CN116635739A - Road side millimeter wave radar calibration method based on vehicle-mounted positioning device - Google Patents
Road side millimeter wave radar calibration method based on vehicle-mounted positioning device Download PDFInfo
- Publication number
- CN116635739A CN116635739A CN202180010978.4A CN202180010978A CN116635739A CN 116635739 A CN116635739 A CN 116635739A CN 202180010978 A CN202180010978 A CN 202180010978A CN 116635739 A CN116635739 A CN 116635739A
- Authority
- CN
- China
- Prior art keywords
- track data
- vehicle
- millimeter wave
- wave radar
- calibration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 72
- 238000012545 processing Methods 0.000 claims abstract description 88
- 238000001514 detection method Methods 0.000 claims abstract description 24
- 238000005070 sampling Methods 0.000 claims description 69
- 238000012952 Resampling Methods 0.000 claims description 56
- 239000011159 matrix material Substances 0.000 claims description 41
- 238000006243 chemical reaction Methods 0.000 claims description 33
- 238000004364 calculation method Methods 0.000 claims description 28
- 238000012795 verification Methods 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 11
- 238000007405 data analysis Methods 0.000 claims description 9
- 238000009434 installation Methods 0.000 claims description 9
- 230000001502 supplementing effect Effects 0.000 claims description 6
- 238000012888 cubic function Methods 0.000 claims description 5
- 230000007547 defect Effects 0.000 claims description 4
- 230000002123 temporal effect Effects 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 2
- 230000001360 synchronised effect Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 238000004148 unit process Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 8
- 230000007812 deficiency Effects 0.000 description 6
- 230000004927 fusion Effects 0.000 description 5
- 238000013507 mapping Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 4
- 230000008447 perception Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000013480 data collection Methods 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 229920001184 polypeptide Polymers 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 102000004196 processed proteins & peptides Human genes 0.000 description 1
- 108090000765 processed proteins & peptides Proteins 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Abstract
A road side millimeter wave radar calibration method based on a vehicle-mounted positioning device belongs to the technical field of mobile vehicle detection and sensor detection target calibration, and the method uses a calibration vehicle with the vehicle-mounted positioning device to run according to a pre-designed route and collect data, and a processing unit processes and analyzes the collected data to realize the calibration of the road side millimeter wave radar, so that the problem of calibrating the road side millimeter wave radar can be solved.
Description
The invention belongs to the technical field of detection of a mobile vehicle and calibration of a detection target of a sensor, and relates to a method for calibrating a millimeter wave radar installed on a road side by using a vehicle-mounted positioning device.
Under the cooperative background of the vehicle and the road, the distance measuring sensor such as millimeter wave radar and the like can sense the distance and the speed of surrounding vehicles sharply, the image sensor can supplement the multidimensional information of the surrounding vehicles visually, and the sensing capability of the environment can be remarkably improved through the fusion of various sensors. Among them, the advantages of the roadside millimeter wave radar are obvious. The millimeter wave radar has the remarkable advantages that the millimeter wave radar is free from the influence of weather and illumination conditions such as rain, snow, fog and the like, can work around the clock, realizes the detection of target positions and speed heading, and is an important environmental perception means. The road side millimeter wave radar is applied to the traffic field, can complete the functions of speed measurement, distance measurement and angle measurement, realizes multi-target detection and track tracking, and can realize diversified functions of traffic operation abnormal event detection, multi-section traffic flow data statistics, intersection holographic state perception and the like. Meanwhile, under the scene of vehicle-road coordination and automatic driving, the millimeter wave radar installed on the road side can assist a laser radar, a video camera and the like to complete holographic perception of multiple sensors, so that holographic fusion target-level scene semantic establishment is realized.
The intelligent road is focused on the requirements of the millimeter wave radar on two aspects in the data layer, namely, the original point cloud data, namely, the original results of signal analysis processing returned to the radar at different spatial positions when the radar is scanned; on the other hand, the real-time data are real-time data, namely the positions, speeds, heading and sizes of all traffic targets and tracking numbers given to the targets, which are obtained by analyzing the point cloud data by the radar, are fused with other sensing means in space and time after the two types of raw data are given to an edge calculator, so that the high-precision and multidimensional sensing of the traffic environment is realized, and then the sensing result is issued by the RSU.
An important premise for realizing the fusion sensing is data calibration. The calibration of the sensor means that the mapping relation between the relative coordinate system of the equipment and the world coordinate system is established through a certain technical means or method.
In a system for information interaction between vehicles and roads, the accurate positioning and fusion of perceived objects can be carried out only by a unified coordinate system at the two ends of the vehicles and the roads, so that the vehicle end and the road side perception equipment are required to be calibrated to ensure that data obtained at the two ends have a unified reference standard and can be mutually converted. At the same time, the pose of the device may be subjected to non-plastic offset and change under the natural influence (wind vibration and bridge vibration) of the road side or vehicle-mounted sensor. At this point recalibration of the external parameters of the device is required.
The millimeter wave radar sensor maps the detection target coordinates in the world coordinate system to the corresponding relative coordinate system during measurement, but the coordinate mapping relationship is also different due to the different installation positions, attitudes and angles of the radar sensor. Meanwhile, due to the influences of natural bridge vibration, wind vibration and external environment, external parameters of the millimeter wave radar are changed, and timely calibration is needed to meet the requirements of accurate positioning and multi-sensor fusion.
In the existing scheme, the calibration research on the road side millimeter wave radar is less, most of documents describe in-factory calibration of the vehicle millimeter wave radar before delivery and after offline, and the calibration technology of other road side sensing devices such as a laser radar, a camera and the like, and the calibration scheme for the road side millimeter wave radar is less proposed.
Existing calibration methods can be broadly classified into a geometry-based method, a motion-based method, a mutual information-based method, and a deep learning-based method according to the difference of principles. The essence of the method is that features are extracted from different sensors according to a certain type of calibrated sensors, and the calibration is carried out according to the coordinate mapping relation of the extracted features under different coordinate systems, wherein the method is a rapid calibration method which does not depend on a calibration object and is based on feature points (lines and planes), but is quite dependent on other calibrated sensors.
One of the conventional calibration methods of the roadside millimeter wave radar device is by calibration by means of a manual calibration object. For example, in order to calibrate a roadside millimeter wave radar, a roadside camera installed at the same position is first calibrated. And in the same visual field range, paving a calibration chessboard diagram on a road surface in the common visual field range of the video camera and the millimeter wave radar, calibrating the internal parameters and the external parameters of the video camera by adopting a Zhang Zhengyou calibration method and an RTK differential positioning technology, calibrating the internal parameters and the external parameters of the camera by establishing a mapping relation between the pixel coordinates of the video image and the longitude and latitude coordinates of the world, and then completing the calibration of the millimeter wave radar by establishing a relation between the pixel coordinate system of the video image and the radar coordinate system. The method has wide application but obvious defects: the calibration method is too dependent on manpower, is only suitable for calibration in a small range, and is difficult to apply in large-scale batch. And for an online autopilot system this would reduce autonomy of the autopilot system; in addition, other sensors are needed in the scheme, so that the indirect calibration of the millimeter wave radar can be completed. In an actual complex traffic environment, the sensor generates vibration deflection due to external factors such as strong wind, bridge vibration and the like, or changes of the gesture due to artificial rotation, so that the mapping relation of the coordinate system established before is invalid, and the parameters of the sensor need to be calibrated again.
Another conventional approach to roadside millimeter wave radar calibration is to select calibration objects or features in the natural environment. And solving parameters through the one-to-one correspondence between the world coordinate system of the selected calibration object or feature and the feature point pairs under the millimeter wave radar coordinate system. The method has definite requirements on the shape, characteristics and position of the object, and can realize accurate calibration. The method can solve the problem that the traditional calibration method depends on manpower, and can solve the problem of recalibration of the sensor in real time, and different methods achieve high precision under different specific application scenes. However, the method is not suitable for large-scale popularization because of the requirement of a static feature or a calibration object with obvious features on the road side. Therefore, a method with strong practicability is needed to solve the problem of calibrating the millimeter wave radar at the road side.
With the continuous development of vehicle road cooperative technology, the permeability of vehicles with high-precision vehicle-mounted positioning equipment is continuously improved. Due to the requirement of an automatic driving technology, the vehicle-mounted positioning device can achieve centimeter-level positioning precision, and can send self-position information to the cloud in real time. Therefore, the method is used as a calibration data source of the road side video and radar sensing equipment, the cost and difficulty of manual and positioning devices required by calibration can be greatly reduced, and the problem of online calibration of the detection target coordinates of the multi-source sensing equipment is effectively solved.
