meter, 1.5 meters × 1.5 meters, 1 meter × 2 meters, etc., to facilitate the subsequent placement of targets for calibration of the vehicle sensor.

[0056] The logic for dividing each calibration zone is as follows: a near-distance threshold and a far-distance threshold are preset, and the near-distance threshold is less than the far-distance threshold. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is no greater than the near-distance threshold is designated as the near-distance calibration zone. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is no less than the far-distance threshold is designated as the far-distance calibration zone. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is between the near-distance threshold and the far-distance threshold is designated as the medium-distance calibration zone.

[0057] It should be noted that when the vehicle body sensor measures the coordinates of objects at different distances, the measurement accuracy varies due to factors such as signal attenuation and reflection characteristics. Using the same sensor parameters may lead to problems where the coordinate measurement accuracy is insufficient in some areas. In order to fully improve the measurement accuracy of the vehicle body sensor, we first divide it into three calibration zones: short distance, medium distance, and long distance. We will then further group them in detail to calibrate the vehicle body sensor separately, so as to ensure that the vehicle body sensor calibration parameters that best fit the calibration group can be found when calibrating the vehicle body sensor in each calibration group.

[0058] As one implementation method, the specific values ​​of the near-distance threshold and the far-distance threshold are set by the staff according to the actual situation. For example, the specific value of the near-distance threshold can be set between 5 meters and 8 meters, and the specific value of the far-distance threshold can be set between 25 meters and 30 meters. There are no restrictions here.

[0059] S3. For each calibration zone, the target test area within it is randomly divided into several calibration groups. The preliminary calibration of the vehicle body sensor parameters is performed in each calibration group. Based on the elimination method, the target test area in the calibration group is divided into normal target test area or abnormal target test area. The normal target test areas in the same calibration group are summarized as a target test area set.

[0060] It should be noted that the number of calibration groups can generally be set to 5-10 groups. The number of target test areas in each calibration group accounts for 10%-20% of the total number of target test areas in the corresponding calibration partition. This avoids the problem of excessive time spent on subsequent comparison processes due to too many calibration groups, and also avoids the problem of complex and difficult single calibration process due to too many targets in a single calibration.

[0061] For any calibration group, the method for preliminary calibration of the vehicle body sensor parameters is as follows: a target is randomly placed in each target test area in the calibration group, the coordinate measurement values ​​of the center of each target are obtained based on the vehicle body sensor, and the vehicle body sensor parameters are preliminarily calibrated by combining the actual coordinate values ​​of the center of each target.

[0062] It should be noted that the parameters of the vehicle body sensor include, but are not limited to, gain and bias. Gain refers to the ratio between the output signal and the input signal of the vehicle body sensor, which reflects the sensitivity of the vehicle body sensor to changes in the input signal. Bias refers to the non-zero value of the output of the vehicle body sensor under zero input conditions, that is, the system error of the output of the vehicle body sensor. Calibrating the parameters of the vehicle body sensor based on the coordinate measurement value and the true coordinate value is an existing technology. Specifically, the optimization objective is to minimize the positional deviation between the coordinate measurement value and the true coordinate value. The parameters of the vehicle body sensor are iteratively adjusted based on the least squares method. Specifically, the positional deviation between the coordinate measurement value and the true coordinate value can be characterized by the Euclidean distance between the two. The specific calibration process of the vehicle body sensor is an existing technology and will not be described in detail here.

[0063] The logic for dividing the target test area in the calibration group into normal target test area or abnormal target test area based on the elimination method is as follows: Based on the coordinate measurement value of the center of each target in the calibration group after the initial calibration of the vehicle body sensor, and combined with the true coordinate value, the measurement deviation of each target is determined. The target corresponding to the maximum measurement deviation is taken as the evaluation target. The other target test areas in the calibration group except the target test area where the evaluation target is located are taken as normal target test areas. Based on the relative reduction of the average measurement deviation of the elimination method compared with the average measurement deviation, it is determined whether the target test area where the evaluation target is located is an abnormal target test area.

