CN114323050B

CN114323050B - Vehicle positioning method and device and electronic equipment

Info

Publication number: CN114323050B
Application number: CN202210012621.6A
Authority: CN
Inventors: 常海玥; 韩志华; 张旭
Original assignee: Suzhou Zhitu Technology Co Ltd
Current assignee: Suzhou Zhitu Technology Co Ltd
Priority date: 2022-01-07
Filing date: 2022-01-07
Publication date: 2024-07-19
Anticipated expiration: 2042-01-07
Also published as: CN114323050A

Abstract

The invention provides a vehicle positioning method, a vehicle positioning device and electronic equipment, wherein at least one target lane line matching pair meeting preset conditions is screened out from a plurality of acquired first lane line matching pairs; acquiring target relative displacement between each target matching point pair in each target lane line matching pair; and inputting the relative displacement of each target, the acquired state transfer equation and the initial parameters into a Kalman filter to obtain a positioning result of the vehicle. According to the method, the obtained plurality of first lane line matching pairs are preprocessed to obtain the target lane line matching pairs, and as the target lane line matching meets the preset condition that the visual perception lane lines and the map lane lines do not cross and/or the relative displacement is smaller than the preset distance, the information utilization rate and the accuracy rate of the target lane lines can be ensured, the vehicle positioning result is corrected according to the relative displacement between each target matching point pair in the target lane line matching pairs, and the accuracy and the precision of the positioning result can be improved.

Description

Vehicle positioning method and device and electronic equipment

Technical Field

The present invention relates to the field of automatic driving technologies, and in particular, to a vehicle positioning method, a vehicle positioning device, and an electronic device.

Background

In an autopilot system, the accuracy of real-time positioning directly affects the implementation of accurate navigation. If the positioning error reaches the centimeter level, the intelligent automobile can accurately judge the surrounding environment and the position of the map where the intelligent automobile is located, and especially after high-precision map data are accessed, the navigation precision of the automobile can be butted to a certain position of a certain lane line, so that the automatic driving result is greatly improved. In the related art, a fusion is usually performed between a GNSS (Global Navigation SATELLITE SYSTEM ) and an IMU (Inertial Measurement Unit) to obtain a relatively accurate position, but when the GNSS signal is poor or lost, the positioning obtained by inertial navigation alone cannot ensure a relatively stable effect, and errors tend to accumulate and diverge over time, resulting in poor positioning accuracy and precision.

Disclosure of Invention

The invention aims to provide a vehicle positioning method, a vehicle positioning device and electronic equipment, so that the improvement of the vehicle positioning precision and accuracy is relieved.

The invention provides a vehicle positioning method, which comprises the following steps: acquiring a plurality of first lane line matching pairs, a state transfer equation and initial parameters of an extended Kalman filter, wherein the first lane line matching pairs and the state transfer equation and the initial parameters are matched with each other, and the first lane line matching pairs and the state transfer equation and the initial parameters are matched with each other; screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line are not crossed, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance; acquiring target relative displacement between each target matching point pair in each target lane line matching pair; and inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle.

Further, the step of screening at least one target lane line matching pair satisfying a preset condition from the plurality of first lane line matching pairs includes: deleting a designated lane line matching pair from the plurality of first lane line matching pairs to obtain at least one target lane line matching pair meeting a preset condition; wherein, the appointed lane line matching pair includes: under the overpass road section scene, the visual perception lane line and the map lane line which are mutually intersected.

Further, the step of screening at least one target lane line matching pair satisfying a preset condition from the plurality of first lane line matching pairs includes: for each first lane line matching pair, acquiring a plurality of first matching point pairs in the current first lane line matching pair; calculating average relative displacement of a plurality of first relative displacements according to the first relative displacement between two matching points in each first matching point pair; if the average relative displacement is larger than a first preset distance, deleting the current first lane line matching pair; and taking the next first lane line matching pair as a new current first lane line matching pair, and repeatedly executing the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair to obtain at least one target lane line matching pair meeting the preset condition.

Further, the step of screening at least one target lane line matching pair satisfying a preset condition from the plurality of first lane line matching pairs includes: for each first lane line matching pair, acquiring a plurality of first matching point pairs in the current first lane line matching pair; calculating a second relative displacement between two matching points in each first matching point pair; selecting a plurality of designated matching point pairs with second relative displacement smaller than a second preset distance from the plurality of first matching point pairs; determining lane line matching sections corresponding to the plurality of specified matching point pairs as target lane line matching pairs meeting preset conditions corresponding to the current first lane line matching; and taking the next first lane line matching pair as a new current first lane line matching pair, and repeatedly executing the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair to obtain at least one target lane line matching pair meeting the preset condition.

Further, before the step of determining the lane line matching segments corresponding to the plurality of specified matching point pairs as the target lane line matching pairs meeting the preset condition corresponding to the current first lane line matching, the method further includes: if the plurality of specified matching point pairs are a plurality of continuous matching point pairs, lane line matching sections corresponding to the plurality of specified matching point pairs are generated based on the plurality of continuous matching point pairs.

Further, the step of inputting the target relative displacement, the state transition equation and the initial parameters between each target matching point pair into the kalman filter to obtain the positioning result of the vehicle includes: inputting target relative displacement, a state transfer equation and initial parameters between each target matching point pair into a Kalman filter, and outputting a first jacobian matrix corresponding to the state transfer equation; outputting a positioning predicted value and an error covariance predicted value of the vehicle based on the state transition equation and the first jacobian matrix; and determining a positioning result of the vehicle based on the positioning predicted value and the error covariance predicted value.

Further, the step of determining the positioning result of the vehicle based on the positioning prediction value and the error covariance prediction value includes: obtaining an observation equation of an extended Kalman filter; based on an observation equation of the extended Kalman filter, determining a second jacobian matrix corresponding to the observation equation; determining updated Kalman yields based on the observation equation, the second jacobian matrix and the error covariance prediction value; and updating the positioning predicted value based on the positioning predicted value, the updated Kalman gain and the observation equation to obtain a positioning estimated value of the vehicle at the current moment, and determining the positioning estimated value as a positioning result of the vehicle.

The invention provides a vehicle positioning device, which comprises: the first acquisition module is used for acquiring a plurality of first lane line matching pairs, a state transfer equation and initial parameters of the extended Kalman filter, wherein the first lane line matching pairs correspond to the visual perception lane lines and the map lane lines; the screening module is used for screening at least one target lane line matching pair meeting the preset condition from the plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line are not crossed, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance; the second acquisition module is used for acquiring target relative displacement between each target matching point pair in each target lane line matching pair; and the third acquisition module is used for inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into the Kalman filter to obtain the positioning result of the vehicle.

