Long corridor environment recognition method, device, equipment and storage medium

Abstract

The present application provides a method, device, equipment and storage medium for identifying a long corridor environment. The recognition system first collects laser data and calculates the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the current frame laser data. Then, the covariance matrix corresponding to each laser point is constructed according to the coordinate mean of each neighborhood, and the eigenvalues ​​of each covariance matrix are solved, and the eigenvector corresponding to the minimum eigenvalue is selected from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each neighborhood. The recognition system obtains the first number of inner points, the second number of inner points and the third number of inner points according to the normal vectors after dimensionality reduction, and judges whether the first number of inner points, the second number of inner points and the third number of inner points meet the corresponding threshold conditions respectively. If the first number of inner points, the second number of inner points and the third number of inner points all meet the corresponding threshold conditions respectively, the current scene is determined to be a long corridor environment.

Description

Technical Field

[0001] The present application relates to the field of laser mapping technology, and in particular to a method, device, equipment, and storage medium for identifying a long corridor environment. Background Art

[0002] Long corridor positioning and mapping present a challenge for existing laser SLAM systems. In actual testing, technicians discovered that robots often misposition themselves in long corridors. This is because when using LiDAR for mapping in long corridors, the robot's scores for each pose remain essentially the same during motion. This prevents the robot from accurately determining its position, significantly impacting both normal positioning and mapping, as well as subsequent planning and control. Therefore, accurately identifying the robot's environment during motion is a pressing technical challenge. Summary of the Invention

[0003] The main purpose of this application is to provide a method, device, equipment and storage medium for identifying a long corridor environment, aiming to solve the problem that existing robots cannot accurately identify the type of environment they are in.

[0004] To achieve the above objectives, this application provides a method for identifying a long corridor environment, comprising:

[0005] Collecting laser data and calculating the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame;

[0006] Constructing covariance matrices corresponding to the respective laser points according to the coordinate means of the respective neighborhoods;

[0007] Solving the eigenvalues ​​of each covariance matrix, and selecting the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each neighborhood;

[0008] Performing a dimensionality reduction operation on each of the normal vectors, and performing step-by-step analysis based on each of the normal vectors after the dimensionality reduction to obtain a first inlier number, a second inlier number, and a third inlier number, and determining in turn whether the first inlier number, the second inlier number, and the third inlier number meet their respective corresponding threshold conditions;

[0009] If the first number of inner points, the second number of inner points, and the third number of inner points all meet their corresponding threshold conditions, it is determined that the current scene is a long corridor environment.

[0010] The present application also provides a device for identifying a long corridor environment, comprising:

[0011] An acquisition module, configured to acquire laser data and calculate the coordinate mean of all first laser points in a neighborhood corresponding to each laser point in the laser data of a current frame;

[0012] A construction module, configured to construct the covariance matrices corresponding to the respective laser points according to the coordinate means of the respective neighborhoods;

[0013] A first solving module is used to solve the eigenvalues ​​of each covariance matrix, and select the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each neighborhood;

[0014] a judgment module, configured to perform a dimensionality reduction operation on each of the normal vectors, and obtain a first number of inliers, a second number of inliers, and a third number of inliers according to each of the normal vectors after the dimensionality reduction, and determine in sequence whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions;

[0015] The first determination module is configured to determine that the current scene is a long corridor environment if the first number of inner points, the second number of inner points, and the third number of inner points all meet their corresponding threshold conditions.

[0016] The present application also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any of the above methods when executing the computer program.

[0017] The present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0018] The present application provides a method, apparatus, device, and storage medium for identifying a long corridor environment. The identification system first collects laser data and calculates the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the current frame laser data. Then, based on the coordinate mean of each neighborhood, the covariance matrix corresponding to each laser point is constructed respectively, and the eigenvalues ​​of each covariance matrix are solved. The eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix is ​​selected as the normal vector corresponding to each neighborhood. The identification system performs a dimensionality reduction operation on each normal vector, and analyzes each normal vector after dimensionality reduction step by step to obtain the first, second, and third inner point numbers, and sequentially determines whether the first, second, and third inner point numbers meet their respective corresponding threshold conditions. If the first, second, and third inner point numbers all meet their respective corresponding threshold conditions, the current scene is determined to be a long corridor environment. In this application, the recognition system solves and obtains the normal vectors corresponding to the neighborhood of each frame of laser data. When the robot is in a long corridor environment, the normal vectors of the neighborhood of most laser points in the acquired frame of laser data should point in a direction perpendicular to the long corridor. Therefore, when the first number of inliers, the second number of inliers, and the third number of inliers obtained by step-by-step analysis of each normal vector all meet their corresponding threshold conditions, it means that the normal vectors of the neighborhood of most laser points in the current frame of laser data point in a direction perpendicular to the long corridor, and thus the robot is determined to be in a long corridor environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a schematic diagram of the steps of a method for identifying a long corridor environment in one embodiment of the present application;

[0020] Figure 2 This is a block diagram of the overall structure of a device for identifying a long corridor environment in one embodiment of the present application;

[0021] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present application.