[0064] It should be noted that, in order to ensure the universality of the vehicle body sensor parameter calibration and improve the calibration efficiency, it is assumed that there is at most one target test area in a calibration group as an abnormal target test area. Therefore, it is assumed that other target test areas in the calibration group other than the target test area where the evaluation target is located are normal target test areas. Only the target test area where the evaluation target is located is judged as an abnormal target test area. In addition, if there are two or more targets in the same calibration group with the maximum measurement deviation, these targets are all regarded as evaluation targets and their measurement deviations are excluded when calculating the measurement deviation in the subsequent calculation.

[0065] It should be noted that when using the vehicle body sensor to measure the position coordinates of the target center, the input data of the vehicle body sensor is recorded simultaneously. For example, for the LiDAR vehicle body sensor, the input data can be the distance and relative angle between the target center and the vehicle body sensor. Later, when the calibrated vehicle body sensor outputs the coordinate measurement values ​​of each target center in the calibration group again, the input data of the vehicle body sensor when measuring the coordinates of each target center can be directly called, without the need for re-measurement, which greatly improves the calibration efficiency.

[0066] Furthermore, for any target, the calculation logic for its measurement deviation is as follows: extract the coordinate measurement value output by the vehicle body sensor after the target center has been preliminarily calibrated, and calculate the Euclidean distance between it and the true coordinate value of the target center, which is used as the measurement deviation of the target.

[0067] Furthermore, the calculation logic for the mean measurement deviation is as follows: the measurement errors of all targets within the same calibration group are averaged to obtain the mean measurement error.

[0068] Furthermore, the calculation logic for the average measurement deviation of the first row is as follows: the measurement errors of other targets in the same calibration group, excluding the evaluation target, are averaged to obtain the average measurement error of the first row.

[0069] Furthermore, the calculation logic for the relative reduction is as follows: calculate the difference between the mean measurement deviation and the mean measurement deviation of the first group, and take the ratio of this difference to the mean measurement deviation as the relative reduction. The larger the value, the more significantly the coordinate measurement accuracy of the pre-calibrated vehicle body sensor for other targets in the same calibration group can be improved after removing the evaluation target. This indicates that the calibrated vehicle body sensor parameters perform well on other targets in the same calibration group, but poorly on the evaluation target. In other words, the pre-calibrated vehicle body sensor parameters may not be applicable to the target test area where the evaluation target is located.

[0070] The logic for determining abnormal target test areas is as follows: a relative reduction threshold is preset. If the relative reduction is greater than the relative reduction threshold, it means that the initially calibrated vehicle body sensor parameters are not applicable to the target test area where the evaluation target is located. In this case, the target test area where the evaluation target is located is regarded as an abnormal target test area. Otherwise, the target test area where the evaluation target is located is regarded as a normal target test area. The specific value of the relative reduction threshold can generally be set between 5% and 10%, and the specific setting is determined by the staff according to the actual situation. No restrictions are imposed here.

[0071] S4. When there is an abnormal target test area in the calibration group, the vehicle body sensor is calibrated a second time based on the target test area set to determine the vehicle body sensor calibration parameters applied to the target test area set. Otherwise, the preliminary calibration result is used as the vehicle body sensor calibration parameters for the target test area set.

[0072] The method for secondary calibration of the vehicle body sensor for any target test area set is as follows: based on the coordinate measurement values ​​of the center of each target in the target test area set output by the vehicle body sensor after preliminary calibration, and combined with the true coordinate values ​​of each target center, the vehicle body sensor parameters are calibrated for the second time. Specifically, with the optimization objective of minimizing the positional deviation between the coordinate measurement value and the true coordinate value, the preliminary calibrated vehicle body sensor parameters are iteratively adjusted based on the least squares method. The specific calibration process is not described in detail here. After removing abnormal target test areas, secondary calibration is performed to make the calibration results more consistent with the normal target test areas in the target test area set. In addition, iterative optimization is performed on the basis of preliminary calibration, which greatly reduces the amount of optimization calculation.

