Vehicle positioning method and system for stationary parking scenarios

Abstract

The present application relates to a vehicle positioning method and system for a fixed parking scenario. The method comprises: a marker detection step, detecting markers and identifying a marker pair; a pose matching step, performing pose matching according to vehicle position information and the marker pair; a pose-free matching step, performing pose-free matching according to the marker pair to obtain the latest vehicle position information; and a position updating step, updating current vehicle position information according to the latest vehicle position information obtained by the pose matching step or the pose-free matching step.

Description

Technical Field

[0001] This invention relates to vehicle control technology, and more specifically to a vehicle positioning method and system for fixed parking scenarios. Background Technology

[0002] Automated parking refers to a vehicle automatically parking itself in a parking space without human intervention. Different automated parking systems typically use different methods to detect objects around the vehicle.

[0003] Image matching aims to identify and align content or structures with the same or similar attributes in two images at the pixel level. Generally, the images to be matched are usually taken from the same or similar scenes or targets, or other types of image pairs with the same shape or semantic information.

[0004] To improve the robustness of matching algorithms, reduce the impact of noise, distortion, and other factors on matching performance, and lower computational complexity, feature-based image matching methods have been extensively studied. Existing feature-based image matching methods mainly include the following steps: extracting physically meaningful salient feature structures from the image, including feature points, feature lines, and salient morphological regions; and applying a matching prediction transformation function to the extracted feature structures and aligning them with other image content.

[0005] In the field of automated parking, rapidly locating vehicle position information is crucial for path planning and safe intelligent driving. Image matching serves as a bridge between vehicle scene information and actual map information. To ensure the real-time performance and accuracy of image matching, acquiring 3D point cloud information from the vehicle scene and performing 3D matching with actual map information has become one of the effective means to solve the positioning problem.

[0006] Common point cloud matching algorithms include RANSAC and ICP. RANSAC can robustly estimate the matching model parameters; however, its drawbacks include a potentially large number of iterations for calculating model parameters, making it time-consuming, and the need for an additional threshold parameter. ICP, on the other hand, is a registration method that solves for the radiative transformation matrix of point sets, achieving accurate matching results, but it involves a large computational load when searching for corresponding points. Summary of the Invention

[0007] In view of the above problems, the present invention aims to propose a vehicle positioning method and a vehicle positioning system for fixed parking scenarios (such as parking garage parking scenarios, battery swapping station parking scenarios, etc.) that can simplify algorithm complexity and provide accurate and robust positioning for vehicles in real time.

[0008] A vehicle positioning method for fixed parking scenarios according to one aspect of the present invention is characterized by comprising:

[0009] The sign detection step detects signs and identifies sign pairs. If the vehicle position information in the fixed parking scenario has been obtained, the following pose matching step continues. If the vehicle position information in the fixed parking scenario has not been obtained, the following poseless matching step is skipped.

[0010] The pose matching step involves performing pose matching based on the obtained vehicle position information and the identified markers. If the latest vehicle position information is obtained, the process jumps to the position update step below; otherwise, the poseless matching step below continues.

[0011] The pose-free matching step involves performing pose-free matching based on the identified marker pairs to obtain the latest vehicle position information; and

[0012] The location update step updates the current vehicle location information based on the latest vehicle location information obtained from the pose matching step or the poseless matching step.

[0013] The vehicle location information in the marker detection step refers to the vehicle location information that has been obtained in the previous time sequence through the vehicle positioning method.

[0014] Optionally, the pair of identifiers may consist of two identical identifiers.

[0015] Optionally, the marker detection step includes:

[0016] Sub-step 1: Obtain the markers based on the detection results obtained from the markers in the image or the tracking results of the markers in the previously detected image;

[0017] Sub-step 2: Iterate through the markers obtained in sub-step 1 and search within a certain range to see if there are any other markers. If there are other markers, obtain multiple markers.

[0018] Sub-step 3: Determine if a pair of identifiers exists among the acquired identifiers; if a pair of identifiers exists, save the identifier information; and

[0019] Sub-step 4: Filter the saved marker information to remove duplicate marker pairs and obtain at least one marker pair as the marker pair in the ground coordinate system.

[0020] Optionally, in sub-step 3, determining whether a pair of identifiers exists is achieved in the following manner:

[0021] By calculating the distance between the center points of two markers and the directional angle between the markers, it is determined whether the distance between the center points of the two markers and the directional angle between the markers meet the specified threshold. If they meet the preset specified threshold, the two markers are determined to be a pair of markers.

[0022] Optionally, in sub-step 4, duplicate marker pairs are filtered out based on the distance between the center points and the directional angle of the marker pairs to obtain marker pairs in the ground coordinate system.

[0023] Optionally, the pose-free matching step includes:

[0024] For the obtained pairs of markers in the ground coordinate system, they are matched sequentially with the actual pairs of markers in the known fixed scene coordinate system;

[0025] The best matching marker pair is selected as the matched marker pair, and the vehicle location information is obtained based on the matched marker pair and used as the latest vehicle location information.

[0026] Optionally, the process of matching the obtained marker pairs in the ground coordinate system with the known actual marker pairs in the fixed scene coordinate system includes:

[0027] For the obtained marker pairs in the ground coordinate system, the SVD decomposition algorithm is used to solve the ICP algorithm to perform the coordinate system transformation matrix. The corresponding positions of the marker pairs in the ground coordinate system are calculated in the fixed scene coordinate system, and the distance difference between these corresponding positions and the actual marker pairs in the known fixed scene coordinate system is obtained.