The invention provides an electronic device comprising a processor and a memory, the memory storing machine executable instructions executable by the processor, the processor executing the machine executable instructions to implement the vehicle locating method of any one of the above.

The present invention provides a machine-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to implement a vehicle positioning method of any of the above.

According to the vehicle positioning method, the vehicle positioning device and the electronic equipment, a plurality of first lane line matching pairs, which are matched with each other, of the visual perception lane lines and the corresponding map lane lines, and state transfer equations and initial parameters of an extended Kalman filter are obtained; screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line are not crossed, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance; acquiring target relative displacement between each target matching point pair in each target lane line matching pair; and inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle. According to the method, the obtained plurality of first lane line matching pairs are preprocessed to obtain the target lane line matching pairs, and as the target lane line matching meets the preset condition that the visual perception lane lines and the map lane lines do not cross and/or the relative displacement is smaller than the preset distance, the information utilization rate and the accuracy rate of the target lane lines can be ensured, the vehicle positioning result is corrected according to the relative displacement between each target matching point pair in the target lane line matching pairs, and the accuracy and the precision of the positioning result can be improved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a vehicle positioning method according to an embodiment of the present invention;

FIG. 2 is a flow chart of another method for locating a vehicle according to an embodiment of the present invention;

fig. 3 is a schematic diagram of a lane matching pair according to an embodiment of the present invention;

FIG. 4 is a flowchart of another vehicle positioning method according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of another lane matching pair according to an embodiment of the present invention;

FIG. 6 is a flowchart of another vehicle positioning method according to an embodiment of the present invention;

FIG. 7 is a schematic diagram of another lane matching pair according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of a positioning system framework according to an embodiment of the present invention;

FIG. 9 is a schematic diagram of a vehicle positioning method according to an embodiment of the present invention;

fig. 10 is a schematic structural diagram of a vehicle positioning device according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

The technical solutions of the present invention will be clearly and completely described in connection with the embodiments, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

In an autopilot system, the accuracy of real-time positioning directly affects the implementation of accurate navigation. If the positioning error reaches the centimeter level, the intelligent automobile can accurately judge the surrounding environment and the position of the map where the intelligent automobile is located, and especially after high-precision map data are accessed, the navigation precision of the automobile can be butted to a certain position of a certain lane line, so that the automatic driving result is greatly improved.

The real-time positioning technology at the present stage is applied to the fields of automatic driving, auxiliary driving, unmanned vehicles, unmanned planes, robots, map acquisition vehicles, high-speed rail cars, ships, aviation and the like. Taking autopilot as an example, the integration of various sensors can be combined to enable the positioning effect to be more ideal, such as a laser radar, a camera, an ultrasonic radar, a millimeter wave radar, an RTK (Real-TIME KINEMATIC, real-time differential positioning), an IMU (inertial sensor) and the like. However, each sensor has own advantages and disadvantages, and an excellent fusion scheme is difficult to find in practical application, so that the positioning technology cannot break through the technical bottleneck.

Although the GPS (Global Positioning System ) signal navigation can realize the starting of the automobile from the self position to the destination, the precision error and the ideal value are in the range of 0-10 m, the satellite positioning precision is also improved to about 5m along with the increasing of available satellites, however, the precision also presents obvious defects along with the development of automatic driving, for example, the width of a lane line is about 3.8 meters, the satellite positioning error can not accurately position the automobile on a certain lane line, the precision of about 5m is likely to lead to the positioning of a wrong lane, the positioning is a basis for automatic driving and navigation, and the positioning error is too large to lead to the positioning of the automobile to be unexpected in the process of turning. The GNSS and the IMU are fused in the industry, so that a more accurate position is obtained, but when the GNSS signal is poor or lost, the positioning obtained by the IMU alone cannot guarantee a stable effect, and errors are accumulated and diverged often with the lapse of time, so that the positioning result is not available.

The intelligent car body often does not use the IMU as a unique navigation sensor and needs to be matched with a chassis, a visual sensor, a radar and the like for use. Radar is also classified into laser radar, millimeter wave radar, ultrasonic radar, and the like. The laser radar has the advantages that the measurement precision and the detection distance are guaranteed, but the laser radar is limited by a movable shielding object, semantic information such as colors on a plane is lost, the cost is high, and the service life is short; the angle detectable by the millimeter wave radar is limited, and a plurality of sensors can realize accurate positioning; the detection distance of the ultrasonic radar is too small, and the precision is low; the use of radar positioning is complicated and complicated by the need to exclude the unavoidable factors, and also takes the cost problem into consideration, so that no major breakthrough in the positioning field exists.

Most of the positioning based on vision is based on two-dimensional images, and more or less information loss and calculation errors, even vision cheating generated under special illumination and other environments, are necessarily generated when the two-dimensional images are converted into three-dimensional worlds. Because the visual positioning is simply used to miss depth information, the situation that the automobile runs on the road and is poor in transverse and longitudinal positioning can occur.

The situation generally appears in the automatic driving field of the passenger car, and for the commercial car, the influence of factors such as high visual angle, heavy tonnage, wide wheel track, long wheel base, whether rear pulling exists or not improves the positioning difficulty of the commercial collection card more.

The following three problems exist in actual positioning due to poor fusion of various sensors when the commercial vehicle runs at a high speed in a straight line: (1) When the GNSS signals are bad or disappear, such as poor rainy weather signals, the vehicle runs in a tunnel, a region above the overpass and the like is shielded, the transverse positioning errors of the commercial vehicle can be accumulated and dispersed, the positioning accuracy is reduced, and even the commercial vehicle cannot be positioned. (2) When the GNSS signal is suddenly lost to recovery, the positioning accuracy needs a period of time to be corrected to a state before the signal is lost. (3) When the GNSS signals are from no to no, the positioning result can generate transverse huge jump and then repair gradually. Based on the above, the embodiment of the invention provides a vehicle positioning method, a device and electronic equipment, and the technology can be applied to a scene where a vehicle needs to be positioned in automatic driving.

For the convenience of understanding the present embodiment, first, a vehicle positioning method disclosed in the embodiment of the present invention will be described in detail; as shown in fig. 1, the method comprises the steps of:

step S102, a plurality of first lane line matching pairs, which are matched with each other, of the visual perception lane lines corresponding to the map lane lines, a state transition equation of the extended Kalman filter and initial parameters are obtained.