The prior patent:
patent CN 111929652A
Patent CN 110703254A
U.S. Pat. No. 4,692,772 B2
U.S. patent 2015/00700207 A1
U.S. patent US 2014/0240190 A1
Patent CN 111060881A
Disclosure of Invention
In order to solve the problem of calibrating the roadside millimeter wave radar, the invention adopts the technical method that: a road side millimeter wave radar calibration method based on a vehicle-mounted positioning device. According to the invention, the calibration vehicle with the vehicle-mounted positioning device is used for running according to a pre-designed route and collecting data, and the collected data is processed and analyzed by the processing unit, so that the calibration of the road side millimeter wave radar is realized.
[ PREPARATION ] A method for producing a polypeptide
● According to the invention, the calibration vehicles with the vehicle-mounted positioning devices can be divided into two types according to different vehicle types, one type refers to embedded automatic driving vehicles, the high-precision positioning devices are embedded and installed, the automatic driving vehicles can accurately position the vehicle, and real-time vehicle positioning data are output; the calibration vehicle with the external positioning equipment is provided with a high-precision positioning device, can accurately position the vehicle position and outputs real-time vehicle positioning data.
● The road side millimeter wave radar disclosed by the invention is a millimeter wave radar which is installed on a road, the installation position of the road side millimeter wave radar can be divided into three types according to different actual scenes and different installation positions, namely the road side millimeter wave radar is installed on the left side of the road, the road right side of the road and the center of the road, and the three types of installation schemes are all the road side ranges disclosed by the scheme, have the capability of identifying a moving target vehicle and can output positioning data of a detection target.
● The calibration parameter calibration in the invention refers to the calibration of a radar calibration parameter matrix, namely the correction and update of conversion parameters from a millimeter wave radar coordinate system to a world coordinate system.
● The pre-designed route in the present invention refers to a travel route designed according to the road line type in the road, and is divided into a conventional route and a complex route: the conventional route is a route for conventional calibration, which is a diagonal, a curve, a loop, and a combination of the above routes, respectively; the complex route is a route for verifying the calibration result, and is divided into an edge route and a central mode, and the specific route is a combination of a diagonal line, a curve line and a ring line.
● The processing unit is a computer with data collection, data processing and data analysis functions. In the invention, a cloud processing unit can be selected to be connected with a road side millimeter wave radar and a calibration vehicle with vehicle-mounted positioning equipment; a base station processing unit is selected and used as an independent base station to be installed indoors or on the road side; the embedded processing unit can be selected and embedded in the millimeter wave radar on the road side or the calibration vehicle.
● The resampling refers to one of the steps of data analysis and processing performed by the processing unit, and the number of sampling points between two tracks is the same and the sampling time is aligned by interpolation or prediction. For two tracks with identical sampling frequencies and aligned sampling times, resampling is not needed; for two tracks with inconsistent sampling frequencies or different sampling times, at least one track needs to be resampled.
● The first track data D of the invention 1 The calibration vehicle with the vehicle-mounted positioning device runs according to a pre-designed conventional route, and the output original vehicle track data is the world coordinate system.
● The resampled first track data D 1 ' refer to the first track data D 1 Track data obtained after resampling, the coordinate system of the data is the world coordinate system
● The second track data D of the invention 2 The method is track data which is output by the roadside millimeter wave radar and is obtained by detecting a target, and a coordinate system of the track data is a radar coordinate system of the roadside millimeter wave radar.
● Third track data D according to the invention 3 Refers to the second track data D 2 And converting the obtained track data according to the calibration parameters of the road side millimeter wave radar, wherein the coordinate system of the track data is a world coordinate system.
● The resampled third track data D 3 ' refer to the third track data D 3 And (3) carrying out resampling to obtain track data, wherein the coordinate system of the track data is a world coordinate system.
● Fourth track data D according to the present invention 4 The method is characterized in that in the verification step, track data which is output by the roadside millimeter wave radar and is obtained by detecting a target is defined as a radar coordinate system in which the roadside millimeter wave radar is located.
● Fifth track data D of the invention 5 In the verification step, the calibration vehicle with the vehicle-mounted positioning device runs according to a pre-designed complex route, and the output original vehicle track data is output, wherein the coordinate system of the data is a world coordinate system.
● Fifth track data D of the invention 5 ' refer to the fifth track dataD 5 And (3) carrying out resampling to obtain track data, wherein the coordinate system of the track data is a world coordinate system.
● Sixth track data D according to the present invention 6 Means that in the verification step, the fourth track data D 4 And converting the obtained track data according to the calibration parameters of the road side millimeter wave radar, wherein the coordinate system of the track data is a world coordinate system.
● Sixth track data D according to the present invention 6 ' means to the sixth track data D 6 And (3) carrying out resampling to obtain track data, wherein the coordinate system of the track data is a world coordinate system.
● The space-time similarity value refers to a value obtained by substituting two pieces of track data into the space-time similarity calculation method.
● The space-time similarity threshold value refers to a numerical value which is obtained through means of manual setting, expert advice or big data analysis and the like and can judge the accuracy degree of the calibration parameters of the radar.
● The judgment of the invention refers to the process of comparing the space-time similarity value with a space-time similarity threshold.
The invention solves the technical problems by adopting the following steps, and the whole flow is as shown in figure 1:
1) Setting a sampling frequency which is not smaller than that of the road side millimeter wave radar for a vehicle-mounted positioning device of the calibration vehicle, and setting clock synchronization for the calibration vehicle with the vehicle-mounted positioning device, the road side millimeter wave radar and the processing unit;
2) The calibration vehicle with the vehicle-mounted positioning device runs in the coverage range of the road side millimeter wave radar according to a pre-designed conventional route, and outputs first track data and second track data;
3) The processing unit collects first track data and second track data generated by a calibration vehicle with a vehicle-mounted positioning device and a road side millimeter wave radar, obtains third track data after coordinate conversion, processes and analyzes the first track data and the third track data, and outputs a space-time similarity value of the first track data and the third track data;
4) The processing unit judges that the parameters of the millimeter wave radar at the open circuit side do not need to be calibrated according to the space-time similarity value and the space-time similarity threshold value, and the calibration is finished; the processing unit judges that parameters of the millimeter wave radar at the circuit side need to be calibrated, and then the parameters of the millimeter wave radar at the circuit side are calibrated;
5) After calibrating the parameters of the road side millimeter wave radar, the processing unit performs the following verification steps:
the calibration vehicle with the vehicle-mounted positioning device is required to run in the coverage range of the roadside millimeter wave radar according to a complex route or another conventional route which is designed in advance, and fourth track data and fifth track data are output;
executing the step 3) to output a space-time similarity value for the output fourth track data and fifth track data; and (4) executing the step 4 on the output space-time similarity value.
If the parameters of the millimeter wave radar on the circuit side are less than or equal to the space-time similarity threshold, the calibration parameters of the millimeter wave radar on the circuit side meet the requirements.
If the calculated spatio-temporal similarity value is greater than the spatio-temporal similarity threshold, the processing may be performed according to the following three alternatives:
(1) calibrating the calibration parameters of the millimeter wave radar, namely temporarily serving as the calibration parameters of the radar according to the radar calibration parameters acquired for the first time;
(2) calibrating the calibration parameters of the millimeter wave radar, namely temporarily serving as the calibration parameters of the radar according to the updated radar calibration parameters;
(3) re-executing the steps 1) to 4), and ending the calibration if the space-time similarity value obtained in the step 4) is smaller than the space-time similarity threshold value; if the space-time similarity value obtained in the step 4) is greater than or equal to the space-time similarity threshold value, calibrating the radar as a fault radar reporting processing unit.
The specific technical scheme in the steps of the invention patent is as follows:
(1) The installation angle of the road side millimeter wave radar is different according to actual scenes, and the installation position of the road side millimeter wave radar is shown in fig. 2. The calibration vehicle with the external positioning equipment is provided with a vehicle-mounted positioning device which is generally a high-precision RTK differential positioning device. Mounted in the interior or roof of a calibration vehicle.