[0073] It should be noted that the preliminary calibration result is the preliminary calibration of the vehicle body sensor parameters. If a calibration group does not have any abnormal target test areas, then this calibration group is a target test area set. The preliminary calibration of the vehicle body sensor parameters obtained based on this calibration group is directly used as the vehicle body sensor calibration parameters of the target test area set.

[0074] S5. For each target test area, calculate the measurement accuracy deviation of each set of vehicle body sensor calibration parameters, and determine the optimal vehicle body sensor calibration parameters for each target test area based on the optimization algorithm with the optimization objectives of minimizing the measurement accuracy deviation and maximizing the universality of the application of vehicle body sensor calibration parameters.

[0075] The mathematical expression for the measurement accuracy deviation is as follows:

[0076]

[0077] In the formula, Δd i,j When applying the j-th set of vehicle body sensor calibration parameters to the target in the i-th target test area, the Euclidean distance Δd between the measured value and the true value of the target center coordinates output by the vehicle body sensor is... max =max{Δd i,j |i∈[1,n],j∈[1,m]} is the maximum Euclidean distance, max{Δd i,j |i∈[1,n],j∈[1,m]} represents the logarithmic sequence {Δd} i,j The element in |i∈[1,n],j∈[1,m]} is taken as the maximum value, Δd min =min{Δd i,j |i∈[1,n],j∈[1,m]} is the minimum Euclidean distance, min{Δd i,j |i∈[1,n],j∈[1,m]} represents the logarithmic sequence {Δd} i,jThe minimum value is taken from the elements in |i∈[1,n],j∈[1,m]}, where i is the index of the target test area within the vehicle calibration area, n is the number of target test areas within the vehicle calibration area, j is the group index of the body sensor calibration parameters, and m is the total number of groups of body sensor calibration parameters, whose value is consistent with the number of target test area sets, ε i,j The measurement accuracy deviation when applying the calibration parameters of the j-th group of vehicle body sensors to the i-th target test area is expressed as Δd by the maximum-minimum normalization method. i,j The larger the value obtained by scaling, the less suitable the i-th target test area is for the j-th group of vehicle body sensor calibration parameters, that is, the worse the accuracy of outputting the coordinate position of the object in the i-th target test area using the j-th group of vehicle body sensor calibration parameters.

[0078] As one implementation method, a genetic algorithm is used to determine the optimal vehicle body sensor calibration parameters for each target test area. The specific steps are as follows:

[0079] 1) Randomly generate multiple sets of calibration parameter configuration schemes as individuals in the initial population. Each individual corresponds to a pairing combination of calibration parameters for each target test area and each vehicle body sensor, represented as: A k ={a i,k |i∈[1,n]},A k Let a represent the k-th individual. i,k This represents the group of vehicle body sensor calibration parameters paired with the i-th target test area in the k-th individual, where k is the index of the individual in the population, and a i,k =1,2,3,…,m respectively represent that in the k-th individual, the i-th target test area is paired with the first group of vehicle body sensor calibration parameters, the second group of vehicle body sensor calibration parameters, the third group of vehicle body sensor calibration parameters, …, the m-th group of vehicle body sensor calibration parameters;

[0080] 2) Based on the optimization objective of minimizing measurement accuracy deviation, the first fitness value of each individual is calculated, and its mathematical expression is as follows:

[0081]

[0082] In the formula, Fit1 k Let be the first fitness value of the k-th individual. To apply the scheme of the kth individual, the mean deviation of measurement accuracy across all target test areas is used. A larger value indicates a worse scheme for the kth individual, making it more difficult to meet the measurement accuracy requirements of the target test areas. Therefore, this is... As an evaluation index of the first fitness value, the larger the value, the smaller the first fitness value, which indicates that the solution of the kth individual is worse.