The visual perception lane line can be collected by the intelligent camera; the map lane line may be a lane line within a preset range acquired from a map based on a current position of the vehicle; the perceived visual perception lane lines can be matched with a plurality of map lane lines in the local map through a related technology such as a KM algorithm (an algorithm for searching for the best matching of the weighted bipartite graph) to obtain a first lane line matching pair; the extended Kalman filter is an extended form of standard Kalman filtering under the nonlinear condition, is a high-efficiency recursive filter, and has the basic ideas that a Taylor series expansion is utilized to linearize a nonlinear system, and then a Kalman filtering frame is adopted to filter signals; the above state transition equation may be s _k＝f(s_k-1)+u_k, which represents the state transition of the vehicle from the state at time k-1 to the state at time k, where s _k＝[x_k y_k z_k θΦγ ], which represents the state vector of the vehicle at time k, x _k、y_k、z_k, θ, Φ, and γ represent the pose of the vehicle in the northeast day coordinate system, respectively corresponding to the northeast direction, the north direction, the sky direction, and the roll angle, pitch angle, and yaw angle (euler angle) of the vehicle in the northeast day direction, s _k-1 represents the state vector of the vehicle at time k-1, u _k to N (Q, Q) are state noises, Q and Q represent the distribution parameters of the noises, and are the expected and variance of the gaussian distribution, respectively; according to Taylor expansionS _k＝f(s'_k-1)+F_k-1(s_k-1-＜s'_k-1＞)+u_k is obtained. The initial parameters generally include an initial state vector s ₀ of the vehicle and an initial error covariance matrix P ₀; in practical implementation, when it is required to accurately locate a determined vehicle, it is generally required to first acquire the above-mentioned first lane line matching pair, the state transition equation of the extended kalman filter, and the initial parameters.

Step S104, screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line do not intersect, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance.

The preset distance can be set according to actual requirements; in actual implementation, the obtained first lane line matching pairs may have a situation of mismatching or bad matching, so that a plurality of first lane line matching pairs are usually required to be preprocessed to select one or more target lane line matching pairs meeting preset conditions; for example, for a road section with a viaduct, as a high-precision map cannot be layered in the elevation direction in a bird's eye view, a situation of transverse and longitudinal cross matching between a visual perception lane line and a map lane line may occur, and the situation belongs to wrong matching and generally needs to be deleted; it is also possible that the visually perceived lane line matches the map lane line correctly, but due to vehicle shake, sudden braking, cornering or deceleration movements, there may be situations where the slope or line type of the visually perceived lane line is inaccurate, etc., where the first lane line matching pair in such a scenario needs to be split to intercept a junction or set of points satisfying the preset condition.

Step S106, obtaining the target relative displacement between each target matching point pair in each target lane line matching pair.

And after the one or more target lane line matching pairs are obtained, each point in the visual perception lane line is obtained from each target lane line matching pair, and a matching point corresponding to the point on the map lane line is obtained, so that the target relative displacement between the two matched points is calculated.

And S108, inputting the target relative displacement, the state transition equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle.

And (3) taking the obtained target relative displacement between each target matching point pair as an observation value, inputting a state value corresponding to the state transition equation and the initial s ₀ and the initial P ₀ set at the time of k=0 into a Kalman filter, and obtaining a positioning result of the vehicle through the Kalman filter processing.

According to the vehicle positioning method, a plurality of first lane line matching pairs, which are matched with each other, of the visual perception lane lines and the corresponding map lane lines, and a state transfer equation and initial parameters of an extended Kalman filter are obtained; screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line are not crossed, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance; acquiring target relative displacement between each target matching point pair in each target lane line matching pair; and inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle. According to the method, the obtained plurality of first lane line matching pairs are preprocessed to obtain the target lane line matching pairs, and as the target lane line matching meets the preset condition that the visual perception lane lines and the map lane lines do not cross and/or the relative displacement is smaller than the preset distance, the information utilization rate and the accuracy rate of the target lane lines can be ensured, the vehicle positioning result is corrected according to the relative displacement between each target matching point pair in the target lane line matching pairs, and the accuracy and the precision of the positioning result can be improved.

The embodiment of the invention also provides another vehicle positioning method, which is realized on the basis of the method of the embodiment; as shown in fig. 2, the method comprises the steps of:

step S202, a plurality of first lane line matching pairs, which are matched with each other, of the visual perception lane lines corresponding to the map lane lines, a state transition equation of the extended Kalman filter and initial parameters are obtained.

Step S204, deleting a designated lane line matching pair from a plurality of first lane line matching pairs to obtain at least one target lane line matching pair meeting preset conditions; wherein, the appointed lane line matching pair includes: under the overpass road section scene, the visual perception lane line and the map lane line which are mutually intersected.

Referring to a schematic diagram of a lane line matching pair shown in fig. 3, wherein a thin solid line is a visually perceived lane line, and a thick solid line is a map lane line; under the scene of a viaduct road section or the scene of shielding by other shielding objects, as the high-precision map cannot be layered in the elevation direction in the aerial view, namely the local map taken in the high-precision map has no height information, the situation that the vehicle is in the upper lane or the lower lane of the viaduct cannot be judged in a layered manner temporarily, and the situation that the lane lines are transversely and longitudinally crossed and matched with each other possibly occurs, and the wrong matching needs to be removed according to the constraint of the extending direction of the lane lines is required.

Step S206, obtaining the target relative displacement between each target matching point pair in each target lane line matching pair.

And step S208, inputting the target relative displacement, the state transition equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle.

According to the vehicle positioning method, the matched pairs of the visual perception lane lines matched with each other and the first lane lines corresponding to the map lane lines, the state transfer equation of the extended Kalman filter and the initial parameters are obtained. Deleting a designated lane line matching pair from the plurality of first lane line matching pairs to obtain at least one target lane line matching pair meeting a preset condition; wherein, the appointed lane line matching pair includes: under the overpass road section scene, the visual perception lane line and the map lane line which are mutually intersected. And acquiring target relative displacement between each target matching point pair in each target lane line matching pair. And inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle. According to the method, the obtained plurality of first lane line matching pairs are preprocessed to obtain the target lane line matching pairs, and as the target lane line matching meets the preset condition that the visual perception lane lines and the map lane lines do not cross and/or the distance is smaller than the preset distance, the information utilization rate and the accuracy rate of the target lane lines can be ensured, the vehicle positioning result is corrected according to the relative displacement between each target matching point pair in the target lane line matching pairs, and the accuracy and the precision of the positioning result can be improved.