The sampling frequency of the calibration vehicle with the vehicle-mounted positioning device is set to be at least not lower than the sampling frequency of the roadside millimeter wave radar. When the higher sampling frequency is maintained, the accuracy of calibrating the roadside millimeter wave radar is higher. When the sampling frequency of the vehicle-mounted positioning device is consistent with the sampling frequency of the roadside millimeter wave radar and the sampling time is the same, resampling processing is not needed. When the sampling frequency of the vehicle-mounted positioning device is inconsistent with the sampling frequency of the road side millimeter wave radar or the sampling time point is different, at least one track needs to be resampled. After resampling, the two tracks can realize the same number of sampling points and aligned sampling time.
The processing unit adjusts the time clocks of the millimeter wave radar on the calibrating vehicle and the road side with the vehicle-mounted positioning device, so that the time clocks of the calibrating vehicle and the road side with the millimeter wave radar are strictly synchronous with the clock time of the processing unit, and clock synchronization is realized.
First track data D collected by the calibration vehicle with the vehicle-mounted positioning device 1 May be represented using the following vectors:
R v ＝[x v ,y v ,z v ]
wherein:
x v representing the X coordinate of the vehicle-mounted positioning device in a world coordinate system;
y v representing the Y coordinate of the vehicle-mounted positioning device in a world coordinate system;
z v representing the Z coordinate of the in-vehicle positioning device in the world coordinate system.
Second track data D acquired by the roadside millimeter wave radar 2 May be represented using the following vectors:
R r ＝[x r ,y r ,z r ]
wherein:
x r representing an X coordinate of the roadside millimeter wave radar in a radar coordinate system;
y r representing Y coordinates of the roadside millimeter wave radar in a reference coordinate system;
z r and the Z coordinate of the road side millimeter wave radar in the reference coordinate system is represented.
(2) In the detection coverage area of the road side millimeter wave radar, a calibration vehicle with a vehicle-mounted positioning device runs in the coverage area of the road side millimeter wave radar according to a pre-designed conventional route to generate first track data D 1 And second track data D 2 . Specifically, a calibration vehicle with a vehicle-mounted positioning device runs according to a preset conventional route, and the vehicle-mounted positioning device of the calibration vehicle continuously collects positioning data of the vehicle to obtain first track data D 1 The method comprises the steps of carrying out a first treatment on the surface of the Continuously acquiring echo data of a calibration vehicle in a road by using a road side millimeter wave radar to obtain second track data D 2 。
As shown in fig. 3, the conventional route refers to a diagonal line, a curve, a circular route, and a combination of the above routes in a road, which are used for conventional calibration of the roadside millimeter wave radar in the present invention.
1) First track data D 1 The calibration vehicle with the vehicle-mounted positioning device runs in the coverage area of the roadside millimeter wave radar according to a pre-designed conventional route (or a complex route), and the vehicle-mounted positioning device continuously collects the positioning data of the calibration vehicle at different moments in real time. The data may be represented by the following set:
S v ＝{tr 1 ,tr 2 ,...tr N }
wherein tr n Indicating that the vehicle-mounted positioning device collects the nth track set at different moments.
Wherein:
the acquisition time of the jth track point in the nth target track acquired by the vehicle-mounted positioning device is represented;
the abscissa of the jth track point in the nth target track acquired by the vehicle-mounted positioning device is represented;
and the ordinate of the jth track point in the nth target track acquired by the vehicle-mounted positioning device is shown.
2) Second track data D 2 Is track echo data of the calibration vehicle generated by the road side millimeter wave radar uninterruptedly. The data may be represented by the following set:
S r ＝{tr 1 ,tr 2 ,...tr M }
Wherein tr m And the method indicates that the roadside millimeter wave radar collects an mth target track set. The expression is as follows:
wherein:
representing the acquisition time of the ith track point in the mth target track acquired by the roadside millimeter wave radar;
representing the abscissa of an ith track point in an mth target track acquired by a roadside millimeter wave radar;
and the ordinate of the ith track point in the mth target track acquired by the roadside millimeter wave radar is represented.
(3) The processing unit acquires D generated by the millimeter wave radar on the road side and the calibration vehicle with the vehicle-mounted positioning device 1 、D 2 And pair D 2 Coordinate conversion is carried out to obtain third track data D 3 For D 1 And D 3 And (5) performing calculation and analysis, and judging whether the millimeter wave radar at the open circuit side needs to be calibrated. The specific flow is shown in fig. 4.
1) And (5) coordinate conversion. Due to D 1 In world coordinate system, D 2 In the radar coordinate system, the angle and the position of the two are different, and data of different scales in the two coordinate systems cannot be directly processed, so that the coordinate conversion is needed. The coordinate conversion is to select corresponding characteristic points in two types of track points to form 4 or more point pair sets, and calculate a calibration parameter matrix from a radar coordinate system to a world coordinate system according to a coordinate conversion formula. Finally, D is obtained through calibrating the parameter matrix 2 Conversion to D 3 . The specific process is as follows:
(1) selecting D 1 Aggregation and D 2 And selecting at least 4 pairs of characteristic points and more than 4 pairs of characteristic points corresponding to the characteristic points in the set. The principle of feature selection is to select between two trace dataPoints with a temporal, spatial correspondence. The selected pairs of feature points are as follows:
one characteristic point of each of the characteristic point pairsD generated from vehicle-mounted positioning device 1 Data, another feature pointD generated by millimeter wave radar at road side 2 。
(2) According to the obtained characteristic point pairs, resolving D 1 Conversion of coordinates to D 3 And the corresponding calibration parameter matrix.
In the present invention, the world coordinate system refers to a reference coordinate system selected in the environment to describe the position of the calibration vehicle, which is referred to as the world coordinate system. In the invention, world coordinates are acquired by a vehicle-mounted positioning device on a calibration vehicle. The vehicle-mounted positioning device can provide three-dimensional positioning results of the measuring station in a specified coordinate system in real time.
In the invention, the coordinates acquired by the millimeter wave radar at the road side are in a radar coordinate system. The radar calibration parameter matrix in the present invention refers to a matrix that converts a radar coordinate system into a world coordinate system. The radar calibration parameter matrix H is shown as follows:
The formula (x) is shown below r ,y r ,z r ) Representation D 2 Three-dimensional coordinates of a point (x) v ,y v ,z v ) Representation D 1 Three-dimensional coordinates of a point in (a).
And (3) solving a linear equation through the manually selected corresponding characteristic point pairs in the step (1), so that the value of the radar calibration parameter matrix H can be solved.
(3) D in a radar coordinate system can be obtained according to the solved radar calibration parameter matrix H 2 Transforming to D under world coordinate system 3 . The following is shown:
wherein, (x) r ,y r ,z r ) Representation D 2 Some point in (x' r ,y′ r ,z′ r ) Representing new three-dimensional coordinate data of the millimeter wave radar at the rear road side converted to the world coordinate system, namely third track data D 3 Is a point in (a).
2) Pair D 1 And D 3 And performing calculation analysis.
(1) Resampling: on the basis of 1), D is 1 And D 3 Resampling is performed such that D 1 And D 3 The two types of data are corresponding to each other in a one-to-one mode.
Although the data acquisition frequencies of the vehicle-mounted positioning device on the calibration vehicle and the roadside millimeter wave radar are higher, the acquisition frequencies are not necessarily the same. Spatially, the sampling point distribution time interval of the two is inconsistent. Thus D 1 And D 3 The two marks cannot be in one-to-one correspondence in space, and the judgment of the accuracy of the marks cannot be completed. For this purpose, the data needs to be resampled by resampling, and the data is revised into two tracks with the same number of sampling points and aligned sampling time. When the sampling frequency of the vehicle-mounted positioning device is consistent with the sampling frequency of the road side millimeter wave radar and the sampling time is the same, resampling processing is not needed. When the sampling frequency of the vehicle-mounted positioning device is inconsistent with the sampling frequency of the road side millimeter wave radar or the sampling time point is different, at least one track needs to be resampled. The method comprises the following specific steps:
First solve D 1 Set of medium time points T 1 And D 3 Set of medium time points T 3 The union T of (2) all . The specific relation is as follows:
T 1 ∪T 3 ＝T all
then sequentially go through D 1 And D 3 Every fourth point as a set, fitting a cubic function. The tertiary function G can be solved by bringing the abscissa of four adjacent points in the track data x Coefficients a, b, c, d of (c). The third function G can be solved by bringing the ordinate of the adjacent three points in the track data y Coefficients e, f, g, h of (c). The expression of the cubic function is shown as follows:
G x ＝ax 3 +bx 2 +cx+d
G y ＝ex 3 +fx 2 +gx+h
finally, with T in the interval all Time point as data resamplingAnd (2) obtaining the cubic function y of the coordinate point g In the final acquisition of T all Resampled roadside millimeter wave radar new data D for time points 1 ' and resampled in-vehicle positioning device trajectory data D 3 ' both maintain a consistent sampling frequency in the time dimension, as shown in fig. 7.