[0028] Among them, selecting the best matching tag pair means selecting the tag pair with the smallest distance difference.

[0029] Optionally, the pose-free matching step includes:

[0030] The coarse matching sub-step involves matching the obtained marker pairs in the ground coordinate system with the actual marker pairs in the known fixed scene coordinate system, using SVD decomposition to solve the ICP algorithm to perform the coordinate system transformation matrix, and obtaining the first transformation matrix and the first error.

[0031] The fine matching sub-step involves performing an inverse operation on the first transformation matrix obtained in the coarse matching sub-step to obtain the virtual marker pairs corresponding to the actual marker pairs in the fixed scene coordinate system in the ground coordinate system. For the obtained virtual marker pairs, the already matched virtual marker pairs are filtered out, and the remaining unmatched virtual marker pairs are combined with the already matched marker pairs in the ground coordinate system to calculate the transformation matrix again, resulting in the second transformation matrix and the second error; and

[0032] In the selection sub-step, from the first transformation matrix and first error obtained in the coarse matching sub-step and the second transformation matrix and second error obtained in the fine matching sub-step, the marker pair with the smaller error of the transformation matrix is ​​selected as the matched marker pair. Based on the matched marker pair, the vehicle position information in the fixed scene coordinate system is obtained as the latest vehicle position information.

[0033] Optionally, the pose matching step includes:

[0034] (a) Obtain the coordinate transformation matrix from the ground coordinate system to the fixed scene coordinate system based on the vehicle location information; and

[0035] (b) Based on the coordinate system transformation matrix obtained in (a), calculate the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system, and match the corresponding position with the actual marker pair in the known fixed scene coordinate system. If the matching error is less than the preset matching threshold, the position of the matched marker pair in the fixed scene coordinate system is taken as the latest vehicle position information.

[0036] A vehicle positioning system for fixed parking scenarios according to one aspect of the present invention is characterized in that it includes:

[0037] The marker detection module is used to detect markers and identify marker pairs;

[0038] There is a pose matching module, which is used to perform pose matching based on vehicle position information and marker pairs;

[0039] A pose-free matching module is used to perform pose-free matching based on marker pairs to obtain the latest vehicle position information; and

[0040] The location update module is used to update the current vehicle location information based on the latest vehicle location information obtained by the pose matching module or the poseless matching module.

[0041] Optionally, the pair of identifiers may consist of two identical identifiers.

[0042] Optionally, the marker detection module includes:

[0043] The detection submodule is used to obtain the markers based on the detection results of the markers in the image or the tracking results of the previously detected markers;

[0044] The traversal submodule is used to traverse the markers obtained in the detection submodule and search within a certain range for whether there are other markers. If there are other markers, multiple markers are obtained.

[0045] The judgment submodule is used to determine whether there are paired identifier pairs among the multiple identifiers obtained by the traversal submodule. If there are paired identifier pairs, the identifier information is saved; and

[0046] The filtering submodule is used to filter the marker information obtained by the judgment submodule to filter out duplicate marker pairs and obtain at least one marker pair as a marker pair in the ground coordinate system.

[0047] Optionally, in the determination submodule, determining whether there are two matching identifiers is achieved in the following way:

[0048] By calculating the distance between the center points of two markers and the directional angle between the markers, it is determined whether the distance between the center points of the two markers and the directional angle between the markers meet the specified threshold. If they meet the preset specified threshold, the two markers are determined to be a pair of markers.

[0049] Optionally, the filtering submodule filters out duplicate marker pairs based on the distance between the center points of the marker pairs and the directional angle to obtain marker pairs in the ground coordinate system.

[0050] Optionally, the pose-free matching module includes:

[0051] The matching submodule is used to sequentially match the obtained pairs of markers in the ground coordinate system with the actual pairs of markers in the known fixed scene coordinate system; and

[0052] The selection submodule is used to select the best matching marker pair as the matched marker pair, obtain vehicle location information based on the matched marker pair, and use it as the latest vehicle location information.

[0053] Optionally, the matching submodule, for the obtained marker pairs in the ground coordinate system, uses the SVD decomposition to solve the ICP algorithm to perform a coordinate system transformation matrix, calculates the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system, and obtains the distance difference between the corresponding position and the known actual marker pair in the fixed scene coordinate system.

[0054] The selection submodule selects the optimal matching marker pair by selecting the marker pair with the smallest distance difference.

[0055] Optionally, the pose-free matching module includes:

[0056] The coarse matching submodule matches the obtained marker pairs in the ground coordinate system with the actual marker pairs in the known fixed scene coordinate system, and uses the SVD decomposition to solve the ICP algorithm to perform the coordinate system transformation matrix, thereby obtaining the first transformation matrix and the first error.

[0057] The fine matching submodule performs an inverse operation on the first transformation matrix obtained in the coarse matching submodule to obtain the virtual marker pairs corresponding to the actual marker pairs in the fixed scene coordinate system in the ground coordinate system. For the obtained virtual marker pairs, it filters out the already matched virtual marker pairs, and calculates the transformation matrix again with the remaining unmatched virtual marker pairs and the already matched marker pairs in the ground coordinate system to obtain the second transformation matrix and the second error; and

[0058] The selection submodule is used to select the marker pair with the smaller error from the first transformation matrix and first error obtained from the coarse matching submodule and the second transformation matrix and second error obtained from the fine matching submodule as the matched marker pair, and to obtain the vehicle position information in the fixed scene coordinate system based on the matched marker pair as the latest vehicle position information.