The embodiment of the invention also provides another vehicle positioning method, which is realized on the basis of the method of the embodiment; as shown in fig. 4, the method comprises the steps of:

Step S402, a plurality of first lane line matching pairs, which are matched with each other, of the visual perception lane lines corresponding to the map lane lines, a state transition equation of the extended Kalman filter and initial parameters are obtained.

Step S404, for each first lane line matching pair, a plurality of first matching point pairs in the current first lane line matching pair are obtained.

In actual implementation, the visual perception lane line in the current first lane line matching pair is composed of a plurality of points, the map lane line is also composed of a plurality of points, and a plurality of first matching point pairs corresponding to the visual perception lane line and the map lane line in the current first lane line matching pair are obtained.

Step S406, calculating the average relative displacement of the first relative displacements according to the first relative displacement between the two matching points in each first matching point pair.

And respectively calculating first relative displacement between each point of the visual perception lane line in the current first lane line matching pair and the point corresponding to the map lane line, and averaging the obtained first relative displacements to obtain average relative displacement of the first relative displacements.

In step S408, if the average relative displacement is greater than the first preset distance, the current first lane line matching pair is deleted.

The first preset distance may be set according to actual requirements, for example, the first preset distance may be 0.5 lane widths or the like; after the average relative displacement is obtained, the average relative displacement can be compared with a first preset distance, if the average relative displacement is smaller than or equal to the first preset distance, the current first lane line matching pair can be considered as an available lane line matching pair, and if the average relative displacement is larger than the first preset distance, the current first lane line matching pair can be considered as an incorrect matching and needs to be deleted. As another schematic diagram of a lane line matching pair shown in fig. 5, the lane line mismatching that occurs due to the inaccuracy of the lane line of the high-precision map may be deleted using, as a constraint, that the average error of the matching pair point set cannot be greater than 0.5 lanes wide.

Step S410, taking the next first lane line matching pair as a new current first lane line matching pair, and repeating the step of obtaining a plurality of first matching point pairs in the current first lane line matching pair, so as to obtain at least one target lane line matching pair meeting the preset condition.

And preprocessing each first lane line matching pair in the mode to finally obtain one or more target lane line matching pairs meeting preset conditions.

In step S412, the target relative displacement between each target matching point pair in each target lane line matching pair is acquired.

In step S414, the target relative displacement, the state transition equation and the initial parameters between each target matching point pair are input into the kalman filter, so as to obtain the positioning result of the vehicle.

According to the vehicle positioning method, the matched pairs of the visual perception lane lines matched with each other and the first lane lines corresponding to the map lane lines, the state transfer equation of the extended Kalman filter and the initial parameters are obtained. And acquiring a plurality of first matching point pairs in the current first lane line matching pair aiming at each first lane line matching pair. An average relative displacement of the plurality of first relative displacements is calculated based on the first relative displacements between the two matching points in each of the first matching point pairs. And if the average relative displacement is larger than the first preset distance, deleting the current first lane line matching pair. And taking the next first lane line matching pair as a new current first lane line matching pair, and repeatedly executing the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair to obtain at least one target lane line matching pair meeting the preset condition. And acquiring target relative displacement between each target matching point pair in each target lane line matching pair. And inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle. According to the method, the obtained plurality of first lane line matching pairs are preprocessed to obtain the target lane line matching pairs, and as the target lane line matching meets the preset condition that the visual perception lane lines and the map lane lines do not cross and/or the distance is smaller than the preset distance, the information utilization rate and the accuracy rate of the target lane lines can be ensured, the vehicle positioning result is corrected according to the relative displacement between each target matching point pair in the target lane line matching pairs, and the accuracy and the precision of the positioning result can be improved.

The embodiment of the invention also provides another vehicle positioning method, which is realized on the basis of the method of the embodiment; as shown in fig. 6, the method includes the steps of:

Step S602, a plurality of first lane line matching pairs, which are matched with each other, of the visual perception lane lines corresponding to the map lane lines, a state transition equation of the extended Kalman filter and initial parameters are obtained.

Step S604, for each first lane line matching pair, a plurality of first matching point pairs in the current first lane line matching pair are obtained.

In step S606, a second relative displacement between two matching points in each of the first matching point pairs is calculated.

In step S608, a plurality of designated matching point pairs with a second relative displacement smaller than a second preset distance are selected from the plurality of first matching point pairs.

The second preset distance may be set according to different road sections or different matching accuracy, for example, the second preset distance may be 1 meter or the like; in actual implementation, the second relative displacement between each point of the visual perception lane line in the current first lane line matching pair and the point corresponding to the map lane line is calculated respectively, for example, any lane line point in the visual perception lane line and the map lane line in the current first lane line matching pair can be integrated, the projection distance, namely the vertical distance, between each point and the other lane line is calculated, and if the projection distance is smaller than the second preset distance, the point and the matching point corresponding to the point can be attributed to the specified matching point pair.

In step S610, if the plurality of specified matching point pairs are consecutive matching point pairs, lane line matching segments corresponding to the plurality of specified matching point pairs are generated based on the consecutive matching point pairs.

When the specified matching point pair is a continuous matching point pair, that is, when the second relative displacement between a plurality of continuous points and corresponding matching points is smaller than the second preset distance, the plurality of continuous points on the corresponding visual perception lane line of the specified matching point pair can be linearly fitted to form a junction section, and the plurality of corresponding continuous points on the map lane line are linearly fitted to form a junction section, so that lane line matching sections corresponding to the specified matching points are obtained.

Step S612, determining lane line matching sections corresponding to the specified matching point pairs as target lane line matching pairs meeting preset conditions corresponding to the current first lane line matching.

Step S614, the next first lane line matching pair is taken as a new current first lane line matching pair, and the step of obtaining a plurality of first matching point pairs in the current first lane line matching pair is repeatedly performed, so as to obtain at least one target lane line matching pair meeting the preset condition.