[0083] In the formula, f k To determine the percentage of target test areas with substandard measurement accuracy when applying the scheme of the k-th individual, specifically, it is the ratio of the number of target test areas with substandard measurement accuracy when applying the scheme of the k-th individual to the total number of target test areas in the vehicle calibration area. A larger value indicates a greater number of target test areas with substandard measurement accuracy when applying the scheme of the k-th individual, meaning a worse scheme for the k-th individual and less likely to meet the measurement accuracy requirements of the target test areas. Therefore, f is used here... k As another evaluation index of the first fitness value, the larger the value, the smaller the first fitness value, which indicates that the solution of the kth individual is worse.

[0084] The logic for determining whether the measurement accuracy of the target test area meets the standard is as follows: a distance threshold is preset. If the Euclidean distance between the measured value of the target center coordinate and the true value of the coordinate is not less than the distance threshold after applying the calibration parameters of the vehicle body sensor paired with the k individual in the i-th target test area, it indicates that the measurement accuracy of the i-th target test area is poor and does not meet the requirements. Therefore, it is determined that the measurement accuracy of the i-th target test area does not meet the standard when applying the scheme of the k individual. Otherwise, it is determined that it meets the standard.

[0085] It should be noted that the smaller the distance threshold value is set, the higher the requirement for measurement accuracy, and vice versa. The specific value can be set by the staff according to the actual situation, such as between 10 cm and 30 cm, without any restrictions.

[0086] In the formula, f0 is a preset percentage threshold, the specific value of which is determined by the required coordinate measurement accuracy. The higher the accuracy requirement, the smaller the percentage threshold, and vice versa. It is generally set between 3% and 8%. When the percentage of target test areas with substandard measurement accuracy is lower than the preset percentage threshold, it indicates that the number of target test areas with substandard measurement accuracy is within acceptable limits. Conversely, when the percentage of target test areas with substandard measurement accuracy is not lower than the preset percentage threshold, it indicates that the number of target test areas with substandard measurement accuracy is not within acceptable limits. Furthermore, as the number of target test areas with substandard measurement accuracy increases, the unacceptability of the individual solution also increases significantly. Therefore, the Logistic function form is used here. To characterize as f k Increase or decrease the first fitness value Fit1 k The trend, and highlight the characteristics of f k If the proportion exceeds the threshold, quickly reduce the first fitness value (Fit1). k The trend, where γ is the Logistic function. The preset slope is used to control the Logistic function. The smaller the value of , the better the growth rate of the Logistic function. The more gradual the growth, the larger the value of the Logistic function. The steeper the growth, the more sensitive the target test area is to the number of targets that do not meet the accuracy requirements, as determined by experts. The larger the value, the more sensitive the target. The specific value is generally set between 0.1 and 5, and no limit is imposed here.

[0087] In the formula, ω1 and ω2 are both preset weighting coefficients, used to characterize... The importance of the first fitness value is determined by the analytic hierarchy process, with a larger value indicating greater importance. The specific values ​​of both are determined by the analytic hierarchy process, and ω1 + ω2 = 1.

[0088] 3) Based on the optimization objective of maximizing the universality of the vehicle body sensor calibration parameters, the second fitness value of each individual is calculated, and its mathematical expression is as follows:

[0089]

[0090] In the formula, Fit2 k Let n be the second fitness value of the k-th individual. k This represents the total number of vehicle body sensor calibration parameter groups appearing in the scheme of the k-th individual. A larger value indicates that the scheme of the k-th individual uses more groups of vehicle body sensor calibration parameters, thus indicating lower universality of the vehicle body sensor calibration parameters in the k-th individual's scheme. This means that more groups of vehicle body sensor calibration parameters are needed to adapt to different target test areas. Therefore, n is used here... k As an evaluation metric for the second fitness value, through The form of n k Scale to between 0 and 1, and via To characterize as n k The trend is that as the fitness value increases, the second fitness value decreases.