(2) Space-time similarity calculation: resampled D 1 ' and D 3 ' one-to-one at each point in time. In order to judge the accuracy of the radar calibration parameter matrix, D is needed to be calculated 1 ' and D 3 ' space-time similarity calculation is carried out, and accuracy evaluation is carried out on the time-space double dimension. The evaluation procedure is as follows:
First, as D 1 ' and D 3 ' set C of set-up matching point pairs 2 . The expression is as follows:
wherein,representation D 3 The ith millimeter wave radar track data point of the u-th road side in the data set, wherein the data point comprises an abscissaOrdinate of the ordinateTime stampingRepresentation D 1 The 'data set' includes the trace data of the u th calibration vehicle with the onboard positioning device, and the data points include the abscissaOrdinate of the ordinateTime stampingAnd the locus point is time-wise with D 3 ' PointMatching.
Then, based on the matching point pair set C 2 The similarity of the pairs is evaluated from the spatial and temporal dual dimensions. The expression for evaluating the similarity is as follows:
wherein sim (p, q) represents the similarity between the point p and the point q,representing set C 2 The sum of the matching similarity of the U point pairs in the model is used for measuring D by using the index 1 ' and D 3 ' degree of similarity. The larger the value of the similarity, the more accurate the radar calibration parameter matrix.
f S Representing the spatial similarity and measuring the matching point pair set C 2 The magnitude of the time difference between the midpoint pair. Calculation ofThe formula is as follows:
f T representing the time similarity and measuring the matching point pair set C 2 The magnitude of the time difference between the midpoint pair. The calculation formula is as follows:
In addition, inWeights representing the degree of spatial similarity are used,weights representing temporal similarity. f (f) S Representing the spatial similarity and measuring the matching point pair set C 2 The distance between the midpoint pair on the abscissa and the ordinate. Measuring the similarity of the point pairs from the space dimension, and directly reflecting the accuracy of the calculated radar calibration parameter matrix; the similarity of the point pairs is measured from the time dimension, the time proximity of the matched point pairs is directly reflected, and the accuracy of the calculated radar calibration parameter matrix is indirectly reflected. By weightingThe proportion of the space similarity and the time similarity can be freely adjusted, and accurate evaluation can be realized when the clock synchronization problem exists.
In addition, where β represents the weight of the similarity of the non-resampled points, and (1- β) represents the weight of the resampled pointsAnd (5) weighting.Representing the similarity between non-resampled points,representing the similarity between resampling points. The similarity of the non-resampling point pair is measured, the reliability is high, and the D is directly reflected 1 ' and D 3 Degree of similarity between'; the similarity of the resampling point pair is measured, the reliability is low, and the D is indirectly reflected 1 ' and D 3 Degree of similarity between'; the weight beta is adopted to freely adjust the proportion of the non-resampling point pair and the resampling point pair, so that more accurate evaluation can be realized under the condition that the reliability is reduced after resampling.
(4) Judging whether the parameters meet a space-time similarity threshold value or not: on the basis of data acquisition and data analysis by the processing unit, the solved D is judged 1 ' and D 3 Whether the' spatiotemporal similarity is less than or equal to a spatiotemporal similarity threshold. If the threshold requirement is met, the roadside millimeter wave radar does not need to be recalibrated; if the threshold requirement is not met, the roadside millimeter wave radar is considered to be required to be recalibrated. The method comprises the following specific steps:
d is calculated through the step (3) 1 ' and D 3 ' similarity ofIf the similarity isIf the space-time similarity threshold delta is larger than or equal to the space-time similarity threshold delta, the radar calibration parameter matrix H is considered to meet the threshold requirement, and recalibration is not needed; if similar toAnd (5) if the similarity threshold value delta is smaller than the similarity threshold value delta, the radar calibration parameter matrix H is considered to not meet the threshold value requirement, and the roadside millimeter wave radar needs to be calibrated according to the step (5).
(5) Calibrating and verifying: and (3) when the radar calibration parameter matrix H is judged to not meet the threshold requirement in the step (4), recalculating the radar calibration parameter matrix of the road side millimeter wave radar, and calibrating. After calibration, to ensure accuracy of the results, the calibration results should be further verified. The specific flow is shown in fig. 10.
(1) Recalibration. Repeating the step (1) in the step (3) to perform coordinate conversion, selecting a new feature matching point pair, and selecting a pair D 1 And D 2 And (5) recalculating a radar calibration parameter matrix H', and updating radar calibration parameter matrix values. The radar calibration parameter matrix H' is shown as follows:
(2) and (5) verifying results. In the detection coverage area of the road side millimeter wave radar, the calibration vehicle with the vehicle-mounted positioning device runs in the coverage area of the road side millimeter wave radar according to a complex route or another conventional route which is designed in advance, and data generated by the calibration vehicle with the vehicle-mounted positioning device is recorded as fifth track data D 5 The data detected by the roadside millimeter wave radar for target detection is recorded as fourth track data D 4 。
In the step of verifying the result, the calibration vehicle may select a complex route which is designed in advance, as shown in fig. 4, and the complex route refers to a straight line, a curve, a circular route and a combination of the above routes at the center or the edge of the road; the calibration vehicle may also select another conventional route designed in advance, as shown in fig. 3, which refers to a diagonal line, a curve, a circular route in a road, and a combination of the above routes. In the aspect of route selection, the characteristic of the conventional route can be selected for the first time, and the complex route which is different from the characteristic of the conventional route or another conventional route can be selected for driving and verification. In the aspect of route selection, the method can also be used for verifying the result by pointedly selecting another conventional route or a complex route which is more similar to the conventional route of the first driving in linearity when the difference between the space-time similarity value calculated by the processing unit and the space-time similarity threshold value is large; or when the difference is small, a different conventional route or a complex route is selected in a targeted manner, which is more different in linearity from the conventional route of the first driving, and the result is verified.
First, the fourth orbit data D in the radar coordinate system can be obtained from the updated radar calibration parameter matrix H 4 Transformation to sixth trajectory data D in world coordinate System 6 . The following is shown:
wherein, (x) r ,y r ,z r ) Representing a fourth trace data point, (x' r ,y′ r ,z′ r ) And the new three-dimensional coordinate data of the millimeter wave radar at the rear road side is converted into the world coordinate system according to the solved radar calibration parameter matrix H', and the new three-dimensional coordinate data is the sixth track data point.
Next, D is carried out 5 And D 6 Uploading to a processing unit for data collection and data analysis, namely carrying out resampling calculation according to the step 2) of the step (3) to obtain resampled track data D 5 ' and D 6 ′。
Then, the trajectory data D obtained after resampling 5 ' and D 6 ' performing space-time similarity numerical calculation to obtain D 5 ' and D 6 ' spatiotemporal similarity value. And (3) judging the calculated space-time similarity value according to the step (4). If the calculated space-time similarity value is smaller than or equal to the space-time similarity threshold value, the calibration parameters of the road side millimeter wave radar meet the requirements; if the calculated spatio-temporal similarity value is greater than the spatio-temporal similarity threshold, the processing may be performed according to the following three alternatives:
1) The calibration parameters of the millimeter wave radar are not calibrated, namely the first acquired radar calibration parameters H are temporarily used as the radar calibration parameters;
2) Calibrating the calibration parameters of the millimeter wave radar, namely temporarily taking the updated radar calibration parameters H' as the calibration parameters of the radar;
3) Re-executing the steps (1) to (4), and ending the calibration if the space-time similarity value obtained in the step (4) is smaller than the space-time similarity threshold value; if the space-time similarity value obtained in the step (4) is greater than or equal to the space-time similarity threshold value, calibrating the radar as a fault radar reporting processing unit.