In practical implementation, the difference matching caused by the information reading error caused by the vision sensor needs to be removed, and particularly, the commercial vehicle is different from the passenger vehicle in that the characteristics of longitudinal shaking and transverse shaking of the vehicle body in the driving process can be increased due to the weight, the length and the like of the commercial vehicle, the influence of centripetal force is serious in the turning process, the selectable mounting position of the vision sensor is limited, the influence of participation of the camera in external parameters is larger in the process of sensing the vision lane lines, and the effect of taking as raw pixel data is larger in the process of taking as the raw pixel dataRemoving distortion to obtainAnd then useCorresponding errors can occur when the three-dimensional image is converted into a world coordinate system or other three-dimensional coordinate systems, wherein u and v are coordinate values of any pixel point in a two-dimensional image acquired by a vision sensor; u 'and v' are coordinate values of the pixel point under a pixel coordinate system after distortion is removed; x _W、Y_W and Z _W are coordinate values of the pixel point in a three-dimensional world coordinate system; r represents a rotation matrix that rotates points from a three-dimensional world coordinate system to a three-dimensional camera coordinate system; t represents a translation vector of a point rotated from a three-dimensional world coordinate system to a three-dimensional camera coordinate system; f _x、f_y、u'₀ and v' ₀ are internal parameters of the camera, where f _x＝f*s_x,fy＝f*s_y, f are focal lengths, and s _x and s _y are the inverse of the pixel cell size (units: pixels/μm). By passing throughThe three-dimensional position of the point under the camera coordinate system can be obtainedIn colloquial terms, zc can be understood as depth information of the same plane point or distance from the camera.

The purpose of converting to the three-dimensional coordinate system is to match with the map lane line, because the conversion between the coordinate systems (R, T above) is determined by the external parameters calibrated by the sensor (camera), and the calibration itself has errors, the matching precision of the lane line is necessarily affected when the coordinate system conversion is made, and the matching precision of the three-dimensional coordinate system based on the conversion and the map lane line is poor, which is specifically as follows:

referring to another schematic diagram of the lane line matching pair shown in fig. 7, as shown in fig. 7 (a), when the commercial vehicle body is in high-speed straight line driving and the shake is serious or the lane is changed suddenly, the slope of the lane line perceived by the camera (corresponding to the above visual perception lane line) is different from that of the lane line of the high-precision map under the influence of inertia. As shown in fig. 7 (b), when the lane information obtained by sensing semantic division exceeds the distortion removal range of the wide-angle camera during emergency braking, the lane is distorted, and the lane which should be sensed as being straight is erroneously divided into curves. As shown in fig. 7 (c), the vehicle turns while performing a deceleration motion, and the commercial vehicle has a large number of axles and a long length, which may cause the vehicle head to tilt upward or temporarily leave the ground, while the front view camera is installed at a high position, so that the length of the perceived lane line is shortened, and the original curve is perceived as a straight line. As shown in fig. 7 (d), when the vehicle turns at a high speed, the vehicle head is deflected in the opposite direction of the turn due to the centripetal force, so that the camera transversely forms an immeasurable angle with the ground, and the curvature of the perceived lane line (corresponding to the visually perceived lane line) is different from that of the lane line on the high-precision map.

In the above four cases, although the matching pair is correct, the problem is that the sensing of the visual sensor is the root cause, and the result after the matching cannot be cut into one piece, in this case, the lane lines obtained by the visual sensing need to be divided, so that the junction sections or the point sets meeting the requirements are involved in fusion positioning.

In step S616, the target relative displacement between each target matching point pair in each target lane line matching pair is acquired.

It should be noted that, if the designated matching point pair is a discrete matching point pair, the target relative displacement between two matching points in the discrete matching point pair may be directly obtained.

Step S618, inputting the target relative displacement, the state transition equation and the initial parameters between each target matching point pair into the Kalman filter to obtain the positioning result of the vehicle.

According to the vehicle positioning method, the matched pairs of the visual perception lane lines matched with each other and the first lane lines corresponding to the map lane lines, the state transfer equation of the extended Kalman filter and the initial parameters are obtained. And acquiring a plurality of first matching point pairs in the current first lane line matching pair aiming at each first lane line matching pair. A second relative displacement between two matching points in each first matching point pair is calculated. And selecting a plurality of designated matching point pairs with second relative displacement smaller than a second preset distance from the plurality of first matching point pairs. If the plurality of specified matching point pairs are a plurality of continuous matching point pairs, lane line matching sections corresponding to the plurality of specified matching point pairs are generated based on the plurality of continuous matching point pairs. And determining lane line matching sections corresponding to the plurality of specified matching point pairs as target lane line matching pairs which meet preset conditions and correspond to the current first lane line matching. And taking the next first lane line matching pair as a new current first lane line matching pair, and repeatedly executing the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair to obtain at least one target lane line matching pair meeting the preset condition. And acquiring target relative displacement between each target matching point pair in each target lane line matching pair. And inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle. According to the method, the obtained plurality of first lane line matching pairs are preprocessed to obtain the target lane line matching pairs, and as the target lane line matching meets the preset condition that the visual perception lane lines and the map lane lines do not cross and/or the distance is smaller than the preset distance, the information utilization rate and the accuracy rate of the target lane lines can be ensured, the vehicle positioning result is corrected according to the relative displacement between each target matching point pair in the target lane line matching pairs, and the accuracy and the precision of the positioning result can be improved.

The embodiment of the invention also provides another vehicle positioning method, which is realized on the basis of the method of the embodiment; the method comprises the following steps:

Step one, a plurality of first lane line matching pairs, which are matched with each other, of the visual perception lane lines corresponding to the map lane lines, and a state transition equation and initial parameters of an extended Kalman filter are obtained.

Screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line do not intersect, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance.

And thirdly, acquiring target relative displacement between each target matching point pair in each target lane line matching pair.

And step four, inputting the target relative displacement, the state transition equation and the initial parameters between each target matching point pair into a Kalman filter, and outputting a first jacobian matrix corresponding to the state transition equation.

In practical implementation, after the target relative displacement, the state transition equation and the initial parameters between each target matching point pair are input into the kalman filter, a first jacobian matrix corresponding to the state transition equation can be obtained, specifically, in combination with the foregoing embodiment, it can be known that, according to the taylor expansion, the obtained state transition equation s _k＝f(s'_k-1)+F_k-1(s_k-1-＜s'_k-1＞)+u_k after the expansion is that the first jacobian matrix at < s' _k-1 > is

And fifthly, outputting a positioning predicted value and an error covariance predicted value of the vehicle based on the state transition equation and the first jacobian matrix.

In actual implementation, the state transition equation and the first jacobian matrix can be combined to obtain a predicted state of the system, that is, a positioning predicted pose (corresponding to the positioning predicted value) s' _k＝F_k-1s'_k-1 of the vehicle; the corresponding error covariance predicted value isWherein P' _k-1 represents the predicted value of the error covariance matrix at time k-1.