[0059] Optionally, the pose matching module includes:

[0060] The transformation submodule obtains the coordinate system transformation matrix from the ground coordinate system to the fixed scene coordinate system based on the vehicle position information; and

[0061] The matching submodule calculates the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system based on the coordinate system transformation matrix obtained by the transformation submodule, and matches the corresponding position with the known actual marker pair in the fixed scene coordinate system. If the matching error is less than a preset matching threshold, the position of the matched marker pair in the fixed scene coordinate system is taken as the latest vehicle position information.

[0062] A computer-readable medium of the present invention stores a computer program thereon, characterized in that the computer program, when executed by a processor, implements the vehicle positioning method for a fixed parking scenario.

[0063] A computer device according to one aspect of the present invention includes a storage module, a processor, and a computer program stored on the storage module and executable on the processor, characterized in that the processor executes the computer program to implement the vehicle positioning method for fixed parking scenarios.

[0064] The vehicle of one aspect of the present invention is characterized by including any one of the vehicle positioning systems for fixed parking scenarios described in the present invention. Attached Figure Description

[0065] Figure 1 This is a schematic diagram illustrating a marker in a vehicle positioning method in a battery swapping station scenario according to one embodiment.

[0066] Figure 2This is a schematic diagram illustrating the process of a vehicle positioning method in a battery swapping station scenario according to one embodiment.

[0067] Figure 3 This is a schematic diagram illustrating the specific process of a transformation example without pose matching steps.

[0068] Figure 4 This is a schematic diagram illustrating the construction of a vehicle positioning system for a fixed parking scenario according to one embodiment.

[0069] Figure 5 This is a schematic diagram illustrating the construction of a transformation example without a pose matching module. Detailed Implementation

[0070] The following are some embodiments of the present invention, intended to provide a basic understanding of the invention. They are not intended to identify key or decisive elements of the invention or to limit the scope of protection sought.

[0071] For purposes of brevity and illustrativeness, the principles of the invention are described herein primarily with reference to exemplary embodiments thereof. However, those skilled in the art will readily recognize that the same principles are equivalently applicable to all types of vehicle positioning methods and systems for fixed parking scenarios, and that these same principles can be implemented therein, with any such variations not departing from the actual spirit and scope of this patent application.

[0072] Furthermore, reference is made in the following description to the accompanying drawings, which illustrate specific exemplary embodiments. Electrical, mechanical, logical, and structural modifications may be made to these embodiments without departing from the spirit and scope of the invention. Moreover, while features of the invention are disclosed in conjunction with only one of several embodiments, this feature may be combined with one or more other features of other embodiments if desired and / or advantageous for any given or identifiable function. Therefore, the following description should not be considered limiting in any sense, and the scope of the invention is defined by the appended claims and their equivalents.

[0073] Terms such as “possessing” and “comprising” indicate that, in addition to having units (modules) and steps that are directly and explicitly stated in the specification and claims, the technical solution of the present invention does not exclude the presence of other units (modules) and steps that are not directly or explicitly stated.

[0074] This invention aims to provide a method for assisted localization within a defined area of ​​a fixed parking scenario, based on visual perception capabilities and aided by markers. The main problems addressed by this invention include: it eliminates the need for strict restrictions on the scene, the vehicle's initial posture, and the parking path, offering a degree of flexibility; it only requires matching ground markers, greatly simplifying the algorithm's complexity; and it can provide accurate and robust real-time localization for vehicles.

[0075] To address the aforementioned technical problems, particularly to enable rapid and accurate matching, this invention proposes an improved visual perception positioning algorithm based on the setting of specific markers. Specifically, by improving the ICP algorithm, it is possible to achieve rapid matching, thereby providing real-time positioning information for vehicles within a certain area of ​​a fixed parking scenario.

[0076] The following is an implementation method for a fixed parking scenario, using parking outside a battery swapping station as an example. This fixed parking scenario is applicable not only to parking outside battery swapping stations but also to other fixed scenarios such as parking lots.

[0077] Figure 1 This is a schematic diagram illustrating a marker in a vehicle positioning method in a battery swapping station scenario according to one embodiment.

[0078] like Figure 1 As shown, several single right-angled markers and several double right-angled markers are installed outside the battery swapping station. Figure 1 The signs shown in the examples are planar signs, but the signs of the present invention can also be implemented using spatial signs.

[0079] Regarding marker pairs, they consist of two identical markers. In this invention, a marker pair specifically refers to two markers that are parallel and within a certain distance to form a marker pair. (See reference) Figure 1 ,like Figure 1 In the ground markings, two signs on the left side of the parking space form a pair, two signs in the middle of the parking space form a pair, and two signs on the right side of the parking space form a pair.

[0080] Regarding the placement of planar and spatial markers, planar markers include one or more of the following combinations: positioning markers placed on the ground side of a fixed parking scenario; positioning markers placed around the fixed parking scenario; and positioning markers placed above the fixed parking scenario. Conversely, spatial markers include one or more of the following combinations: positioning markers placed on the ground side of a fixed parking scenario; positioning markers placed around the fixed parking scenario; and positioning markers placed above the fixed parking scenario.

[0081] Regarding the shapes of planar and spatial signs, planar signs can be a combination of one or more of the following: arrow-shaped signs; single right-angled signs; triangular signs; and polygonal signs. On the other hand, spatial signs can be a combination of one or more of the following: QR code signs; arrow-shaped signs; single right-angled signs; triangular signs; polygonal signs; and objects already present in a fixed parking scenario. The term "objects already present in a fixed parking scenario" refers to objects inherent in the fixed parking scenario (such as a battery swapping station), in addition to additional signs, such as front V-grooves, rear flat grooves, warning stickers to remind users, etc.

