Mobile robot position location confidence assessment method, device and storage medium

Abstract

The application discloses a mobile robot position location reliability evaluation method and device and a storage medium, wherein the method comprises the following steps: processing a grid map, extracting features to obtain feature containers; extracting features of a laser frame to obtain feature data; judging whether the number of the feature data exceeds a first preset threshold to obtain a first judgment result; when the first judgment result is yes, performing feature matching on the feature data according to the feature containers to obtain a matching result; when the first judgment result is no, performing correlation matching and calculating a positioning accuracy and a pose confidence; judging whether the matching result exceeds a second preset threshold to obtain a second judgment result; and when the second judgment result is yes, calculating the positioning accuracy and the pose confidence according to the matched features. The application provides a pose confidence evaluation method independent of a positioning algorithm, can timely find an error situation caused by an accident of the positioning algorithm, helps the mobile robot to stop running in time, and avoids the occurrence of a dangerous situation.

Description

Technical Field

[0001] This invention relates to the field of robotics, and in particular to a method, apparatus, and storage medium for evaluating the reliability of positioning of a mobile robot. Background Technology

[0002] Current mobile robots, such as home service robots and industrial logistics robots, mostly employ Kalman oscillators or nonlinear optimization methods to output their position information and positioning reliability in a grid map. They use the concept of confidence in probability to express the correctness of the current pose as positioning confidence. These two main types of methods typically output the robot's current pose at a certain frequency. Filtering methods are often affected by wheel odometer slippage or inaccurate assumptions in uniform motion models, resulting in a mismatch between the output pose and the actual pose. Nonlinear optimization methods may suffer from batch optimization or mismatches, causing the output pose to deviate from the actual pose. The covariance inherent in the method cannot effectively express this "lost" positioning information, leading to dangerous robot behavior during use. To balance real-time performance and motion continuity, these methods typically optimize the output pose at a certain frequency. Furthermore, to maintain motion continuity, they use uniform motion models or wheel odometer information. These two characteristics lead to occasional "lost" poses in mobile robots—that is, the output position is not the actual position. In such cases, the positioning confidence fails to express the "lost" information, causing dangerous robot behavior during use. Summary of the Invention

[0003] Purpose of the invention: In order to overcome the shortcomings of the prior art, the present invention provides a method, device and storage medium for evaluating the positioning reliability of mobile robots, which aims to monitor occasional positioning loss of mobile robots and avoid dangerous situations.

[0004] Technical solution: To achieve the above objectives, the present invention provides a mobile robot positioning reliability evaluation method, which includes:

[0005] The raster map is processed to extract features and obtain feature containers;

[0006] Extract the features of the laser frame to obtain feature data;

[0007] Determine whether the number of the feature data exceeds a first preset threshold, and obtain a first determination result;

[0008] When the first judgment result is yes, feature matching is performed on the feature data according to the feature container to obtain the matching result;

[0009] If the first judgment result is negative, perform correlation matching and calculate the positioning accuracy and pose confidence.

[0010] Determine whether the matching result exceeds a second preset threshold to obtain a second determination result;

[0011] If the second judgment result is yes, the positioning accuracy and pose confidence are calculated based on the matched features.

[0012] Furthermore, the method also includes:

[0013] If the second judgment result is negative, the following steps are executed:

[0014] Cluster the features that did not match successfully to obtain the clustering results;

[0015] Determine whether the clustering result exceeds a third preset threshold to obtain a third determination result;

[0016] If the third judgment result is yes, a warning message is output.

[0017] Furthermore, the process of processing the raster map and extracting features to obtain a feature container includes:

[0018] The raster map is binarized to obtain a grayscale image;

[0019] The grayscale image is used to perform edge detection using an edge detection algorithm to generate valid boundaries, resulting in a processed map.

[0020] The ORB features of the processed map are extracted, and the ORB features are sorted according to the pixel coordinates to obtain the feature container.

[0021] Furthermore, the extraction of features from the laser frame to obtain feature data includes:

[0022] The collected laser data is converted into raster images;

[0023] ORB features are extracted from the raster image to obtain feature data.

[0024] Furthermore, the process of converting the acquired laser data into a raster image includes:

[0025] With the mobile robot as the origin, the area between the laser hit point and the mobile robot is assigned a value of 255, the remaining effective area of ​​laser data is assigned a value of 0, and other areas are assigned a value of -1.

[0026] Further, the step of performing feature matching on the feature data based on the feature container to obtain the matching result includes:

[0027] Based on the pose of the mobile robot, the feature data is retrieved within the corresponding pixel region of the feature container;

[0028] The number of matching features is counted to obtain the matching results.