And step six, determining a positioning result of the vehicle based on the positioning predicted value and the error covariance predicted value.

The sixth step can be specifically realized by the following steps a to D:

and step A, obtaining an observation equation of the extended Kalman filter.

And B, determining a second jacobian matrix corresponding to the observation equation based on the observation equation of the extended Kalman filter.

The state transition equation and the observation mode in the embodiment may be determined according to an update method of the system and an observation method of the sensor, where the observation equation may be z _k＝h(s_k)+v_k, where h (s _k) represents a process equation for mapping a state vector to a control in which the measured value is located; v _k -N (R, R) represent observation noise, and z _k＝h(s'_k)+H_k(s_k-s'_k)+v_k is obtained according to Taylor expansion, wherein h (s '_k) represents a predicted observation value obtained by inputting a predicted pose s' _k into an observation equation; the second jacobian matrix at s' _k is

And C, determining updated Kalman yields based on the observation equation, the second jacobian matrix and the error covariance predicted value.

And D, updating the positioning predicted value based on the positioning predicted value, the updated Kalman gain and the observation equation to obtain a positioning estimated value of the vehicle at the current moment, and determining the positioning estimated value as a positioning result of the vehicle.

After the positioning predicted value and the error covariance predicted value of the vehicle are obtained, the updated Kalman gain can be obtained asAccording to the updated Kalman gain, the positioning predicted value and the observation equation, the positioning estimated value of the vehicle at the current moment is less than s' _k＞＝s'_k+K_k(z_k-h(s'_k), and the relative displacement between the corresponding line pair or the corresponding point pair of the visual perception and the high-precision map lane line matching pair is substituted into calculation. And finally updating the error covariance matrix to be P _k＝P'_k-K_kH_kP'_k.

According to the vehicle positioning method, the removed wrong lane line matching result is added into the filter of the positioning system to obtain the visual fusion positioning result, so that the transverse deviation of lane line level positioning in the vehicle driving process is ensured to be within an acceptable range.

In order to further understand the above embodiment, referring to a schematic diagram of a positioning system frame shown in fig. 8, sensors in automatic driving are divided into sensors, radars, chassis and visual perception information combined by GNSS and IMU, and lane line feature matching can be performed based on the visual perception information and high-precision map information for matching and fusion, and positioning results of the vehicle are obtained by combining the fusion of GNSS and IMU, radar fusion and chassis fusion. When the method is specifically implemented, the general fusion technical scheme needs to be confirmed firstly, and the fusion scheme is determined according to different types, different numbers, different precision and different working modes of sensors installed in a commercial vehicle:

The sensor with higher precision, such as radar distance observation precision, can properly improve the distance confidence coefficient with the surrounding wall surface or road edge in running, the chassis wheel speed can correct the longitudinal pose better, and the semantic segmentation result obtained by visual perception can be matched with a high-precision map to correct the transverse error greatly.

The multiple sensors or a plurality of sensors run simultaneously, the time for acquiring data according to each sensor is divided successively, the frequency for acquiring the data cannot be unified, the data calculation mode and the complexity are different, the compatibility problem is clear, and a foundation is laid for the subsequent positioning result output of fusion vision and high-precision map lane line matching. Compatibility problems include: whether the data acquired by the sensors are processed by an internal system or not and whether the time sequence is reasonable or not are the situation that each sensor repels or not, for example, visual perception tells the vehicle to deviate to the right side of a lane line at the moment, a differential model of the wheel speed positions the vehicle to the left side of the lane line, and after the respective precision of the two sensors is measured, observation noise can be adjusted, so that the problem of repulsion is solved. The reliability of data obtained by each sensor can be coordinated by reasonably adjusting the observation noise of an observation equation in the Kalman filtering.

Referring to a schematic diagram of a vehicle positioning mode shown in fig. 9, a plurality of lane line matching pairs (corresponding to the first lane line matching pair) obtained after matching the visual perception and the high-precision map lane lines are preprocessed to obtain available matching pairs (corresponding to the target lane line matching pairs), the relative displacement between two matching point pairs corresponding to each available matching pair is input as an observed value to a kalman filter, state noise corresponds to u _k, a positioning system state corresponds to f (s _k-1), the state noise and the positioning system state are combined to obtain a state transition equation s _k＝f(s_k-1)+u_k, s _k is input as a state value to the kalman filter, an initialization s ₀ and an error covariance matrix P ₀ when k=0 are set are input to the kalman filter, a first jacobian matrix corresponding to the state transition equation can be output after the input data is processed by the kalman filter, a positioning predicted pose s' _k＝F_k-1s'_k-1 of the vehicle can be obtained by a prediction module, and an error covariance predicted value is obtainedThe state observer corresponds to h (s _k), the observation noise is v _k -N (R, R), after the state observer is combined with the observation noise, an observation equation z _k＝h(s_k)+v_k is obtained, the observation equation is input into the optimal state estimator, a second jacobian matrix corresponding to the observation equation can be determined, the positioning prediction pose and the error covariance predicted value are updated based on the positioning prediction pose, the observation equation, the second jacobian matrix and the error covariance predicted value, updated Kalman gain and a vision fusion positioning result < s '_k＞＝s'_k+K_k(z_k-h(s'_k) can be obtained, the updated error covariance matrix is P _k, the obtained result is input into the positioning system state again, the vehicle positioning at the next moment is improved, the vision fusion positioning result < s' _k＞＝s'_k+K_k(z_k-h(s'_k) is taken as output, and the optimal estimation of the vehicle positioning system state at the current moment is realized according to the positioning result of the vehicle.

The vehicle positioning method aims at improving the transverse positioning when the satellite signal is weak or not, the positioning after the IMU or the wheel speed is calculated is in the transverse inaccuracy or divergence, and the relative displacement between two matching points in the matching point pair can be used for improving the transverse positioning, namely, the intelligent vehicle can learn how far the positioning is from the real positioning according to how far the visual perception lane line perceived by the intelligent vehicle is from the high-precision map lane line, so that the positioning is corrected according to how far the positioning is from the real positioning.