[0022] The realization of the objectives, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0024] Reference Figure 1 In one embodiment of the present application, a method for identifying a long corridor environment is provided, comprising:

[0025] S1: Collect laser data and calculate the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame;

[0026] S2: constructing the covariance matrices corresponding to the respective laser points according to the coordinate means of the respective neighborhoods;

[0027] S3: solving the eigenvalues ​​of each covariance matrix, and selecting the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each of the neighborhoods;

[0028] S4: performing a dimensionality reduction operation on each of the normal vectors, and performing step-by-step analysis based on each of the normal vectors after the dimensionality reduction to obtain a first number of inliers, a second number of inliers, and a third number of inliers, and sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions;

[0029] S5: If the first number of inner points, the second number of inner points, and the third number of inner points all meet their corresponding threshold conditions, it is determined that the current scene is a long corridor environment.

[0030] In this embodiment, the recognition method is described by taking its application to a sweeping robot as an example. The sweeping robot is equipped with a laser radar, and the recognition system of the sweeping robot obtains laser data through the scanning collection of the laser radar. One frame of laser data contains multiple laser points. Taking 1 degree as the quantized distance as an example, the laser radar scans one circle (i.e., scans 360 degrees) to obtain a frame of laser data, and the frame of laser data contains 360 laser points. The recognition system takes each laser point as the middle position, selects a preset number of second laser points on both sides of the laser point, and the preset number of second laser points and the laser point constitute the neighborhood of the laser point; thus, the recognition system obtains the neighborhood corresponding to each laser point in the current frame of laser data. Then, the recognition system performs an average calculation based on the coordinate values ​​of all the first laser points contained in each neighborhood, thereby obtaining the coordinate mean corresponding to each neighborhood. The recognition system constructs the covariance matrix corresponding to each laser point based on the coordinate mean corresponding to each neighborhood (it can also be considered as the covariance matrix corresponding to each neighborhood), and then solves each covariance matrix to obtain multiple eigenvalues ​​corresponding to each covariance matrix (that is, a single covariance matrix is ​​solved to obtain multiple eigenvalues). The recognition system selects a minimum eigenvalue from the eigenvalues ​​of each covariance matrix, and uses the eigenvectors corresponding to each minimum eigenvalue as the normal vector corresponding to each neighborhood (that is, a minimum eigenvalue is selected from the multiple eigenvalues ​​of a single covariance matrix, and the eigenvector corresponding to the minimum eigenvalue is the normal vector of the neighborhood corresponding to the covariance matrix). The recognition system maps two normal vectors with the same direction to the same value to reduce the dimension of each normal vector to several first normal vectors (for example, two normal vectors corresponding to -90 degrees and 90 degrees with the same direction are mapped to the same value, thereby reducing the number of normal vectors and achieving the purpose of dimensionality reduction). The recognition system votes for each target angle within a preset range adjacent to each first normal vector, obtaining a corresponding number of votes for each target angle. The system then selects the angle with the largest number of votes from each target angle as the first angle, and uses the number of votes for the first angle as the first inlier number. The recognition system then eliminates angles within a first angular range adjacent to the first angle from each target angle to obtain a number of second inlier numbers. The system then selects the angle with the largest number of votes from each second inlier number as the second angle, and uses the number of votes for the second angle as the second inlier number. Finally, the recognition system eliminates second angles within a second angular range adjacent to the second angle from each second inlier number to obtain a number of third inlier numbers. The system then selects the angle with the largest number of votes from each third inlier number as the third angle, and uses the number of votes for the third angle as the third inlier number. This achieves a step-by-step analysis based on the normal vectors of each neighborhood to obtain the first, second, and third inlier numbers.The recognition system sequentially determines whether the first, second, and third inlier point counts meet their respective threshold conditions. Specifically, the recognition system calculates a first inlier point ratio based on the first inlier point count and the total number of laser points in the current frame of laser data, and determines whether the first inlier point ratio is greater than a first threshold. If the first inlier point ratio is greater than the first threshold, the recognition system calculates a second inlier point ratio based on the second inlier point count and the total number of laser points in the current frame of laser data, and determines whether the second inlier point ratio is less than a second threshold. If the second inlier point ratio is less than the second threshold, the recognition system calculates a third inlier point ratio based on the third inlier point count and the total number of laser points in the current frame of laser data, and determines whether the third inlier point ratio is less than a third threshold. If the third inlier point ratio is less than the third threshold, the first, second, and third inlier point counts all meet their respective threshold conditions. This indicates that the normal vectors of the neighborhoods of most laser points in the current frame of laser data point perpendicularly to a long corridor, and the recognition system then determines that the current scene of the sweeping robot is a long corridor environment.

[0031] In this embodiment, the recognition system solves and obtains the normal vectors corresponding to the neighborhood of each frame of laser data. When the robot is in a long corridor environment, the normal vectors of the neighborhoods of most laser points in a frame of laser data should point in a direction perpendicular to the long corridor. Therefore, when the first number of inliers, the second number of inliers, and the third number of inliers obtained by step-by-step analysis of each normal vector after dimensionality reduction all meet their corresponding threshold conditions, it means that the normal vectors of the neighborhoods of most laser points in the current frame of laser data point in a direction perpendicular to the long corridor, thereby determining that the current scene in which the robot is located is a long corridor environment. This recognition method has high accuracy. After recognizing that the robot is in a long corridor environment, the data types based on which the robot is mapped, positioned, and other subsequent planning and control can be adjusted, thereby improving the accuracy and stability of mapping, positioning, and other subsequent planning and control.

[0032] Furthermore, the step of performing a dimensionality reduction operation on each of the normal vectors and obtaining a first number of inner points, a second number of inner points, and a third number of inner points by step-by-step analysis based on each of the normal vectors after dimensionality reduction includes:

[0033] S401: mapping the two normal vectors having the same direction to the same value, so as to reduce the dimension of each normal vector into a plurality of first normal vectors;

[0034] S402: Voting for each target angle within a preset range adjacent to each of the first normal vectors to obtain a number of votes corresponding to each target angle;

[0035] S403: Selecting a first candidate angle with the largest number of votes from the target angles as the first angle, and using the number of votes for the first angle as the first inlier number;

[0036] S404: Eliminate angles in a first angle range adjacent to the first angle from the target angles to obtain a plurality of second selected angles;

[0037] S405: Selecting a second angle to be selected with the largest number of votes from the second angles to be selected as the second angle, and using the number of votes received by the second angle as the second inner point number;

[0038] S406: Eliminate the second angles in the second angle range adjacent to the second angle from the second angles to obtain a plurality of third angles to be selected;

[0039] S407: Filter out the third candidate angle with the largest number of votes from each of the third candidate angles as the third angle, and use the number of votes for the third angle as the third inner point number.

[0040] In this embodiment, the recognition system maps two normal vectors with the same direction to the same value, thereby reducing the total number of normal vectors and reducing their dimensionality to a number of first normal vectors. For example, when the laser radar scans with an interval of 1 degree, a frame of laser data contains 360 laser points, corresponding to 360 neighborhoods, so the number of normal vectors corresponding to a frame of laser data is 360. Two normal vectors with the same direction, such as the two normal vectors corresponding to -90 degrees and 90 degrees, are mapped to the same value in the same direction, which is the first normal vector; thus, the 360 ​​normal vectors can be reduced to 180 first normal vectors, and the amount of data is greatly reduced, which can improve the subsequent data processing speed.

[0041] The recognition system votes for each target angle within a preset range adjacent to each first normal vector, determining the number of votes for each target angle. For example, the first normal vector corresponding to 90 degrees is used. The preset range is plus or minus 8 degrees (the angle value within the preset range can be selected based on actual needs and is not specifically limited here). With 90 degrees as the center, the number of votes for all target angles within the angle range of [82, 98] is increased by 1. This process continues in this manner, resulting in the number of votes for each target angle value.

[0042] The recognition system first selects the first candidate angle with the most votes from each target angle and uses the number of votes for this first angle as the first angle. The system then removes angles within the first angle range adjacent to the first angle from each target angle to obtain several second candidate angles. For example, the first angle range is ±15 degrees (θ = th * 2 - 1°, which can be directly expressed as ±θ, where θ represents the first angle range and th represents the angle within the preset range. Both are values ​​set in this embodiment, but in practice, θ does not need to be associated with th. In theory, as long as the angles closer to the sides receive fewer votes (designing two gradients, each toward the sides), the first angle is 90 degrees. This means that the angles within the range [75, 105] are removed from the 180 angles, and the remaining angles are selected as the second candidate angles. The recognition system selects the second candidate angle with the most votes from each second candidate angle and uses the number of votes for this second angle as the second angle. The system uses the number of votes for this second angle as the second inlier number. Finally, the recognition system removes the second candidate angles within the second angle range adjacent to the second angle from each second candidate angle (the logic for this removal is the same as that for obtaining the second candidate angle and is not further described here). The remaining second candidate angle is the third candidate angle, wherein the second angle range is preferably the same as the first angle range. Furthermore, the second angle range and the first angle range can also be different. The specific setting can be based on needs and is not limited here. The recognition system selects the third candidate angle with the most votes from each third candidate angle as the third angle, and uses the number of votes for the third angle as the third inner point number.

[0043] Furthermore, the step of mapping the two normal vectors having the same direction to the same value to reduce the dimension of each normal vector to a plurality of first normal vectors includes:

[0044] S40101: Substitute each of the normal vectors into the calculation formula for mapping processing to map the two normal vectors with the same direction into one value, wherein the calculation formula is: X2 = (X1 + 90) % 180, X1 represents the normal vector, and X2 represents the value of the normal vector after mapping.

[0045] In this embodiment, the recognition system performs a mapping process by substituting each normal vector into the formula X2 = (X1 + 90) % 180. After the mapping process, two normal vectors with the same direction are mapped to the same value; where X1 represents the normal vector and X2 represents the mapped value of the normal vector. This embodiment significantly reduces the amount of normal vector data through the mapping process, thereby achieving dimensionality reduction, effectively increasing the speed of subsequent data processing, and reducing processing complexity.

[0046] Furthermore, the step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions includes:

[0047] S408: Calculate a first inlier ratio based on the first number of inliers and the total number of laser points included in the laser data of the current frame, and determine whether the first inlier ratio is greater than a first threshold;

[0048] S409: If the first inlier ratio is greater than the first threshold, calculating a second inlier ratio based on the second number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the second inlier ratio is less than a second threshold;

[0049] S4010: If the second inlier ratio is less than the second threshold, calculating a third inlier ratio based on the third number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the third inlier ratio is less than a third threshold;

[0050] S4011: If the third inlier ratio is less than the third threshold, determining that the first inlier number, the second inlier number, and the third inlier number all meet their corresponding threshold conditions;

[0051] S4012: Otherwise, determine that the first inlier number, the second inlier number, and the third inlier number do not meet their corresponding threshold conditions.

[0052] In this embodiment, the recognition system calculates the ratio of the first inlier number to the total number of laser points in the current frame of laser data (i.e., divides the first inlier number by the total number of laser points) to obtain a first inlier ratio. The system then uses a first threshold to determine whether the first inlier ratio is greater than the first threshold. If the first inlier ratio is greater than the first threshold, further analysis is performed. Specifically, the recognition system calculates the ratio of the second inlier number to the total number of laser points in the current frame of laser data to obtain a second inlier ratio. The system then uses a second threshold to determine whether the second inlier ratio is less than the second threshold. If the second inlier ratio is less than the second threshold, the threshold condition corresponding to the third inlier number is identified. Specifically, the recognition system calculates the ratio of the third inlier number to the total number of laser points in the current frame of laser data to obtain a third inlier ratio. The system then uses a third threshold to determine whether the third inlier ratio is less than the third threshold. If the third inlier ratio is also less than the third threshold, the first, second, and third inlier numbers all meet their respective threshold conditions.

[0053] In this embodiment, the specific values ​​of the first threshold, the second threshold, and the third threshold are empirical data and need to be adjusted according to actual conditions. No specific restrictions are made here. In this embodiment, in a step-by-step manner, the first number of inner points, the second number of inner points, and the third number of inner points are judged step by step to see whether they meet their corresponding threshold conditions. Only when all three threshold conditions are met can the current scene be judged as a long corridor environment. If one of the threshold conditions is not met, for example, the ratio of the second inner points is greater than the second threshold (that is, the second number of inner points does not meet its corresponding threshold condition), there is no need to judge the threshold condition corresponding to the third number of inner points, and the current scene can be directly judged as not a long corridor environment, thereby reducing the amount of data processing and improving the processing speed.

[0054] Furthermore, in the step of calculating the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame, the step of calculating the coordinate mean corresponding to a single neighborhood includes:

[0055] S101: Taking the laser point as the center position, selecting a preset number of second laser points on both sides of the laser point to form a neighborhood corresponding to the laser point;

[0056] S102: Obtain a first laser coordinate value of the laser point and second laser coordinate values ​​corresponding to each of the second laser points, and add the coordinate values ​​corresponding to the same coordinate axis to obtain a total laser coordinate value;

[0057] S103: performing an average calculation based on the total number of the first laser points included in the neighborhood and the sum of the laser coordinate values ​​to obtain the coordinate mean.

[0058] In this embodiment, the processing logic for selecting the neighborhood corresponding to each laser point in the current frame laser data and the coordinate mean corresponding to the neighborhood is the same. The coordinate mean corresponding to a single neighborhood is used as an example for explanation. When scanning along a circle, the laser radar scans. The recognition system uses the currently selected laser point as the center position and selects a preset number of second laser points on both sides of the laser point (for example, 3 second laser points before and after the laser point) to form the neighborhood corresponding to the laser point (that is, the neighborhood of the laser point includes 6 second laser points and the laser point itself). The recognition system obtains the first laser coordinate value of the laser point itself and the second laser coordinate values ​​corresponding to each second laser point, and then adds them together to obtain the total laser coordinate value (the total laser coordinate value includes the Y value sum and the X value sum, that is, the horizontal coordinate value of the first laser coordinate value is added to the horizontal coordinate value of each second laser coordinate value to obtain the X value sum of the total laser coordinate value; the vertical coordinate value of the first laser coordinate value is added to the vertical coordinate value of each second laser coordinate value to obtain the Y value sum of the total laser coordinate value). The recognition system calculates the average of the total number of all first laser points contained in the neighborhood (that is, the number of laser points plus the number of second laser points. In the above example, the total number of first laser points is 7) and the sum of the laser point coordinates to obtain the coordinate mean corresponding to the neighborhood.

[0059] Furthermore, the identification method is applied to a sweeping robot equipped with an odometer. After the step of determining that the current scene is a long corridor environment, the method includes:

[0060] S6: Retrieve odometer data;

[0061] S7: Positioning the sweeping robot according to the odometer data.

[0062] In this embodiment, the robot vacuum is equipped with an odometer. When the recognition system determines that the current scene is a long corridor, the reliability of the laser data collected by the laser scanner is significantly reduced. To ensure the reliability of the robot vacuum's subsequent mapping, positioning, and planning control, the odometer data collected by the odometer is used as the basis. Specifically, the recognition system retrieves the odometer data collected by the odometer and uses it to locate the robot vacuum, build maps, and perform planning control.

[0063] Furthermore, after the step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions, the method further includes:

[0064] S8: If the first number of inner points and / or the second number of inner points and / or the third number of inner points do not meet their respective corresponding threshold conditions, then it is determined that the current scene is a non-long corridor environment;

[0065] S9: Positioning the sweeping robot according to the laser data.

[0066] In this embodiment, if any of the first, second, or third inliers do not meet their corresponding thresholds, the recognition system determines that the current scene is not a long corridor environment. Under this premise, the laser data collected by laser scanning is more reliable than odometry data. Therefore, to ensure the reliability of the robot vacuum's subsequent mapping, positioning, and planning control, the recognition system uses the laser data collected by laser scanning as the basis for positioning, mapping, and planning control of the robot vacuum, thereby ensuring the accuracy and reliability of positioning, mapping, and planning control.

[0067] Reference Figure 2 In one embodiment of the present application, a device for identifying a long corridor environment is provided, comprising:

[0068] Acquisition module 1, for collecting laser data and calculating the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame;

[0069] Construction module 2, for constructing covariance matrices corresponding to the respective laser points according to the coordinate mean of each of the neighborhoods;

[0070] A first solving module 3 is used to solve the eigenvalues ​​of each covariance matrix, and select the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each neighborhood;

[0071] a judgment module 4, configured to perform a dimensionality reduction operation on each of the normal vectors, and obtain a first number of inliers, a second number of inliers, and a third number of inliers according to each of the normal vectors after the dimensionality reduction, and determine in turn whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions;

[0072] The first determination module 5 is configured to determine that the current scene is a long corridor environment if the first number of inner points, the second number of inner points, and the third number of inner points all meet their corresponding threshold conditions.

[0073] Furthermore, the judgment module 4 includes:

[0074] a mapping unit, configured to map the two normal vectors having the same direction to the same value, so as to reduce the dimension of each normal vector into a plurality of first normal vectors;

[0075] A voting unit is configured to vote for target angles within a preset range adjacent to each of the first normal vectors, respectively, to obtain a number of votes corresponding to each of the target angles;

[0076] a first screening unit, configured to screen out a first candidate angle with the largest number of votes from the target angles as the first angle, and use the number of votes for the first angle as the first inlier number;

[0077] a first elimination unit, configured to eliminate angles in a first angle range adjacent to the first angle from the target angles, to obtain a plurality of second selected angles;

[0078] a second screening unit, configured to screen out a second angle to be selected with the largest number of votes from the second angles to be selected as the second angle, and use the number of votes received by the second angle as the second inlier number;

[0079] a second elimination unit, configured to eliminate angles in a second angle range adjacent to the second angle from each of the second angles to be selected, to obtain a plurality of third angles to be selected;

[0080] The third screening unit is used to screen out a third candidate angle with the largest number of votes from each of the third candidate angles as the third angle, and use the number of votes of the third angle as the third inner point number.

[0081] Furthermore, the mapping unit includes:

[0082] The processing subunit is used to substitute each of the normal vectors into the calculation formula for mapping processing so as to map the two normal vectors with the same direction into one value, wherein the calculation formula is: X2=(X1+90)%180, X1 represents the normal vector, and X2 represents the value after the normal vector is mapped.

[0083] Furthermore, the judgment module 4 further includes:

[0084] a first calculation unit, configured to calculate a first inlier ratio based on the first number of inliers and a total number of laser points included in the laser data of a current frame, and determine whether the first inlier ratio is greater than a first threshold;

[0085] a second calculating unit, configured to calculate a second inlier ratio based on the second number of inliers and a total number of laser points included in the laser data of a current frame, if the first inlier ratio is greater than the first threshold, and determine whether the second inlier ratio is less than a second threshold;

[0086] a third calculating unit, configured to calculate a third inlier ratio based on the third number of inliers and a total number of laser points included in the laser data of a current frame, if the second inlier ratio is less than the second threshold, and determine whether the third inlier ratio is less than a third threshold;

[0087] a first determining unit, configured to determine that the first number of inliers, the second number of inliers, and the third number of inliers all meet their corresponding threshold conditions if the third inlier ratio is less than the third threshold;

[0088] The second determination unit is configured to determine that the first number of inliers, the second number of inliers, and the third number of inliers do not meet their respective corresponding threshold conditions if the first inlier ratio is smaller than the first threshold, if the second inlier ratio is larger than the second threshold, and if the third inlier ratio is larger than the third threshold.

[0089] Furthermore, the acquisition module 1 includes:

[0090] A selection unit is configured to take the laser point as the middle position and select a preset number of second laser points on both sides of the laser point to form a neighborhood corresponding to the laser point;

[0091] a fourth calculation unit, configured to obtain the first laser coordinate value of the laser point and the second laser coordinate values ​​corresponding to each of the second laser points, and to calculate the sum of the laser coordinate values ​​by adding the coordinate values ​​corresponding to the same coordinate axis;

[0092] A fifth calculation unit is configured to perform an averaging calculation based on the total number of the first laser points included in the neighborhood and the sum of the laser coordinate values ​​to obtain the coordinate mean.

[0093] Furthermore, the identification method is applied to a sweeping robot, the sweeping robot is equipped with an odometer, and the identification device further includes:

[0094] Retrieval module 6, used to retrieve odometer data;

[0095] The first positioning module 7 is used to position the sweeping robot according to the odometer data.

[0096] Furthermore, the identification device further includes:

[0097] A second determination module 8 is configured to determine that the current scene is a non-long corridor environment if the first number of inner points and / or the second number of inner points and / or the third number of inner points do not meet their respective corresponding threshold conditions;

[0098] The second positioning module 9 is used to position the sweeping robot according to the laser data.

[0099] This embodiment provides a device for identifying a long corridor environment. The recognition system first collects laser data and calculates the coordinate mean of all first laser points within the neighborhood corresponding to each laser point in the current frame of laser data. Then, based on the coordinate mean of each neighborhood, it constructs a covariance matrix corresponding to each laser point. The eigenvalues ​​of each covariance matrix are then calculated, and the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix is ​​selected as the normal vector corresponding to each neighborhood. The recognition system then performs a dimensionality reduction operation on each normal vector and, based on the reduced dimensionality, sequentially analyzes each normal vector to obtain the first, second, and third inliers. The recognition system then determines whether the first, second, and third inliers meet their respective threshold conditions. If the first, second, and third inliers all meet their respective threshold conditions, the current scene is determined to be a long corridor environment. In this application, the recognition system calculates the normal vectors corresponding to the neighborhood of each frame of laser data. When the robot is in a long corridor environment, the normal vectors of the neighborhoods of most laser points in a frame of laser data should point perpendicular to the long corridor. Therefore, when the first number of inliers, the second number of inliers, and the third number of inliers obtained by analyzing each normal vector step by step meet their corresponding threshold conditions, it means that the normal vectors of the neighborhood of most laser points in the current frame laser data point to the direction perpendicular to the long corridor, thereby determining that the current scene in which the robot is located is a long corridor environment.

[0100] Reference Figure 3 In the embodiment of the present application, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as laser data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for identifying a long corridor environment is implemented.

[0101] The processor executes the following steps of the method for identifying a long corridor environment:

[0102] S1: Collect laser data and calculate the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame;

[0103] S2: constructing the covariance matrices corresponding to the respective laser points according to the coordinate means of the respective neighborhoods;

[0104] S3: solving the eigenvalues ​​of each covariance matrix, and selecting the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each of the neighborhoods;

[0105] S4: performing a dimensionality reduction operation on each of the normal vectors, and performing step-by-step analysis based on each of the normal vectors after the dimensionality reduction to obtain a first number of inliers, a second number of inliers, and a third number of inliers, and sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions;

[0106] S5: If the first number of inner points, the second number of inner points, and the third number of inner points all meet their corresponding threshold conditions, it is determined that the current scene is a long corridor environment.

[0107] Furthermore, the step of performing a dimensionality reduction operation on each of the normal vectors and obtaining a first number of inner points, a second number of inner points, and a third number of inner points by step-by-step analysis based on each of the normal vectors after dimensionality reduction includes:

[0108] S401: mapping the two normal vectors having the same direction to the same value, so as to reduce the dimension of each normal vector into a plurality of first normal vectors;

[0109] S402: Voting for each target angle within a preset range adjacent to each of the first normal vectors to obtain a number of votes corresponding to each target angle;

[0110] S403: Selecting a first candidate angle with the largest number of votes from the target angles as the first angle, and using the number of votes for the first angle as the first inlier number;

[0111] S404: Eliminate angles in a first angle range adjacent to the first angle from the target angles to obtain a plurality of second selected angles;

[0112] S405: Selecting a second angle to be selected with the largest number of votes from the second angles to be selected as the second angle, and using the number of votes received by the second angle as the second inner point number;

[0113] S406: Eliminate the second angles in the second angle range adjacent to the second angle from the second angles to obtain a plurality of third angles to be selected;

[0114] S407: Filter out the third candidate angle with the largest number of votes from each of the third candidate angles as the third angle, and use the number of votes for the third angle as the third inner point number.

[0115] Furthermore, the step of mapping the two normal vectors having the same direction to the same value to reduce the dimension of each normal vector to a plurality of first normal vectors includes:

[0116] S40101: Substitute each of the normal vectors into the calculation formula for mapping processing to map the two normal vectors with the same direction into one value, wherein the calculation formula is: X2 = (X1 + 90) % 180, X1 represents the normal vector, and X2 represents the value of the normal vector after mapping.

[0117] Furthermore, the step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions includes:

[0118] S408: Calculate a first inlier ratio based on the first number of inliers and the total number of laser points included in the laser data of the current frame, and determine whether the first inlier ratio is greater than a first threshold;

[0119] S409: If the first inlier ratio is greater than the first threshold, calculating a second inlier ratio based on the second number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the second inlier ratio is less than a second threshold;

[0120] S4010: If the second inlier ratio is less than the second threshold, calculating a third inlier ratio based on the third number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the third inlier ratio is less than a third threshold;

[0121] S4011: If the third inlier ratio is less than the third threshold, determining that the first inlier number, the second inlier number, and the third inlier number all meet their corresponding threshold conditions;

[0122] S4012: Otherwise, determine that the first inlier number, the second inlier number, and the third inlier number do not meet their corresponding threshold conditions.

[0123] Furthermore, in the step of calculating the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame, the step of calculating the coordinate mean corresponding to a single neighborhood includes:

[0124] S101: Taking the laser point as the center position, selecting a preset number of second laser points on both sides of the laser point to form a neighborhood corresponding to the laser point;

[0125] S102: Obtain a first laser coordinate value of the laser point and second laser coordinate values ​​corresponding to each of the second laser points, and add the coordinate values ​​corresponding to the same coordinate axis to obtain a total laser coordinate value;

[0126] S103: performing an average calculation based on the total number of the first laser points included in the neighborhood and the sum of the laser coordinate values ​​to obtain the coordinate mean.

[0127] Furthermore, the identification method is applied to a sweeping robot equipped with an odometer. After the step of determining that the current scene is a long corridor environment, the method includes:

[0128] S6: Retrieve odometer data;

[0129] S7: Positioning the sweeping robot according to the odometer data.

[0130] Furthermore, after the step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions, the method further includes:

[0131] S8: If the first number of inner points and / or the second number of inner points and / or the third number of inner points do not meet their respective corresponding threshold conditions, then it is determined that the current scene is a non-long corridor environment;

[0132] S9: Positioning the sweeping robot according to the laser data.

[0133] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, a method for identifying a long corridor environment is implemented. The method for identifying a long corridor environment is specifically as follows:

[0134] S1: Collect laser data and calculate the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame;

[0135] S2: constructing the covariance matrices corresponding to the respective laser points according to the coordinate means of the respective neighborhoods;

[0136] S3: solving the eigenvalues ​​of each covariance matrix, and selecting the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each of the neighborhoods;

[0137] S4: performing a dimensionality reduction operation on each of the normal vectors, and performing step-by-step analysis based on each of the normal vectors after the dimensionality reduction to obtain a first number of inliers, a second number of inliers, and a third number of inliers, and sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions;

[0138] S5: If the first number of inner points, the second number of inner points, and the third number of inner points all meet their corresponding threshold conditions, it is determined that the current scene is a long corridor environment.

[0139] Furthermore, the step of performing a dimensionality reduction operation on each of the normal vectors and obtaining a first number of inner points, a second number of inner points, and a third number of inner points by step-by-step analysis based on each of the normal vectors after dimensionality reduction includes:

[0140] S401: mapping the two normal vectors having the same direction to the same value, so as to reduce the dimension of each normal vector into a plurality of first normal vectors;

[0141] S402: Voting for each target angle within a preset range adjacent to each of the first normal vectors to obtain a number of votes corresponding to each target angle;

[0142] S403: Selecting a first candidate angle with the largest number of votes from the target angles as the first angle, and using the number of votes for the first angle as the first inlier number;

[0143] S404: Eliminate angles in a first angle range adjacent to the first angle from the target angles to obtain a plurality of second selected angles;

[0144] S405: Selecting a second angle to be selected with the largest number of votes from the second angles to be selected as the second angle, and using the number of votes received by the second angle as the second inner point number;

[0145] S406: Eliminate the second angles in the second angle range adjacent to the second angle from the second angles to obtain a plurality of third angles to be selected;

[0146] S407: Filter out the third candidate angle with the largest number of votes from each of the third candidate angles as the third angle, and use the number of votes for the third angle as the third inner point number.

[0147] Furthermore, the step of mapping the two normal vectors having the same direction to the same value to reduce the dimension of each normal vector to a plurality of first normal vectors includes:

[0148] S40101: Substitute each of the normal vectors into the calculation formula for mapping processing to map the two normal vectors with the same direction into one value, wherein the calculation formula is: X2 = (X1 + 90) % 180, X1 represents the normal vector, and X2 represents the value of the normal vector after mapping.

[0149] Furthermore, the step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions includes:

[0150] S408: Calculate a first inlier ratio based on the first number of inliers and the total number of laser points included in the laser data of the current frame, and determine whether the first inlier ratio is greater than a first threshold;

[0151] S409: If the first inlier ratio is greater than the first threshold, calculating a second inlier ratio based on the second number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the second inlier ratio is less than a second threshold;

[0152] S4010: If the second inlier ratio is less than the second threshold, calculating a third inlier ratio based on the third number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the third inlier ratio is less than a third threshold;

[0153] S4011: If the third inlier ratio is less than the third threshold, determining that the first inlier number, the second inlier number, and the third inlier number all meet their corresponding threshold conditions;

[0154] S4012: Otherwise, determine that the first inlier number, the second inlier number, and the third inlier number do not meet their corresponding threshold conditions.

[0155] Furthermore, in the step of calculating the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame, the step of calculating the coordinate mean corresponding to a single neighborhood includes:

[0156] S101: Taking the laser point as the center position, selecting a preset number of second laser points on both sides of the laser point to form a neighborhood corresponding to the laser point;

[0157] S102: Obtain a first laser coordinate value of the laser point and second laser coordinate values ​​corresponding to each of the second laser points, and add the coordinate values ​​corresponding to the same coordinate axis to obtain a total laser coordinate value;

[0158] S103: performing an average calculation based on the total number of the first laser points included in the neighborhood and the sum of the laser coordinate values ​​to obtain the coordinate mean.

[0159] Furthermore, the identification method is applied to a sweeping robot equipped with an odometer. After the step of determining that the current scene is a long corridor environment, the method includes:

[0160] S6: Retrieve odometer data;

[0161] S7: Positioning the sweeping robot according to the odometer data.

[0162] Furthermore, after the step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions, the method further includes:

[0163] S8: If the first number of inner points and / or the second number of inner points and / or the third number of inner points do not meet their respective corresponding threshold conditions, then it is determined that the current scene is a non-long corridor environment;

[0164] S9: Positioning the sweeping robot according to the laser data.

[0165] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0166] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, apparatus, first object, or method comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, apparatus, first object, or method. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, apparatus, first object, or method comprising the element.

[0167] The above description is only a preferred embodiment of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for identifying a long corridor environment, characterized in that: include: Collecting laser data and calculating the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame; Constructing covariance matrices corresponding to the respective laser points according to the coordinate means of the respective neighborhoods; Solving the eigenvalues ​​of each covariance matrix, and selecting the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each of the neighborhoods; Mapping the two normal vectors having the same direction to the same value, so as to reduce the dimension of each normal vector into a plurality of first normal vectors; Voting for target angles within a preset range adjacent to each of the first normal vectors to obtain a number of votes corresponding to each of the target angles; selecting a first candidate angle with the largest number of votes from among the target angles as the first angle, and using the number of votes for the first angle as the first inlier number; Eliminate angles within a first angle range adjacent to the first angle from each target angle to obtain a plurality of second candidate angles; select the second candidate angle with the largest number of votes from each of the second candidate angles as the second angle, and use the number of votes received by the second angle as the second inner point number; Eliminate angles within a second angle range adjacent to the second angle from each of the second candidate angles to obtain a plurality of third candidate angles; select the third candidate angle with the largest number of votes from each of the third candidate angles as the third angle, and use the number of votes for the third angle as the third inner point number; Determine in sequence whether the first number of inner points, the second number of inner points, and the third number of inner points meet their respective corresponding threshold conditions; if the first number of inner points, the second number of inner points, and the third number of inner points all meet their respective corresponding threshold conditions, determine that the current scene is a long corridor environment.

2. The method for identifying a long corridor environment according to claim 1, characterized in that: The step of mapping the two normal vectors having the same direction to the same value to reduce the dimension of each normal vector to a plurality of first normal vectors includes: The direction angles of the two normal vectors are respectively substituted into the calculation formula for mapping processing to map the direction angles of the two normal vectors with the same direction to one value, wherein the calculation formula is: X2 = (X1 + 90)% 180, X1 represents the normal vector, and X2 represents the value of the normal vector after mapping.

3. The method for identifying a long corridor environment according to claim 1, characterized in that: The step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions includes: Calculating a first inlier ratio based on the first number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the first inlier ratio is greater than a first threshold; If the first inlier ratio is greater than the first threshold, calculating a second inlier ratio based on the second number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the second inlier ratio is less than a second threshold; If the second inlier ratio is less than the second threshold, calculating a third inlier ratio based on the third number of inliers and the total number of laser points included in the laser data of the current frame, and determining whether the third inlier ratio is less than a third threshold; If the third inlier ratio is less than the third threshold, determining that the first inlier number, the second inlier number, and the third inlier number all meet their corresponding threshold conditions; Otherwise, it is determined that the first number of inliers, the second number of inliers, and the third number of inliers do not meet their corresponding threshold conditions.

4. The method for identifying a long corridor environment according to claim 1, characterized in that: In the step of calculating the coordinate mean of all first laser points in the neighborhood corresponding to each laser point in the laser data of the current frame, the step of calculating the coordinate mean corresponding to a single neighborhood includes: Taking the laser point as the center position, a preset number of second laser points on both sides of the laser point are selected to form a neighborhood corresponding to the laser point; Obtaining a first laser coordinate value of the laser point and second laser coordinate values ​​corresponding to each of the second laser points, and summing the coordinate values ​​corresponding to the same coordinate axis to obtain a total laser coordinate value; The coordinate mean is obtained by performing an average calculation based on the total number of the first laser points included in the neighborhood and the sum of the laser coordinate values.

5. The method for identifying a long corridor environment according to claim 1, wherein: The identification method is applied to a sweeping robot equipped with an odometer. After determining that the current scene is a long corridor environment, the method includes: Retrieve odometer data; The sweeping robot is positioned according to the odometer data.

6. The method for identifying a long corridor environment according to claim 1, characterized in that: After the step of sequentially determining whether the first number of inliers, the second number of inliers, and the third number of inliers meet their respective corresponding threshold conditions, the method further includes: If the first number of inner points and / or the second number of inner points and / or the third number of inner points do not meet their respective corresponding threshold conditions, determining that the current scene is a non-long corridor environment; The sweeping robot is positioned according to the laser data.

7. A device for identifying a long corridor environment, characterized in that: include: An acquisition module, configured to acquire laser data and calculate the coordinate mean of all first laser points in a neighborhood corresponding to each laser point in the laser data of a current frame; A construction module, configured to construct the covariance matrices corresponding to the respective laser points according to the coordinate means of the respective neighborhoods; A first solving module is used to solve the eigenvalues ​​of each covariance matrix, and select the eigenvector corresponding to the minimum eigenvalue from the eigenvalues ​​of each covariance matrix as the normal vector corresponding to each neighborhood; a judgment module, configured to map the two normal vectors having the same direction to the same value, so as to reduce the dimension of each normal vector into a plurality of first normal vectors; Voting is performed for target angles within a preset range adjacent to each first normal vector respectively to obtain the number of votes corresponding to each target angle; a first candidate angle with the largest number of votes is selected from each target angle as the first angle, and the number of votes for the first angle is used as the first inner point number; angles within a first angle range adjacent to the first angle are eliminated from each target angle to obtain a number of second candidate angles; a second candidate angle with the largest number of votes is selected from each second candidate angle as the second angle, and the number of votes for the second angle is used as the second inner point number; angles within a second angle range adjacent to the second angle are eliminated from each second candidate angle to obtain a number of third candidate angles; a third candidate angle with the largest number of votes is selected from each third candidate angle as the third angle, and the number of votes for the third angle is used as the third inner point number, and whether the first inner point number, the second inner point number and the third inner point number meet their respective corresponding threshold conditions is determined in turn; The first determination module is configured to determine that the current scene is a long corridor environment if the first number of inner points, the second number of inner points, and the third number of inner points all meet their corresponding threshold conditions.

8. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images