[0091] In the formula, Q k,q This represents the number of target test areas using the q-th group of vehicle body sensor calibration parameters within the scheme of the k-th individual. A larger value indicates a wider application of the q-th group of vehicle body sensor calibration parameters; therefore, Q is used here. k,q As an evaluation metric for the second fitness value, through The applicability of the calibration parameters of the qth group of vehicle body sensors is scaled to between 0 and 1, and then... The value is used to characterize the average applicability of all groups of vehicle body sensor calibration parameters within the scheme of the kth individual. The larger the value, the higher the universality of the vehicle body sensor calibration parameters within the scheme of the kth individual. q is the index of the group of vehicle body sensor calibration parameters within the individual.

[0092] In the formula, Q k,max =max{Q k,q |q∈[1,n k Q represents the maximum number of target test areas using each group of vehicle body sensor calibration parameters within the scheme of the k-th individual. A larger Q value indicates higher applicability of the dominant vehicle body sensor calibration parameters within the scheme of the k-th individual. Therefore, Q is defined here. k,max As an evaluation index of the second fitness value, the larger the value, the larger the second fitness value.

[0093] In the formula, T0 is a preset proportional threshold. Its specific value is determined by the universality requirements of the application of the vehicle body sensor calibration parameters. The higher the universality requirements for the application of the vehicle body sensor calibration parameters, the larger the proportional threshold value, and vice versa. It is generally set between 20% and 40%. When the ratio is below the preset threshold, it indicates that the universality requirement for the application of vehicle body sensor calibration parameters has not been met. When the ratio is not lower than the preset threshold, it indicates that the universality requirement for the application of vehicle body sensor calibration parameters has been met, and as... The increased number of parameters significantly improves the applicability of vehicle body sensor calibration parameters in individual solutions; therefore, a Logistic function is used here. To characterize as Increase, increase the second fitness value Fit2 k The trend, and highlight the characteristics of the trend as If the fitness value exceeds the proportional threshold, rapidly increase the second fitness value (Fit2). k The trend, δ is the Logistic function The preset slope is used to control the Logistic function. The smaller the value of , the better the growth rate of the Logistic function. The more gradual the growth, the larger the value of the Logistic function. The steeper the growth, the more sensitive the value will be, depending on the sensitivity of the calibration parameters of the vehicle body sensor, which is determined by experts based on the general requirements of the application. The larger the value, the more sensitive the sensor will be. The specific value is generally set between 0.1 and 2, and no restrictions are imposed here.

[0094] In the formula, ω3, ω4, and ω5 are all preset weighting coefficients, used to characterize... and The importance of the second fitness value is determined by the analytic hierarchy process, with a larger value indicating greater importance. The specific values ​​of the three are determined by the analytic hierarchy process, and ω3+ω4+ω5=1.

[0095] 5) The weighted weights of the first fitness value and the second fitness value are determined based on the analytic hierarchy process (AHP). The first fitness value and the second fitness value of each individual are multiplied by their respective weights and then summed to obtain the comprehensive fitness value of each individual. This balances the two optimization objectives of minimizing measurement accuracy deviation and maximizing the universality of the application of vehicle body sensor calibration parameters. The larger the comprehensive fitness value of an individual, the better the individual is. The weighted weights of the first fitness value and the weighted weights of the second fitness value satisfy the constraint that the sum is 1.

[0096] 6) Select superior individuals from the population based on the comprehensive fitness value as parents, perform crossover and mutation operations on the parents to generate offspring, combine the parents and offspring to form a new population and recalculate the comprehensive fitness value of each individual, repeat the selection, crossover and mutation operations until the predetermined number of iterations threshold is reached, and determine the optimal vehicle body sensor calibration parameters for each target test area based on the individual with the largest comprehensive fitness value. When performing the mutation operation, perform mutation on the parent individuals based on the preset mutation probability. The mutation operation is to randomly mutate the gene values ​​in the parent individuals to integer values ​​between 1 and m to ensure the rationality of the mutated individuals.