The invention has the technical key points and advantages that:
the data collected by the vehicle-mounted positioning device on the calibration vehicle is used as true value data for calibrating the road side millimeter wave radar, an automatic and high-precision data source is provided, and the calibration precision of the road side millimeter wave radar is effectively improved. Meanwhile, resampling is used as a preprocessing method for matching data with different sampling frequencies, so that the problem that multi-source track data with different sampling frequencies are difficult to compare is solved. In addition, a measurement method of the time-space similarity is used for judging whether the millimeter wave radar on the open circuit side needs to be calibrated or not, and the judgment method is simple, convenient and feasible and does not need to be calibrated manually. And finally, designing a conventional route and a complex route running scheme for calibrating the vehicle, and performing multi-angle result verification on the calibration result of the road side millimeter wave radar.
The above symbols and their meaning are summarized in the following table:
FIG. 1 is a schematic flow chart of the present invention;
FIG. 2 is a schematic diagram of the relative positions of the road side millimeter radar layout and the vehicle-mounted positioning device of the invention;
FIG. 3 is a schematic diagram of a conventional driving route for data collection;
FIG. 4 is a schematic diagram of a complex driving route with result verification;
FIG. 5 is a flow chart of the process unit for collecting data and analyzing the data;
FIG. 6 is a schematic diagram of trace feature point selection;
FIG. 7 is a schematic diagram of trajectory space matching;
FIG. 8 is a schematic diagram of a track of a road side millimeter radar layout at different sampling frequencies than a vehicle-mounted positioning device;
FIG. 9 is a schematic diagram of resampling;
FIG. 10 is a flow chart of recalibration and result verification.
The invention is described in detail below with reference to the drawings and the detailed description.
The invention relates to a road side millimeter wave radar calibration method based on a vehicle-mounted positioning device. As shown in fig. 1, the present invention can be divided into five main steps:
firstly, layout and clock synchronization of the road side millimeter wave radar.
The vehicle-mounted positioning device of the calibration vehicle can assist an automatic driving vehicle or a common vehicle to perform high-precision positioning, has centimeter-level positioning precision, and can output the running track point of the calibration vehicle in real time. The method takes the millimeter wave radar as a data reference value, thereby carrying out calibration and updating of calibration parameters on the road side millimeter wave radar. The vehicle-mounted positioning device of the calibration vehicle can output real-time position data of the calibration vehicle, and the sampling frequency is set to be not less than the sampling frequency of the roadside millimeter wave radar.
The road side millimeter wave radar is installed on the road side and is distributed at different positions of the road, the distributed positions cover the complete range of the road surface of the road, as shown in fig. 2, the road field is as open as possible, no high building or tree is shielded, and the single detection range is about 300 meters, so that the accurate measurement of the distance and the speed of the detection target can be realized. And outputting the position data of the moving target on the detection road according to the set sampling frequency.
The processing unit is located in the virtual cloud or in the base station, or is installed in the roadside millimeter wave radar or the calibration vehicle in an embedded mode. The processing unit and the road side millimeter wave radar are connected with a calibration vehicle with a vehicle-mounted positioning device in real time. And the road side millimeter wave radar, the processing unit and the vehicle-mounted positioning device on the calibration vehicle are connected by a base station or are subjected to network time service to realize clock synchronization.
And secondly, the calibration vehicle runs according to a pre-designed route and uploads the data to the processing unit.
First, in a road environment actually monitored by a roadside millimeter wave radar, a calibration vehicle with a vehicle-mounted positioning device runs along a road according to a pre-designed conventional route or a complex route, the conventional route is shown in fig. 3, and the complex route is shown in fig. 4.
In the running process, a calibration vehicle with a vehicle-mounted positioning device generates target vehicle track data with a time stamp in real time, the target vehicle track data is recorded as first track data, the data is located under a world coordinate system, and the first track data is uploaded to a processing unit; the road side millimeter wave radar synchronously acquires target vehicle track detection data with a time stamp, marks the target vehicle track detection data as second track data, the data are positioned in a radar coordinate system, and the second track data are uploaded to the processing unit. Data acquisition is achieved by calibrating the travel of the vehicle.
And thirdly, the processing unit performs data processing and data analysis.
The processing unit receives the uploaded first track data and second track data, and processes and analyzes the data. Coordinate conversion, resampling and space-time similarity calculation are performed respectively. As shown in fig. 5.
First, since the first track data and the second track data are not in the same coordinate system, the second track data is subjected to coordinate conversion. In the track points of the first track data and the second track data, as shown in fig. 6, more than four sets of characteristic point pairs are selected, and a radar calibration parameter matrix of the road side millimeter wave radar is calculated. And converting the second track data points under the radar coordinate system into the world coordinate system according to the calculated radar calibration parameter matrix to obtain third track data, and realizing coordinate conversion, wherein the two tracks are in the same coordinate system.
Next, in order to solve the problem that the sampling frequencies of the first track data and the second track data are not identical, as shown in fig. 8, the first track data and the third track data are resampled. Resampling obtains a union of sampling time points in the first track data and the third track data point set. And checking the time points which are respectively missing in the first track number and the third track data by utilizing the union of the obtained sampling time points, supplementing the deficiency abscissa to the checked deficiency time points based on a cubic interpolation algorithm, and realizing resampling, wherein the two types of tracks are in the same coordinate system and the sampling time points are consistent, and the first track data are matched with the third track data corresponding to the road side millimeter wave radar one to one. See fig. 9.
Finally, in order to quantitatively judge whether the radar calibration parameters of the millimeter wave radar at the open circuit side are accurate, space-time similarity calculation is carried out on the first track data and the third track data after coordinate conversion and resampling. And according to the expression of the space-time similarity, substituting the first track data point and the third track data point pair under the same sampling time point respectively, and calculating the space-time similarity value of the first track data point and the second track data point.
Fourth, judging the accuracy of the calibration parameters.
And after the processing unit solves the space-time similarity value of the first track data and the third track data, comparing the solving result with a preset precision threshold value. If the solved space-time similarity value is smaller than or equal to a preset precision threshold value, the roadside millimeter wave radar does not need to be recalibrated, and the process is ended; if the solved space-time similarity value is larger than a preset precision threshold value, the roadside millimeter wave radar is considered to need to be recalibrated, a fifth step is executed, and result verification is carried out.
And fifthly, recalibrating and verifying results.
And (3) after determining that the roadside millimeter wave radar needs to be recalibrated in the step (4), executing the current step (5) to perform recalibration and result verification. A specific flow chart is shown in fig. 10.
And (3) firstly, recalibrating the roadside millimeter wave radar. And recalculating the calibration parameter matrix for the first track data and the second track data in the primary calibration. In the track points of the first track data and the second track data, as shown in fig. 6, more than four sets of characteristic point pairs different from those in the initial calibration are newly selected, and a radar calibration parameter matrix of the roadside millimeter wave radar is updated.
A calibration experiment to calibrate the vehicle is then performed. In the road environment actually monitored by the roadside millimeter wave radar, the calibration vehicle with the vehicle-mounted positioning device runs along the road again according to a complex pre-designed route, and the complex route is shown in fig. 4. In the running process, the calibration vehicle with the vehicle-mounted positioning device generates target vehicle track data with a time stamp in real time, marks the target vehicle track data as fifth track data, the data are positioned under a world coordinate system, and the fifth track data are uploaded to the processing unit; the road side millimeter wave radar synchronously acquires target vehicle track detection data with a time stamp, marks the target vehicle track detection data as fourth track data, the data are positioned under a radar coordinate system, and the fourth track data are uploaded to the processing unit. And the processing unit converts fourth track data points in the radar coordinate system into the world coordinate system according to the updated radar calibration parameter matrix to obtain sixth track data, and coordinate conversion is realized, wherein the two types of tracks are in the same coordinate system.
And finally, the processing unit performs result verification on the updated radar calibration parameter matrix. The method comprises the following three steps of:
(1) in order to solve the problem that the sampling frequencies of the fifth track data and the sixth track data are inconsistent, resampling is performed on the fifth track data and the sixth track data. Resampling obtains a union of the fifth track data and the sampling time point in the sixth track data point set. And checking the time points which are respectively missing in the fifth track number and the sixth track data by utilizing the union of the obtained sampling time points, supplementing the missing time points which are checked with the missing abscissa based on a cubic interpolation algorithm, and realizing resampling, wherein the two types of tracks are in the same coordinate system and the sampling time points are consistent, and the fifth track data are matched with the sixth track data corresponding to the road side millimeter wave radar one to one.