[0082] Figure 2 This is a schematic diagram illustrating the process of a vehicle positioning method in a battery swapping station scenario according to one embodiment.

[0083] like Figure 2 As shown, a vehicle positioning method in a battery swapping station scenario according to one embodiment includes the following steps:

[0084] Step S100: After initialization, obtain the detection result or tracking result of the marker. Preferably, some markers that are too large or too small can be further filtered out according to the size information. If the vehicle location information has been obtained, continue to step S200. If the vehicle location information has not been obtained, continue to step S300.

[0085] Step S200: Perform pose matching based on vehicle position information. If pose matching is successful, proceed to step S400. If pose matching fails, continue to step S300.

[0086] Step S300: Perform poseless matching based on the markers;

[0087] Step S400: Update the vehicle position information based on the result of pose matching or no pose matching. Based on the matched vehicle position information, predict the vehicle position information of the next frame in combination with the vehicle odometer, and repeat step S100.

[0088] Regarding vehicle location information, from a temporal perspective, there is no vehicle location information initially. However, the vehicle location information in the ground coordinate system is obtained based on the positioning results. In subsequent positioning, the current frame can be matched based on the previous matching results and the odometer results, which is called "pose-based matching".

[0089] Before describing the specific steps of this invention, a brief explanation of the process of solving the ICP algorithm by SVD decomposition in the prior art will be given.

[0090] Singular Value Decomposition (SVD) is an algorithm widely used in the field of machine learning. SVD is an important matrix decomposition in linear algebra, and it is a generalization of eigenvalue decomposition to any matrix.

[0091] The ICP algorithm (Iterative Closest Points Algorithm) is mainly used for matching problems of 3D objects. It can be understood as follows: given two sets of 3D data points from different coordinate systems, find a spatial transformation between the two sets of points so that they can be spatially matched.

[0092] The process of solving the ICP algorithm through SVD decomposition includes the following steps:

[0093] For two sets of matching points P = {p1, p2, ..., p...} n} and Q = {q1, q2, ..., q n}, where p i q i ∈R d Let be the set of matching points in space. The rotation matrix R and translation vector t are calculated using the ICP algorithm to minimize the error between the transformed point set and the target point set. The formula is as follows:

[0094]

[0095] Step 1: Calculate the weighted center of the two sets of matching points:

[0096]

[0097] Step 2: Obtain the decentralized point set:

[0098]

[0099] Step 3: Calculate the d×d covariance matrix:

[0100] S = XWY T

[0101] Where X and Y are d×n matrices, x i y i These are their column elements.

[0102] W = diag(ω1, ω2, ..., ω) n )

[0103] Step 4: Perform SVD decomposition on S: S = U∑Y T The rotation matrix R is obtained as follows:

[0104]

[0105] Step 5: Calculate the translation amount:

[0106]

[0107] The transformation matrix (R, t) is calculated through the above steps. However, directly searching for matching points is very time-consuming and computationally intensive. To solve this technical problem, this invention first designs... Figure 1 The special marker pairs shown are determined by calculating the distance and marker direction based on the detection or tracking results of the markers during the poseless matching process. The determined marker pairs are then matched with the actual marker pairs in the map to obtain the optimal matching result.

[0108] In this invention, the reason why matching using marker pairs can reduce the amount of computation is, for example, in Figure 1 In this algorithm, there are only three pairs of markers. Matching is performed based on the detected marker pairs, assuming a match with a specific marker pair in the map. This requires significantly less computation than traversing all matching points. For example, if four marker pairs are detected, the computational cost is at most 4*3=12 coordinate transformation matrices, which is much faster than the ICP algorithm, which performs extensive calculations on the obtained point locations. In practical applications, if the detection results are good, the number of detected marker pairs is generally close to the number of marker pairs in the actual map.

[0109] Next, the process of obtaining paired markers in this invention will be explained.

[0110] The process of obtaining paired identifiers includes the following steps:

[0111] Step S1: Based on the image detection results and the markers in the temporal frame tracking results, filter according to the marker size information to filter out markers that are too large or too small. The image detection results are obtained from the detection of markers in the image, and the temporal frame tracking results are the tracking results of the previously detected markers. There may be some overlap of markers.

[0112] Step S2: Traverse the markers obtained in Step S1 and search within a certain range for other markers. If there are other markers, proceed to Step S3 (if there are other markers, then multiple markers are obtained); otherwise, continue traversing Step S2.

[0113] Step S3: Calculate the distance between the center points of the two markers and the angle between their directions. Here, the direction of a marker is defined as the sum of the vectors from its two endpoints to its center point. Normalize this sum vector and determine if the distance between the center points of the two markers and the angle between their directions meet a certain threshold. If they do, find a pair of markers and save the marker pair information.

[0114] Step S4: Filter the acquired marker pair information. Based on the distance and orientation angle information of the paired marker pairs, filter out some duplicate marker pairs (because the tracking results and detection results are not completely consistent, which may cause one marker to form a pair with multiple markers). Finally, obtain at least one pair of markers, including the case of only one pair of markers, as well as the case of multiple pairs of markers and matching each marker pair to find the marker pair with the smallest matching error.

[0115] In this embodiment, two coordinate systems are used: the ground coordinate system (also known as the image coordinate system); and the battery swapping station coordinate system (also known as the map coordinate system, corresponding to the "fixed scene coordinate system" in the claims).