[0097] It should be noted that the specific value of the mutation probability is within the conventional range of 0.01-0.1, and the specific value is selected by the staff according to the actual situation. The specific value of the iteration number threshold is also determined by the staff according to the actual situation, and it can generally be between 100-500. When performing the selection operation, conventional selection methods such as roulette wheel selection, tournament selection, or ranking selection can be used to select superior individuals as parents. When performing the crossover operation, conventional crossover methods such as single-point crossover, multi-point crossover, or uniform crossover can be selected. These are conventional techniques in the application of genetic algorithms and will not be elaborated here.

[0098] Furthermore, during the iterative optimization process, the average comprehensive fitness value after each iteration is calculated. For any given iteration, the corresponding average comprehensive fitness value is the average comprehensive fitness value of each individual in the newly generated population after that iteration. The iterative optimization is terminated early when the increasing trend of the average comprehensive fitness value is not obvious, thus avoiding unnecessary waste of computational resources. As shown in Figure 2, in the early stage of iterative optimization, the average comprehensive fitness value after each iteration increases with the number of iterations. However, around the 400th iteration, the increasing trend of the average comprehensive fitness value tends to level off, indicating that further iterative optimization at this point is not very meaningful. Therefore, the iteration can be terminated early to avoid unnecessary waste of computational resources.

[0099] Therefore, the above scheme determines the optimal vehicle body sensor calibration parameters for each target test area, taking into account both the measurement accuracy requirements and the universality of the application of the vehicle body sensor calibration parameters. When performing subsequent external object position coordinate measurements, the preliminary position coordinate measurement is first performed based on the most widely used vehicle body sensor calibration parameters among the individuals with the highest comprehensive fitness values ​​to determine the relative position between the external object and the vehicle body sensor. Then, it is determined which target test area this relative position corresponds to, and the corresponding vehicle body sensor calibration parameters are selected to output accurate position coordinates. This achieves targeted setting of vehicle body sensor calibration parameters to improve the measurement accuracy of position coordinates.

[0100] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0101] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0102] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0103] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for calibrating a vehicle body sensor, characterized in that, The process includes the following steps: S1, establishing a vehicle calibration coordinate system with the rear axle as the Y-axis, based on a detachable calibration coordinate system positioning module installed at the rear wheel hub, with the X-axis perpendicular to the Y-axis and passing through the center of the vehicle body; S2, defining the vehicle calibration area and dividing it into three calibration zones—near-distance, medium-distance, and far-distance—according to the order of distance from the vehicle body sensors from near to far, and further dividing each calibration zone into several target test areas; S3, for each calibration zone, randomly dividing the target test areas within it into several calibration groups, performing preliminary calibration of the vehicle body sensor parameters in each calibration group, and classifying the target test areas in the calibration groups into normal target test areas based on the elimination method. S4. When abnormal target test areas exist in the calibration group, the vehicle body sensors are calibrated a second time based on the target test area set to determine the vehicle body sensor calibration parameters applicable to the target test area set. Otherwise, the preliminary calibration results are used as the vehicle body sensor calibration parameters for the target test area set. S5. For each target test area, the measurement accuracy deviation of each group of vehicle body sensor calibration parameters is calculated. With the optimization objectives of minimizing the measurement accuracy deviation and maximizing the universality of the application of vehicle body sensor calibration parameters, the optimal vehicle body sensor calibration parameters applicable to each target test area are determined based on the optimization algorithm.

2. The vehicle body sensor calibration method according to claim 1, characterized in that, The logic for dividing each calibration zone is as follows: a near-distance threshold and a far-distance threshold are preset, and the near-distance threshold is less than the far-distance threshold. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is no greater than the near-distance threshold is designated as the near-distance calibration zone. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is no less than the far-distance threshold is designated as the far-distance calibration zone. The area within the vehicle calibration zone where the distance between the vehicle and the body sensor is between the near-distance threshold and the far-distance threshold is designated as the medium-distance calibration zone.