(2) In order to quantitatively judge whether the radar calibration parameters of the millimeter wave radar at the open circuit side are accurate or not, space-time similarity calculation is carried out on fifth and sixth track data after coordinate conversion and resampling. And according to the expression of the space-time similarity, substituting the fifth track data point and the sixth track data point pair under the same sampling time point respectively, and calculating the space-time similarity value of the fifth track data point and the second track data point.
(3) And after the processing unit solves the space-time similarity value of the fifth track data and the sixth track data, comparing the solving result with a preset precision threshold value. If the solved space-time similarity value is smaller than or equal to a preset precision threshold value, the roadside millimeter wave radar does not need to be recalibrated, and the process is ended; if the solved space-time similarity value is larger than a preset precision threshold value, the roadside millimeter wave radar is considered to need to be recalibrated, a fifth step is executed, and result verification is carried out.
The invention has the technical key points and advantages that:
the data collected by the vehicle-mounted positioning device on the calibration vehicle is used as true value data for calibrating the road side millimeter wave radar, an automatic and high-precision data source is provided, and the calibration precision of the road side millimeter wave radar is effectively improved. Meanwhile, resampling is used as a preprocessing method for matching data with different sampling frequencies, so that the problem that multi-source track data with different sampling frequencies are difficult to compare is solved. In addition, a measurement method of the time-space similarity is used for judging whether the millimeter wave radar on the open circuit side needs to be calibrated or not, and the judgment method is simple, convenient and feasible and does not need to be calibrated manually. And finally, designing a conventional route and a complex route running scheme for calibrating the vehicle, and performing multi-angle result verification on the calibration result of the road side millimeter wave radar.
An example is as follows:
(1) Layout and clock synchronization setting of road side millimeter wave radar, and cloud server is used as processing unit
Setting an experimental scene in a steam-created new harbor park in a Shanghai city jaggies. In an experimental scene, a road side millimeter wave radar is arranged at the road side 250 m at equal intervals, the installation height is 5.0 m, the depression angle is 10 degrees, the road side millimeter wave radar covers the whole area of a road, and the field is open and does not reach a building for shielding. A high-precision RTK positioning device is installed on a common vehicle for calibration, the RTK is installed on the top end of the vehicle for calibration, and the angular relationship between the RTK and the vehicle is kept stable. The cloud server is accessed to serve as a processing unit, and meanwhile, a network is adopted to perform unified time service on the road side millimeter wave radar, the RTK positioning equipment for calibrating the vehicle and the processing unit, so that the three keep clock synchronization.
(2) The calibration vehicle travels along a pre-designed regular route (curve) and uploads the data to the cloud processing unit.
In the experimental road environment of the campus, a calibration vehicle with a high-precision RTK positioning device normally runs along the road of the campus according to a curve route in a pre-designed conventional route. Meanwhile, the first track data acquired by the RTK on the calibration vehicle with high precision are uploaded to the cloud processing unit in real time; and uploading the second track data acquired by the road side millimeter wave radar to the cloud processing unit in real time. And data acquisition and uploading are realized.
(3) And the cloud processing unit performs data processing and data analysis.
Through the step (2), the sampling frequency of the acquired second track data is 10Hz; the acquisition frequency of the first track data of the acquired calibration vehicle is 100Hz. The cloud processing unit receives the uploaded first track data and second track data, and processes and analyzes the data. Coordinate conversion, resampling and space-time similarity calculation are performed respectively.
(1) Coordinate conversion: after the cloud processing unit realizes data acquisition, the first track data and the second track data are not in the same coordinate system, so that the second track data are subjected to coordinate conversion. Selecting four groups of characteristic point pairs from track points of the first track data and the second track data, and calculating a radar calibration parameter matrix H of the road side millimeter wave radar:
and converting the second track data points under the radar coordinate system into the world coordinate system according to the calculated radar calibration parameter matrix to obtain third track data, and realizing coordinate conversion, wherein the two tracks are in the same coordinate system.
(2) Resampling: in order to solve the problem that the sampling frequencies of the first track data and the second track data are inconsistent, resampling is carried out on the first track data and the third track data. Resampling to obtain a union of sampling time points in the first track data and the third track data point set, wherein the union is [1516783680.132207, … 1516783954.932218]. And checking the time points which are respectively missing in the first track data and the third track data by utilizing the union of the obtained sampling time points, supplementing the deficiency abscissa to the checked deficiency time points based on a cubic interpolation algorithm, and realizing resampling, wherein the two types of tracks are in the same coordinate system and the sampling time points are consistent, and the first track data and the third track data points corresponding to the road side millimeter wave radar are matched one by one.
(3) Space-time similarity calculation: in order to quantitatively judge whether the radar calibration parameters of the millimeter wave radar at the open circuit side are accurate or not, space-time similarity calculation is carried out on the first track data and the third track data after coordinate conversion and resampling. According to the expression of the space-time similarity, substituting the first track data point and the third track data point pair under the same sampling time point respectively to calculate the space-time similarity value of the first track data point and the second track data point
Specifically, weighting of spatial similarity is takenWeight of similarity of timeAnd calculating resampled points and non-resampled points in the matched points according to different weights. Taking the weight β=0.6 for the non-resampled point pair, the weight (1- β) =0.4 for the resampled point pair. Calculation sim (p, q):
the calculation method of the similarity function f is calculated according to the average value of Euclidean distances.
(4) And judging the accuracy of the calibration parameters.
After the space-time similarity index is solved, the solving result is compared with a preset threshold delta. And finding that the solving result is smaller than a preset threshold value indicates that the road side millimeter wave radar does not need to be recalibrated.
Example two is as follows:
(1) Layout and clock synchronization setting of road side millimeter wave radar, and road side computer is used as processing unit
Setting an experimental scene in a steam-created new harbor park in a Shanghai city jaggies. In an experimental scene, a road side millimeter wave radar is arranged at the road side 250 m at equal intervals, the installation height is 5.0 m, the depression angle is 10 degrees, the road side millimeter wave radar covers the whole area of a road, and the field is open and does not reach a building for shielding. A high-precision RTK positioning device is installed on a common vehicle for calibration, the RTK is installed on the top end of the vehicle for calibration, and the angular relationship between the RTK and the vehicle is kept stable. The cloud server is accessed to serve as a processing unit, and meanwhile, a network is adopted to perform unified time service on the road side millimeter wave radar, the RTK positioning equipment for calibrating the vehicle and the processing unit, so that the three keep clock synchronization.
(2) The calibration vehicle travels along a pre-designed regular route (loop) and uploads the data to the roadside processing unit.
In the experimental road environment of the campus, a calibration vehicle with a high-precision RTK positioning device normally travels along the road of the campus according to a circular route in a pre-designed conventional route. Meanwhile, the first track data acquired by the RTK on the calibration vehicle with high precision are uploaded to the cloud processing unit in real time; and uploading the second track data acquired by the road side millimeter wave radar to the cloud processing unit in real time. And data acquisition and uploading are realized.
(3) The road side processing unit performs data processing and data analysis.
Through the step (2), the sampling frequency of the acquired second track data is 10Hz; the acquisition frequency of the first track data of the acquired calibration vehicle is 100Hz. The cloud processing unit receives the uploaded first track data and second track data, and processes and analyzes the data. Coordinate conversion, resampling and space-time similarity calculation are performed respectively.
(1) Coordinate conversion: after the cloud processing unit realizes data acquisition, the first track data and the second track data are not in the same coordinate system, so that the second track data are subjected to coordinate conversion. Selecting four groups of characteristic point pairs from track points of the first track data and the second track data, and calculating a radar calibration parameter matrix H of the road side millimeter wave radar:
and converting the second track data points under the radar coordinate system into the world coordinate system according to the calculated radar calibration parameter matrix to obtain third track data, and realizing coordinate conversion, wherein the two tracks are in the same coordinate system.
(2) Resampling: in order to solve the problem that the sampling frequencies of the first track data and the second track data are inconsistent, resampling is carried out on the first track data and the third track data. Resampling to obtain a union of sampling time points in the first track data and the third track data point set, wherein the union is [1516783680.132207, … 1516783954.932218]. And checking the time points which are respectively missing in the first track data and the third track data by utilizing the union of the obtained sampling time points, supplementing the deficiency abscissa to the checked deficiency time points based on a cubic interpolation algorithm, and realizing resampling, wherein the two types of tracks are in the same coordinate system and the sampling time points are consistent, and the first track data and the third track data points corresponding to the road side millimeter wave radar are matched one by one.
