Vehicle pose determination method and device, readable storage medium and program product

Abstract

One or more embodiments of the invention provide a vehicle pose determination method and device, a readable storage medium and a program product. The method comprises the following steps: obtaining geographic coordinate values of at least two lane points corresponding to a vehicle; the lane point is determined based on an associated lane line of the vehicle, and the geographic coordinate value is a coordinate value of the lane point in a geographic coordinate system; generating at least two particles according to the geographic coordinate values of the at least two lane points; performing particle filtering operation on the at least two lane points based on candidate particles in the at least two particles to obtain pose prediction error parameters corresponding to the candidate particles; and determining the actual pose of the vehicle based on the pose prediction error parameter corresponding to each candidate particle.

Description

Technical Field

[0001] One or more embodiments of this specification relate to the field of autonomous driving technology, and in particular, to a method, device, readable storage medium, and program product for determining a vehicle posture. Background Art

[0002] In the scenario of autonomous driving, semantic positioning algorithms can help vehicles identify the current road conditions and accurately determine their position on the road. Specifically, this algorithm matches the environmental perception data acquired by the vehicle in real time, such as lane point clouds, with the corresponding environmental reference data in pre-built high-precision maps, such as lane lines and roadside point clouds, to determine the vehicle's current actual position and posture, i.e., actual posture, from multiple predicted postures.

[0003] In the related art, the semantic positioning algorithm usually constructs a corresponding residual function based on the environmental perception data and the environmental reference data during the matching process, and calculates the actual posture that satisfies the minimum residual through the least squares optimization algorithm; or, uses statistical or machine learning methods to estimate the similarity between the environmental perception data and the environmental reference data, thereby determining the actual posture with the highest corresponding similarity. However, no matter which method is used, the amount of calculation required is very large, the equipment cost is higher, and the efficiency of determining the vehicle posture is limited. Summary of the invention

[0004] In view of this, one or more embodiments of this specification provide the following technical solutions:

[0005] According to a first aspect of one or more embodiments of this specification, a method for determining a vehicle posture is proposed, the method comprising:

[0006] Obtaining geographic coordinate values ​​of at least two lane points corresponding to the vehicle; the lane points are determined based on the associated lane lines of the vehicle, and the geographic coordinate values ​​are coordinate values ​​of the lane points in the geographic coordinate system;

[0007] generating at least two particles according to the geographic coordinate values ​​of the at least two lane points, wherein different particles are used to represent different predicted postures of the vehicle;

[0008] Performing particle filtering operations on the at least two lane points based on candidate particles among the at least two particles respectively, to obtain a posture prediction error parameter corresponding to each candidate particle; the posture prediction error parameter is determined based on a deviation between a basic vehicle-borne coordinate value and a predicted vehicle-borne coordinate value of the lane point under the action of the corresponding candidate particle, the basic vehicle-borne coordinate value is obtained by converting the geographic coordinate value of the lane point to the vehicle-borne coordinate system of the vehicle through the corresponding candidate particle, and the predicted vehicle-borne coordinate value is obtained by estimating the basic vehicle-borne coordinate value through a lane line equation;

[0009] Based on the pose prediction error parameters corresponding to each candidate particle, the actual pose of the vehicle is determined.

[0010] Optionally, the particle filtering operation for the candidate particle includes:

[0011] The geographic coordinate value of each lane point in the at least two lane points is converted into a vehicle-borne coordinate system of the vehicle by using the candidate particles to obtain a basic vehicle-borne coordinate value of each lane point;

[0012] The basic vehicle coordinate value of each lane point is estimated by the lane line equation corresponding to each lane point, and the predicted vehicle coordinate value of each lane point is obtained;

[0013] According to the difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of each lane point in at least two lane points, a posture prediction error parameter corresponding to the candidate particle is determined.

[0014] Optionally, the basic vehicle-mounted coordinate value includes at least a non-target coordinate value and a target coordinate value;

[0015] The method of estimating the basic vehicle-borne coordinate value of each lane point by using the lane line equation corresponding to each lane point to obtain the predicted vehicle-borne coordinate value of each lane point includes: processing the non-target coordinate value in the basic vehicle-borne coordinate value of each lane point by using the lane line equation corresponding to each lane point to obtain the predicted target coordinate value of each lane point;

[0016] Determine the pose prediction error parameter corresponding to the candidate particle according to the difference between the basic vehicle-mounted coordinate value and the predicted vehicle-mounted coordinate value of each lane point in at least two lane points, including: determine the pose prediction error parameter corresponding to the candidate particle according to the difference between the target coordinate value in the basic vehicle-mounted coordinate value and the predicted target coordinate value of each lane point in at least two lane points.

[0017] Optionally, determining the actual posture of the vehicle based on the posture prediction error parameter corresponding to each candidate particle includes:

[0018] The candidate particle corresponding to the minimum value in the posture prediction error parameter is determined as the target particle, and the predicted posture represented by the target particle is determined as the actual posture of the vehicle.

[0019] Optionally, before obtaining the geographic coordinate values ​​of at least two lane points corresponding to the vehicle, the method further includes:

[0020] Determine the geographical location range of the vehicle;

[0021] The lane line within the geographical location range is obtained from a preset electronic map as the associated lane line of the vehicle;

[0022] At least two lane points are determined based on the associated lane line.

[0023] Optionally, determining the pose prediction error parameter corresponding to the candidate particle according to the difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of each lane point in at least two lane points includes:

[0024] Determine a sub-prediction deviation corresponding to each lane point of at least two lane points; the sub-prediction deviation is determined based on a difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of the corresponding lane point;

[0025] The pose prediction error parameter corresponding to the candidate particle is determined based on the sub-prediction deviation corresponding to each lane point.

[0026] Optionally, performing a particle filtering operation on the at least two lane points based on candidate particles in the at least two particles to obtain a pose prediction error parameter corresponding to each candidate particle includes:

[0027] Determine each particle of the at least two particles as a candidate particle; or

[0028] A particle located in front of the vehicle among the at least two particles is determined as the candidate particle.

[0029] According to a second aspect of one or more embodiments of this specification, an electronic device is proposed, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor implements the steps of the method described in the first aspect by running the executable instructions.

[0030] According to a third aspect of one or more embodiments of the present specification, a computer-readable storage medium is provided, on which computer instructions are stored, and when the instructions are executed by a processor, the steps of the method described in the first aspect are implemented.

[0031] According to a fourth aspect of one or more embodiments of the present specification, a computer program product is provided, comprising a computer program / instruction, which implements the steps of the method described in the first aspect when executed by a processor.

[0032] In the technical solution provided in the present specification, first, the coordinate values ​​of at least two lane points corresponding to the vehicle in the geographic coordinate system can be obtained as the geographic coordinate values, and at the same time, at least two particles can be generated according to the geographic coordinate values ​​of the at least two lane points, wherein different particles are used to characterize different predicted postures of the above-mentioned vehicle; thereafter, particle filtering operations can be performed on the at least two lane points based on candidate particles in the at least two particles, so as to obtain posture prediction error parameters corresponding to each candidate particle. On the one hand, the geographic coordinate values ​​of the lane points can be converted to the vehicle-mounted coordinate system of the vehicle by the corresponding candidate particles to obtain basic vehicle-mounted coordinate values; on the other hand, the basic vehicle-mounted coordinate values ​​can be estimated by the lane line equation to obtain predicted vehicle-mounted coordinate values, and then the posture prediction error parameters are determined based on the deviation between the basic vehicle-mounted coordinate values ​​and the predicted vehicle-mounted coordinate values ​​of the lane points under the action of the corresponding candidate particles; finally, the actual posture of the vehicle is determined based on the posture prediction error parameters corresponding to each candidate particle. Among them, the above-mentioned perception lane line equation is used as a simple calculation method. The basic vehicle coordinate value quickly determined by it can be deviated together with the basic vehicle coordinate value obtained by coordinate system conversion, so as to obtain the posture prediction error parameter to determine the actual posture of the vehicle. The whole process does not require the acquisition or construction of other additional data. Compared with traditional residual function calculations or statistical or machine learning algorithms, the amount of calculation required is smaller, the equipment cost is lower, and the limitation and influence of computing power resources on the efficiency of vehicle posture determination are effectively reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments or prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 is a flow chart of a method for determining a vehicle posture provided by an exemplary embodiment;

[0035] Figure 2 is a schematic diagram of particle generation provided by an exemplary embodiment;

[0036] Figure 3 is a flow chart of another method for determining a vehicle posture provided by an exemplary embodiment;

[0037] Figure 4 is a schematic structural diagram of an electronic device provided by an exemplary embodiment;

[0038] Figure 5 It is a schematic structural diagram of a device for determining a vehicle posture provided by an exemplary embodiment. DETAILED DESCRIPTION

[0039] Here, exemplary embodiments will be described in detail, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are only examples of devices and methods consistent with some aspects of the present invention.

[0040] The terms used in the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "the" and "the" used in the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0041] It should be understood that although the terms first, second, third, etc. may be used in the present invention to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present invention, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0042] Figure 1 This is a flowchart of a method for determining vehicle posture provided by an exemplary embodiment. The method can be executed by any execution subject or any execution subject in cooperation with the vehicle, a processor and / or controller deployed on the vehicle, a service platform and / or an intelligent driving platform and / or a navigation platform communicating with the vehicle, an Internet of Things device, an Internet of Vehicles device, other terminals, an independent physical server, a server cluster, a cloud, a distributed system, etc. Those skilled in the art can deploy the method on the corresponding execution subject based on actual needs; Figure 1 As shown, the above method includes:

[0043] Step S102, obtaining geographic coordinate values ​​of at least two lane points corresponding to the vehicle; the lane points are determined based on the associated lane lines of the vehicle, and the geographic coordinate values ​​are coordinate values ​​of the lane points in the geographic coordinate system.

[0044] When the vehicle posture needs to be determined, accurate and reliable environmental reference data can first be obtained through the reference electronic map (such as but not limited to high-precision maps, standard-precision maps or navigation maps provided by various map manufacturers, etc.), that is, the geographic coordinate values ​​of at least two lane points corresponding to the vehicle can be obtained through the electronic map. Among them, the lane point is determined based on the associated lane line of the vehicle. The above lane line refers to the marking line on the road that separates the lanes, which can usually be a solid or dotted line in white or yellow; the above lane point is not a standardized physical entity, but a data point or logical concept used to describe a specific position on the corresponding lane line, which is used to digitally represent the position and shape of the corresponding lane line in the above electronic map.

[0045] The acquired geographic coordinate value refers to the coordinate value of the lane point in the geographic coordinate system. At this time, the lane point can be composed of the first dimension coordinates and the second dimension coordinates in the geographic coordinate system, wherein the coordinate types of the first dimension coordinates and the second dimension coordinates in the geographic coordinate system are respectively any one of the longitude coordinates and the latitude coordinates, that is, the first dimension coordinates are the longitude coordinates in the geographic coordinate system and the second dimension coordinates are the latitude coordinates in the geographic coordinate system, or the first dimension coordinates are the longitude coordinates in the geographic coordinate system and the second dimension coordinates are the latitude coordinates in the geographic coordinate system. Of course, in addition to longitude and latitude, the above-mentioned geographic coordinate values ​​can also be represented by other forms such as Universal Transverse Mercator (UTM) coordinates and Universal Polar Stereographic (UPS) coordinates, which are not limited in this specification.

[0046] In this specification, the method for obtaining lane points can be determined based on the positioning information of the vehicle.

[0047] In one embodiment, the vehicle can determine the geographical location range where it is located, and determine the associated lane line of the vehicle from the lane lines within the geographical location range obtained from a preset electronic map, and then determine at least two lane points based on the associated lane lines. Specifically, after the vehicle obtains its own positioning information, it can determine the geographical location range where the vehicle is located based on the positioning information and its corresponding error range, and then obtain the lane lines and corresponding lane points within the geographical location range from a preset high-precision electronic map through a map engine. Among them, the above positioning information can be obtained through a satellite navigation system such as the Global Positioning System (GPS) configured by the vehicle, so the positioning information can also be expressed in a geographic coordinate system, and the above error range can be dynamically adjusted by the vehicle manufacturer or user according to the actual positioning quality. For example, when the accuracy of the positioning quality is lower than the threshold of 95% within a period of time, the error range can be expanded from the preset 100-meter radius to the 150-meter radius, and then the geographical location range of the vehicle is determined as a circular range with the geographical coordinates corresponding to the above positioning information as the center and the above error range as the radius. Of course, the information contained in the above error range can also be diversified so that the above geographical location range can be a rectangular, elliptical, etc. range, and the geographical location range can also be combined with the national region boundary to perform subsequent irregular processing on the geographical location range, so that the corresponding lane lines and lane points are more consistent with the current geographical location of the vehicle. The above electronic map can be a vector map loaded by the vehicle host through the cloud or offline, and the lane lines and lane points within the corresponding range can be obtained from the high-precision map through the above geographical location range.

[0048] Step S104: generating at least two particles according to the geographic coordinate values ​​of the at least two lane points, wherein different particles are used to represent different predicted postures of the vehicle.

[0049] After obtaining the above-mentioned lane lines and corresponding lane points, multiple predicted postures that the vehicle may correspond to can be determined in advance, so as to subsequently determine the actual posture of the vehicle. Among them, the predicted posture of the vehicle can be represented by particles in the particle filtering algorithm. Specifically, multiple particles can be generated based on the lane points according to probability distribution methods such as normal distribution and uniform distribution; or, multiple particles can be generated based on the lane points according to deterministic particle generation methods such as grid division and spiral pattern. Different particle generation methods meet the posture acquisition requirements in different situations. For example, the normal distribution of particles has a faster generation speed, and the grid division of particles has a more comprehensive coverage rate. Each particle represents the predicted posture of a hypothetical vehicle, and the predicted postures corresponding to different particles are different. The following is combined with Figure 2 , taking the above uniform distribution method as an example, the principle of generating particles based on lane points is introduced: Figure 2 As shown, for each lane point, multiple particles corresponding to the predicted postures can be randomly distributed within the corresponding range with the lane point as the center and the preset distribution distance as the radius, where each particle has an equal probability of being selected, and their respective positions and directions can be used to characterize a predicted posture.

[0050] Of course, in addition to the initialization link of generating particles, the above-mentioned particle filter algorithm actually also includes but is not limited to the prediction link of determining the position of each particle at the next time based on the control input of the vehicle obtained from the vehicle sensors such as the Inertial Measurement Unit (IMU) and the wheel speed meter, as well as the motion noise caused by uncertain factors such as slip or wind resistance, and the resampling link of selecting particles with a certain probability to form a new particle set according to the weight of the particles. Considering that one of the key points in this specification is to select particles representing the actual posture of the vehicle from multiple particles used to represent different predicted postures of the vehicle through the perception lane line equation below, the prediction and resampling links originally included in the above-mentioned particle filter algorithm will not be further elaborated.

[0051] Step S106, performing particle filtering operations on the at least two lane points based on candidate particles among the at least two particles, respectively, to obtain pose prediction error parameters corresponding to each candidate particle; the pose prediction error parameters are determined based on the deviation between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of the lane point under the action of the corresponding candidate particle, the basic vehicle-borne coordinate value is obtained by converting the geographic coordinate value of the lane point to the vehicle-borne coordinate system of the vehicle through the corresponding candidate particle, and the predicted vehicle-borne coordinate value is obtained by estimating the basic vehicle-borne coordinate value through the lane line equation.

[0052] Based on the at least two generated particles, a particle filtering operation can be performed on the at least two lane points based on the candidate particles, respectively, so as to obtain the pose prediction error parameter corresponding to each candidate particle, and the pose prediction error parameter is determined based on the deviation between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of the lane point under the action of the corresponding candidate particle. Among them, the so-called candidate particle can be either each particle of the at least two particles or a particle located in front of the vehicle. The particle filtering operation can specifically include: converting the geographic coordinate value of each lane point in the at least two lane points to the vehicle-borne coordinate system of the vehicle through the candidate particles, so as to obtain the basic vehicle-borne coordinate value of each lane point, and at the same time, estimating the basic vehicle-borne coordinate value of each lane point through the lane line equation corresponding to each lane point, so as to obtain the predicted vehicle-borne coordinate value of each lane point, and finally determining the pose prediction error parameter corresponding to the candidate particle according to the difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of each lane point in the at least two lane points.

[0053] Specifically, for the coordinate system conversion process mentioned in the above particle filter operation, the vehicle coordinate system is a coordinate system used to describe the position and direction of the internal or external points of the vehicle relative to the vehicle itself. It is similar to the geographic coordinate system, and is usually defined with three mutually perpendicular axes: X-axis, Y-axis and Z-axis, and an origin, wherein the origin is generally located at the center of mass of the vehicle or a fixed reference point, such as the geometric center of the bottom of the vehicle; the Z-axis is perpendicular to the ground, and the positive direction is upward, indicating the height; as for the X-axis used to represent the longitudinal distance coordinate and the Y-axis used to represent the lateral distance coordinate, the two can be formulated according to the specific provisions of the vehicle coordinate system. For example, in the "front left upper" vehicle coordinate system, the positive direction along the X-axis can be regarded as the front of the vehicle, and the positive direction along the Y-axis can be regarded as the left side of the vehicle; in the "right front upper" vehicle coordinate system, the positive direction along the X-axis can be regarded as the right side of the vehicle, and the positive direction along the Y-axis can be regarded as the front of the vehicle. Each of the lane points mentioned above is originally in the geographic coordinate system, corresponding to a unique longitude and latitude coordinate. For at least two particles generated, it is assumed that one particle p is located at the position (x p ,y p ), and has a heading angle θ p , then for a lane point (X, Y) in a geographic coordinate system, it can be converted to the vehicle coordinate system corresponding to particle p by, for example, the following transformation formula:

[0054] x local =(Xx p )·cos(θ p )+(Yy p )·sin(θ p )

[0055] y local =-(Xx p )·sin(θ p )+(Yy p )·cos(θ p )

[0056] Among them, (x local ,y local ) is the coordinate position of the lane point in the vehicle coordinate system corresponding to particle p. After the coordinate system is converted, the basic vehicle coordinate values ​​of the at least two lane points mentioned above will have different longitudinal distance coordinates and / or lateral distance coordinates. For example, assuming that the vehicle has f lane points and n particles are generated, then a total of n*f lane points need to be converted from the geographic coordinate system to the vehicle coordinate system, and a total of n*f lateral distance coordinates and longitudinal distance coordinates in the vehicle coordinate system are obtained, where f is an arbitrary positive integer and n is an integer greater than 1.

[0057] Regarding the lane line equation estimation process mentioned in the above particle filter operation, after obtaining the basic vehicle-mounted coordinate values ​​of the above lane points in the corresponding vehicle-mounted coordinate system, they can be regarded as accurate environmental reference data. At this time, the particles closest to the actual posture of the vehicle can be matched in combination with the environmental perception data obtained by the vehicle in real time. The above environmental perception data can be expressed by the above perceived lane line equation. Specifically, the road image can be captured according to the visual sensor of the above vehicle, the lane line can be identified through image processing and feature extraction technology, and the perception lane line equation corresponding to the lane line corresponding to the lane point can be generated using a mathematical fitting method, so as to substitute the non-target coordinate values ​​of the above lane points in the corresponding vehicle-mounted coordinate system into the above perceived lane line equation to calculate the above predicted target coordinate values. Among them, the basic vehicle-mounted coordinate values ​​of the above lane points at least include non-target coordinate values ​​and target coordinate values. The above perceived lane line equation is a model for mathematically describing the geometric shape of the lane line. It is usually used to fit the position and direction of the lane line. When the lane line is approximately a straight line, it can be expressed using a simple linear equation:

[0058] y=mx+b

[0059] Among them, the value of x as a non-target coordinate value can be determined based on the longitudinal distance coordinate of the vehicle-mounted coordinate system, and the value of y as a target coordinate value can be determined based on the lateral distance coordinate of the vehicle-mounted coordinate system. m is the slope, indicating the direction of the lane line, and b is the intercept, indicating the intersection of the lane line and the y-axis. Since the geographic coordinate values ​​of each lane point are converted to the corresponding vehicle-mounted coordinate system, the corresponding non-target coordinate value can be directly substituted into the perception lane line equation based on the vehicle-mounted coordinate system on the premise that the slope and intercept have been determined, thereby obtaining the predicted target coordinate value corresponding to the above non-target coordinate value. Of course, the value of x as a non-target coordinate value can also be determined based on the lateral distance coordinate of the vehicle-mounted coordinate system, and the value of y as a target coordinate value can be determined based on the longitudinal distance coordinate of the vehicle-mounted coordinate system. This is not limited in this specification.

[0060] The above lane line equation is usually only used for short distances or straight sections, but may not be accurate enough for curved roads or long distances. Therefore, in order to more accurately capture the actual shape of the lane line, especially for curved roads, this specification can use a high-order polynomial to fit the lane line. For example, the quadratic polynomial

[0061] y=ax 2 +bx+c

[0062] or a cubic polynomial:

[0063] y=ax 3 +bx 2 +cx+d

[0064] Among them, a, b, c, d, as coefficients of the polynomial, can be estimated from the vehicle's sensor data by the least squares method or other optimization algorithms. Of course, the lane line equation can also be used to mathematically describe the three-dimensional model of the lane line. Accordingly, the height distance coordinates of the lane point in the corresponding vehicle coordinate system except the longitudinal distance coordinates and the lateral distance coordinates can be brought into the above lane line equation as target coordinate values ​​or non-target coordinate values, and this is not limited in this specification.

[0065] On this basis, if the vehicle obtains multiple lane lines on the road, the perception lane line equation corresponding to each lane line can be determined according to the vehicle's visual sensor, and then the non-target coordinate values ​​in the basic vehicle-mounted coordinate values ​​of each lane point are processed to obtain the predicted target coordinate values ​​of each lane point. In other words, the non-target coordinate values ​​of the above lane points in the corresponding vehicle-mounted coordinate system are substituted into the perception lane line equation of the corresponding lane line to calculate the above predicted target coordinate values. Among them, the lane points, lane lines and perception lane line equations are one-to-one corresponding, so the substitution method of the above perception lane line equation is actually not affected by the number of lane lines or lane points, which improves the accuracy of the above predicted target coordinate values. Of course, it can be understood that considering that the lane line corresponding to the perception lane line equation is usually obtained from the processing of the video sensor in front of the vehicle, the lane point substituted into the above perception lane line equation is also within a certain range in front of the vehicle, such as within 100 in front of the vehicle, to avoid the situation where the lane line corresponding to the lane point is inconsistent with the perception lane line equation, which is also the reason why the candidate particle is determined as the particle in front of the vehicle in the previous article.

[0066] For the calculated predicted target coordinate value, the difference between the predicted target coordinate value and the target coordinate value corresponding to the non-target coordinate value in the corresponding vehicle coordinate system can be determined, thereby determining the posture prediction error parameter corresponding to the above-mentioned candidate particle, wherein the difference can be the difference between the predicted target coordinate value and the target coordinate value.

[0067] Of course, the above differences are only applicable to the case where there is only one lane point. At this time, the difference between the predicted target coordinate value and the target coordinate value is the pose prediction error parameter corresponding to the candidate particle. When there are at least two lane points corresponding to the lane line, the vehicle coordinate system corresponding to the same particle involves the difference between the predicted target coordinate value and the target coordinate value of at least two lane points. At this time, the sub-prediction deviation corresponding to each lane point in the at least two lane points can be determined, and the pose prediction error parameter corresponding to the candidate particle can be determined based on the sub-prediction deviation corresponding to each lane point, wherein the sub-prediction deviation is determined based on the difference between the basic vehicle coordinate value and the predicted vehicle coordinate value of the corresponding lane point. In other words, the difference between the predicted target coordinate value and the target coordinate value of at least two lane points in the corresponding vehicle coordinate system can be determined according to the preset difference statistical strategy and the multiple sub-prediction deviations of at least two lane points in each vehicle coordinate system, and then used as the pose prediction error parameter, wherein the difference statistical strategy can be used to flexibly determine the statistical method of the multiple differences of the multiple lane points in each vehicle coordinate system, such as calculating the average difference, variance or mean square error of the multiple differences as the pose prediction error parameter. For example, assuming that the vehicle simultaneously acquires a total of f lane points, the vehicle can substitute the non-target coordinate value of the j-th lane point in the corresponding vehicle coordinate system, such as the lateral coordinate value, into the perceived lane line equation of the lane line in the vehicle coordinate system corresponding to the i-th particle to obtain the predicted target coordinate value of the first lane point in the vehicle coordinate system, that is, the longitudinal coordinate value bj. Assuming that the original longitudinal coordinate value corresponding to the non-target coordinate value in the basic vehicle coordinate value of the j-th lane point and serving as the target coordinate value is aj, the difference between the predicted target coordinate value and the target coordinate value of the j-th lane point can be obtained. j -b j |, and when the difference statistics strategy decides to calculate the average difference of the above multiple differences, the pose prediction error parameter M of the i-th particle can be obtained. i :

[0068]

[0069] Step S108, determining the actual posture of the vehicle based on the posture prediction error parameters corresponding to each candidate particle.

[0070] For the posture prediction error parameters corresponding to each candidate particle, they are different in size. If the posture prediction error parameter is smaller, it means that the basic vehicle coordinate value in the corresponding vehicle coordinate system is more in line with the predicted vehicle coordinate value calculated in the perception lane line equation, that is, the predicted posture represented by the corresponding candidate particle is closer to the actual posture of the vehicle; conversely, if the posture prediction error parameter is larger, it means that the basic vehicle coordinate value in the corresponding vehicle coordinate system deviates more from the predicted vehicle coordinate value calculated in the perception lane line equation, that is, the predicted posture represented by the corresponding candidate particle is farther away from the actual posture of the vehicle. Therefore, the candidate particle corresponding to the minimum value in the posture prediction error parameter is determined as the target particle, and the predicted posture represented by the above target particle is determined as the actual posture of the above vehicle.

[0071] Among them, it is necessary to explain that the accuracy of the perception lane line equation used in the embodiment of the present application meets the quality requirements, that is, the perception lane line equation used in the embodiment of the present application itself is relatively accurate, that is, the calculation error of the result should be less than the preset error threshold, otherwise the prediction error parameters determined based on the perception lane line equation will also lose accuracy, thereby increasing the probability of misjudgment of the actual posture.

[0072] Figure 3 FIG. 1 is a flow chart of another method for determining vehicle posture provided by an exemplary embodiment. Figure 3 As shown, the above method includes:

[0073] Step S302, obtaining multiple lane points of the lane line where the vehicle is located from the high-precision map.

[0074] In one embodiment, assuming that the current position of the vehicle is longitude 116.407527, latitude 39.904034 (a location in Beijing), after being determined by the GPS positioning system, the vehicle obtains lane line information within 100 meters before and after it from the high-precision map loaded in the cloud, including multiple lane points on the left and right lane lines. The coordinates of each lane point in the geographic coordinate system are recorded. For example, the left lane line has the following three lane points:

[0075] Lane point A: longitude 116.407528, latitude 39.904035;

[0076] Lane point B: longitude 116.407529, latitude 39.904036;

[0077] Lane point C: longitude 116.407530, latitude 39.904037.

[0078] Step S304: transform the acquired lane points of the lane line into the vehicle coordinate system through a particle filtering algorithm to obtain the coordinates of each lane point in the vehicle coordinate system.

[0079] In one embodiment, based on the current position of the vehicle, 100 particles representing different predicted positions are generated. These particles are evenly distributed in a circular area with a radius of 5 meters centered on the current position of the vehicle. For each particle, the geographic coordinates of the three lane points are converted to the vehicle coordinate system corresponding to the particle. Assume that the converted coordinates are as follows (unit: meter):

[0080] For particle P1:

[0081] Lane point A: X = 0.1, Y = -0.5

[0082] Lane point B: X=1.1, Y=-0.6

[0083] Lane point C: X=2.1, Y=-0.7

[0084] For particle P2:

[0085] Lane point A: X = 0.2, Y = -0.6

[0086] Lane point B: X=1.2, Y=-0.7

[0087] Lane point C: X=2.2, Y=-0.8

[0088] ...(other particles)

[0089] Step S306, substitute the non-target coordinate values ​​of each lane point in the vehicle coordinate system into the known perceived lane line equation for the lane line, and obtain the predicted target coordinate values ​​corresponding to each non-target coordinate value.

[0090] In one embodiment, it is assumed that the lane line identified by the visual sensor can be expressed by a simple linear equation y=mx+b, where m is the slope and b is the intercept. Through image processing technology, the vehicle calculates that the slope of the current lane line is m=0.1 and the intercept is b=-0.5. Then, for each particle, this equation is used to calculate the expected lateral distance coordinate of the lane point, that is, the Y coordinate.

[0091] For example, for lane point A of particle P1, the X coordinate is 0.1, and the expected Y coordinate is obtained by substituting into the equation:

[0092] Expected Y coordinate = 0.1*0.1-0.5 = -0.49.

[0093] Step S308, subtract all target coordinate values ​​in the vehicle coordinate system corresponding to each particle from the predicted target coordinate value, and calculate the average value, and determine the predicted posture represented by the particle with the smallest corresponding average value as the actual posture of the vehicle.

[0094] In one embodiment, the difference between the actual Y coordinates and the expected Y coordinates of all lane points corresponding to each particle can be calculated. For particle P1, the differences are:

[0095] Difference A = |-0.5-(-0.49)| = 0.01;

[0096] Difference B = |-0.6-(0.1*1.1-0.5)| = 0.01;

[0097] The difference C = |-0.7-(0.1*2.1-0.5)| = 0.01.

[0098] Compute the mean of the differences:

[0099] Average difference = (0.01 + 0.01 + 0.01) / 3 = 0.01

[0100] This process is repeated to evaluate all particles, and the particle with the smallest average difference is selected as the final estimation result. If particle P1 has the smallest average difference, the predicted pose it represents is considered to be the actual pose of the vehicle.

[0101] Figure 4 is a schematic structural diagram of a device provided by an exemplary embodiment. Figure 4 At the hardware level, the device includes a processor 402, an internal bus 404, a network interface 406, a memory 408, and a non-volatile memory 410, and may also include hardware required for other functions. One or more embodiments of this specification may be implemented based on software, such as the processor 402 reading the corresponding computer program from the non-volatile memory 410 into the memory 408 and then running it. Of course, in addition to the software implementation, one or more embodiments of this specification do not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0102] Corresponding to the above-mentioned embodiment of the method for determining the vehicle posture, this specification also provides an embodiment of a device for determining the vehicle posture. Figure 5 ; The device comprises:

[0103] The lane point acquisition unit 502 is used to obtain the geographic coordinate values ​​of at least two lane points corresponding to the vehicle; the lane points are determined based on the associated lane lines of the vehicle, and the geographic coordinate values ​​are the coordinate values ​​of the lane points in the geographic coordinate system.

[0104] The particle generation unit 504 is used to generate at least two particles according to the geographic coordinate values ​​of the at least two lane points, wherein different particles are used to represent different predicted postures of the vehicle.

[0105] The pose prediction error parameter acquisition unit 506 is used to perform particle filtering operations on the at least two lane points based on candidate particles among the at least two particles, respectively, to obtain pose prediction error parameters corresponding to each candidate particle; the pose prediction error parameters are determined based on the deviation between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of the lane point under the action of the corresponding candidate particle, the basic vehicle-borne coordinate value is obtained by converting the geographic coordinate value of the lane point to the vehicle-borne coordinate system of the vehicle through the corresponding candidate particle, and the predicted vehicle-borne coordinate value is obtained by estimating the basic vehicle-borne coordinate value through the lane line equation.

[0106] The actual posture determination unit 508 is used to determine the actual posture of the vehicle based on the posture prediction error parameters corresponding to each candidate particle.

[0107] Optionally, the particle filtering operation for the candidate particle includes:

[0108] The geographic coordinate value of each lane point in the at least two lane points is converted into a vehicle-borne coordinate system of the vehicle by using the candidate particles to obtain a basic vehicle-borne coordinate value of each lane point;

[0109] The basic vehicle coordinate value of each lane point is estimated by the lane line equation corresponding to each lane point, and the predicted vehicle coordinate value of each lane point is obtained;

[0110] According to the difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of each lane point in at least two lane points, a posture prediction error parameter corresponding to the candidate particle is determined.

[0111] Optionally, the basic vehicle-mounted coordinate value includes at least a non-target coordinate value and a target coordinate value;

[0112] The pose prediction error parameter acquisition unit 506 is specifically used to process the non-target coordinate value in the basic vehicle coordinate value of each lane point through the lane line equation corresponding to each lane point, so as to obtain the predicted target coordinate value of each lane point;

[0113] According to the difference between the target coordinate value in the basic vehicle-borne coordinate value of each lane point in at least two lane points and the predicted target coordinate value, a posture prediction error parameter corresponding to the candidate particle is determined.

[0114] Optionally, the actual posture determination unit 508 is specifically used to:

[0115] The candidate particle corresponding to the minimum value in the posture prediction error parameter is determined as the target particle, and the predicted posture represented by the target particle is determined as the actual posture of the vehicle.

[0116] Optionally, before obtaining the geographic coordinate values ​​of at least two lane points corresponding to the vehicle, the device further includes:

[0117] A lane point determination unit, used to determine the geographical location range of the vehicle;

[0118] The lane line within the geographical location range is obtained from a preset electronic map as the associated lane line of the vehicle;

[0119] At least two lane points are determined based on the associated lane line.

[0120] Optionally, the posture prediction error parameter acquisition unit 506 is specifically used to:

[0121] Determine a sub-prediction deviation corresponding to each lane point of at least two lane points; the sub-prediction deviation is determined based on a difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of the corresponding lane point;

[0122] The pose prediction error parameter corresponding to the candidate particle is determined based on the sub-prediction deviation corresponding to each lane point.

[0123] Optionally, the posture prediction error parameter acquisition unit 506 is specifically used to:

[0124] Determine each particle of the at least two particles as a candidate particle; or

[0125] A particle located in front of the vehicle among the at least two particles is determined as the candidate particle.

[0126] Based on the same concept as the above method, this specification also provides an electronic device, including: a processor; a memory for storing processor executable instructions; wherein the processor implements the steps of the method described in any of the above embodiments by running the executable instructions.

[0127] Based on the same concept as the above method, this specification also provides a computer-readable storage medium on which computer instructions are stored. When the instructions are executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0128] Based on the same concept as the above method, this specification also provides a computer program product, including a computer program / instruction, which implements the steps of the method described in any of the above embodiments when executed by a processor.

[0129] Although the present invention includes many specific implementation details, these should not be interpreted as limiting the scope of any invention or the scope of protection claimed, but are mainly used to describe the features of the specific embodiments of specific inventions. Certain features described in multiple embodiments of the present invention may also be implemented in combination in a single embodiment. On the other hand, the various features described in a single embodiment may also be implemented separately in multiple embodiments or in any suitable sub-combination. In addition, although features may work in certain combinations as described above and even initially claimed as such, one or more features from the claimed combination may be removed from the combination in some cases, and the claimed combination may point to a sub-combination or a variation of a sub-combination.

[0130] Similarly, although operations are depicted in a particular order in the accompanying drawings, this should not be understood as requiring that these operations be performed in the particular order shown or performed sequentially, or requiring that all illustrated operations be performed to achieve the desired results. In some cases, multitasking and parallel processing may be advantageous. In addition, the separation of various system modules and components in the above-described embodiments should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product, or packaged into multiple software products.

[0131] Thus, specific embodiments of the subject matter have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the particular order or sequential order shown to achieve the desired results. In some implementations, multitasking and parallel processing may be advantageous.

[0132] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for determining a vehicle posture, characterized in that: The method comprises: Obtaining geographic coordinate values ​​of at least two lane points corresponding to the vehicle; the lane points are determined based on the associated lane lines of the vehicle, and the geographic coordinate values ​​are coordinate values ​​of the lane points in the geographic coordinate system; generating at least two particles according to the geographic coordinate values ​​of the at least two lane points, wherein different particles are used to represent different predicted postures of the vehicle; Performing particle filtering operations on the at least two lane points based on candidate particles among the at least two particles respectively, to obtain a posture prediction error parameter corresponding to each candidate particle; the posture prediction error parameter is determined based on a deviation between a basic vehicle-borne coordinate value and a predicted vehicle-borne coordinate value of the lane point under the action of the corresponding candidate particle, the basic vehicle-borne coordinate value is obtained by converting the geographic coordinate value of the lane point to the vehicle-borne coordinate system of the vehicle through the corresponding candidate particle, and the predicted vehicle-borne coordinate value is obtained by estimating the basic vehicle-borne coordinate value through a lane line equation; Based on the pose prediction error parameters corresponding to each candidate particle, the actual pose of the vehicle is determined.

2. The method according to claim 1, characterized in that: The particle filtering operation for the candidate particles includes: The geographic coordinate value of each lane point in the at least two lane points is converted into a vehicle-borne coordinate system of the vehicle by using the candidate particles to obtain a basic vehicle-borne coordinate value of each lane point; The basic vehicle coordinate value of each lane point is estimated by the lane line equation corresponding to each lane point, and the predicted vehicle coordinate value of each lane point is obtained; According to the difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of each lane point in at least two lane points, a posture prediction error parameter corresponding to the candidate particle is determined.

3. The method according to claim 2, characterized in that The basic vehicle-mounted coordinate values ​​at least include non-target coordinate values ​​and target coordinate values; The method of estimating the basic vehicle-borne coordinate value of each lane point by using the lane line equation corresponding to each lane point to obtain the predicted vehicle-borne coordinate value of each lane point includes: processing the non-target coordinate value in the basic vehicle-borne coordinate value of each lane point by using the lane line equation corresponding to each lane point to obtain the predicted target coordinate value of each lane point; Determine the pose prediction error parameter corresponding to the candidate particle according to the difference between the basic vehicle-mounted coordinate value and the predicted vehicle-mounted coordinate value of each lane point in at least two lane points, including: determine the pose prediction error parameter corresponding to the candidate particle according to the difference between the target coordinate value in the basic vehicle-mounted coordinate value and the predicted target coordinate value of each lane point in at least two lane points.

4. The method according to claim 1, characterized in that: The determining the actual posture of the vehicle based on the posture prediction error parameters corresponding to each candidate particle includes: The candidate particle corresponding to the minimum value in the posture prediction error parameter is determined as the target particle, and the predicted posture represented by the target particle is determined as the actual posture of the vehicle.

5. The method according to claim 1, characterized in that Before obtaining the geographic coordinate values ​​of at least two lane points corresponding to the vehicle, the method further includes: Determine the geographical location range of the vehicle; The lane line within the geographical location range is obtained from a preset electronic map as the associated lane line of the vehicle; At least two lane points are determined based on the associated lane line.

6. The method according to claim 2, characterized in that The step of determining the pose prediction error parameter corresponding to the candidate particle according to the difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of each lane point in at least two lane points comprises: Determine a sub-prediction deviation corresponding to each lane point of at least two lane points; the sub-prediction deviation is determined based on a difference between the basic vehicle-borne coordinate value and the predicted vehicle-borne coordinate value of the corresponding lane point; The pose prediction error parameter corresponding to the candidate particle is determined based on the sub-prediction deviation corresponding to each lane point.

7. According to any one of claims 1 to 6, the performing of a particle filter operation on the at least two lane points based on candidate particles in the at least two particles to obtain a pose prediction error parameter corresponding to each candidate particle comprises: Determine each particle of the at least two particles as a candidate particle; or A particle located in front of the vehicle among the at least two particles is determined as the candidate particle.

8. An electronic device, characterized in that: include: processor; A memory for storing processor-executable instructions; wherein the processor implements the steps of the method according to any one of claims 1 to 7 by executing the executable instructions.

9. A computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and when the instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product, characterized in that The method comprises a computer program / instruction, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 7.

Images