[0116] Here, "ground coordinate system" refers to the traditional image coordinate system, a two-dimensional coordinate system constructed with the top left corner of the image as the origin, and the unit is pixels. "Battery swapping station coordinate system" refers to the coordinate system in the actual 3D scene with a certain location in the battery swapping station as the origin, and the unit is, for example, meters.

[0117] The following section explains the specifics of pose-free matching and pose-based matching.

[0118] 1. Poseless matching includes the following steps:

[0119] (a) Find the marker pairs that can form a pair (i.e., marker pairs in the ground coordinate system);

[0120] (b) Traverse the marker pairs obtained in (a) and match the marker pairs in the ground coordinate system with the actual marker pairs (i.e., the actual marker pairs in the battery swapping station coordinate system). Specifically, use the SVD decomposition to solve the ICP algorithm to obtain the coordinate system transformation matrix, and calculate the distance difference between the actual marker pairs (i.e., the actual marker pairs in the battery swapping station coordinate system) and the marker pairs transformed into the battery swapping station coordinate system by the transformation matrix (if they are not matched, the distance between the marker pairs is replaced by a larger value).

[0121] (c) Select the optimal match (i.e. the smallest distance difference) from (b). If the optimal match meets the preset matching threshold, the no-pose matching is completed. Update the vehicle's position information in the battery swapping station coordinate system according to the matched marker pair. If the optimal match does not meet the preset matching threshold, no actual marker pair is matched, and wait for the subsequent time frame to perform the matching.

[0122] 2. Pose matching includes the following steps:

[0123] (a) Obtain the coordinate system transformation matrix based on vehicle position information; and

[0124] (b) If the matrix matching obtained in (a) is good (i.e. the distance difference meets the preset matching threshold), then update the vehicle position information according to the matched marker pair; otherwise, perform poseless matching.

[0125] In (a), the vehicle position information, as an example, refers to the position information of the center point of the front and rear axles of the vehicle in the coordinate system of the battery swapping station. Since the coordinates of the front and rear axles of the vehicle in the image are fixed, the transformation matrix from the coordinates of the front and rear axles of the vehicle in the image to the coordinates of the front and rear axles of the vehicle in the coordinate system of the battery swapping station can be calculated, that is, the transformation matrix from the ground coordinate system to the coordinate system of the battery swapping station.

[0126] In (b), the coordinate transformation matrix obtained in (a) is used to calculate the corresponding position of the marker pair in the battery swapping station coordinate system. The corresponding position of the marker pair in the battery swapping station coordinate system is compared with the actual marker pair in the battery swapping station coordinate system. If the matching error is less than the preset matching threshold, the vehicle position information is considered to be relatively accurate, and there is no need to match the markers again. Only the error of the matching result needs to be calculated once. The vehicle position information can be updated according to the position of the actual marker pair in the battery swapping station coordinate system that has been matched. If the matching error is greater than the preset matching threshold, it means that the current vehicle position information is inaccurate, and poseless matching is performed.

[0127] Furthermore, a transformation example for poseless matching is explained.

[0128] Figure 3 This is a schematic diagram illustrating the specific process of a transformation example without pose matching.

[0129] In this example, pose-free matching can be divided into two processes: coarse matching and fine matching.

[0130] like Figure 3 As shown, the poseless matching of this transformation example includes:

[0131] (1) Coarse matching:

[0132] Match the marker pairs in the ground coordinate system with the actual marker pairs (i.e., the actual marker pairs in the battery swapping station coordinate system) to obtain the first transformation matrix RT1 (the transformation matrix from the ground coordinate system to the battery swapping station coordinate system) and the first error cost1 at this time.

[0133] (2) Precise matching:

[0134] (a) By performing an inverse operation on the transformation matrix RT1 obtained from the coarse matching, the corresponding virtual markers in the ground coordinate system for the actual marker pairs in the map (i.e., the actual marker pairs in the battery swapping station coordinate system) are obtained.

[0135] (b) Based on the virtual marker pairs obtained in (a), filter out the marker pairs that have been matched for detection or tracking, and calculate the transformation matrix again with the remaining unmatched map marker pairs (i.e., virtual marker pairs) and the matched marker pairs for detection or tracking to obtain the second transformation matrix RT2 and the second error cost2.

[0136] (3) Based on the first error cost1 and the second error cost2 of process (1) and process (2), select the transformation matrix corresponding to the one with smaller error as the final transformation matrix.

[0137] The reason for performing coarse and fine matching in this example is that during matching, the actual detected or tracked markers are matched according to the coordinate transformation matrix. However, due to matching errors, the real markers in the map (i.e., the actual markers in the battery swapping station coordinate system) may not be completely matched. The unmatched real map markers (i.e., the actual markers in the battery swapping station coordinate system) are inversely transformed into "virtual markers" (i.e., markers in the ground coordinate system) in the image through the coordinate transformation matrix. These "virtual markers" are not detected or tracked, but are hypothetical. Then, these "virtual markers" are added to the image marker-map actual marker pairs that can be matched, forming a set of marker matching pairs. The marker matching pairs are specifically manifested as: (1) the matching pair of image markers (i.e., markers in the ground coordinate system) - map actual markers (i.e., actual markers in the battery swapping station coordinate system); and (2) the matching pair of virtual markers - map actual markers (i.e., actual markers in the battery swapping station coordinate system). The coordinate transformation matrix is ​​recalculated based on these new marker matching pairs, and the transformation matrix is ​​updated.

[0138] Since the transformation matrix is ​​obtained through SVD decomposition, generally speaking, more matching points mean a better transformation matrix. Moreover, the detection and tracking results are not completely accurate. By back-calculating the positions of map marker pairs in the ground coordinate system, the matching marker pairs can be supplemented to a certain extent. Here, the method of improving matching accuracy from coarse matching to fine matching is a further optimization method.

[0139] Next, the vehicle positioning system for fixed parking scenarios of the present invention will be described.

[0140] Figure 4 This is a schematic diagram illustrating the construction of a vehicle positioning system for a fixed parking scenario according to one embodiment.

[0141] like Figure 4 As shown, the vehicle positioning system for fixed parking scenarios of the present invention includes:

[0142] The marker detection module 100 is used to detect markers and identify marker pairs;

[0143] A pose matching module 200 is used to perform pose matching based on vehicle position information and marker pairs;

[0144] The poseless matching module 300 is used to perform poseless matching based on marker pairs to obtain the latest vehicle position information; and

[0145] The location update module 400 is used to update the current vehicle location information based on the latest vehicle location information obtained by the pose matching module or the poseless matching module.

[0146] The pair of identifiers consists of two identical identifiers.

[0147] Furthermore, the marker detection module 100 includes:

[0148] The detection submodule 110 is used to obtain the marker based on the detection result of the marker in the image or the tracking result of the previously detected marker;

[0149] The traversal submodule 120 is used to traverse the markers obtained in the detection submodule and search within a certain range whether there are other markers. If there are other markers, multiple markers are obtained.

[0150] The judgment submodule 130 is used to determine whether there are paired identifier pairs among the multiple identifiers obtained by the traversal submodule; if there are paired identifier pairs, the identifier information is saved; and

[0151] The filtering submodule 140 is used to filter the marker information obtained by the judgment submodule to filter out duplicate marker pairs and obtain at least one marker pair as a marker pair in the ground coordinate system.

[0152] In the judgment submodule 130, the determination of whether two markers are a pair of markers is achieved in the following way: by calculating the distance between the center points of the two markers and the directional angle between the markers, it is determined whether the distance between the center points of the two markers and the directional angle between the markers meet the specified threshold. If the specified threshold is met, the two markers are determined to be a pair of markers.

[0153] The filtering submodule 140 filters out duplicate marker pairs based on the distance between the center points of the marker pairs and the directional angle to obtain marker pairs in the ground coordinate system.

[0154] Furthermore, the pose matching module 200 includes:

[0155] Transformation submodule 210 obtains the coordinate system transformation matrix from the ground coordinate system to the fixed scene coordinate system based on the vehicle position information; and

[0156] The matching submodule 220 calculates the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system according to the coordinate system transformation matrix obtained by the transformation submodule, and matches the corresponding position with the known actual marker pair in the fixed scene coordinate system. If the matching error is less than a preset matching threshold, the position of the matched marker pair in the fixed scene coordinate system is taken as the latest vehicle position information.

[0157] On the other hand, if the matching error in the matching submodule 220 is greater than the preset matching threshold, it indicates that the current vehicle position information is inaccurate, and the poseless matching module 300 performs poseless matching.

[0158] Furthermore, the poseless matching module 300 includes:

[0159] The matching submodule 310 is used to match the obtained marker pairs in the ground coordinate system with the actual marker pairs in the known fixed scene coordinate system in turn; and the selection submodule 320 is used to select the best matching marker pair as the matched marker pair, obtain the vehicle position information based on the matched marker pair and use it as the latest vehicle position information.

[0160] Specifically, the matching submodule 310 uses the SVD decomposition to solve the ICP algorithm for the coordinate system transformation matrix of the obtained marker pair in the ground coordinate system, calculates the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system, and obtains the distance difference between the corresponding position and the actual marker pair in the known fixed scene coordinate system. The selection submodule 320 selects the marker pair with the best matching, which means selecting the marker pair with the smallest distance difference.

[0161] Next, another transformation example without a pose matching module will be explained.

[0162] Figure 5 This is a schematic diagram illustrating the construction of a transformation example without a pose matching module.

[0163] like Figure 5 As shown, in a transformation example without a pose matching module, the pose-free matching module 300 includes:

[0164] The coarse matching submodule 311 matches the obtained pairs of markers in the ground coordinate system with the actual pairs of markers in the known fixed scene coordinate system, and uses the SVD decomposition to solve the ICP algorithm to perform the coordinate system transformation matrix, thereby obtaining the first transformation matrix and the first error.

[0165] The fine matching submodule 312 performs an inverse operation on the first transformation matrix obtained in the coarse matching submodule to obtain the virtual marker pairs corresponding to the actual marker pairs in the fixed scene coordinate system in the ground coordinate system. For the obtained virtual marker pairs, it filters out the already matched virtual marker pairs, and calculates the transformation matrix again with the remaining unmatched virtual marker pairs and the already matched marker pairs in the ground coordinate system to obtain the second transformation matrix and the second error; and

[0166] Selection submodule 313 is used to select the marker pair with the smaller error from the first transformation matrix and first error obtained from the coarse matching submodule and the second transformation matrix and second error obtained from the fine matching submodule as the matched marker pair, and obtain the vehicle position information in the fixed scene coordinate system based on the matched marker pair as the latest vehicle position information.

[0167] As described above, the vehicle positioning method and system for fixed parking scenarios according to the present invention can achieve fast and accurate matching. In particular, the present invention improves the ICP algorithm by using specific markers (i.e., marker pairs) to achieve fast matching and provide real-time positioning information for vehicles within a certain planar area outside the battery swapping station.

[0168] Moreover, the vehicle positioning method and system for fixed parking scenarios of the present invention utilize specific markers for positioning, which has high flexibility and is not limited by parking path. By combining the matched ground marker pairs with the actual marker pairs in the known map (e.g., the cloud-based map), the position of the vehicle in the coordinate system of the battery swapping station is calculated, thereby achieving the purpose of real-time vehicle positioning.

[0169] The above lists the vehicle positioning method and system for battery swapping stations. Of course, the vehicle positioning method and system for fixed parking scenarios of the present invention are not only applicable to battery swapping stations, but also to other fixed parking scenarios, such as parking garages.

[0170] The present invention also provides a computer-readable medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the above-described vehicle positioning method for fixed parking scenarios.

[0171] The present invention also provides a computer device, including a storage module, a processor, and a computer program stored on the storage module and executable on the processor, characterized in that the processor executes the computer program to implement the above-described vehicle positioning method for fixed parking scenarios.

[0172] The present invention also provides a vehicle, characterized in that it includes the above-described vehicle positioning system for fixed parking scenarios.

[0173] The above examples primarily illustrate the vehicle positioning method and system for fixed parking scenarios according to the present invention. Although only some specific embodiments of the invention have been described, those skilled in the art should understand that the invention can be implemented in many other forms without departing from its spirit and scope. Therefore, the examples and embodiments shown are to be considered illustrative rather than restrictive, and the invention may encompass various modifications and substitutions without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. A vehicle positioning method for fixed parking scenarios, characterized in that, include: The sign detection step detects signs and identifies sign pairs. If the vehicle position information in the fixed parking scenario has been obtained, the following pose matching step continues. If the vehicle position information in the fixed parking scenario has not been obtained, the following poseless matching step is skipped. The pose matching step involves performing pose matching based on the obtained vehicle position information and the identified markers. If the latest vehicle position information is obtained based on the pose matching, the process jumps to the position update step below; otherwise, the poseless matching step below continues. The pose-free matching step involves performing pose-free matching based on the identified marker pairs to obtain the latest vehicle position information; and The location update step updates the current vehicle location information based on the latest vehicle location information obtained from the pose matching step or the poseless matching step. The vehicle location information in the marker detection step refers to the vehicle location information that has been obtained in a previous time sequence through the vehicle positioning method. The pair of identifiers includes two identical identifiers. The marker detection step includes: Sub-step 1: Obtain the markers based on the detection results obtained from the markers in the image or the tracking results of the markers in the previously detected image; Sub-step 2: Iterate through the markers obtained in sub-step 1 and search within a certain range to see if there are any other markers. If there are other markers, obtain multiple markers. Sub-step 3: Determine if a pair of identifiers exists among the acquired identifiers; if a pair of identifiers exists, save the identifier information; and Sub-step 4: Filter the saved marker information to remove duplicate marker pairs and obtain at least one marker pair as the marker pair in the ground coordinate system.

2. The vehicle positioning method for fixed parking scenarios as described in claim 1, characterized in that, In sub-step 3, the determination of whether a pair of identifiers exists is achieved in the following manner: By calculating the distance between the center points of two markers and the directional angle between the markers, it is determined whether the distance between the center points of the two markers and the directional angle between the markers meet the specified threshold. If they meet the preset specified threshold, the two markers are determined to be a pair of markers.

3. The vehicle positioning method for fixed parking scenarios as described in claim 2, characterized in that, In sub-step 4, duplicate marker pairs are filtered out based on the distance between the center points and the directional angle of the marker pairs to obtain the marker pairs in the ground coordinate system.

4. The vehicle positioning method for fixed parking scenarios as described in claim 1, characterized in that, The pose-free matching step includes: For the obtained pairs of markers in the ground coordinate system, they are matched sequentially with the actual pairs of markers in the known fixed scene coordinate system; The best matching marker pair is selected as the matched marker pair, and the vehicle location information is obtained based on the matched marker pair and used as the latest vehicle location information.

5. The vehicle positioning method for fixed parking scenarios as described in claim 4, characterized in that, in, For the obtained marker pairs in the ground coordinate system, they are matched sequentially with the actual marker pairs in the known fixed scene coordinate system, including: For the obtained marker pairs in the ground coordinate system, the SVD decomposition algorithm is used to solve the ICP algorithm to perform the coordinate system transformation matrix. The corresponding positions of the marker pairs in the ground coordinate system are calculated in the fixed scene coordinate system, and the distance difference between these corresponding positions and the actual marker pairs in the known fixed scene coordinate system is obtained. Among them, selecting the best matching tag pair means selecting the tag pair with the smallest distance difference.

6. The vehicle positioning method for a fixed parking scenario as described in claim 1, characterized in that, The pose-free matching step includes: The coarse matching sub-step involves matching the obtained marker pairs in the ground coordinate system with the actual marker pairs in the known fixed scene coordinate system, using SVD decomposition to solve the ICP algorithm to perform the coordinate system transformation matrix, and obtaining the first transformation matrix and the first error. The fine matching sub-step involves performing an inverse operation on the first transformation matrix obtained in the coarse matching sub-step to obtain the virtual marker pairs corresponding to the actual marker pairs in the fixed scene coordinate system in the ground coordinate system. For the obtained virtual marker pairs, the already matched virtual marker pairs are filtered out, and the remaining unmatched virtual marker pairs are combined with the already matched marker pairs in the ground coordinate system to calculate the transformation matrix again, resulting in the second transformation matrix and the second error; and In the selection sub-step, from the first transformation matrix and first error obtained in the coarse matching sub-step and the second transformation matrix and second error obtained in the fine matching sub-step, the marker pair with the smaller error of the transformation matrix is ​​selected as the matched marker pair. Based on the matched marker pair, the vehicle position information in the fixed scene coordinate system is obtained as the latest vehicle position information.

7. The vehicle positioning method for fixed parking scenarios as described in claim 1, characterized in that, The pose matching step includes: (a) Obtain the coordinate transformation matrix from the ground coordinate system to the fixed scene coordinate system based on the vehicle location information; and (b) Based on the coordinate system transformation matrix obtained in (a), calculate the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system, and match the corresponding position with the actual marker pair in the known fixed scene coordinate system. If the matching error is less than the preset matching threshold, the position of the matched marker pair in the fixed scene coordinate system is taken as the latest vehicle position information.

8. A vehicle positioning system for fixed parking scenarios, characterized in that, include: The marker detection module is used to detect markers and identify marker pairs; There is a pose matching module, which is used to perform pose matching based on vehicle position information and marker pairs; A pose-free matching module is used to perform pose-free matching based on marker pairs to obtain the latest vehicle position information; and The location update module is used to update the current vehicle location information based on the latest vehicle location information obtained by the pose matching module or the poseless matching module. The pair of identifiers consists of two identical identifiers. The marker detection module includes: The detection submodule is used to obtain the markers based on the detection results of the markers in the image or the tracking results of the previously detected markers; The traversal submodule is used to traverse the markers obtained in the detection submodule and search within a certain range for whether there are other markers. If there are other markers, multiple markers are obtained. The judgment submodule is used to determine whether there are paired identifier pairs among the multiple identifiers obtained by the traversal submodule. If there are paired identifier pairs, the identifier information is saved; and The filtering submodule is used to filter the marker information obtained by the judgment submodule to filter out duplicate marker pairs and obtain at least one marker pair as a marker pair in the ground coordinate system.

9. The vehicle positioning system for fixed parking scenarios as described in claim 8, characterized in that, In the judgment submodule, the determination of whether there are two markers that are a pair of markers is achieved in the following way: By calculating the distance between the center points of two markers and the directional angle between the markers, it is determined whether the distance between the center points of the two markers and the directional angle between the markers meet the specified threshold. If they meet the preset specified threshold, the two markers are determined to be a pair of markers.

10. The vehicle positioning system for fixed parking scenarios as described in claim 9, characterized in that, The filtering submodule filters out duplicate marker pairs based on the distance between their center points and the angle between their directions to obtain marker pairs in the ground coordinate system.

11. The vehicle positioning system for fixed parking scenarios as described in claim 8, characterized in that, The pose-free matching module includes: The matching submodule is used to sequentially match the obtained pairs of markers in the ground coordinate system with the actual pairs of markers in the known fixed scene coordinate system; and The selection submodule is used to select the best matching marker pair as the matched marker pair, obtain vehicle location information based on the matched marker pair, and use it as the latest vehicle location information.

12. The vehicle positioning system for fixed parking scenarios as described in claim 11, characterized in that, The matching submodule, for the obtained marker pairs in the ground coordinate system, uses the SVD decomposition to solve the ICP algorithm to perform coordinate system transformation matrix, calculates the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system, and obtains the distance difference between the corresponding position and the actual marker pair in the known fixed scene coordinate system. The selection submodule selects the optimal matching marker pair by selecting the marker pair with the smallest distance difference.

13. The vehicle positioning system for fixed parking scenarios as described in claim 8, characterized in that, The pose-free matching module includes: The coarse matching submodule matches the obtained marker pairs in the ground coordinate system with the actual marker pairs in the known fixed scene coordinate system, and uses the SVD decomposition to solve the ICP algorithm to perform the coordinate system transformation matrix, thereby obtaining the first transformation matrix and the first error. The fine matching submodule performs an inverse operation on the first transformation matrix obtained in the coarse matching submodule to obtain the virtual marker pairs corresponding to the actual marker pairs in the fixed scene coordinate system in the ground coordinate system. For the obtained virtual marker pairs, it filters out the already matched virtual marker pairs, and calculates the transformation matrix again with the remaining unmatched virtual marker pairs and the already matched marker pairs in the ground coordinate system to obtain the second transformation matrix and the second error; and The selection submodule is used to select the marker pair with the smaller error from the first transformation matrix and first error obtained from the coarse matching submodule and the second transformation matrix and second error obtained from the fine matching submodule as the matched marker pair, and to obtain the vehicle position information in the fixed scene coordinate system based on the matched marker pair as the latest vehicle position information.

14. The vehicle positioning system for fixed parking scenarios as described in claim 8, characterized in that, The pose matching module includes: The transformation submodule obtains the coordinate system transformation matrix from the ground coordinate system to the fixed scene coordinate system based on the vehicle position information; and The matching submodule calculates the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system based on the coordinate system transformation matrix obtained by the transformation submodule, and matches the corresponding position with the known actual marker pair in the fixed scene coordinate system. If the matching error is less than a preset matching threshold, the position of the matched marker pair in the fixed scene coordinate system is taken as the latest vehicle position information.

15. A computer-readable medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the vehicle positioning method for fixed parking scenarios as described in any one of claims 1 to 7.

16. A computer device comprising a storage module, a processor, and a computer program stored on the storage module and executable on the processor, characterized in that, When the processor executes the computer program, it implements the vehicle positioning method for fixed parking scenarios as described in any one of claims 1 to 7.

17. A vehicle, characterized in that, Includes the vehicle positioning system for fixed parking scenarios as described in any one of claims 8 to 14.

Images