3. The vehicle body sensor calibration method according to claim 1, characterized in that, For any calibration group, the method for preliminary calibration of the vehicle body sensor parameters is as follows: a target is randomly placed in each target test area in the calibration group, the coordinate measurement values ​​of the center of each target are obtained based on the vehicle body sensor, and the vehicle body sensor parameters are preliminarily calibrated by combining the actual coordinate values ​​of the center of each target.

4. The vehicle body sensor calibration method according to claim 1, characterized in that, The logic for dividing the target test area in the calibration group into normal target test area or abnormal target test area based on the elimination method is as follows: Based on the coordinate measurement values ​​of the center of each target in the calibration group output by the vehicle body sensor after preliminary calibration, and combined with the true coordinate values, the measurement deviation of each target is determined. The target corresponding to the maximum measurement deviation is taken as the evaluation target. The other target test areas in the calibration group except for the target test area where the evaluation target is located are taken as normal target test areas. Based on the relative decrease of the average measurement deviation of the elimination method compared with the average measurement deviation, it is determined whether the target test area where the evaluation target is located is an abnormal target test area.

5. The vehicle body sensor calibration method according to claim 4, characterized in that, For any target, the calculation logic for its measurement deviation is as follows: extract the coordinate measurement value output by the vehicle body sensor after preliminary calibration of the target center, and calculate the Euclidean distance between it and the true coordinate value of the target center, which is used as the measurement deviation of the target; the calculation logic for the mean measurement deviation is as follows: average the measurement errors of all targets in the same calibration group to obtain the mean measurement error; the calculation logic for the mean measurement deviation of the first-ranked target is as follows: average the measurement errors of other targets in the same calibration group except for the evaluation target to obtain the mean measurement error of the first-ranked target; the calculation logic for the relative reduction is as follows: calculate the difference between the mean measurement deviation and the mean measurement deviation of the first-ranked target, and use the ratio of this difference to the mean measurement deviation as the relative reduction.

6. The vehicle body sensor calibration method according to claim 4, characterized in that, The logic for determining abnormal target test areas is as follows: a relative reduction threshold is preset. If the relative reduction is greater than the relative reduction threshold, the target test area where the evaluation target is located is regarded as an abnormal target test area; otherwise, the target test area where the evaluation target is located is regarded as a normal target test area.

7. The vehicle body sensor calibration method according to claim 1, characterized in that, The mathematical expression for the measurement accuracy deviation is as follows: In the formula, Δd i,j When applying the j-th set of vehicle body sensor calibration parameters to the target in the i-th target test area, the Euclidean distance Δd between the measured value and the true value of the target center coordinates output by the vehicle body sensor is... max =max{Δd i,j |i∈[1,n],j∈[1,m]} represents the maximum Euclidean distance, Δd min =min{Δd i,j |i∈[1,n],j∈[1,m]} represents the minimum Euclidean distance, i is the index of the target test area within the vehicle calibration area, n is the number of target test areas within the vehicle calibration area, j is the group index of the vehicle body sensor calibration parameters, m is the total number of groups of vehicle body sensor calibration parameters, and ε i,j The measurement accuracy deviation when applying the calibration parameters of the j-th group of vehicle body sensors to the i-th target test area.

8. The vehicle body sensor calibration method according to claim 7, characterized in that, The optimal vehicle body sensor calibration parameters for each target test area are determined using a genetic algorithm. The specific steps are as follows: 1) Randomly generate multiple sets of calibration parameter configuration schemes as individuals in the initial population. Each individual corresponds to a pairing combination of calibration parameters for each target test area and each vehicle body sensor, represented as: A k ={a i,k |i∈[1,n]},A k Let a represent the k-th individual. i,k This represents the group of vehicle body sensor calibration parameters paired with the i-th target test area in the k-th individual, where k is the index of the individual in the population, and a i,k =1,2,3,…,m respectively represent the pairing of the i-th target test area with the first group of vehicle body sensor calibration parameters, the second group of vehicle body sensor calibration parameters, the third group of vehicle body sensor calibration parameters, …, the m-th group of vehicle body sensor calibration parameters in the k-th individual; 2) Based on the optimization objective of minimizing the measurement accuracy deviation, the first fitness value of each individual is calculated, and its mathematical expression is as follows: In the formula, Fit1 k Let be the first fitness value of the k-th individual. To apply the scheme of the kth individual, the mean measurement accuracy deviation of all target test areas, f k To apply the scheme of the kth individual, the percentage of target test areas with substandard measurement accuracy is determined, where f0 is a preset percentage threshold and γ is the logistic function. The preset slope, ω1 and ω2 are preset weight coefficients, and their specific values ​​are determined by the analytic hierarchy process, and ω1 + ω2 = 1; 3) Based on the optimization objective of maximizing the universality of the vehicle body sensor calibration parameters, the second fitness value of each individual is calculated, and its mathematical expression is as follows: In the formula, Fit2 k Let n be the second fitness value of the k-th individual. k Q represents the total number of groups of vehicle body sensor calibration parameters that appear in the scheme of the k-th individual. k,q This represents the number of target test areas using the q-th group of vehicle body sensor calibration parameters within the scheme of the k-th individual, where q is the index of the group of vehicle body sensor calibration parameters within that individual. k,max =max{Q k,q |q∈[1,n k ]} represents the maximum number of target test areas using each group of vehicle body sensor calibration parameters within the scheme of the k-th individual, T0 is the preset proportional threshold, and δ is the Logistic function. The preset slopes, ω3, ω4, and ω5 are preset weight coefficients. The specific values ​​of the three are determined by the analytic hierarchy process (AHP), and ω3 + ω4 + ω5 = 1; 5) Based on the AHP, the weighted weights of the first fitness value and the second fitness value are determined, and the first fitness value and the second fitness value of each individual are multiplied by their respective weighted weights and then summed to obtain the comprehensive fitness value of each individual; 6) Based on the comprehensive fitness value, superior individuals are selected from the population as parents, and crossover and mutation operations are performed on the parents to generate offspring. The parents and offspring are combined to form a new population, and the comprehensive fitness value of each individual is calculated again. The selection, crossover, and mutation operations are repeated until the predetermined number of iterations is reached. Based on the individual with the largest comprehensive fitness value, the optimal vehicle body sensor calibration parameters applied to each target test area are determined. When performing the mutation operation, the parent individuals are mutated based on the preset mutation probability. The mutation operation is to randomly mutate the gene values ​​in the parent individuals to integer values ​​between 1 and m.

9. A vehicle body sensor calibration method according to claim 8, characterized in that, The logic for determining whether the measurement accuracy of the target test area meets the standard is as follows: a distance threshold is preset. If the Euclidean distance between the measured value of the target center coordinate and the true value of the coordinate is not less than the distance threshold after applying the calibration parameters of the vehicle body sensor paired with the k individual in the i-th target test area, then it is determined that the measurement accuracy of the i-th target test area does not meet the standard when applying the scheme of the k individual; otherwise, it is determined that it meets the standard.

10. A calibration coordinate system positioning module, applied to the vehicle body sensor calibration method according to any one of claims 1-9, characterized in that: The calibration coordinate system positioning module consists of two vehicle body positioning units, each installed at one of the rear wheel hubs of the vehicle. Each vehicle body positioning unit includes an axle balance module and a laser ranging module. The axle balance module includes a fixed plate, bearings, and a balance shaft. The fixed plate is detachably fixed to the rear wheel hubs using three positioning bolts. The bearing is fixed to the center hole of the fixed plate by a first retaining ring, and the balance shaft is fixed to the inner ring of the bearing by a second retaining ring. The laser ranging module includes a laser rangefinder base fixed to the fixed plate by screws, several laser ranging heads fixed to the laser rangefinder base by screws, a wireless transmission module fixed to the laser rangefinder base by adhesive and communicating with the laser ranging heads, and a laser emitter fixed to the laser rangefinder base by screws and emitting laser light.

Images