(3) Space-time similarity calculation: in order to quantitatively judge whether the radar calibration parameters of the millimeter wave radar at the open circuit side are accurate or not, space-time similarity calculation is carried out on the first track data and the third track data after coordinate conversion and resampling. According to the expression of the space-time similarity, substituting the first track data point and the third track data point pair under the same sampling time point respectively to calculate the space-time similarity value of the first track data point and the second track data point
Specifically, weighting of spatial similarity is takenWeight of similarity of timeAnd calculating resampled points and non-resampled points in the matched points according to different weights. Taking weights for non-resampled point pairsβ=0.5, then the weight of the resample point pair (1- β) =0.5. Calculation sim (p, q):
the calculation method of the similarity function f is calculated according to the average value of Euclidean distances.
(4) And judging the accuracy of the calibration parameters.
After the space-time similarity index is solved, the solving result is compared with a preset threshold delta. Finding that the solving result is larger than a preset threshold value, considering that the roadside millimeter wave radar needs to be recalibrated, executing the fifth step, and verifying the result.
(5) Recalibration and result verification.
And (3) after determining that the roadside millimeter wave radar needs to be recalibrated in the step (4), executing the current step (5) to perform recalibration and result verification. A specific flow chart is shown in fig. 10.
And (3) recalibrating the roadside millimeter wave radar. And recalculating the calibration parameter matrix for the first track data and the second track data in the primary calibration. And re-selecting more than four sets of characteristic point pairs different from those in the primary calibration from the track points of the first track data and the second track data, and updating a radar calibration parameter matrix H' of the road side millimeter wave radar.
A calibration experiment to calibrate the vehicle is then performed. In the road environment actually monitored by the roadside millimeter wave radar, the calibration vehicle with the on-vehicle positioning device is again driven along the road according to a complex route (route combination 1) designed in advance, see fig. 4. In the running process, the calibration vehicle with the vehicle-mounted positioning device generates target vehicle track data with a time stamp in real time, marks the target vehicle track data as fifth track data, the data are positioned under a world coordinate system, and the fifth track data are uploaded to the processing unit; the road side millimeter wave radar synchronously acquires target vehicle track detection data with a time stamp, marks the target vehicle track detection data as fourth track data, the data are positioned under a radar coordinate system, and the fourth track data are uploaded to the road side processing unit. And the processing unit converts fourth track data points in the radar coordinate system into the world coordinate system according to the updated radar calibration parameter matrix to obtain sixth track data, and coordinate conversion is realized, wherein the two types of tracks are in the same coordinate system.
And finally, the processing unit performs result verification on the updated radar calibration parameter matrix. The method comprises the following three steps of:
(1) in order to solve the problem that the sampling frequencies of the fifth track data and the sixth track data are inconsistent, resampling is performed on the fifth track data and the sixth track data. Resampling sums the fifth trajectory data with the sampling time points in the sixth trajectory data point set, and the union is [1516783680.132207, … 1516784063.852348]. And checking the time points which are respectively missing in the fifth track number and the sixth track data by utilizing the union of the obtained sampling time points, supplementing the missing time points which are checked with the missing abscissa based on a cubic interpolation algorithm, and realizing resampling, wherein the two types of tracks are in the same coordinate system and the sampling time points are consistent, and the fifth track data are matched with the sixth track data corresponding to the road side millimeter wave radar one to one.
(2) In order to quantitatively judge whether the radar calibration parameters of the millimeter wave radar at the open circuit side are accurate or not, space-time similarity calculation is carried out on fifth and sixth track data after coordinate conversion and resampling. And according to the expression of the space-time similarity, substituting the fifth track data point and the sixth track data point pair under the same sampling time point respectively, and calculating the space-time similarity value of the fifth track data point and the second track data point.
Specifically, weighting of spatial similarity is takenWeight of similarity of timeAnd calculating resampled points and non-resampled points in the matched points according to different weights. Taking the weight β=0.5 for the non-resampled point pair, the weight (1- β) =0.5 for the resampled point pair. Calculation sim (p, q):
the calculation method of the similarity function f is calculated according to the average value of Euclidean distances.
(3) And after the processing unit solves the space-time similarity value of the fifth track data and the sixth track data, comparing the solving result with a preset precision threshold value. And if the solved space-time similarity value is smaller than or equal to a preset precision threshold delta, the recalibration of the road side millimeter wave radar is accurate, and the process is finished.
The present invention is not limited to the above-mentioned embodiments, and any changes or substitutions that can be easily understood by those skilled in the art within the technical scope of the present invention are intended to be included in the scope of the present invention. Therefore, the protection scope of the present invention should be subject to the claims.
Claims (12)
- The method relates to a calibrating vehicle, a road side millimeter wave radar and a processing unit, and comprises the following steps:1) Setting a sampling frequency which is not smaller than that of the road side millimeter wave radar for a vehicle-mounted positioning device of the calibration vehicle, and setting clock synchronization for the calibration vehicle with the vehicle-mounted positioning device, the road side millimeter wave radar and the processing unit;2) The calibration vehicle with the vehicle-mounted positioning device runs in the coverage area of the roadside millimeter wave radar according to a preset conventional route, and outputs first track data and second track data;3) The processing unit collects first track data and second track data generated by the calibration vehicle with the vehicle-mounted positioning device and the road side millimeter wave radar, obtains third track data after coordinate conversion, processes and analyzes the first track data and the third track data, and outputs a space-time similarity value of the first track data and the third track data;4) The processing unit judges whether the parameters of the millimeter wave radar at the open circuit side do not need to be calibrated according to the space-time similarity value and the space-time similarity threshold value, and if so, the calibration is ended; if the processing unit judges that the parameters of the millimeter wave radar at the open circuit side need to be calibrated, the parameters of the millimeter wave radar at the circuit side are calibrated;5) After calibrating the parameters of the road side millimeter wave radar, the processing unit performs the following verification steps:5.1 The calibration vehicle with the vehicle-mounted positioning device runs in the coverage area of the roadside millimeter wave radar according to a complex route or another conventional route which is designed in advance, and fourth track data and fifth track data are output;5.2 Executing the step 3) to output the space-time similarity value for the fourth track data and the fifth track data; executing step 4) on the output space-time similarity value;if the parameters of the millimeter wave radar on the circuit side are less than or equal to the space-time similarity threshold, the calibration parameters of the millimeter wave radar on the circuit side meet the requirements;if the calculated spatio-temporal similarity value is greater than the spatio-temporal similarity threshold, the processing may be performed according to the following three alternatives:5.2.1 No calibration is carried out on the calibration parameters of the millimeter wave radar, namely the calibration parameters are temporarily used as the calibration parameters of the radar according to the radar calibration parameters acquired for the first time;5.2.2 Calibrating the calibration parameters of the millimeter wave radar, namely temporarily serving as the calibration parameters of the radar according to the updated radar calibration parameters;5.2.3 Re-executing the steps 1) to 4), and ending the calibration if the space-time similarity value obtained in the step 4) is smaller than the space-time similarity threshold value; if the space-time similarity value obtained in the step 4) is greater than or equal to the space-time similarity threshold value, the millimeter wave radar is calibrated as a fault radar reporting processing unit.
- The method of claim 1, wherein the installation locations of the roadside millimeter wave radar are classified into three types, i.e., installed on the left side of the road, installed on the right side of the road, and installed in the center of the road; the millimeter wave radar has the capability of identifying a moving vehicle target and can output positioning data of the detected target.
- The method of claim 1, wherein the calibration vehicles with the vehicle-mounted positioning devices can be divided into two types according to different vehicle types, one type refers to an automatic driving vehicle with a high-precision positioning device embedded therein, and the automatic driving vehicle can accurately position the vehicle and output real-time vehicle positioning data; the other type of calibration vehicle with the positioning equipment outside the finger is provided with a high-precision positioning device, so that the vehicle position can be precisely positioned, and real-time vehicle positioning data is output through the processing unit; and the sampling frequency of the vehicle-mounted positioning device of the calibration vehicle is not smaller than that of the roadside millimeter wave radar.
- The method of claim 1, wherein the calibration of the roadside millimeter wave radar parameters refers to calibration of a roadside millimeter wave radar calibration parameter matrix, that is, correction and update of conversion parameters of a millimeter wave radar coordinate system into world coordinate system coordinates.