Compared with the classical single GNSS and IMU fusion, the vehicle positioning mode has the advantage that after visual perception and high-precision map matching fusion are added, the positioning effect is improved. When the GNSS signals are weak or the signals disappear, the visual perception of the high-precision map fusion can enable the positioning errors not to accumulate and diverge. When GNSS signals suddenly change, the positioning effect is better maintained by adding visual perception into high-precision map fusion, so that the GNSS signals are in a relatively stable positioning state. The invention can fully utilize various knowledge of the sensing system and the high-precision map, increases the utilization rate of the sensing sensor and the information, improves the transverse positioning precision of the commercial vehicle in the high-speed straight running process, enhances the environment sensing capability of intelligent driving and improves the self-positioning effect. Visual perception and high-precision map matching fusion are added into the GNSS and IMU fusion system, so that the transverse error of the commercial collector card is not accumulated and not diverged when the commercial collector card runs at a high speed in a straight line under the condition that the GNSS signal is weak or no error exists in an IMU gyroscope, and the error can be corrected rapidly when the GNSS signal is recovered.

An embodiment of the present invention provides a vehicle positioning device, as shown in fig. 10, including: the first obtaining module 100 is configured to obtain a plurality of first lane line matching pairs, a state transition equation of an extended kalman filter, and initial parameters, where the first lane line matching pairs correspond to the visual perception lane lines and the map lane lines; a screening module 101, configured to screen at least one target lane line matching pair that meets a preset condition from a plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line are not crossed, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance; a second obtaining module 102, configured to obtain a target relative displacement between each target matching point pair in each target lane line matching pair; and a third obtaining module 103, configured to input the target relative displacement, the state transition equation and the initial parameters between each target matching point pair into the kalman filter, so as to obtain a positioning result of the vehicle.

The vehicle positioning device acquires a plurality of first lane line matching pairs, a state transfer equation and initial parameters of an extended Kalman filter, wherein the first lane line matching pairs correspond to the visual perception lane lines and the map lane lines; screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; the preset conditions comprise: the visual perception lane line in the target lane line matching pair and the map lane line are not crossed, and/or the relative displacement between the visual perception lane line in the target lane line matching pair and the map lane line is smaller than the preset distance; acquiring target relative displacement between each target matching point pair in each target lane line matching pair; and inputting the target relative displacement, the state transfer equation and the initial parameters between each target matching point pair into a Kalman filter to obtain a positioning result of the vehicle. The device obtains the target lane line matching pairs by preprocessing the obtained plurality of first lane line matching pairs, and as the target lane line matching meets the preset condition that the visual perception lane lines and the map lane lines do not cross and/or the relative displacement is smaller than the preset distance, the information utilization rate and the accuracy rate of the target lane lines can be ensured, the vehicle positioning result can be corrected according to the relative displacement between each target matching point pair in the target lane line matching pairs, and the precision and the accuracy of the positioning result can be improved.

Further, the screening module is further configured to: deleting a designated lane line matching pair from the plurality of first lane line matching pairs to obtain at least one target lane line matching pair meeting a preset condition; wherein, the appointed lane line matching pair includes: under the overpass road section scene, the visual perception lane line and the map lane line which are mutually intersected.

Further, the screening module is further configured to: for each first lane line matching pair, acquiring a plurality of first matching point pairs in the current first lane line matching pair; calculating average relative displacement of a plurality of first relative displacements according to the first relative displacement between two matching points in each first matching point pair; if the average relative displacement is larger than a first preset distance, deleting the current first lane line matching pair; and taking the next first lane line matching pair as a new current first lane line matching pair, and repeatedly executing the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair to obtain at least one target lane line matching pair meeting the preset condition.

Further, the screening module is further configured to: for each first lane line matching pair, acquiring a plurality of first matching point pairs in the current first lane line matching pair; calculating a second relative displacement between two matching points in each first matching point pair; selecting a plurality of designated matching point pairs with second relative displacement smaller than a second preset distance from the plurality of first matching point pairs; determining lane line matching sections corresponding to the plurality of specified matching point pairs as target lane line matching pairs meeting preset conditions corresponding to the current first lane line matching; and taking the next first lane line matching pair as a new current first lane line matching pair, and repeatedly executing the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair to obtain at least one target lane line matching pair meeting the preset condition.

Further, the screening module is further configured to: if the plurality of specified matching point pairs are a plurality of continuous matching point pairs, lane line matching sections corresponding to the plurality of specified matching point pairs are generated based on the plurality of continuous matching point pairs.

Further, the third obtaining module is further configured to: inputting target relative displacement, a state transfer equation and initial parameters between each target matching point pair into a Kalman filter, and outputting a first jacobian matrix corresponding to the state transfer equation; outputting a positioning predicted value and an error covariance predicted value of the vehicle based on the state transition equation and the first jacobian matrix; and determining a positioning result of the vehicle based on the positioning predicted value and the error covariance predicted value.

Further, the third obtaining module is further configured to: obtaining an observation equation of an extended Kalman filter; based on an observation equation of the extended Kalman filter, determining a second jacobian matrix corresponding to the observation equation; determining updated Kalman yields based on the observation equation, the second jacobian matrix and the error covariance prediction value; and updating the positioning predicted value based on the positioning predicted value, the updated Kalman gain and the observation equation to obtain a positioning estimated value of the vehicle at the current moment, and determining the positioning estimated value as a positioning result of the vehicle.

The implementation principle and the technical effects of the vehicle positioning device provided by the embodiment of the invention are the same as those of the vehicle positioning method embodiment, and for the sake of brevity, reference is made to corresponding contents in the vehicle positioning method embodiment to the point that the embodiment of the vehicle positioning device is not mentioned.

The embodiment of the present invention further provides an electronic device, referring to fig. 11, where the electronic device includes a processor 130 and a memory 131, where the memory 131 stores machine executable instructions that can be executed by the processor 130, and the processor 130 executes the machine executable instructions to implement the vehicle positioning method described above.

Further, the electronic device shown in fig. 11 further includes a bus 132 and a communication interface 133, and the processor 130, the communication interface 133, and the memory 131 are connected through the bus 132.

The memory 131 may include a high-speed random access memory (RAM, random Access Memory), and may further include a non-volatile memory (non-volatile memory), such as at least one disk memory. The communication connection between the system network element and at least one other network element is implemented via at least one communication interface 133 (which may be wired or wireless), and may use the internet, a wide area network, a local network, a metropolitan area network, etc. Bus 132 may be an ISA bus, a PCI bus, an EISA bus, or the like. The buses may be classified as address buses, data buses, control buses, etc. For ease of illustration, only one bi-directional arrow is shown in FIG. 11, but not only one bus or type of bus.

The processor 130 may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuitry in hardware or instructions in software in processor 130. The processor 130 may be a general-purpose processor, including a central processing unit (Central Processing Unit, abbreviated as CPU), a network processor (Network Processor, abbreviated as NP), etc.; but may also be a digital signal Processor (DIGITAL SIGNAL Processor, DSP), application Specific Integrated Circuit (ASIC), field-Programmable gate array (FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components. The disclosed methods, steps, and logic blocks in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in the memory 131, and the processor 130 reads the information in the memory 131, and in combination with its hardware, performs the steps of the method of the foregoing embodiment.

The embodiment of the invention also provides a machine-readable storage medium, which stores machine-executable instructions that, when being called and executed by a processor, cause the processor to implement the vehicle positioning method, and the specific implementation can be referred to the method embodiment and will not be described herein.

The computer program product of the vehicle positioning method, the apparatus and the electronic device provided in the embodiments of the present invention include a computer readable storage medium storing program codes, where the instructions included in the program codes may be used to execute the method described in the foregoing method embodiment, and specific implementation may refer to the method embodiment and will not be described herein.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.

Claims

1. A vehicle positioning method, the method comprising:

Acquiring a plurality of first lane line matching pairs, a state transfer equation and initial parameters of an extended Kalman filter, wherein the first lane line matching pairs and the state transfer equation and the initial parameters are matched with each other, and the first lane line matching pairs and the state transfer equation and the initial parameters are matched with each other; the initial parameters comprise initial state vectors and initial error covariance matrixes of the vehicles;

Screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; wherein, the preset conditions include: the visual perception lane lines in the target lane line matching pair do not cross with the map lane lines, and/or the relative displacement between the visual perception lane lines in the target lane line matching pair and the map lane lines is smaller than a preset distance;

Acquiring target relative displacement between each target matching point pair of each target lane line matching pair;

Inputting the target relative displacement between each target matching point pair, the state transfer equation and the initial parameters into a Kalman filter to obtain a positioning result of the vehicle;

the step of screening at least one target lane line matching pair meeting the preset condition from the plurality of first lane line matching pairs comprises the following steps:

Deleting a designated lane line matching pair from the plurality of first lane line matching pairs to obtain at least one target lane line matching pair meeting a preset condition; wherein the specified lane line matching pair includes: under the scene of the viaduct section, the visual perception lane lines and the map lane lines which are intersected with each other;

The step of inputting the target relative displacement between each target matching point pair, the state transfer equation and the initial parameters into a Kalman filter to obtain a positioning result of the vehicle comprises the following steps:

Inputting the target relative displacement between each target matching point pair, the state transition equation and the initial parameters into a Kalman filter, and outputting a first jacobian matrix corresponding to the state transition equation; the Kalman filter linearizes a nonlinear system by utilizing Taylor series expansion, and filters signals by adopting a Kalman filtering frame;

Outputting a positioning predicted value and an error covariance predicted value of the vehicle based on the state transition equation and the first jacobian matrix;

Determining a positioning result of the vehicle based on the positioning prediction value and the error covariance prediction value;

For each first lane line matching pair, acquiring a plurality of first matching point pairs in the current first lane line matching pair;

Calculating second relative displacement between two matching points in each first matching point pair;

selecting a plurality of appointed matching point pairs with second relative displacement smaller than a second preset distance from the plurality of first matching point pairs;

determining lane line matching sections corresponding to the specified matching point pairs as target lane line matching pairs meeting preset conditions corresponding to the current first lane line matching;

The next first lane line matching pair is used as a new current first lane line matching pair, and the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair is repeatedly executed to obtain at least one target lane line matching pair meeting preset conditions;

Before the step of determining the lane line matching sections corresponding to the specified matching point pairs as the target lane line matching pair meeting the preset condition corresponding to the current first lane line matching, the method further includes:

If the specified matching point pairs are continuous matching point pairs, generating lane line matching sections corresponding to the specified matching points based on the continuous matching point pairs;

the step of determining a positioning result of the vehicle based on the positioning prediction value and the error covariance prediction value includes:

obtaining an observation equation of the extended Kalman filter;

Determining a second jacobian matrix corresponding to an observation equation based on the observation equation of the extended Kalman filter;

Determining updated kalman return based on the observation equation, the second jacobian matrix, and the error covariance prediction value;

And updating the positioning predicted value based on the positioning predicted value, the updated Kalman gain and the observation equation to obtain a positioning estimated value of the vehicle at the current moment, and determining the positioning estimated value as a positioning result of the vehicle.

2. The method of claim 1, wherein the step of screening at least one target lane-line matching pair satisfying a preset condition from a plurality of the first lane-line matching pairs comprises:

Calculating average relative displacement of a plurality of first relative displacements according to first relative displacement between two matching points in each first matching point pair;

Deleting the current first lane line matching pair if the average relative displacement is greater than a first preset distance;

And taking the next first lane line matching pair as a new current first lane line matching pair, and repeatedly executing the step of acquiring a plurality of first matching point pairs in the current first lane line matching pair to obtain at least one target lane line matching pair meeting the preset condition.

3. A vehicle positioning device, the device comprising:

The first acquisition module is used for acquiring a plurality of first lane line matching pairs, a state transfer equation and initial parameters of the extended Kalman filter, wherein the first lane line matching pairs correspond to the visual perception lane lines and the map lane lines; the initial parameters comprise initial state vectors and initial error covariance matrixes of the vehicles;

The screening module is used for screening at least one target lane line matching pair meeting preset conditions from a plurality of first lane line matching pairs; wherein, the preset conditions include: the visual perception lane lines in the target lane line matching pair do not cross with the map lane lines, and/or the relative displacement between the visual perception lane lines in the target lane line matching pair and the map lane lines is smaller than a preset distance;

The second acquisition module is used for acquiring target relative displacement between each target matching point pair of each target lane line matching pair;

The third acquisition module is used for inputting the target relative displacement between each target matching point pair, the state transfer equation and the initial parameters into a Kalman filter to obtain a positioning result of the vehicle;

the third acquisition module is further configured to:

The screening module is further configured to:

the third acquisition module is further configured to:

obtaining an observation equation of the extended Kalman filter;

4. An electronic device comprising a processor and a memory, the memory storing machine executable instructions executable by the processor, the processor executing the machine executable instructions to implement the vehicle localization method of any of claims 1-2.

5. A machine-readable storage medium storing machine-executable instructions which, when invoked and executed by a processor, cause the processor to implement the vehicle locating method of any one of claims 1-2.