[0029] Furthermore, the matching feature calculation positioning accuracy and pose confidence include:

[0030] The percentage of the number of matched features relative to the total number of feature data is used as the pose confidence score.

[0031] The resulting pose of the mobile robot is calculated using the matched features;

[0032] The positioning accuracy is obtained by subtracting the result pose from the input pose.

[0033] Furthermore, the process of performing correlation matching and calculating positioning accuracy and pose confidence includes:

[0034] The laser data of the laser frame is mapped to the grid map according to the input pose, and the probability value of the laser data corresponding to the grid map is calculated.

[0035] Determine whether the probability value exceeds the fourth preset threshold. If it does, the pose information is considered accurate, and the subsequent steps are continued.

[0036] Calculate the matching similarity between the laser data and the grid map, and use it as the pose confidence score;

[0037] The resulting pose of the mobile robot is calculated, and the difference between the resulting pose and the input pose is used to obtain the positioning accuracy.

[0038] A mobile robot positioning reliability assessment device, comprising:

[0039] The first extraction module is used to process the raster map and extract features to obtain a feature container;

[0040] The second extraction module is used to extract features from the laser frame to obtain feature data.

[0041] The first judgment module is used to determine whether the number of the feature data exceeds a first preset threshold and obtain a first judgment result.

[0042] The first processing module is configured to perform feature matching on the feature data according to the feature container when the first judgment result is yes, and obtain a matching result;

[0043] The second processing module is used to perform correlation matching and calculate positioning accuracy and pose confidence when the first judgment result is negative.

[0044] The second judgment module is used to determine whether the matching result exceeds a second preset threshold, and to obtain a second judgment result.

[0045] The third processing module is used to calculate the positioning accuracy and pose confidence based on the matched features when the second judgment result is yes.

[0046] A storage medium containing a computer program that is executed by a processor to implement the above-described mobile robot location reliability assessment method.

[0047] Beneficial effects: The mobile robot positioning reliability assessment method, device, and storage medium of the present invention address the problem of dangerous robot behavior caused by environmental changes or abnormal positioning algorithms during operation. It proposes a pose confidence assessment method independent of the positioning algorithm, which comprehensively monitors the positioning status of mobile robots on grid maps in daily production and life by using image feature information and probability information expressed by the grid map. This allows for timely detection of errors in the positioning algorithm due to unforeseen circumstances, helping the mobile robot to stop operation promptly and avoiding dangerous situations. Attached Figure Description

[0048] Figure 1 A flowchart illustrating the method for assessing the reliability of mobile robot positioning;

[0049] Figure 2 A schematic diagram of the device for assessing the reliability of positioning of a mobile robot. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0051] It should be noted that the descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0052] Furthermore, in the following description, the use of suffixes such as "module," "part," or "unit" to denote elements is solely for the purpose of illustrative purposes and has no specific meaning in itself. Therefore, "module," "part," or "unit" may be used interchangeably.

[0053] It is understood that the flowchart step numbers in the embodiments of the present invention are not used to limit the order of execution steps.

[0054] The mobile robot positioning reliability assessment method of this application is implemented by a processor, which can be the mobile robot's own processor or an external scheduling center or control center that coordinates the operation of the mobile robot. The mobile robot is equipped with a lidar for collecting environmental data.

[0055] like Figure 1 The mobile robot positioning reliability assessment method shown includes the following steps S101-S107:

[0056] Step S101: Process the raster map and extract features to obtain feature containers;

[0057] In this step, the grid map is a map of the mobile robot's operating area. The grid map can be obtained manually, by a dedicated data acquisition device, or by the mobile robot itself during its previous task execution. Additionally, because information such as obstacles in the mobile robot's operating area may change, the grid map may contain some outdated data.

[0058] Step S102: Extract the features of the laser frame to obtain feature data;

[0059] Step S103: Determine whether the number of feature data exceeds the first preset threshold. If yes, proceed to step S104; otherwise, proceed to step S105.

[0060] In this step, if the number of feature data exceeds a first preset threshold, it indicates that relying solely on the feature data may be sufficient to assess the confidence level of the robot's input pose and calculate the positioning accuracy. If the number of feature data is less than the first preset threshold, it indicates that relying solely on the feature data may be insufficient to assess the confidence level of the position and calculate the positioning accuracy, and other methods need to be introduced as supplements to assess the confidence level of the robot's input pose and calculate the positioning accuracy. The aforementioned robot input pose refers to the pose information fed back by the robot to the processor. The robot is currently navigating based on this pose information, and the purpose of this invention is to assess the confidence level of this pose.

[0061] Step S104: Perform feature matching on the feature data according to the feature container to obtain the matching result;

[0062] Step S105: Perform correlation matching and calculate positioning accuracy and pose confidence.

[0063] Step S106: Determine whether the matching result exceeds the second preset threshold; if yes, proceed to step S107.

[0064] Step S107: Calculate the positioning accuracy and pose confidence based on the matched features.

[0065] The purpose of this step is to determine whether the matched features are sufficient for calculating the positioning accuracy and pose confidence. Generally, after step S103, the amount of feature data is sufficient for calculation. However, a mismatch occurred in step S106, indicating that the actual environment may have changed compared to the content recorded in the grid map. Therefore, preferably, the method further includes the following:

[0066] If the second judgment result is negative, proceed with the following steps S201-S203:

[0067] Step S201: Cluster the features that did not match successfully to obtain the clustering results;

[0068] In this application, Euclidean distance clustering is used for clustering.

[0069] Step S202: Determine whether the clustering result exceeds a third preset threshold, and obtain a third determination result;

[0070] Step S203: If the third judgment result is yes, output a warning message.

[0071] In this step, if the clustering result exceeds the third preset threshold, it indicates that the low matching result in step S106 is caused by environmental changes. Outputting a warning message can prompt the user to prevent the mobile robot from engaging in dangerous behavior, trigger the mobile robot to stop operating promptly, or trigger the process of updating the grid map to supplement the grid map with environmental change data. The content of the warning message may include the warning content (i.e., the environment has changed), the location information of the change, and the changed environmental data.

[0072] The process of processing the raster map and extracting features to obtain a feature container in step S101 above includes the following steps S301-S303:

[0073] Step S301: Binarize the raster map to obtain a grayscale image;

[0074] Step S302: Use an edge detection algorithm to perform edge detection on the grayscale image, generate effective boundaries, and obtain the processed map;

[0075] Step S303: Extract the ORB features of the processed map, sort the ORB features according to pixel coordinates, and obtain the feature container.

[0076] In this step, the ORB features are sorted according to the pixel coordinates, that is, the ORB features corresponding to each pixel are sorted according to the (x,y) coordinates in the pixel coordinate system.

[0077] Through the above steps S301-S303, the grid map can be processed into a feature container composed of ORB features that is easy to retrieve. Subsequently, regions can be delineated based on the pixel coordinates of the image, and the corresponding feature elements can be queried within the corresponding regions based on the input pose of the mobile robot.

[0078] The extraction of features from the laser frame in step S102 above, to obtain feature data, includes the following steps S401-S402:

[0079] Step S401: Convert the acquired laser data into a raster image;

[0080] In this step, the specific method is as follows: taking the mobile robot as the origin, the area between the laser impact point and the mobile robot is assigned a value of 255, the remaining valid laser data area is assigned a value of 0, and other areas are assigned a value of -1. That is, the grid image is divided into areas that the laser can scan and areas that the laser cannot scan. All grids in the unscanned areas are directly assigned a value of -1. The areas that the laser can scan are the valid laser data areas. In this part of the area, some lasers are emitted without returning a signal, and the grids in these areas are assigned a value of 0. In the other part, the lasers are emitted with returning signals, which are valid data. The position where the laser hits is the laser impact point, and the grids in the area between the laser impact point and the mobile robot are assigned a value of 255.

[0081] Step S402: Extract ORB features from the raster image to obtain feature data.

[0082] Further, the step S104 above, which involves performing feature matching on the feature data based on the feature container to obtain the matching result, includes the following steps S501-S502:

[0083] Step S501: Based on the pose of the mobile robot, retrieve the feature data within the corresponding pixel region of the feature container;

[0084] Step S502: Count the number of matching features to obtain the matching results.

[0085] Based on this, the matching feature calculation positioning accuracy and pose confidence in step S107 above includes the following steps S601-S603:

[0086] Step S601: Calculate the percentage of the number of matched features to the total number of feature data, and use it as the pose confidence.

[0087] Step S602: Calculate the resulting pose of the mobile robot using the matched features;

[0088] In this step, the matching features can be used to determine which part of the grid map the feature matches, and the resulting pose of the mobile robot can be calculated accordingly.

[0089] Step S603: Calculate the difference between the resulting pose and the input pose to obtain the positioning accuracy.

[0090] Furthermore, the correlation matching and calculation of positioning accuracy and pose confidence described in step S105 above includes the following steps S701-S704:

[0091] Step S701: Map the laser data of the laser frame to the grid map according to the input pose, and count the probability value of the laser data corresponding to the grid map;

[0092] Step S702: Determine whether the probability value exceeds the fourth preset threshold. If yes, the pose information is considered accurate, and the subsequent steps are executed.

[0093] Step S703: Calculate the matching similarity between the laser data and the grid map, and use it as the pose confidence score;

[0094] Step S704: Calculate the resulting pose of the mobile robot and calculate the difference between the resulting pose and the input pose to obtain the positioning accuracy.

[0095] In the above steps, correlation matching is a supplementary operation performed when the matched features are insufficient. It utilizes the probability representation of pixels in a grid map to map the laser data onto the map and calculates the probability values ​​corresponding to the laser data on the grid map. If the probability exceeds a fourth preset threshold, the pose information is considered accurate. Correlation matching is a pre-operation performed before evaluating the pose confidence level to increase the algorithm's robustness. Its purpose is to reduce false alarms caused by minor perturbations and improve the accuracy of the algorithm's output. Step S704 also uses matched features to calculate the resulting pose.

[0096] The aforementioned technical solution separates the evaluation of location reliability from the calculation of location accuracy into a single module. This module acts as an external monitor, providing a more comprehensive evaluation of location reliability and preventing dangerous robot behaviors caused by inherent flaws in the localization algorithm. Furthermore, this solution employs a region-ranked feature representation method, performing computationally intensive feature extraction only once, improving real-time performance and reducing resource usage. For grid maps, where there are many line features (such as corridors) but few corner features, current methods handle this aspect rather crudely, either treating missing features as "missing" or using line features for matching. Both methods are ineffective in scenarios with limited features. Therefore, a correlation matching method is used to evaluate location reliability even with insufficient features, providing a more robust approach to everyday production and living environments.

[0097] This invention also provides a mobile robot positioning reliability assessment device 800. The mobile robot positioning reliability assessment device 800 may include or be divided into one or more program modules. One or more program modules are stored in a storage medium and executed by one or more processors to complete this invention and implement the aforementioned mobile robot positioning reliability assessment method. The program module referred to in this embodiment of the invention refers to a series of computer program instruction segments capable of performing a specific function, which is more suitable than the program itself for describing the execution process of the mobile robot positioning reliability assessment device 800 in the storage medium. The following description will specifically introduce the functions of each program module in this embodiment, such as... Figure 2 As shown, it includes:

[0098] The first extraction module 801 is used to process the raster map and extract features to obtain a feature container;

[0099] The second extraction module 802 is used to extract features of the laser frame to obtain feature data;

[0100] The first judgment module 803 is used to judge whether the number of the feature data exceeds a first preset threshold and obtain a first judgment result.

[0101] The first processing module 804 is used to perform feature matching on the feature data according to the feature container when the first judgment result is yes, and obtain a matching result.

[0102] The second processing module 805 is used to perform correlation matching and calculate positioning accuracy and pose confidence when the first judgment result is negative.

[0103] The second judgment module 806 is used to determine whether the matching result exceeds a second preset threshold and obtain a second judgment result.

[0104] The third processing module 807 is used to calculate the positioning accuracy and pose confidence based on the matched features when the second judgment result is yes.

[0105] Preferably, the mobile robot positioning reliability assessment device 800 further includes:

[0106] The fourth processing module 808 is used to perform the following steps when the second judgment result is negative:

[0107] Cluster the features that did not match successfully to obtain the clustering results;

[0108] Determine whether the clustering result exceeds a third preset threshold to obtain a third determination result;

[0109] If the third judgment result is yes, a warning message is output.

[0110] Other aspects of implementing the above-mentioned mobile robot positioning reliability assessment method based on the mobile robot positioning reliability assessment device 800 have been described in detail in previous embodiments. Please refer to the corresponding content in the previous embodiments. They will not be repeated here.

[0111] This embodiment also provides a storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, server, App store, etc., which stores a computer program. When the program is executed by a processor, it performs the corresponding function. In this embodiment, the storage medium stores a computer program, which, when executed by a processor, can implement the above-described mobile robot positioning reliability assessment method.

[0112] The present invention relates to a mobile robot positioning confidence assessment method, apparatus, and storage medium. Addressing the problem of dangerous robot behavior caused by environmental changes or positioning algorithm malfunctions during operation, this invention proposes a pose confidence assessment method independent of the positioning algorithm. By utilizing image feature information and probability information expressed by the grid map, it comprehensively monitors the positioning status of mobile robots on grid maps in daily production and life. This allows for timely detection of positioning algorithm errors due to unforeseen circumstances, helping the mobile robot to stop operation promptly and avoiding dangerous situations.

[0113] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for evaluating the reliability of mobile robot positioning, characterized in that, It includes: The raster map is processed to extract features and obtain feature containers; Extract the features of the laser frame to obtain feature data; Determine whether the number of the feature data exceeds a first preset threshold, and obtain a first determination result; When the first judgment result is yes, feature matching is performed on the feature data according to the feature container to obtain the matching result; When the first judgment result is negative, correlation matching is performed and positioning accuracy and pose confidence are calculated. Determine whether the matching result exceeds a second preset threshold to obtain a second determination result; When the second judgment result is yes, the positioning accuracy and pose confidence are calculated based on the matched features; The steps of performing correlation matching and calculating positioning accuracy and pose confidence include: The laser data of the laser frame is mapped to the grid map according to the input pose, and the probability value of the laser data corresponding to the grid map is calculated. Determine whether the probability value exceeds the fourth preset threshold. If it does, the pose information is considered accurate, and the subsequent steps are continued. Calculate the matching similarity between the laser data and the grid map, and use it as the pose confidence score; The resulting pose of the mobile robot is calculated, and the difference between the resulting pose and the input pose is used to obtain the positioning accuracy.

2. The mobile robot positioning reliability assessment method according to claim 1, characterized in that, The method further includes: If the second judgment result is negative, the following steps are executed: Cluster the features that did not match successfully to obtain the clustering results; Determine whether the clustering result exceeds a third preset threshold to obtain a third determination result; If the third judgment result is yes, a warning message is output.

3. The mobile robot positioning reliability assessment method according to claim 1, characterized in that, The process of processing the raster map and extracting features to obtain a feature container includes: The raster map is binarized to obtain a grayscale image; The grayscale image is used to perform edge detection using an edge detection algorithm to generate valid boundaries, resulting in a processed map. The ORB features of the processed map are extracted, and the ORB features are sorted according to the pixel coordinates to obtain the feature container.

4. The mobile robot positioning reliability assessment method according to claim 1, characterized in that, The extraction of features from the laser frame to obtain feature data includes: The collected laser data is converted into raster images; ORB features are extracted from the raster image to obtain feature data.

5. The mobile robot positioning reliability assessment method according to claim 4, characterized in that, The process of converting the acquired laser data into a raster image includes: With the mobile robot as the origin, the area between the laser hit point and the mobile robot is assigned a value of 255, the remaining effective area of ​​laser data is assigned a value of 0, and other areas are assigned a value of -1.

6. The mobile robot positioning reliability assessment method according to claim 1, characterized in that, The step of performing feature matching on the feature data based on the feature container to obtain the matching result includes: Based on the pose of the mobile robot, the feature data is retrieved within the corresponding pixel region of the feature container; The number of matching features is counted to obtain the matching results.

7. The mobile robot positioning reliability assessment method according to claim 1, characterized in that, The matching features The calculation of positioning accuracy and pose confidence includes: The percentage of the number of matched features relative to the total number of feature data is used as the pose confidence score. The resulting pose of the mobile robot is calculated using the matched features; The positioning accuracy is obtained by subtracting the result pose from the input pose.

8. A mobile robot positioning reliability assessment device, characterized in that, It includes: The first extraction module is used to process the raster map and extract features to obtain a feature container; The second extraction module is used to extract features from the laser frame to obtain feature data. The first judgment module is used to determine whether the number of the feature data exceeds a first preset threshold and obtain a first judgment result. The first processing module is configured to perform feature matching on the feature data according to the feature container to obtain a matching result when the first judgment result is yes. The second processing module is used to perform correlation matching and calculate positioning accuracy and pose confidence when the first judgment result is negative. The steps of performing correlation matching and calculating positioning accuracy and pose confidence include: The laser data of the laser frame is mapped to the grid map according to the input pose, and the probability value of the laser data corresponding to the grid map is calculated. Determine whether the probability value exceeds the fourth preset threshold. If it does, the pose information is considered accurate, and the subsequent steps are continued. Calculate the matching similarity between the laser data and the grid map, and use it as the pose confidence score; Calculate the resulting pose of the mobile robot and calculate the difference between the resulting pose and the input pose to obtain the positioning accuracy; The second judgment module is used to determine whether the matching result exceeds a second preset threshold, and to obtain a second judgment result. The third processing module is used to calculate the positioning accuracy and pose confidence based on the matched features when the second judgment result is yes.

9. A storage medium, characterized in that, It contains a computer program that is executed by a processor to implement the mobile robot location reliability assessment method according to any one of claims 1-7.

Images