Search

CN-113619569-B - Vehicle positioning method and system for fixing parking scene

CN113619569BCN 113619569 BCN113619569 BCN 113619569BCN-113619569-B

Abstract

The invention relates to a vehicle positioning method and a system for fixing a parking scene. The method comprises the steps of detecting the marker, identifying the marker pair, carrying out pose matching according to the vehicle position information and the marker pair, carrying out pose-free matching according to the marker pair to obtain the latest vehicle position information, and updating the current vehicle position information according to the pose matching step or the latest vehicle position information obtained in the pose-free matching step.

Inventors

  • ZHANG HENG
  • ZHANG KAI
  • DAI JUNYI
  • WANG SHITING

Assignees

  • 蔚来汽车科技(安徽)有限公司

Dates

Publication Date
20260508
Application Date
20210802

Claims (17)

  1. 1. A vehicle positioning method for fixing a parking scene, comprising: A marker detection step of detecting a marker and identifying a marker pair, if the vehicle position information in the fixed parking scene is obtained, continuing the following pose matching step, and if the vehicle position information in the fixed parking scene is not obtained, jumping to the following pose-free matching step; a pose matching step, namely carrying out pose matching according to the obtained vehicle position information and the identified marker pairs, if the latest vehicle position information is obtained according to the pose matching, jumping to a position updating step, otherwise, continuing a pose matching step; a pose-free matching step of performing pose-free matching based on the identified marker pair to obtain the latest vehicle position information, and A position updating step of updating current vehicle position information according to the latest vehicle position information obtained in the pose matching step or the pose-free matching step, Wherein the vehicle position information in the marker detecting step refers to vehicle position information that has been obtained by performing the vehicle positioning method at a previous timing, The pair of markers comprises two identical markers, The marker detection step includes: step 1, acquiring a marker according to a detection result obtained by the marker in the image or a tracking result of the marker in the image detected before; step 2, traversing the markers obtained in the sub-step 1, searching whether other markers exist in a certain range, and obtaining a plurality of markers if the other markers exist; a substep 3 of judging whether a pair of markers exists among the acquired plurality of markers, storing the marker information if the condition of the pair of markers is satisfied, and And 4, filtering the stored identifier information to filter out repeated identifier pairs to obtain at least one pair of identifier pairs serving as the identifier pairs in the ground coordinate system.
  2. 2. The vehicle locating method for securing a parking scene as claimed in claim 1, characterized in that, In said sub-step 3, the determination of whether a marker pair is present is achieved by: And judging whether the center point distance and the direction included angle of the two markers meet the specified threshold value or not by calculating the center point distance and the direction included angle of the two markers, and judging that the two markers are paired markers if the center point distance and the direction included angle of the two markers meet the preset specified threshold value.
  3. 3. A vehicle locating method for a stationary parking scenario as claimed in claim 2, wherein, In sub-step 4, duplicate pairs of markers are filtered out based on the center point distance and the direction angle of the pairs of markers to obtain pairs of markers in the ground coordinate system.
  4. 4. The vehicle positioning method for fixing a parking scene as set forth in claim 1, wherein the pose-free matching step includes: sequentially matching the obtained marker pairs in the ground coordinate system with the actual marker pairs in the known fixed scene coordinate system; And selecting the matched optimal identifier pair as the matched identifier pair, and obtaining vehicle position information based on the matched identifier pair and serving as the latest vehicle position information.
  5. 5. The vehicle locating method for securing a parking scene as claimed in claim 4, characterized in that, Wherein, for the marker pairs in the obtained ground coordinate system, sequentially matching with the actual marker pairs in the known fixed scene coordinate system comprises: for the obtained marker pairs in the ground coordinate system, carrying out coordinate system transformation matrix by utilizing SVD decomposition to solve ICP algorithm, calculating the corresponding positions of the marker pairs in the ground coordinate system in the fixed scene coordinate system, obtaining the distance difference between the corresponding positions and the actual marker pairs in the known fixed scene coordinate system, Wherein selecting the identifier pair that matches optimally refers to selecting the identifier pair for which the distance difference is the smallest.
  6. 6. The vehicle positioning method for fixing a parking scene as set forth in claim 1, wherein the pose-free matching step includes: A rough matching sub-step, namely matching the obtained identifier pairs in the ground coordinate system with the actual identifier pairs in the known fixed scene coordinate system, and carrying out coordinate system transformation matrix by utilizing SVD decomposition to solve an ICP algorithm to obtain a first transformation matrix and a first error; A fine matching sub-step of performing inverse operation on the first transformation matrix acquired in the coarse matching sub-step to obtain a virtual identifier pair corresponding to the actual identifier pair in the ground coordinate system in the fixed scene coordinate system, filtering the virtual identifier pair which is already matched for the obtained virtual identifier pair, calculating a transformation matrix again for the remaining unmatched virtual identifier pair together with the identifier pair in the matched ground coordinate system to obtain a second transformation matrix and a second error, and And a selection sub-step of selecting a pair of identifiers of the transformation matrix with small error from the first transformation matrix obtained in the rough matching sub-step, the first error and the second transformation matrix obtained in the fine matching sub-step as the matched pair of identifiers, and obtaining the vehicle position information in the fixed scene coordinate system as the latest vehicle position information based on the matched pair of identifiers.
  7. 7. The vehicle locating method for securing a parking scene as claimed in claim 1, characterized in that, The pose matching step comprises the following steps: (a) Acquiring a coordinate system transformation matrix from a ground coordinate system to a fixed scene coordinate system according to the vehicle position information, and (B) According to the coordinate system transformation matrix obtained in the step (a), calculating the corresponding position of the marker pair in the ground coordinate system in the fixed scene coordinate system, matching the corresponding position with the actual marker pair in the known fixed scene coordinate system, and taking the position of the marker pair in the matched fixed scene coordinate system as the latest vehicle position information if the matching error is smaller than a preset matching threshold value.
  8. 8. A vehicle locating system for securing a parking scene, comprising: The marker detection module is used for detecting markers and identifying marker pairs; The pose matching module is used for carrying out pose matching according to the vehicle position information and the marker pairs; a pose-free matching module for performing pose-free matching according to the marker pair to obtain the latest vehicle position information, and A position updating module for updating the current vehicle position information according to the latest vehicle position information obtained by the pose matching module or the pose-free matching module, As the pair of markers, two identical markers are used, Wherein the marker detection module comprises: the detection sub-module is used for obtaining the marker according to a detection result obtained by the marker in the image or a tracking result of the marker detected before; The traversing sub-module is used for traversing the markers acquired in the detection sub-module, searching whether other markers exist in a certain range, and acquiring a plurality of markers if the other markers exist; A judging sub-module for judging whether there is a paired identifier pair in the plurality of identifiers obtained by the traversing sub-module, if there is a paired identifier pair, storing the identifier information, and And the filtering sub-module is used for filtering the identifier information acquired by the judging sub-module to filter out repeated identifier pairs to obtain at least one pair of identifier pairs which are used as the identifier pairs in the ground coordinate system.
  9. 9. The vehicle locating system for securing a parking scene as recited in claim 8, wherein, In the judging submodule, judging whether two markers are paired marker pairs or not is realized by the following modes: And judging whether the center point distance and the direction included angle of the two markers meet the specified threshold value or not by calculating the center point distance and the direction included angle of the two markers, and judging that the two markers are paired markers if the center point distance and the direction included angle of the two markers meet the preset specified threshold value.
  10. 10. The vehicle locating system for securing a parking scene as recited in claim 9, wherein, And the filtering submodule filters out repeated marker pairs according to the center point distance and the direction included angle of the marker pairs to obtain the marker pairs in the ground coordinate system.
  11. 11. The vehicle locating system for securing a parking scene of claim 8, wherein the pose-free matching module comprises: A matching sub-module for matching the obtained marker pairs in the ground coordinate system with the actual marker pairs in the known fixed scene coordinate system in turn, and And the selecting sub-module is used for selecting the matched optimal identifier pair as the matched identifier pair, and obtaining the vehicle position information based on the matched identifier pair and taking the vehicle position information as the latest vehicle position information.
  12. 12. The vehicle locating system for securing a parking scene as recited in claim 11, characterized in that, The matching submodule performs coordinate system transformation matrix on the obtained marker pairs in the ground coordinate system by utilizing SVD decomposition to solve ICP algorithm, calculates the corresponding positions of the marker pairs in the ground coordinate system in the fixed scene coordinate system, obtains the distance difference between the corresponding positions and the actual marker pairs in the known fixed scene coordinate system, The selecting submodule selects the identifier pair with the optimal matching, which means that the identifier pair with the minimum distance difference is selected.
  13. 13. The vehicle locating system for securing a parking scene of claim 8, wherein the pose-free matching module comprises: the rough matching submodule is used for matching the obtained identifier pairs in the ground coordinate system with the actual identifier pairs in the known fixed scene coordinate system, and carrying out coordinate system transformation matrix by utilizing SVD decomposition to solve an ICP algorithm to obtain a first transformation matrix and a first error; A fine matching sub-module for performing inverse operation on the first transformation matrix obtained in the coarse matching sub-module to obtain virtual identifier pairs corresponding to the actual identifier pairs in the ground coordinate system in the fixed scene coordinate system, filtering the matched virtual identifier pairs for the obtained virtual identifier pairs, calculating the transformation matrix again for the remaining unmatched virtual identifier pairs together with the identifier pairs in the matched ground coordinate system to obtain a second transformation matrix and a second error, and And the selection submodule is used for selecting the identifier pair of the transformation matrix with small error from the first transformation matrix and the first error obtained from the rough matching submodule and the second transformation matrix and the second error obtained from the fine matching submodule as the matched identifier pair, and obtaining the vehicle position information in the fixed scene coordinate system based on the matched identifier pair as the latest vehicle position information.
  14. 14. The vehicle locating system for securing a parking scene as recited in claim 8, wherein, The pose matching module comprises: a transformation submodule for obtaining a coordinate system transformation matrix from a ground coordinate system to a fixed scene coordinate system according to the vehicle position information, and And the matching sub-module is used for calculating 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 acquired by the transformation sub-module, matching the corresponding position with the actual marker pair in the known fixed scene coordinate system, and taking the position of the marker pair in the matched fixed scene coordinate system as the latest vehicle position information if the matching error is smaller than a preset matching threshold value.
  15. 15. A computer readable medium having a computer program stored thereon, characterized in that, The computer program, when executed by a processor, implements the vehicle positioning method for fixing a parking scene according to any one of claims 1 to 7.
  16. 16. A computer device comprising a memory module, a processor and a computer program stored on the memory module and executable on the processor, characterized in that the processor implements the vehicle positioning method for stationary parking scenes according to any of claims 1-7 when executing the computer program.
  17. 17. A vehicle comprising a vehicle positioning system for securing a parking scene according to any one of claims 8 to 14.

Description

Vehicle positioning method and system for fixing parking scene Technical Field The present invention relates to a vehicle control technology, and in particular, to a vehicle positioning method for fixing a parking scene and a vehicle positioning system for fixing a parking scene. Background Automatic parking refers to the automatic parking of a vehicle into a parking space without manual control. Different automated parking systems typically employ different methods to detect objects around the vehicle. Image matching (IMAGE MATCHING) aims at pixel-level identification and alignment of content or structure in two images that have the same/similar properties, in general, the images to be matched are usually taken from the same or similar scene or object, or other types of image pairs that have the same shape or semantic information. In order to improve the robustness of the matching algorithm, reduce the influence of other factors such as noise, distortion and the like on the matching performance and reduce the computational complexity, a feature-based image matching method is widely studied. The image matching method based on the features in the prior art mainly comprises the steps of extracting significant feature structures with physical significance from images, wherein the significant feature structures comprise feature points, feature lines and morphological areas with significance, and matching and predicting transformation functions on the extracted feature structures and aligning other image contents. In the field of automatic parking, the rapid positioning of vehicle position information has important significance for path planning and safe and intelligent driving of vehicles. The image matching is a bridge for communicating the vehicle scene information and the actual map information, and for the purpose of real-time performance and accuracy of the image matching, obtaining three-dimensional point cloud information in the vehicle scene and the actual map information for three-dimensional matching becomes one of effective means for solving the positioning problem. Common point cloud matching algorithms include RANSAC algorithm, ICP algorithm. The RANSAC method can robustly estimate matching model parameters, but has the disadvantage that the number of iterations of calculating model parameters may be large, time consuming, and additional threshold parameters are required. The ICP algorithm is a registration method for solving the radiation transformation matrix by using a point set-to-point set, and can achieve an accurate matching effect, but the calculation amount is large when searching the corresponding point. Disclosure of Invention In view of the above, the present invention aims to propose a vehicle positioning method for a stationary parking scenario (e.g., a parking garage parking scenario, a battery-exchange parking scenario, etc.) and a vehicle positioning system for a stationary parking scenario, which can simplify algorithm complexity and can provide a precise and robust positioning requirement for a vehicle in real time. A vehicle positioning method for fixing a parking scene according to an aspect of the present invention is characterized by comprising: A marker detection step of detecting a marker and identifying a marker pair, if the vehicle position information in the fixed parking scene is obtained, continuing the following pose matching step, and if the vehicle position information in the fixed parking scene is not obtained, jumping to the following pose-free matching step; A pose matching step, namely carrying out pose matching according to the obtained vehicle position information and the identified marker pair, if the latest vehicle position information is obtained, jumping to a position updating step, otherwise, continuing the pose matching step; a pose-free matching step of performing pose-free matching based on the identified marker pair to obtain the latest vehicle position information, and A position updating step of updating current vehicle position information according to the latest vehicle position information obtained in the pose matching step or the pose-free matching step, Wherein the vehicle position information in the marker detection step refers to vehicle position information that has been obtained by performing the vehicle positioning method at a previous timing. Alternatively, as the pair of markers, two identical markers are used. Optionally, the marker detection step includes: step 1, acquiring a marker according to a detection result obtained by the marker in the image or a tracking result of the marker in the image detected before; step 2, traversing the markers obtained in the sub-step 1, searching whether other markers exist in a certain range, and obtaining a plurality of markers if the other markers exist; a substep 3 of judging whether a pair of markers exists among the acquired plurality of markers, storing the marker information if the condition o