- The method of claim 1, wherein the processing unit has functions of collecting, processing, analyzing and uploading data; and meanwhile, the processing unit adjusts the time clocks of the calibrating vehicle with the vehicle-mounted positioning device and the millimeter wave radar at the road side, so that the time clocks are strictly synchronous with the clock time of the processing unit, and the clock synchronization is realized.
- The method of claim 5, wherein the processing unit employs one of three means:6.1 The cloud processing unit is connected with the road side millimeter wave radar and a calibration vehicle with a vehicle-mounted positioning device;6.2 A base station processing unit installed as an independent base station in a processing center;6.3 An embedded processing unit embedded in the roadside millimeter wave radar or the calibration vehicle.
- The method of claim 1, wherein the first track data and the second track data are obtained by: in the running process, the calibration vehicle of the vehicle-mounted positioning device generates target vehicle track data with a time stamp in real time, marks the target vehicle track data as first track data, and uploads the first track data to the processing unit, wherein the first track data is positioned under a world coordinate system; the road side millimeter wave radar synchronously acquires target vehicle track detection data with a time stamp, marks the target vehicle track detection data as second track data, the second track data is positioned under a radar coordinate system, and uploads the second track data to the processing unit.
- The method of claim 1, wherein the processing unit performs data processing and data analysis, comprising three steps: 1) coordinate conversion, 2) resampling, 3) space-time similarity calculation.
- The method of claim 8, wherein the method of coordinate transformation comprises:(1) selecting four groups of corresponding characteristic point pairs in the first track data and the second track data in a manual or automatic mode;(2) based on the four sets of feature point pairs, a first trajectory data point (x r ,y r ,z r ) And a second track number point (x v ,y v ,z v ) Is substituted into the followingCalculating a radar calibration parameter matrix H:(3) based on the solved radar calibration parameter matrix H, a second trajectory data point (x r ,y r ,z r ) Conversion to (x 'in world coordinate system' r ,y′ r ,z′ r ) As third track data:wherein the parameter x r Representing X coordinate of the roadside millimeter wave radar in a radar coordinate system and parameter X' r Representing new coordinate data after X coordinates of the roadside millimeter wave radar in a radar coordinate system are converted into a world coordinate system;parameter y r Representing Y coordinate of the roadside millimeter wave radar in a reference coordinate system and parameter Y' r Representing new coordinate data after Y coordinates of the roadside millimeter wave radar in a radar coordinate system are converted into a world coordinate system;Parameter z r Z coordinate of the road side millimeter wave radar in a reference coordinate system is represented, and the parameter Z' r Z coordinate of millimeter wave radar on road side in radar coordinate system is converted into world coordinate systemCoordinate data of (a) is provided.
- The method of claim 8, wherein the method of resampling comprises:(1) for the first track data and the third track data, a union T of sampling time points in the first track data and the third track data point set is obtained all ：T 1 ∪T 2 ＝T all(2) Using the union T of the determined sampling time points all The points in time at which the first track data and the third track data are missing, respectively, are checked. Supplementing the defect abscissa and the defect ordinate of the detected defect time point based on a cubic interpolation algorithm; fitting a cubic function G by known trajectory points x 、G y ：G x ＝ax 3 +bx 2 +cx+dG y ＝ex 3 +fx 2 +gx+h(3) By T all The time point is taken as the resampling time point and is obtained by a cubic function G x 、G y Resampling is achieved; at the moment, the two types of tracks are in the same coordinate system and the sampling time points are consistent, and the first track data are matched with the third track data corresponding to the road side millimeter wave radar one to one;wherein the parameters a, b, c, d represent a cubic interpolation function G for fitting the X-coordinate of the trajectory x Coefficients of (2);the parameters e, f, G, h represent the cubic interpolation function G for fitting the trajectory coordinate Y y Is a coefficient of (a).
- The method of claim 8, wherein the space-time similarity calculation means that space-time similarity calculation is performed on the first track data and the third track data after coordinate conversion and resampling; expression according to space-time similarity:substituting the first track data point and the third track data point pair under the same sampling time point respectively, and calculating a space-time similarity value of the first track data point and the second track data point;wherein the parameters areWeights representing spatial similarity, using weight parametersFreely adjusting the proportion of the space similarity and the time similarity;the parameter beta represents the weight of the similarity of the non-resampling point, and the weight parameter beta is adopted to freely adjust the proportion of the similarity of the non-resampling point and the similarity of the resampling point;parameter f S Representing spatial similarity, parameter f T Representing temporal similarity, parametersRepresenting similarity, parameters between non-resampled pointsRepresenting the similarity between resampling points.
- The method of claim 11, wherein if the similarity isLess than or equal to the similarity threshold deltaThe radar calibration parameter matrix H is considered to meet the precision requirement; if similar toIf the similarity threshold delta is larger than the similarity threshold delta, the radar calibration parameter matrix H is considered to not meet the precision requirement, and the roadside millimeter wave radar needs to be recalibrated.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110000327 | 2021-01-01 | ||
| CN2021100003279 | 2021-01-01 | ||
| PCT/CN2021/085149 WO2022141913A1 (en) | 2021-01-01 | 2021-04-01 | On-board positioning device-based roadside millimeter-wave radar calibration method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116635739A true CN116635739A (en) | 2023-08-22 |
Family
ID=87621634
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202180010978.4A Pending CN116635739A (en) | 2021-01-01 | 2021-04-01 | Road side millimeter wave radar calibration method based on vehicle-mounted positioning device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116635739A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117471461A (en) * | 2023-12-26 | 2024-01-30 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Road side radar service device and method for vehicle-mounted auxiliary driving system |
-
2021
- 2021-04-01 CN CN202180010978.4A patent/CN116635739A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117471461A (en) * | 2023-12-26 | 2024-01-30 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Road side radar service device and method for vehicle-mounted auxiliary driving system |
| CN117471461B (en) * | 2023-12-26 | 2024-03-08 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Road side radar service device and method for vehicle-mounted auxiliary driving system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117836653A (en) | Road side millimeter wave radar calibration method based on vehicle-mounted positioning device | |
| CN103176185B (en) | Method and system for detecting road barrier | |
| CN112836737A (en) | Roadside combined sensing equipment online calibration method based on vehicle-road data fusion | |
| CN113340325B (en) | System, method and medium for verifying vehicle-road cooperative roadside perception fusion precision | |
| CN108414238A (en) | Automatic parking function real steering vectors system and test method | |
| CN102208013A (en) | Scene matching reference data generation system and position measurement system | |
| CN114444158B (en) | Underground roadway deformation early warning method and system based on three-dimensional reconstruction | |
| CN107796373B (en) | Distance measurement method based on monocular vision of front vehicle driven by lane plane geometric model | |
| CN110864696A (en) | Three-dimensional high-precision map drawing method based on vehicle-mounted laser inertial navigation data | |
| CN112363167A (en) | Extended target tracking method based on fusion of millimeter wave radar and monocular camera | |
| CN112924955B (en) | Roadside laser radar point cloud coordinate dynamic correction method | |
| CN111325187B (en) | Lane position identification method and device | |
| JP6465421B1 (en) | Structural deformation detector | |
| CN114758504A (en) | Online vehicle overspeed early warning method and system based on filtering correction | |
| CN114488094A (en) | Vehicle-mounted multi-line laser radar and IMU external parameter automatic calibration method and device | |
| CN116635739A (en) | Road side millimeter wave radar calibration method based on vehicle-mounted positioning device | |
| CN114663473A (en) | Personnel target positioning and tracking method and system based on multi-view information fusion | |
| CN114485658A (en) | Device and method for precision evaluation of roadside sensing system | |
| CN112726351B (en) | Vehicle-mounted portable lightweight intelligent inspection method and system | |
| CN117198057A (en) | Experimental method and system for road side perception track data quality inspection | |
| CN206096937U (en) | Unmanned distribution vehicle | |
| CN117029840A (en) | Mobile vehicle positioning method and system | |
| CN117310627A (en) | Combined calibration method applied to vehicle-road collaborative road side sensing system | |
| CN113947141B (en) | Roadside beacon sensing system of urban intersection scene | |
| Zhang et al. | A Roadside Millimeter-Wave Radar Calibration Method Based on Connected Vehicle Technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |