Method, system, robot and readable storage medium for judging robot positioning quality

Abstract

The present invention provides a method for judging the positioning quality of a robot. The method matches the positioning data obtained by the positioning module of the robot with the map information of the robot, and judges the positioning quality of the robot by the matching result. Compared with the independent high-precision positioning system used in the prior art, the judgment method of the present invention does not require any additional device, which not only reduces the cost but also saves space. It is applied to the positioning quality evaluation of the robot on-site deployment link to improve the quality of robot deployment. The present invention also provides a system for judging the positioning quality of a robot, a robot and a non-volatile computer-readable storage medium.

Description

Technical Field

[0001] The present invention relates to the field of robot positioning, and in particular to a method and system for judging the robot positioning quality, a robot and a readable storage medium. Background Art

[0002] With the rapid development of robot-related technologies, people's demand for robots is also increasing. Wheeled mobile robots rely on rollers to move and automatically perform work. They can accept human commands, run pre-programmed programs, or act according to principles formulated by artificial intelligence technology. Their mission is to assist or replace human work, such as delivering meals, receiving and sending express deliveries, etc.

[0003] The accurate and efficient operation of a wheeled mobile robot depends on precise positioning, so it is particularly important to judge and evaluate the positioning accuracy and positioning quality of a wheeled mobile robot. In the prior art, the evaluation of positioning accuracy generally requires the use of other independent, higher-precision positioning systems, which use the positioning output results of the high-precision positioning system as the true value, and then compare the positioning data results of the wheeled mobile robot's own positioning system with the true value, and evaluate the positioning quality of the wheeled mobile robot through the absolute error between the two.

[0004] The above-mentioned prior art evaluates the positioning quality of the robot by calculating the absolute error between the positioning output results of the high-precision positioning system and the wheeled mobile robot. Although this method has high accuracy, it relies on an independent high-precision positioning system and has a high cost. It is usually only placed in fixed scenes such as laboratories and factories, such as the VICON optical motion capture system. Moreover, the high-precision positioning system requires a certain amount of space, resulting in a waste of space. Summary of the invention

[0005] The present invention aims to solve at least one of the above technical problems to a certain extent or at least provide a useful commercial option. To this end, the present invention proposes a method for judging the positioning quality of a robot, which matches the positioning data obtained by the positioning module of the robot with the map information of the robot, and judges the positioning quality of the robot by the matching result. Compared with the independent high-precision positioning system used in the prior art, the judgment method of the present invention does not require any additional devices, which not only reduces costs but also saves space. It is applied to the positioning quality evaluation of the robot on-site deployment link to improve the quality of robot deployment.

[0006] According to the method for judging the positioning quality of a robot of the present invention, the method comprises the following steps:

[0007] Using a positioning module of the robot to obtain positioning data of the robot in a predetermined area, and processing the positioning data to obtain processed positioning data, wherein the positioning module is a laser positioning module;

[0008] Acquiring map information of the robot in the predetermined area;

[0009] Matching the processed positioning data with the map information to obtain a matching result;

[0010] The positioning quality of the robot in the predetermined area is determined according to the matching result.

[0011] According to the method for judging the positioning quality of a robot of the present invention, the method matches the positioning data acquired by the positioning module of the robot with the map information of the robot, and judges the positioning quality of the robot by the matching result. Compared with the use of an independent high-precision positioning system in the prior art, the judgment method of the present invention does not require any additional devices, which not only reduces costs but also saves space. It is applied to the positioning quality evaluation of the robot on-site deployment link to improve the quality of robot deployment.

[0012] In addition, the method for determining the positioning quality of the robot according to the present invention may also have the following additional technical features:

[0013] The specific steps of using the positioning module of the robot to obtain the positioning data of the robot in a predetermined area and processing the positioning data to obtain processed positioning data include:

[0014] The laser positioning module on the robot is used to obtain the posture data and laser radar point cloud data of each frame of the robot during operation. The posture data and laser radar point cloud data in the same frame correspond one to one.

[0015] The pose data and lidar point cloud data of each frame are summarized to obtain the overall pose data and the overall lidar point cloud data, and then the overall lidar point cloud data is processed by voxel filtering and density-based clustering algorithm to obtain the processed lidar point cloud data.

[0016] The specific steps of matching the processed positioning data with the map information to obtain a matching result include:

[0017] Converting the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a first spatial position, and the posture data corresponding to the laser radar point cloud data in the world coordinate system in the same frame has a second spatial position, and connecting the first spatial position and the second spatial position to form a spatial straight line segment;

[0018] Mapping the spatial straight line segment into a digital grid map in the map information;

[0019] Determine whether the spatial straight line segment collides with the obstacle of the digital grid map, and if a collision occurs, mark the posture data corresponding to the spatial straight line segment colliding with the digital grid map as abnormal posture data;

[0020] Calculate whether the posture data of different detection points in the same frame are abnormal posture data, and count the first ratio of the number of abnormal posture data to the number of all posture data in the frame. If the first ratio exceeds a first preset threshold, the posture of the robot in the frame is determined to be an abnormal posture; if the first ratio is lower than the first preset threshold, the posture of the robot in the frame is determined to be a normal posture;

[0021] Determine respectively whether the posture of the robot in different frames is an abnormal posture, and count the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot is a first ratio.

[0022] The specific step of judging the positioning quality of the robot in the predetermined area according to the matching result comprises:

[0023] The positioning quality of the robot in the predetermined area is judged according to the first ratio, wherein if the first ratio is greater than or equal to a second preset threshold, it is determined that the positioning quality of the robot in the predetermined area is poor; if the first ratio is less than the second preset threshold, it is determined that the positioning quality of the robot in the predetermined area is good.

[0024] The specific steps of matching the processed positioning data with the map information to obtain a matching result include:

[0025] Converting the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a third spatial position;

[0026] Using a first fixed threshold, converting the digital grid map in the map information into a binary map, and then converting the binary map into a distance map;

[0027] Match the third spatial position with the distance map, and determine whether the distance between the nearest obstacle in the distance map and the third spatial position is greater than or equal to a first preset distance; if it is greater than or equal to the first preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the third spatial position as abnormal pose data; if it is less than the first preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the third spatial position as normal pose data;

[0028] Calculate whether the posture data of different detection points in the same frame are abnormal posture data, and count the second ratio of the number of abnormal posture data to the number of all posture data in the frame, if the second ratio exceeds the third preset threshold, determine that the posture of the robot in the frame is an abnormal posture; if the second ratio is lower than the third preset threshold, determine that the posture of the robot in the frame is a normal posture;

[0029] Determine respectively whether the posture of the robot in different frames is an abnormal posture, and count the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot is the second ratio.

[0030] The specific step of judging the positioning quality of the robot in the predetermined area according to the matching result comprises:

[0031] The positioning quality of the robot in the predetermined area is judged according to the second ratio, wherein if the second ratio is greater than or equal to a fourth preset threshold, it is determined that the positioning quality of the robot in the predetermined area is poor; if the second ratio is less than the fourth preset threshold, it is determined that the positioning quality of the robot in the predetermined area is good.

[0032] The specific step of matching the processed positioning data with the map information to obtain a matching result also includes:

[0033] Converting the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a fourth spatial position;

[0034] Using a second fixed threshold value to convert the digital grid map in the map information into a binary map, and then converting the binary map into a distance map;

[0035] Match the fourth spatial position with the distance map, and determine whether the distance between the nearest obstacle in the distance map and the fourth spatial position is greater than or equal to a second preset distance; if it is greater than or equal to the second preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the fourth spatial position as abnormal pose data; if it is less than the second preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the fourth spatial position as normal pose data;

[0036] Calculate whether the posture data of different detection points in the same frame are abnormal posture data, and count the third ratio of the number of abnormal posture data to the number of all posture data in the frame, if the third ratio exceeds the fifth preset threshold, determine that the posture of the robot in the frame is an abnormal posture; if the third ratio is lower than the fifth preset threshold, determine that the posture of the robot in the frame is a normal posture;

[0037] Determine whether the posture of the robot in different frames is an abnormal posture, and count the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot is the third ratio.

[0038] The specific step of judging the positioning quality of the robot in the predetermined area according to the matching result comprises:

[0039] Calculating a matching result value according to the first ratio and the third ratio, wherein the matching result value is in direct proportion to the first ratio and the third ratio respectively;

[0040] The smaller the matching result value is, the better the positioning quality of the robot in the predetermined area is.

[0041] The specific steps of successively subjecting the overall laser radar point cloud data to voxel filtering and density-based clustering algorithm processing to obtain processed laser radar point cloud data include:

[0042] Using a grid of a certain size to convert the overall lidar point cloud data into filtered lidar point cloud data through a voxel filtering algorithm;

[0043] Then, a density-based clustering algorithm is used to cluster different points in the filtered lidar point cloud data into multiple clusters of different sizes; the number of points in each cluster is determined respectively, and if the number is less than a certain number, all points in the cluster with a number less than a certain number are deleted to obtain the processed lidar point cloud data.

[0044] The present invention also provides a system for judging the positioning quality of a robot. The system matches the positioning data obtained by the positioning module of the robot with the map information of the robot, and judges the positioning quality of the robot based on the matching result. Compared with the independent high-precision positioning system used in the prior art, the judgment system of the present invention does not require any additional devices, which not only reduces costs but also saves space. It is applied to the positioning quality evaluation of the robot on-site deployment link to improve the quality of robot deployment.

[0045] The present invention also provides a robot, comprising one or more processors, a memory; and one or more programs, wherein the one or more programs are stored in the memory and executed by the one or more processors, and the program includes instructions for executing any of the above-mentioned methods for determining the positioning quality of the robot.

[0046] The present invention also provides a non-volatile computer-readable storage medium containing a computer program. When the computer program is executed by one or more processors, the processors execute any of the above-mentioned methods for determining the positioning quality of the robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0048] Figure 1 is a flow chart of a method for determining robot positioning quality according to an embodiment of the present invention;

[0049] Figure 2 is a structural block diagram of a robot positioning quality judgment system according to an embodiment of the present invention;

[0050] Figure 3 is a schematic diagram of a module of a robot according to an embodiment of the present invention; and

[0051] Figure 4 It is a schematic diagram of the connection relationship between a computer-readable storage medium and a processor according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0053] The present invention provides a method for judging the positioning quality of a robot. Compared with the traditional method using a high-precision positioning system, the method saves cost and space, is applied to the positioning quality evaluation of the robot on-site deployment link, and improves the quality of the robot deployment.

[0054] Figure 1 FIG. 1 is a flow chart of a method for determining robot positioning quality according to an embodiment of the present invention. Figure 1 The present invention provides a method for judging the positioning quality of a robot, and the method is used to evaluate and judge the positioning quality of a robot in a predetermined area. In this embodiment, the robot is a wheeled mobile robot operating indoors. In other embodiments, the robot may also be other robots, such as a production robot for manufacturing or a window cleaning robot that replaces humans in dangerous work.

[0055] The method for judging the robot positioning quality comprises the following steps:

[0056] S10. Obtaining positioning data of the robot in a predetermined area using a positioning module of the robot, and processing the positioning data to obtain processed positioning data, wherein the positioning module is a laser positioning module.

[0057] Specifically, before a wheeled food delivery robot delivers food for the first time in a restaurant, or when the layout of the restaurant changes, or when the positioning system of the food delivery robot is upgraded or the positioning algorithm is iterated, it is necessary to traverse and update the positioning data such as the posture information and path planning of the food delivery robot. The quality of positioning directly affects the work progress and efficiency of the food delivery robot.

[0058] The wheeled food delivery robot is provided with a laser positioning module. The laser positioning module takes a laser radar as an example. The wheeled food delivery robot is provided with a laser radar, which can emit a laser beam outward. It compares the received signal reflected from the target with the transmitted signal, and obtains relevant information of the target after appropriate processing, such as target distance, direction, height, speed, posture and other parameters, so as to detect and identify targets in the restaurant.

[0059] The wheeled food delivery robot moves back and forth in the restaurant and traverses every place in the restaurant. While the robot is moving, the laser radar on the robot continuously emits laser beams outward, thereby obtaining relevant parameters of the target according to the received feedback signal, and then summarizing the relevant parameters to obtain positioning data in the restaurant. Usually, the positioning data is also subjected to noise reduction processing to obtain processed positioning data, thereby improving the accuracy of the positioning data.

[0060] S20: Obtain map information of the robot in the predetermined area.

[0061] Specifically, the map information in the restaurant may be map data and map information displayed by a pre-drawn map, or may be map information generated by the robot during its traversal in the restaurant.

[0062] In this embodiment, the map information is preferably the map information generated while the robot obtains positioning data during the traversal process in the restaurant. Specifically, the food delivery robot of the present invention is provided with a SLAM (Simultaneous Localization and Mapping) system, which can enable the robot to gradually draw a complete map of the environment while moving in the restaurant, and locate itself according to the position estimation and map during the movement, and build an incremental map based on its own positioning, so that it can move to every accessible corner of the restaurant without obstacles, so as to realize the autonomous positioning and navigation of the food delivery robot. By setting a SLAM system on the food delivery robot of the present invention, while obtaining the positioning data of the robot in the restaurant, it also obtains the map information in the restaurant.

[0063] S30: Match the processed positioning data with the map information to obtain a matching result.

[0064] Specifically, after obtaining the processed positioning data of the food delivery robot in the restaurant and the map information in the restaurant, the present invention matches and compares the processed positioning data with the map information, and obtains a matching and comparison result, thereby determining the degree of correlation between the processed positioning data and the map information.

[0065] S40: Determine the positioning quality of the robot in the predetermined area according to the matching result.

[0066] Specifically, in the above step S30, after obtaining the matching result of the processed positioning data and the map information in the restaurant, the robot positioning quality judgment method of the present invention judges the positioning quality of the food delivery robot in the restaurant through the matching result. If the degree of correlation between the processed positioning data and the map information in the restaurant is high after matching, it is considered that the positioning quality of the food delivery robot in the restaurant is good; on the contrary, if the degree of correlation between the processed positioning data and the map information in the restaurant is low after matching, it is considered that the positioning quality of the food delivery robot in the restaurant is poor.

[0067] The method for judging the robot positioning quality of the present invention matches the positioning data obtained by the robot's positioning module with the robot's map information, and judges the robot's positioning quality based on the matching result. Compared with the use of an independent high-precision positioning system in the prior art, the judgment method of the present invention does not require any additional devices, which not only reduces costs but also saves space. It is applied to the positioning quality evaluation of the robot on-site deployment link to improve the quality of robot deployment.

[0068] In a specific implementation, the specific steps of step S10 include the following steps:

[0069] S11. Obtain the posture data and laser radar point cloud data of each frame of the robot during operation through the laser positioning module on the robot, wherein the posture data and the laser radar point cloud data in the same frame correspond one to one.

[0070] Specifically, when the food delivery robot is running in the restaurant, its laser radar is also turned on. The food delivery robot runs regularly or irregularly in the restaurant, and its movement trajectory includes straight, oblique, reverse, turning on the spot, turning while walking, etc.

[0071] During the operation of the food delivery robot, its laser radar obtains the posture data and laser radar point cloud data in each frame through calculation. The posture data in each frame is mapped to the laser radar point cloud data in the frame, one by one. Usually, the posture data includes position data and posture data, such as the x-coordinate and y-coordinate in the plane coordinate system, and the posture data includes the rotation angle, etc. The laser radar point cloud data is the data obtained by laser radar scanning, which is a set of vectors in a three-dimensional coordinate system. Each point contains three-dimensional coordinate information (X, Y, Z three elements), as well as color information, reflection intensity information, echo number information, etc.

[0072] In this embodiment, the pose data is recorded as Ti, where Ti represents the pose data in the i-th frame, i∈(1, N), where N is a positive integer greater than 2, and the value of N is usually very large; the lidar point cloud data is recorded as Ci, and similarly, i∈(1, N), where N is a positive integer greater than 2.

[0073] S12. Summarize the pose data and lidar point cloud data of each frame to obtain overall pose data and overall lidar point cloud data, and then process the overall lidar point cloud data through voxel filtering and density-based clustering algorithm to obtain processed lidar point cloud data.

[0074] Specifically, the collection of pose data in each frame obtained by the food delivery robot traversing every corner in the restaurant forms the overall pose data, and the collection of lidar point cloud data in each frame forms the overall lidar point cloud data.

[0075] The method for judging the positioning quality of the robot of the present invention, after obtaining the above-mentioned overall laser radar point cloud data, also processes the overall laser radar point cloud data to obtain the processed laser radar point cloud data. Specifically, firstly, the original laser radar point cloud data Ci under each frame is filtered by a voxel filtering algorithm using a grid of a certain size to obtain the filtered laser radar point cloud data FCi. In this embodiment, the size of the grid is 5 cm; in other embodiments, grids of other sizes can also be used; then, for each frame FCi, the density-based clustering algorithm DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is used to cluster different points in the point cloud data FCi into multiple clusters of different sizes, and the number of points in each cluster in the point cloud under each frame is determined respectively. If the number is less than a certain number, all points of this cluster are deleted from FCi to obtain the processed laser radar point cloud data DCi. The purpose of this step is mainly to remove some non-fixed obstacles such as human legs, table and chair legs, etc. Optionally, since obstacles such as human legs and table and chair legs occupy a smaller area, the point cloud data FCi formed is also smaller. And since this application is for matching the map, that is, some fixed obstacles, such as walls, etc., and human legs and table and chair legs are not fixed obstacles in this map, on the one hand, they may cause an impact, and on the other hand, they will increase the amount of calculation. Therefore, by deleting some clusters, the calculation efficiency can be effectively enhanced and the calculation accuracy can be improved.

[0076] In a specific implementation, the specific steps of step S30 include the following steps:

[0077] S31. Convert the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a first spatial position, and the posture data corresponding to the laser radar point cloud data in the world coordinate system in the same frame has a second spatial position, and connect the first spatial position and the second spatial position to form a spatial straight line segment; map the spatial straight line segment to the digital grid map in the map information; determine whether the spatial straight line segment collides with an obstacle in the digital grid map, and if a collision occurs, mark the posture data corresponding to the spatial straight line segment that collides with the digital grid map as abnormal posture data.

[0078] Specifically, the processed lidar point cloud data DCi in each frame is converted from the body coordinate system to the world coordinate system, so as to obtain the lidar point cloud data WCi in the world coordinate system, that is, WCi=Ti⊙DCi, WCi represents the point cloud in the world coordinate system, Ti is the posture data, and the relationship between the body coordinate system and the world coordinate system can be described specifically by the rotation matrix and the translation vector.

[0079] The laser radar point cloud data WCi in the world coordinate system has a first spatial position, and the posture data Ti corresponding to the WCi in the same frame has a second spatial position. The first spatial position and the second spatial position are connected to form a spatial straight line segment, and the spatial straight line segment is mapped to the digital grid map in the map information. It is determined whether the spatial straight line segment collides with the digital grid map. That is, if the spatial straight line segment collides with an obstacle in the digital grid map, it means that there is an obstacle in the digital grid map passed by the spatial straight line segment, so that the digital grid map collides with the digital grid map. The pose data Ti corresponding to the spatial straight line segment that collides with the grid map is marked as abnormal pose data bTi. This step is mainly to determine whether there is an obstacle in the grid in the map passed by this straight line segment, and to determine whether the pose in this frame is abnormal. Because if the pose is accurate, the laser beam emitted by the lidar should illuminate the obstacle and form point cloud data WCi at the position of the obstacle, instead of passing through the obstacle and forming point cloud data WCi on the side of the obstacle away from the lidar. Therefore, the pose data can be screened in this way to determine whether it is abnormal pose data.

[0080] S32. Calculate respectively whether the posture data of different detection points in the same frame are abnormal posture data, and count the first ratio of the number of abnormal posture data to the number of all posture data in the frame; if the first ratio exceeds the first preset threshold, the posture of the robot in the frame is determined to be an abnormal posture; if the first ratio is lower than the first preset threshold, the posture of the robot in the frame is determined to be a normal posture.

[0081] Specifically, the food delivery robot includes multiple detection points in each frame, for example, at least one detection point is set at the base, arm, and rotating joint of the food delivery robot, and the posture data of each detection point in each frame are different. The judgment method of the present invention determines whether the overall posture of the robot in each frame is a normal posture by calculating the ratio of the number of abnormal posture detection points in each frame to the total detection points in the frame. The ratio is defined as a first ratio. For example, when the first ratio exceeds 0.75, that is, among all the detection points in a certain frame, if the number of detection points determined to be abnormal postures accounts for more than 0.75, then the posture of the food delivery robot in the frame is considered to be an abnormal posture; if among all the detection points in the frame, the number of detection points determined to be abnormal postures accounts for less than 0.75, then the posture of the food delivery robot in the frame is considered to be a normal posture.

[0082] S33. Determine whether the posture of the robot in different frames is an abnormal posture, and count the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot is a first ratio.

[0083] Specifically, when the food delivery robot traverses every corner of the restaurant, its laser radar obtains posture data and posture information in multiple frames. The judgment method of the present invention judges the posture in each frame, judges and counts whether the posture in each frame is an abnormal posture or a normal posture, and counts the proportion of abnormal postures in all postures, and the proportion is defined as the first ratio S1.

[0084] In a specific implementation, the specific steps of step S40 include: judging the positioning quality of the robot in the predetermined area based on the first ratio, wherein if the first ratio is greater than or equal to a second preset threshold, it is judged that the positioning quality of the robot in the predetermined area is poor; if the first ratio is less than the second preset threshold, it is judged that the positioning quality of the robot in the predetermined area is good.

[0085] Specifically, the judgment method of the present invention evaluates the quality of robot positioning by calculating the proportion of all abnormal postures to all postures. For example, when S1≥0.1, the positioning quality of the robot is considered to be poor, that is, after the robot traverses every corner of the restaurant through laser radar, if the number of abnormal postures exceeds 10% of the total number of postures, the positioning quality of the robot is considered to be poor; on the contrary, if the number of abnormal postures is less than 10% of the total number of postures, the positioning quality of the robot is considered to be good.

[0086] The method for judging the robot positioning quality of the present invention judges whether the posture is an abnormal posture by connecting the laser point and the posture, and further can judge the quality of the robot's positioning. Compared with the prior art that requires the use of an additional high-precision positioning system, it saves cost and space, and is effective in improving the quality of robot deployment.

[0087] In a specific implementation, as a second embodiment, the specific steps of step S30 further include the following steps:

[0088] S311. Convert the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a third spatial position; use a first fixed threshold to convert the digital grid map in the map information into a binary map, and then convert the binary map into a distance map; match the third spatial position with the distance map, and determine whether the distance between the nearest obstacle in the distance map and the third spatial position is greater than or equal to a first preset distance; if it is greater than or equal to the first preset distance, mark the posture data corresponding to the laser radar point cloud data in the world coordinates at the third spatial position as abnormal posture data; if it is less than the first preset distance, mark the posture data corresponding to the laser radar point cloud data in the world coordinates at the third spatial position as normal posture data.

[0089] Specifically, the processed lidar point cloud data DCi in each frame is converted from the body coordinate system to the world coordinate system, so as to obtain the lidar point cloud data WCi in the world coordinate system, that is, WCi=Ti⊙DCi, WCi represents the point cloud in the world coordinate system, Ti is the posture data, and the relationship between the body coordinate system and the world coordinate system can be described specifically by the rotation matrix and the translation vector.

[0090] The laser radar point cloud data WCi in the world coordinate system has a third spatial position. First, a fixed threshold (for example, 120) is used to convert the digital grid map in the map information into a binary image, and then the binary image is transformed into a distance map by a method such as Euclidean distance transformation, wherein the value of each point on the distance map represents the distance between the point and the nearest obstacle.

[0091] Match the third spatial position of WCi with the distance map in the map information, and determine whether the distance between the nearest obstacle in the distance map and the third spatial position is greater than or equal to the first preset distance. In this embodiment, the first preset distance is set to 0.1 meters; in other embodiments, the first preset distance can also be defined as other values, which can be set specifically according to the needs of actual positioning accuracy. That is, when it is greater than or equal to 0.1 meters, the posture data Ti corresponding to WCi is marked as abnormal posture data fTi; when it is less than 0.1 meters, the posture data Ti corresponding to WCi is marked as normal posture data. This step is mainly to determine whether the posture is abnormal by calculating the distance between the laser point and the obstacle, because if the posture is correct, the laser beam emitted by the laser radar of the food delivery robot should be irradiated on the obstacle, and point cloud data WCi is formed at the position corresponding to the obstacle or in the range that meets the preset distance, and this distance should be very small.

[0092] S312. Calculate whether the posture data of different detection points in the same frame are abnormal posture data, and count the second ratio of the number of abnormal posture data to the number of all posture data in the frame. If the second ratio exceeds the third preset threshold, the posture of the robot in the frame is determined to be an abnormal posture; if the second ratio is lower than the third preset threshold, the posture of the robot in the frame is determined to be a normal posture.

[0093] Specifically, the food delivery robot includes multiple detection points in each frame, for example, at least one detection point is set at the base, arm, and rotating joint of the food delivery robot, and the posture data of each detection point in each frame are different. The judgment method of the present invention determines whether the overall posture of the robot in each frame is a normal posture by calculating the ratio of the number of abnormal posture detection points in each frame to the total detection points in the frame. The ratio is defined as the second ratio. For example, when the second ratio exceeds 0.75, that is, among all the detection points in a certain frame, if the number of detection points determined to be abnormal postures accounts for more than 0.75, then the posture of the food delivery robot in the frame is considered to be an abnormal posture; if among all the detection points in the frame, the number of detection points determined to be abnormal postures accounts for less than 0.75, then the posture of the food delivery robot in the frame is considered to be a normal posture.

[0094] S313. Determine whether the posture of the robot in different frames is an abnormal posture, and count the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot is the second ratio.

[0095] Specifically, when the food delivery robot traverses every corner of the restaurant, its laser radar obtains posture data and posture information in multiple frames. The judgment method of the present invention judges the posture in each frame, judges and counts whether the posture in each frame is an abnormal posture or a normal posture, and counts the proportion of abnormal postures in all postures, and the proportion is defined as the second ratio S2.

[0096] In a specific implementation, after step S313, the following steps are further included:

[0097] The positioning quality of the robot in the predetermined area is judged according to the second ratio, wherein if the second ratio is greater than or equal to a fourth preset threshold, it is determined that the positioning quality of the robot in the predetermined area is poor; if the second ratio is less than the fourth preset threshold, it is determined that the positioning quality of the robot in the predetermined area is good.

[0098] The judgment method of the present invention evaluates the quality of robot positioning by calculating the ratio of all abnormal postures to all postures. For example, when S2≥0.1, the positioning quality of the robot is considered to be poor, that is, after the robot traverses every corner of the restaurant through the laser radar, if the number of abnormal postures exceeds 10% of the total number of postures, the positioning quality of the robot is considered to be poor; on the contrary, if the number of abnormal postures is less than 10% of the total number of postures, the positioning quality of the robot is considered to be good.

[0099] The method for judging the robot positioning quality of the present invention determines whether the posture is an abnormal posture by calculating the distance between the laser point and the obstacle, and then can judge the quality of the robot's positioning. Compared with the prior art that requires the use of an additional high-precision positioning system, it saves cost and space, and is useful for improving the quality of robot deployment.

[0100] In a specific implementation, as a third embodiment, the determination method of the present invention further includes the following steps after step S33:

[0101] S34. Convert the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a fourth spatial position; use a second fixed threshold to convert the digital grid map in the map information into a binary map, and then convert the binary map into a distance map; match the fourth spatial position with the distance map, and determine whether the distance between the nearest obstacle in the distance map and the fourth spatial position is greater than or equal to a second preset distance; if it is greater than or equal to the second preset distance, mark the posture data corresponding to the laser radar point cloud data in the world coordinates at the fourth spatial position as abnormal posture data; if it is less than the second preset distance, mark the posture data corresponding to the laser radar point cloud data in the world coordinates at the fourth spatial position as normal posture data.

[0102] Specifically, the processed lidar point cloud data DCi in each frame is converted from the body coordinate system to the world coordinate system, so as to obtain the lidar point cloud data WCi in the world coordinate system, that is, WCi=Ti⊙DCi, WCi represents the point cloud in the world coordinate system, Ti is the posture data, and the relationship between the body coordinate system and the world coordinate system can be described specifically by the rotation matrix and the translation vector.

[0103] The laser radar point cloud data WCi in the world coordinate system has a fourth spatial position. First, a fixed threshold (for example, 120) is used to convert the digital grid map in the map information into a binary image, and then the binary image is transformed into a distance map by a method such as Euclidean distance transformation, wherein the value of each point on the distance map represents the distance between the point and the nearest obstacle.

[0104] Match the fourth spatial position of WCi with the distance map in the map information, and determine whether the distance between the nearest obstacle in the distance map and the fourth spatial position is greater than or equal to the second preset distance. In this embodiment, the second preset distance is set to 0.1 meters; in other embodiments, the second preset distance can also be defined as other values, which can be set according to the actual positioning accuracy requirements. That is, when it is greater than or equal to 0.1 meters, the posture data Ti corresponding to WCi is marked as abnormal posture data fTi; when it is less than 0.1 meters, the posture data Ti corresponding to WCi is marked as normal posture data. This step is mainly to determine whether the posture is abnormal by calculating the distance between the laser point and the obstacle, because if the posture is correct, the laser beam emitted by the laser radar of the food delivery robot should be irradiated on the obstacle, and point cloud data WCi is formed at the position corresponding to the obstacle or in the range that meets the preset distance, and this distance should be very small.

[0105] S35. Calculate whether the posture data of different detection points in the same frame are abnormal posture data, and count the third ratio of the number of abnormal posture data to the number of all posture data in the frame. If the third ratio exceeds the fifth preset threshold, the posture of the robot in the frame is determined to be an abnormal posture; if the third ratio is lower than the fifth preset threshold, the posture of the robot in the frame is determined to be a normal posture.

[0106] Specifically, the food delivery robot includes multiple detection points in each frame, for example, at least one detection point is set at the base, arm, and rotating joint of the food delivery robot, and the posture data of each detection point in each frame are different. The judgment method of the present invention determines whether the overall posture of the robot in each frame is a normal posture by calculating the ratio of the number of abnormal posture detection points in each frame to the total detection points in the frame. The ratio is defined as a third ratio. For example, when the third ratio exceeds 0.75, that is, among all the detection points in a certain frame, if the number of detection points determined to be abnormal postures accounts for more than 0.75, then the posture of the food delivery robot in the frame is considered to be an abnormal posture; if among all the detection points in the frame, the number of detection points determined to be abnormal postures accounts for less than 0.75, then the posture of the food delivery robot in the frame is considered to be a normal posture.

[0107] S36. Determine whether the posture of the robot in different frames is an abnormal posture respectively, and count the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot is the third ratio.

[0108] Specifically, when the food delivery robot traverses every corner of the restaurant, its laser radar obtains posture data and posture information in multiple frames. The judgment method of the present invention judges the posture in each frame, judges and counts whether the posture in each frame is an abnormal posture or a normal posture, and counts the proportion of abnormal postures in all postures, and the proportion is defined as the third ratio S3.

[0109] In a specific implementation, the step S36 further includes the following steps:

[0110] S37. Calculate a matching result value according to the first ratio and the third ratio, wherein the matching result value is in direct proportion to the first ratio and the third ratio respectively; wherein, the smaller the matching result value is, the better the positioning quality of the robot in the predetermined area is.

[0111] Specifically, the judgment method of the present invention uses two schemes for judging abnormal postures. The first scheme is to judge whether the posture is an abnormal posture by connecting the laser point and the posture. The second scheme is to judge whether the posture is an abnormal posture by the distance between the laser point and the obstacle. The positioning quality of the robot is evaluated by associating the ratios of the number of two abnormal postures to the total number of postures. The positioning quality judgment method is objective and accurate, and is applied to the positioning quality evaluation of the robot on-site deployment link, which can improve the quality of robot deployment. It can also be used in the development and iteration of the positioning algorithm, so as to evaluate whether the iteration of the positioning algorithm is effective.

[0112] In a specific implementation, the matching result value score is in direct proportion to the first ratio S1 and the third ratio S3, that is, the numerical value of score is in direct proportion to the values ​​of S1 and S3. When the values ​​of S1 and S3 are large, the numerical value of score is also large; when the values ​​of S1 and S3 are small, the numerical value of score is also small, that is, score = λ1*S1+λ2*S3, where 0<λ1<1, 0<λ2<1, and λ1+λ2=1. In this embodiment, λ1 is 0.6 and λ2 is 0.4; in other embodiments, the values ​​of λ1 and λ2 can also be other values, which can be specifically set according to the actual positioning accuracy requirements. When the numerical value of score is small, it is determined that the positioning quality of the food delivery robot in the restaurant is good; when the numerical value of score is large, it is determined that the positioning quality of the food delivery robot in the restaurant is poor. Specifically, a reference threshold value may be set for the score value. When the score value exceeds the reference threshold value, the positioning data of the food delivery robot needs to be corrected or repositioned.

[0113] refer to Figure 2 The present invention also provides a robot positioning quality judgment system, which corresponds to the above-mentioned robot positioning quality judgment method. The robot positioning quality judgment system includes a positioning data acquisition module 100, a map information acquisition module 200, a matching module 300 and a positioning quality judgment module 400.

[0114] The positioning data acquisition module 100 is used to obtain the positioning data of the robot in a predetermined area according to the positioning module of the robot, and process the positioning data to obtain processed positioning data, wherein the positioning module is a laser positioning module.

[0115] The map information acquisition module 200 is used to acquire the map information of the robot in the predetermined area.

[0116] The matching module 300 is used to match the processed positioning data with the map information and obtain a matching result.

[0117] The positioning quality determination module 400 is used to determine the positioning quality of the robot in the predetermined area according to the matching result.

[0118] The robot positioning quality judgment system of the present invention matches the positioning data obtained by the robot's positioning module with the robot's map information, and judges the robot's positioning quality based on the matching result. Compared with the use of an independent high-precision positioning system in the prior art, the judgment system of the present invention does not require any additional devices, which not only reduces costs but also saves space. It is applied to the positioning quality evaluation of the robot on-site deployment link to improve the quality of robot deployment.

[0119] See also Figure 1 , Figure 3 The present invention also provides a robot, comprising:

[0120] One or more processors 10, memory 20; and

[0121] One or more programs, wherein the one or more programs are stored in the memory 20 and executed by the one or more processors 10, and the programs include instructions for executing any of the above-mentioned methods for determining the positioning quality of the robot.

[0122] The memory 20 is used to store a computer program that can be run on the processor 10. When the processor 10 executes the program, the method for determining the robot positioning quality in any of the above-mentioned embodiments is implemented.

[0123] The memory 20 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory. Further, the robot may also include a communication interface 30, which is used for communication between the memory 20 and the processor 10.

[0124] If the memory 20, the processor 10 and the communication interface 30 are implemented independently, the communication interface 30, the memory 20 and the processor 10 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0125] Optionally, in a specific implementation, if the memory 20, the processor 10 and the communication interface 30 are integrated on a chip, the memory 20, the processor 10 and the communication interface 30 can communicate with each other through an internal interface.

[0126] The processor 10 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0127] See also Figure 4 The present invention also provides a non-volatile computer-readable storage medium 500 containing a computer program. When the computer program 501 is executed by one or more processors 10, the processor 10 executes any of the above-mentioned methods for determining the robot positioning quality.

[0128] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, results, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, results, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0129] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and intent of the present invention.

Claims

1. A method for judging the positioning quality of a robot, characterized in that: The following steps are involved: Using a positioning module of the robot to obtain positioning data of the robot in a predetermined area, and processing the positioning data to obtain processed positioning data, wherein the positioning module is a laser positioning module; Acquiring map information of the robot in the predetermined area; Matching the processed positioning data with the map information to obtain a matching result; Determining the positioning quality of the robot in the predetermined area according to the matching result; The specific steps of using the positioning module of the robot to obtain the positioning data of the robot in a predetermined area, and processing the positioning data to obtain the processed positioning data include: obtaining the posture data and laser radar point cloud data of each frame of the robot during operation through the laser positioning module on the robot, wherein the posture data and the laser radar point cloud data under the same frame correspond one to one; respectively summarizing the posture data and the laser radar point cloud data under each frame to obtain the overall posture data and the overall laser radar point cloud data, and then successively subjecting the overall laser radar point cloud data to voxel filtering and density-based clustering algorithm processing to obtain the processed laser radar point cloud data; The specific steps of matching the processed positioning data with the map information to obtain a matching result include: converting the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, the laser radar point cloud data in the world coordinate system has a first spatial position, and the posture data corresponding to the laser radar point cloud data in the world coordinate system in the same frame has a second spatial position, connecting the first spatial position and the second spatial position to form a spatial straight line segment; mapping the spatial straight line segment to the digital grid map in the map information; determining whether the spatial straight line segment collides with an obstacle in the digital grid map, and if a collision occurs, mapping the spatial straight line segment to the digital grid map; and The posture data corresponding to the spatial straight line segments that collide with the grid map are marked as abnormal posture data; the posture data of different detection points in the same frame are calculated respectively to see whether they are abnormal posture data, and the first ratio of the number of abnormal posture data to the number of all posture data in the frame is counted; if the first ratio exceeds the first preset threshold, the posture of the robot in the frame is determined to be an abnormal posture; if the first ratio is lower than the first preset threshold, the posture of the robot in the frame is determined to be a normal posture; whether the posture of the robot in different frames is an abnormal posture is determined respectively, and the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot is counted, wherein the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot is the first ratio.

2. The method for determining the robot positioning quality according to claim 1, characterized in that: The specific step of judging the positioning quality of the robot in the predetermined area according to the matching result comprises: The positioning quality of the robot in the predetermined area is judged according to the first ratio, wherein if the first ratio is greater than or equal to a second preset threshold, it is determined that the positioning quality of the robot in the predetermined area is poor; if the first ratio is less than the second preset threshold, it is determined that the positioning quality of the robot in the predetermined area is good.

3. The method for determining the robot positioning quality according to claim 1, characterized in that: The specific steps of matching the processed positioning data with the map information to obtain a matching result include: Converting the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a third spatial position; Using a first fixed threshold, converting the digital grid map in the map information into a binary map, and then converting the binary map into a distance map; Match the third spatial position with the distance map, and determine whether the distance between the nearest obstacle in the distance map and the third spatial position is greater than or equal to a first preset distance; if it is greater than or equal to the first preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the third spatial position as abnormal pose data; if it is less than the first preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the third spatial position as normal pose data; Calculate whether the posture data of different detection points in the same frame are abnormal posture data, and count the second ratio of the number of abnormal posture data to the number of all posture data in the frame, if the second ratio exceeds the third preset threshold, determine that the posture of the robot in the frame is an abnormal posture; if the second ratio is lower than the third preset threshold, determine that the posture of the robot in the frame is a normal posture; Determine respectively whether the posture of the robot in different frames is an abnormal posture, and count the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot is the second ratio.

4. The method for determining the robot positioning quality according to claim 3, characterized in that: The specific step of judging the positioning quality of the robot in the predetermined area according to the matching result comprises: The positioning quality of the robot in the predetermined area is judged according to the second ratio, wherein if the second ratio is greater than or equal to a fourth preset threshold, it is determined that the positioning quality of the robot in the predetermined area is poor; if the second ratio is less than the fourth preset threshold, it is determined that the positioning quality of the robot in the predetermined area is good.

5. The method for determining the robot positioning quality according to claim 1, characterized in that: The specific step of matching the processed positioning data with the map information to obtain a matching result also includes: Converting the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, wherein the laser radar point cloud data in the world coordinate system has a fourth spatial position; Using a second fixed threshold value to convert the digital grid map in the map information into a binary map, and then converting the binary map into a distance map; Match the fourth spatial position with the distance map, and determine whether the distance between the nearest obstacle in the distance map and the fourth spatial position is greater than or equal to a second preset distance; if it is greater than or equal to the second preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the fourth spatial position as abnormal pose data; if it is less than the second preset distance, mark the pose data corresponding to the laser radar point cloud data in the world coordinates of the fourth spatial position as normal pose data; Calculate whether the posture data of different detection points in the same frame are abnormal posture data, and count the third ratio of the number of abnormal posture data to the number of all posture data in the frame, if the third ratio exceeds the fifth preset threshold, determine that the posture of the robot in the frame is an abnormal posture; if the third ratio is lower than the fifth preset threshold, determine that the posture of the robot in the frame is a normal posture; Determine whether the posture of the robot in different frames is an abnormal posture, and count the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot, wherein the ratio of the number of abnormal postures of the robot in different frames to all the postures of the robot is the third ratio.

6. The method for determining the robot positioning quality according to claim 5, characterized in that: The specific step of judging the positioning quality of the robot in the predetermined area according to the matching result comprises: Calculating a matching result value according to the first ratio and the third ratio, wherein the matching result value is in direct proportion to the first ratio and the third ratio respectively; The smaller the matching result value is, the better the positioning quality of the robot in the predetermined area is.

7. The method for determining the robot positioning quality according to claim 1, characterized in that: The specific steps of successively subjecting the overall laser radar point cloud data to voxel filtering and density-based clustering algorithm processing to obtain processed laser radar point cloud data include: Using a grid of a certain size to convert the overall lidar point cloud data into filtered lidar point cloud data through a voxel filtering algorithm; Then, a density-based clustering algorithm is used to cluster different points in the filtered lidar point cloud data into multiple clusters of different sizes; the number of points in each cluster is determined respectively, and if the number is less than a certain number, all points in the cluster with a number less than a certain number are deleted to obtain the processed lidar point cloud data.

8. A robot positioning quality judgment system, characterized in that: include: A positioning data acquisition module, used to obtain positioning data of the robot in a predetermined area using a positioning module of the robot, and process the positioning data to obtain processed positioning data, wherein the positioning module is a laser positioning module; A map information acquisition module, used to acquire map information of the robot in the predetermined area; a matching module, used to match the processed positioning data with the map information and obtain a matching result; and A positioning quality determination module is used to determine the positioning quality of the robot in the predetermined area according to the matching result; the specific steps of using the positioning module of the robot to obtain the positioning data of the robot in the predetermined area, and processing the positioning data to obtain the processed positioning data include: obtaining the posture data and laser radar point cloud data of each frame of the robot during operation through the laser positioning module on the robot, wherein the posture data and the laser radar point cloud data under the same frame correspond one to one; respectively summarizing the posture data and the laser radar point cloud data under each frame to obtain the overall posture data and the overall laser radar point cloud data, and then successively subjecting the overall laser radar point cloud data to voxel filtering and density-based clustering algorithm processing to obtain the processed laser radar point cloud data; The specific steps of matching the processed positioning data with the map information to obtain a matching result include: converting the processed laser radar point cloud data from the body coordinate system to the world coordinate system to obtain the laser radar point cloud data in the world coordinate system, the laser radar point cloud data in the world coordinate system has a first spatial position, and the posture data corresponding to the laser radar point cloud data in the world coordinate system in the same frame has a second spatial position, connecting the first spatial position and the second spatial position to form a spatial straight line segment; mapping the spatial straight line segment to the digital grid map in the map information; determining whether the spatial straight line segment collides with an obstacle in the digital grid map, and if a collision occurs, mapping the spatial straight line segment to the digital grid map; and The posture data corresponding to the spatial straight line segments that collide with the grid map are marked as abnormal posture data; the posture data of different detection points in the same frame are calculated respectively to see whether they are abnormal posture data, and the first ratio of the number of abnormal posture data to the number of all posture data in the frame is counted; if the first ratio exceeds the first preset threshold, the posture of the robot in the frame is determined to be an abnormal posture; if the first ratio is lower than the first preset threshold, the posture of the robot in the frame is determined to be a normal posture; whether the posture of the robot in different frames is an abnormal posture is determined respectively, and the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot is counted, wherein the ratio of the number of abnormal postures of the robot in different frames to the number of all postures of the robot is the first ratio.

9. A robot, characterized in that: include: One or more processors, memory; and One or more programs, wherein the one or more programs are stored in the memory and executed by the one or more processors, the programs comprising instructions for executing the method for determining the robot positioning quality according to any one of claims 1 to 7.

10. A non-volatile computer-readable storage medium containing a computer program, which, when executed by one or more processors, enables the processors to execute the method for determining the robot positioning quality according to any one of claims 1 to 7.

Images