Local map-based confidence determination method and apparatus, device, and storage medium

Abstract

This disclosure provides a method, apparatus, device, and storage medium for determining confidence levels based on local maps, relating to the field of artificial intelligence, particularly to autonomous driving and mapping technologies. The specific implementation involves: determining relative pose information between multiple local maps; the relative pose information includes the initial relative poses of the multiple local maps before point cloud registration processing, and the current relative poses after point cloud registration processing; determining relative pose errors based on the relative pose information and point cloud data; the relative pose errors include the relative pose errors of the initial and current relative poses; determining confidence level information based on the relative pose errors of the initial and current relative poses; and the confidence level information indicating the accuracy of the point cloud registration processing performed on the multiple local maps. By utilizing the relative poses before and after registration to examine the confidence level of the point cloud registration results, the accuracy of confidence level determination is improved, thereby improving mapping accuracy.

Description

Technical Field

[0001] This disclosure relates to autonomous driving technology and mapping technology in artificial intelligence, and more particularly to a method, apparatus, device and storage medium for determining confidence based on local maps. Background Technology

[0002] The vehicle is equipped with a lidar system, which can be used to collect point cloud data for mapping while the vehicle is in motion.

[0003] In the process of mapping using LiDAR, the accuracy of the map depends on the accuracy of point cloud registration. Errors in point cloud registration can severely affect the quality of the map. Therefore, it is necessary to accurately determine the accuracy of the point cloud registration process. Summary of the Invention

[0004] This disclosure provides a method, apparatus, device, and storage medium for determining confidence based on local maps.

[0005] According to a first aspect of this disclosure, a confidence determination method based on a local map is provided, comprising:

[0006] Determine the relative pose information between at least two local maps; wherein the local maps are used to represent maps constructed from point cloud data collected within a preset time period; the relative pose information includes the initial relative pose of at least two local maps before point cloud registration processing, and the current relative pose of at least two local maps after point cloud registration processing.

[0007] The relative pose error is determined based on the relative pose information and the point cloud data in the local map; wherein the relative pose error includes the relative pose error of the initial relative pose and the relative pose error of the current relative pose.

[0008] Confidence information is determined based on the relative pose error of the initial relative pose and the relative pose error of the current relative pose; wherein the confidence information is used to indicate the accuracy of point cloud registration processing performed on at least two local maps.

[0009] According to a second aspect of this disclosure, a confidence determination apparatus based on a local map is provided, comprising:

[0010] The first determining unit is used to determine the relative pose information between at least two local maps; wherein the local map is used to represent a map constructed from point cloud data collected within a preset time period; the relative pose information includes the initial relative pose of the at least two local maps before point cloud registration processing, and the current relative pose of the at least two local maps after point cloud registration processing.

[0011] The second determining unit is used to determine the relative pose error based on the relative pose information and the point cloud data in the local map; wherein the relative pose error includes the relative pose error of the initial relative pose and the relative pose error of the current relative pose.

[0012] The third determining unit is used to determine confidence information based on the relative pose error of the initial relative pose and the relative pose error of the current relative pose; wherein the confidence information is used to indicate the accuracy of point cloud registration processing performed on at least two local maps.

[0013] According to a third aspect of this disclosure, an electronic device is provided, comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the confidence determination method based on the local map described in the first aspect.

[0017] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing the computer to perform the confidence determination method based on a local map as described in the first aspect.

[0018] According to a fifth aspect of this disclosure, a computer program product is provided, the computer program product comprising: a computer program stored in a readable storage medium, wherein at least one processor of an electronic device can read the computer program from the readable storage medium, and the at least one processor executes the computer program to cause the electronic device to perform the confidence determination method based on a local map as described in the first aspect.

[0019] According to the technology disclosed herein, the relative pose information of multiple local maps before and after registration is determined. Based on the point cloud data in the local maps, the relative pose error of the relative pose information before and after registration is determined. Based on the relative pose error of the relative pose information before and after registration, confidence information is determined, that is, the accuracy of the point cloud registration process is determined. By using the relative pose before and after registration to examine the confidence of the point cloud registration result, the accuracy of confidence determination is improved, and the accuracy of the point cloud registration process is accurately determined.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0021] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0022] Figure 1 This is a flowchart illustrating a confidence determination method based on a local map according to an embodiment of the present disclosure;

[0023] Figure 2 This is a flowchart illustrating a confidence determination method based on a local map according to an embodiment of the present disclosure;

[0024] Figure 3 This is a flowchart illustrating a confidence determination method based on a local map according to an embodiment of the present disclosure;

[0025] Figure 4 This is a structural block diagram of a confidence determination device based on a local map, according to an embodiment of the present disclosure.

[0026] Figure 5 This is a structural block diagram of a confidence determination device based on a local map, according to an embodiment of the present disclosure.

[0027] Figure 6 This is a structural block diagram of an electronic device used to implement a method for determining confidence based on a local map according to an embodiment of this disclosure;

[0028] Figure 7 This is a structural block diagram of an electronic device used to implement a method for determining confidence based on a local map, according to an embodiment of this disclosure. Detailed Implementation

[0029] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0030] In the process of mapping using LiDAR, the accuracy of the map depends on the accuracy of point cloud registration. However, the registration result may get trapped in a local optimum or produce registration errors due to errors in point cloud correction. In addition, in some noisy or insufficiently constrained scenarios, the reliability of point cloud registration results will also decrease, resulting in ghosting in the constructed point cloud map, affecting the quality of mapping, and consequently affecting the accuracy and safety of autonomous driving.

[0031] Therefore, it is necessary to determine the confidence level of the point cloud registration results. For example, the root mean square error (RMSE) between points of a registered point cloud pair can be used to judge the quality of the registration. However, RMSE is easily affected by outliers, and different point cloud structures in different environments can lead to significant differences in the calculation results. The accuracy of determining the confidence level of point cloud registration is low, making it difficult to determine whether the point cloud registration is correct. Another example is to predict the risk of registration failure before registration. However, predicting risk before registration does not assess the reliability of the registration result, and it cannot guarantee the accuracy of the point cloud registration, affecting mapping accuracy and consequently impacting the safety of vehicles in autonomous driving scenarios.

[0032] This disclosure provides a method, apparatus, device, and storage medium for determining confidence based on local maps, which can be applied to autonomous driving technology and mapping technology in the field of artificial intelligence to achieve accurate determination of the confidence of point cloud registration.

[0033] It should be noted that the point cloud data and local maps used in this disclosure are not point cloud data and local maps for a specific scenario and cannot reflect a specific environmental scenario.

[0034] To help readers gain a deeper understanding of the implementation principles of this disclosure, the following will be discussed in conjunction with... Figures 1-7 The illustrated embodiments are further refined.

[0035] Figure 1 This is a flowchart illustrating a confidence determination method based on a local map according to an embodiment of the present disclosure. This method can be executed by a confidence determination device based on a local map. Figure 1 As shown, the method includes the following steps:

[0036] S101. Determine the relative pose information between at least two local maps; wherein, the local map is used to represent the map constructed from point cloud data collected within a preset time period; the relative pose information includes the initial relative pose of at least two local maps before point cloud registration processing, and the current relative pose of at least two local maps after point cloud registration processing.

[0037] For example, a LiDAR can acquire point cloud data in real time. For instance, a single scan by the LiDAR can capture one frame of point cloud data. By acquiring point cloud data over a continuous period, such as one minute, the LiDAR can collect point cloud data within that minute. A map is then constructed based on this point cloud data, serving as a local map within the global map. This local map can be a submap. The global map is the map of the entire scene corresponding to the point cloud data collected by the LiDAR, while the local map is a portion of the global map. For example, the global map might be the overall map of a street, while the local map might be the map of the area within ten meters of that street. A local map can contain point cloud data collected by the LiDAR at multiple times; that is, a single local map can include point cloud data collected at multiple times.

[0038] Multiple local maps can be constructed based on point cloud data collected within a preset time period. For example, point cloud data collected over two consecutive minutes can be used to construct a local map based on the first minute's data, and another local map based on the second minute's data. That is, each local map represents a different scene. For instance, the first local map might represent the scene within 0 to 5 meters of the road, while the second local map might represent the scene within 5 to 10 meters of the road. The LiDAR can move forward while collecting point cloud data, and since the point cloud data is collected at different times in each local map, the scene content represented by each local map will differ.

[0039] The relative pose information between at least two local maps is calculated, and the relative pose information is the relative pose between each local map. This disclosure does not specifically limit the algorithm for determining the relative pose information. For example, a pairwise method can be used to calculate the relative pose between two local maps, or a multiview method can be used to calculate the relative pose between three local maps.

[0040] Before performing point cloud registration between multiple local maps, the relative pose information between the local maps can be calculated as the initial relative pose. After performing point cloud registration between the multiple local maps, the relative pose information between the registered local maps is calculated as the current relative pose. That is, the relative pose information can include the initial relative pose of at least two local maps before point cloud registration, and the current relative pose of at least two local maps after point cloud registration. In this embodiment, the point cloud registration processing algorithm is not specifically limited. For example, algorithms such as NDT (Normal Distribution Transformation), ICP (Iterative Closest Point), and GICP (Generalized-ICP) can be used for point cloud registration.

[0041] S102. Determine the relative pose error based on the relative pose information and the point cloud data in the local map; wherein, the relative pose error includes the relative pose error of the initial relative pose and the relative pose error of the current relative pose.

[0042] For example, the relative pose error of the initial relative pose can be determined based on the initial relative pose and all point cloud data in the local map, and the relative pose error of the current relative pose can be determined based on the current relative pose and all point cloud data in the local map.

[0043] A local map can include point cloud data from multiple time points, i.e., it can include multiple frames of point cloud data. A keyframe can be selected from each local map, and the relative pose error of the initial relative pose can be determined based on the initial relative pose and the point cloud data in each keyframe. Similarly, the relative pose error of the current relative pose can be determined based on the current relative pose and the point cloud data in each keyframe. The point cloud data collected by a LiDAR scan in one loop is called a frame. Keyframes can be selected according to actual needs; for example, a frame can be randomly selected as a keyframe.

[0044] The differential entropy of a local map can be determined based on the coordinates of each point cloud data point in the local map. Alternatively, the differential entropy of a keyframe can be determined based on the coordinates of its point cloud data point in the local map, and this can be used as the differential entropy of the local map. The differential entropy of the local map is then defined as the differential entropy of the point cloud data within the local map. The differential entropy of the local map can be the sum of the differential entropies of each point cloud data point in the local map. If the differential entropy of the local map is the differential entropy of a keyframe, then the differential entropy of the local map can be the sum of the differential entropies of each point cloud data point in the keyframe. In other words, the individual differential entropy information of each point cloud data point can be calculated first, and then the differential entropy of the point cloud data of the local map can be calculated.

[0045] After obtaining the differential entropy of the point cloud data of the local map, the relative pose error of the initial relative pose and the relative pose error of the current relative pose can be calculated separately. The relative pose error can be used to measure the accuracy of the relative pose before and after registration.

[0046] When determining the relative pose error, the separation differential entropy and joint differential entropy between multiple local maps can be determined. Subtracting the separation differential entropy from the joint differential entropy yields the relative pose error. The separation differential entropy information can represent the ratio of the sum of the differential entropies of the point cloud data from at least two local maps to the sum of the number of point cloud data points in at least two local maps. For example, if there are two local maps, the differential entropies of the point cloud data from these two local maps are added together, and the number of point cloud data points from these two local maps is added together. The two sums are then divided to obtain the separation differential entropy; alternatively, the differential entropies of the point cloud data from keyframes in the two local maps are added together, and the number of point cloud data points from the keyframes in the two local maps is added together. The two sums are then divided to obtain the separation differential entropy.

[0047] Joint differential entropy can be represented as the ratio of the differential entropy of the point cloud data after fusing point cloud data from at least two local maps to the sum of the number of point cloud data in the at least two local maps. For example, given two local maps, the point cloud data from these two local maps is fused into a single coordinate system. The differential entropy of the fused point cloud data is determined, and then the joint differential entropy is obtained by dividing the fused point cloud data differential entropy by the sum of the number of point cloud data in the two local maps. Alternatively, the point cloud data from the keyframes of these two local maps is fused into a single coordinate system. The differential entropy of the fused point cloud data is determined, and then the joint differential entropy is obtained by dividing the fused point cloud data differential entropy by the sum of the number of point cloud data in the keyframes of the two local maps.

[0048] Ideally, the separation differential entropy and joint differential entropy of the two local maps are equal. The greater the error in relative pose, the more significant the difference between the separation differential entropy and joint differential entropy. Therefore, the quality of point cloud fusion can be measured by evaluating the difference between the separation differential entropy and joint differential entropy, i.e., the relative pose error. Ideally, this means that the relative pose is true and the point cloud is noise-free.

[0049] S103. Determine confidence information based on the relative pose error of the initial relative pose and the relative pose error of the current relative pose; wherein, the confidence information is used to indicate the accuracy of the point cloud registration process performed on at least two local maps.

[0050] For example, relative pose error can reflect the accuracy of the relative pose of a point cloud to some extent; the smaller the relative pose error, the higher the accuracy of point cloud registration. However, due to changes in scene and point cloud noise information, judging the confidence level of a point cloud based on the magnitude of the relative pose error is unreliable. Therefore, the change in relative pose error before and after registration can be used to examine whether registration makes the relative pose more or worse, thereby effectively determining the confidence information of the point cloud registration process.

[0051] Confidence information can be used to indicate the accuracy of point cloud registration processing performed on at least two local maps. For example, the relative pose error of the initial relative pose can be subtracted from the relative pose error of the current relative pose. If the current relative pose obtained through registration is more accurate than the initial relative pose, then the relative pose error of the current relative pose is smaller, the resulting confidence information is positive, the registration effect is better, and the score is higher. Conversely, if the current relative pose obtained through registration is worse than the initial relative pose, then the relative pose error of the current relative pose is larger, the resulting confidence information is negative, the registration effect is worse, and the score is lower. In other words, the higher the confidence information value, the better the registration effect; the lower the confidence information value, the worse the registration effect.

[0052] If we subtract the relative pose error of the initial relative pose from the relative pose error of the current relative pose, then the higher the confidence information value, the worse the configuration effect; the lower the confidence information value, the better the configuration effect.

[0053] This disclosure determines the relative pose information of multiple local maps before and after registration, and determines the relative pose error of the relative pose information before and after registration based on the point cloud data in the local maps. Based on the relative pose error of the relative pose information before and after registration, confidence information is determined, that is, the accuracy of the point cloud registration process is determined. By using the relative pose before and after registration to examine the confidence of the point cloud registration result, the accuracy of confidence determination is improved, and the accuracy of the point cloud registration process is accurately determined.

[0054] Figure 2 This is a flowchart illustrating a confidence determination method based on a local map, which is an optional embodiment based on the above embodiments.

[0055] In this embodiment, the relative pose error is determined based on the relative pose information and the point cloud data in the local map. This can be further refined as follows: the differential entropy of the point cloud data in the local map is determined based on the point cloud data in the local map; wherein, the differential entropy of the point cloud data is used to represent the sum of the differential entropies of the point cloud data; the relative pose error of the initial relative pose is determined based on the initial relative pose and the differential entropy of the point cloud data in the local map; and the relative pose error of the current relative pose is determined based on the current relative pose and the differential entropy of the point cloud data in the local map.

[0056] like Figure 2 As shown, the method includes the following steps:

[0057] S201. Determine the relative pose information between at least two local maps; wherein, the local map is used to represent the map constructed from point cloud data collected within a preset time period; the relative pose information includes the initial relative pose of at least two local maps before point cloud registration processing, and the current relative pose of at least two local maps after point cloud registration processing.

[0058] For example, this step can refer to step S101 above, and will not be repeated here.

[0059] S202. Based on the point cloud data in the local map, determine the differential entropy of the point cloud data in the local map; wherein, the differential entropy of the point cloud data is used to represent the sum of the differential entropies of the point cloud data.

[0060] For example, the differential entropy information of each point cloud data in the local map is determined, and the differential entropy information of each point cloud data is added together to obtain the differential entropy of the point cloud data in the local map. Alternatively, a frame can be selected from the local map as a keyframe, the differential entropy information of each point cloud data in the keyframe is determined, and the differential entropy information of each point cloud data in the keyframe is added together to obtain the differential entropy of the point cloud data in the local map.

[0061] By converting point cloud data into differential entropy, the sensitivity to noise and outliers in the point cloud data can be reduced, thereby improving the robustness of the confidence information determination process.

[0062] In this embodiment, determining the differential entropy of the point cloud data in the local map based on the point cloud data in the local map includes: determining the neighboring point cloud data of the point cloud data in the local map according to a preset radius range; determining the differential entropy information of the point cloud data in the local map based on the coordinate information of the neighboring point cloud data in the local map; wherein, the differential entropy information represents the differential entropy corresponding to the coordinate information of the point cloud data; and determining the differential entropy of the point cloud data in the local map based on the differential entropy information of the point cloud data in the local map.

[0063] Specifically, a local map is denoted as Or, a keyframe in a local map can be denoted as , The point cloud data in .from Choose any point Preset a certain radius range, so as to Using a circle as the center, determine the point cloud data within a preset radius, as... The neighboring point cloud data. Determine the coordinate information of the neighboring point cloud data; this coordinate information can be the coordinates of the neighboring point cloud data within the coordinate system of the local map. Based on... The coordinate information of each neighboring point cloud data in the local map is used to determine Differential entropy information can represent point cloud data. The differential entropy corresponding to the coordinate information. After obtaining... Point cloud data After obtaining the differential entropy information, we can get The differential entropy of point cloud data. For example, it can be used to differentiate the entropy of each point cloud data. Add the differential entropy information together to obtain The differential entropy of point cloud data.

[0064] The advantage of this setup is that it utilizes the differential entropy information of point cloud data to determine the reliability of point cloud registration. By determining the differential entropy information of each point cloud data point, the differential entropy of the point cloud data for the local map is obtained, ensuring overall consideration of point cloud registration for the local map. Furthermore, by converting point cloud data into differential entropy, the stability advantage of differential entropy can be leveraged to reduce sensitivity to noise and outliers in the point cloud data, thereby improving the robustness of confidence determination.

[0065] In this embodiment, the differential entropy information of the point cloud data in the local map is determined based on the coordinate information of the neighboring point cloud data in the local map, including: determining the covariance information of the point cloud data in the local map based on the coordinate information of the neighboring point cloud data in the local map; and determining the differential entropy information of the point cloud data in the local map based on the covariance information of the point cloud data in the local map.

[0066] Specifically, a coordinate system is established in each local map. This can be done by constructing a Cartesian coordinate system with the LiDAR as the origin, serving as the coordinate system for the local map. The coordinate information of each point cloud data point within each local map is then determined, that is, the coordinates of each point cloud data point within its respective local map. Based on... The coordinate information of the neighboring point cloud data is used to calculate the coordinates of the neighboring point cloud data. The covariance information. A covariance determination formula can be preset; the covariance determination formula can be:

[0067] ;

[0068] in, Represented as Covariance information, The coordinates of the neighboring point cloud data are represented by N, where N is the number of neighboring point cloud data. Represented as The average coordinates of the neighboring point cloud data.

[0069] In obtaining After obtaining the covariance information, the differential entropy information can be calculated using a pre-defined formula to obtain... The differential entropy information. The formula for calculating differential entropy information can be:

[0070] ;

[0071] in, Represented in a local map The differential entropy information.

[0072] The beneficial effect of this setup is that it calculates the covariance using the coordinates of the point cloud data and transforms the covariance into differential entropy, thereby reducing the sensitivity to noise and outliers in the point cloud data, avoiding interference from the external environment, and improving the accuracy of confidence determination.

[0073] In this embodiment, the differential entropy of the point cloud data in the local map is determined based on the differential entropy information of the point cloud data in the local map, including: accumulating the differential entropy information of the point cloud data in the local map to obtain the differential entropy of the point cloud data in the local map.

[0074] Specifically, it is possible to determine the point cloud data of each part of the local map. The covariance information is then used to obtain the point cloud data. The differential entropy information. Based on point cloud data from a local map. The differential entropy information is used to obtain the differential entropy of the point cloud data for the entire local map. If multiple local maps exist, the differential entropy of the point cloud data for each local map can be determined.

[0075] For a local map, the differential entropy of point cloud data can be the individual point cloud data in that local map. The sum of differential entropy information. That is, the differential entropy information of the point cloud data in the local map is accumulated, and the accumulated result is the differential entropy of the point cloud data of the local map. For example, there exists a local map. and local maps Local map Midpoint Cloud Data The differential entropy information is Local map Zhongdian Cloud Data The differential entropy information is Local map The differential entropy of the point cloud data is Local map The differential entropy of the point cloud data is The formula for calculating the differential entropy of point cloud data can be:

[0076] ;

[0077] in, Represents a local map Differential entropy of point cloud data, For local maps The amount of point cloud data. If the local map is... ,but for If the local map is ,but for .

[0078] The advantage of this setting is that it allows for the calculation of the differential entropy of the local map as a whole, which facilitates a comprehensive consideration of the accuracy of point cloud registration and improves the efficiency of determining the accuracy of point cloud registration.

[0079] S203. Determine the relative pose error of the initial relative pose based on the differential entropy of the point cloud data of the local map; and determine the relative pose error of the current relative pose based on the differential entropy of the point cloud data of the current map.

[0080] For example, the relative pose error of the initial relative pose and the relative pose error of the current relative pose are calculated separately. The relative pose error of the initial relative pose is determined based on the initial relative pose and the differential entropy of the point cloud data for each local map, and the relative pose error of the current relative pose is determined based on the current relative pose and the differential entropy of the point cloud data for each local map. For example, there are two local maps, namely... and Then, based on the initial relative pose, and Calculate the relative pose error of the initial relative pose; based on the current relative pose, and Calculate the relative pose error of the current relative pose.

[0081] In this embodiment of the disclosure, the calculation order of the relative pose error of the initial relative pose and the relative pose error of the current relative pose is not specifically limited.

[0082] In this embodiment, determining the relative pose error of the initial relative pose based on the differential entropy of the point cloud data of the local map includes: determining the separation differential entropy information between at least two local maps based on the differential entropy of the point cloud data of the local map and the number of point cloud data in the local map; wherein, the separation differential entropy information represents the ratio of the sum of the differential entropy of the point cloud data of at least two local maps to the sum of the number of point cloud data in at least two local maps; fusing the point cloud data in the local map based on the initial relative pose to obtain first fused point cloud data; determining the first joint differential entropy information between at least two local maps based on the first fused point cloud data and the number of point cloud data in the local map; wherein, the first joint differential entropy information represents the magnitude of the joint differential entropy of at least two local maps before registration; and determining the relative pose error of the initial relative pose based on the separation differential entropy information and the first joint differential entropy information.

[0083] Specifically, based on the differential entropy of the point cloud data in each local map and the number of point cloud data in each local map, the separation differential entropy information between at least two local maps is determined. For example, the local map includes... and ,according to , , and It can be calculated and The separation differential entropy information. Separation differential entropy information can be expressed as the ratio of the sum of the differential entropies of point cloud data from at least two local maps to the sum of the number of point cloud data points in at least two local maps. The formula for calculating separation differential entropy information can be:

[0084] ;

[0085] in, express and The information of separation differential entropy between them.

[0086] Based on the initial relative pose, the joint differential entropy between the local maps before point cloud registration is determined and used as the first joint differential entropy information. The joint differential entropy information can represent the ratio between the differential entropy of the fused point cloud data after the point cloud data from multiple local maps are fused into one coordinate system and the sum of the number of point clouds in the multiple local maps. The first joint differential entropy information represents the magnitude of the joint differential entropy of the multiple local maps before registration.

[0087] For each partial map The point cloud data is fused using the initial relative pose. Point cloud fusion refers to combining point cloud data from various local maps into a single coordinate system. For example, a local map may include... and Based on the initial relative pose, Point cloud data from the middle is fused into In the coordinate system, or Point cloud data from the middle is fused into In the coordinate system, all point cloud data in the fused coordinate system are identified as the first fused point cloud data, that is, the first fused point cloud data represents the fused point cloud data before registration. Based on the first fused point cloud data and the number of point cloud data in each local map, and based on the preset joint differential entropy information calculation formula, the first joint differential entropy information between at least two local maps is determined.

[0088] The relative pose error of the initial relative pose is calculated based on the separated differential entropy information and the first joint differential entropy information. For example, the relative pose error of the initial relative pose can be obtained by subtracting the separated differential entropy information from the first joint differential entropy information.

[0089] The advantage of this setup is that it determines the separation differential entropy information and the first joint differential entropy information, thereby obtaining the error of the relative pose before registration. This facilitates the subsequent determination of the confidence level based on the error of the relative pose before and after registration, thus improving the accuracy of the confidence level determination.

[0090] In this embodiment, point cloud data in a local map is fused according to the initial relative pose to obtain first fused point cloud data, including: determining a target coordinate system from the coordinate systems of point cloud data from at least two local maps; and mapping all point cloud data of the local maps to the target coordinate system according to the initial relative pose to obtain first fused point cloud data in the target coordinate system.

[0091] Specifically, point cloud fusion is performed on multiple local maps, where the point cloud data in each local map resides in the coordinate system of its own local map. One local map is selected as the target map from the multiple local maps, and its coordinate system is used as the target coordinate system. Alternatively, any local map can be selected from the multiple local maps, and the coordinate system of the selected local map can be used as the target coordinate system.

[0092] Based on the initial relative pose, the point cloud data from each local map is fused into a single coordinate system, i.e., fused into the target coordinate system. The original point cloud data in the target coordinate system is already within the target coordinate system; therefore, it is necessary to map point cloud data that was not originally in the target coordinate system to the target coordinate system. For example, if there are two local maps, respectively... and Select If the coordinate system is the target coordinate system, then according to and The initial relative pose will The point cloud data is mapped to the target coordinate system. The fused point cloud data is denoted as... The fused point cloud data obtained by fusing based on the initial relative pose is determined as the first fused point cloud data, and the first fused point cloud data is denoted as... , .

[0093] The advantage of this setup is that it allows for the calculation of joint differential entropy information between local maps after the point cloud data is fused, thereby measuring the error in relative pose and improving the accuracy of confidence determination.

[0094] In this embodiment, determining the first joint differential entropy information between at least two local maps based on the number of point cloud data in the first fused point cloud data and the local map includes: determining the point cloud data differential entropy of the first fused point cloud data based on the first fused point cloud data; and determining the first joint differential entropy information between at least two local maps based on the number of point cloud data in the local map and the point cloud data differential entropy of the first fused point cloud data.

[0095] Specifically, based on the preset formula for calculating the differential entropy of point cloud data, the differential entropy of the first fused point cloud data is determined, and the differential entropy of the fused point cloud data is: The differential entropy of the first fused point cloud data is: .

[0096] The number of point cloud data points in each local map is determined. Based on the number of point cloud data points in each local map and the point cloud data differential entropy of the first fused point cloud data, the first joint differential entropy information among multiple local maps is determined. A pre-defined formula for calculating the joint differential entropy information is set, and the joint differential entropy information is denoted as... The first joint differential entropy information is denoted as The formula for calculating joint differential entropy information can be:

[0097] ;

[0098] in, This indicates the number of point cloud data points after fusion.

[0099] The beneficial effect of this setting is that it determines the first joint differential entropy information before point cloud registration, and by separating the difference between the differential entropy information and the first joint differential entropy information, it determines the error of the relative pose before point cloud registration, thereby determining whether the relative pose after registration is more accurate and improving the accuracy of determining the point cloud registration accuracy.

[0100] In this embodiment, determining the relative pose error of the current relative pose based on the differential entropy of the point cloud data of the local map includes: determining the separation differential entropy information between at least two local maps based on the differential entropy of the point cloud data of the local map and the number of point cloud data in the local map; fusing the point cloud data in the local map based on the current relative pose to obtain second fused point cloud data; determining the second joint differential entropy information between at least two local maps based on the second fused point cloud data and the number of point cloud data in the local map; wherein, the second joint differential entropy information characterizes the magnitude of the joint differential entropy of at least two local maps after registration; and determining the relative pose error of the current relative pose based on the separation differential entropy information and the second joint differential entropy information.

[0101] Specifically, based on the differential entropy of the point cloud data in each local map and the number of point cloud data in each local map, the separation differential entropy information between at least two local maps is determined. For example, the local map includes... and ,according to It can be calculated The separation differential entropy information. Separation differential entropy information can be expressed as the ratio of the sum of the differential entropies of point cloud data from at least two local maps to the sum of the number of point cloud data points in at least two local maps. The formula for calculating separation differential entropy information can be:

[0102] ;

[0103] in, The information of separation differential entropy between them.

[0104] Based on the current relative pose, the joint differential entropy between local maps after point cloud registration is determined as the second joint differential entropy information. The second joint differential entropy information can be expressed as the magnitude of the joint differential entropy of multiple local maps after registration.

[0105] For each partial map The point cloud data is fused using the current relative pose. All point cloud data in the fused coordinate system are identified as the second fused point cloud data, meaning the second fused point cloud data represents the fused point cloud data after registration. Based on the second fused point cloud data and the number of point cloud data in each local map, and using a preset joint differential entropy information calculation formula, the second joint differential entropy information between at least two local maps is determined.

[0106] The relative pose error of the current relative pose is calculated based on the separate differential entropy information and the second joint differential entropy information. For example, the relative pose error of the current relative pose can be obtained by subtracting the separate differential entropy information from the second joint differential entropy information.

[0107] The advantage of this setup is that it determines the separation differential entropy information and the second joint differential entropy information, thereby obtaining the error of the relative pose after registration. This facilitates the subsequent determination of the confidence level based on the error of the relative pose before and after registration, thus improving the accuracy of the confidence level determination.

[0108] In this embodiment, point cloud data in local maps are fused according to the initial relative pose to obtain second fused point cloud data, including: determining a target coordinate system from the coordinate systems of point cloud data from at least two local maps; and mapping all point cloud data of the local maps to the target coordinate system according to the current relative pose to obtain second fused point cloud data in the target coordinate system.

[0109] Specifically, point cloud fusion is performed on multiple local maps, with the point cloud data in each local map residing in its own coordinate system. One local map is selected as the target map from these multiple maps, and its coordinate system is then established as the target coordinate system. This allows us to determine the target coordinate system chosen before point cloud registration, and ultimately, to establish it as the target coordinate system after point cloud registration.

[0110] Based on the current relative pose, the point cloud data from each local map is merged into a single coordinate system, i.e., merged into the target coordinate system. The original point cloud data in the target coordinate system is already within the target coordinate system; therefore, it is necessary to map point cloud data that was not originally in the target coordinate system to the target coordinate system. For example, if there are two local maps, respectively... Select If the coordinate system is the target coordinate system, then according to The current relative pose will The point cloud data is mapped to the target coordinate system.

[0111] The merged point cloud data is denoted as The fused point cloud data, obtained by fusing based on the current relative pose, is defined as the second fused point cloud data, and is denoted as... .

[0112] The advantage of this setup is that it allows for the calculation of joint differential entropy information between local maps after the point cloud data is fused, thereby measuring the error in relative pose and improving the accuracy of confidence determination.

[0113] In this embodiment, determining the second joint differential entropy information between at least two local maps based on the number of point cloud data in the second fused point cloud data and the local map includes: determining the point cloud data differential entropy of the second fused point cloud data based on the second fused point cloud data; and determining the second joint differential entropy information between at least two local maps based on the number of point cloud data in the local map and the point cloud data differential entropy of the second fused point cloud data.

[0114] Specifically, based on the preset formula for calculating the differential entropy of point cloud data, the differential entropy of the second fused point cloud data is determined, and the differential entropy of the fused point cloud data is: The differential entropy of the second fused point cloud data is: .

[0115] The number of point cloud data points in each local map is determined. Based on the number of point cloud data points in each local map and the differential entropy of the point cloud data from the second fused point cloud data, the second joint differential entropy information among multiple local maps is determined. A pre-defined formula for calculating the joint differential entropy information is set, and the joint differential entropy information is denoted as... The second joint differential entropy information is denoted as .

[0116] The beneficial effect of this setting is that it determines the second joint differential entropy information after point cloud registration, and by separating the difference between the differential entropy information and the second joint differential entropy information, it determines the error of the relative pose after point cloud registration, thereby determining whether the relative pose after registration is more accurate and improving the accuracy of determining the point cloud registration accuracy.

[0117] The embodiments disclosed herein improve the robustness of determining the point cloud registration effect by reducing the sensitivity to noise and outliers in the point cloud data through the process of transforming differential entropy.

[0118] S204. Determine confidence information based on the relative pose error of the initial relative pose and the relative pose error of the current relative pose; wherein, the confidence information is used to indicate the accuracy of the point cloud registration process performed on at least two local maps.

[0119] For example, this step can refer to step S103 above, and will not be repeated here.

[0120] This disclosure determines the relative pose information of multiple local maps before and after registration, and determines the relative pose error of the relative pose information before and after registration based on the point cloud data in the local maps. Based on the relative pose error of the relative pose information before and after registration, confidence information is determined, that is, the accuracy of the point cloud registration process is determined. By using the relative pose before and after registration to examine the confidence of the point cloud registration result, the accuracy of confidence determination is improved, and the accuracy of the point cloud registration process is accurately determined.

[0121] Figure 3 This is a flowchart illustrating a confidence determination method based on a local map, which is an optional embodiment based on the above embodiments.

[0122] In this embodiment, the confidence information is determined based on the relative pose error of the initial relative pose and the relative pose error of the current relative pose. This can be further refined as follows: the confidence information is obtained by subtracting the relative pose error of the initial relative pose from the relative pose error of the current relative pose.

[0123] like Figure 3 As shown, the method includes the following steps:

[0124] S301. Determine the relative pose information between at least two local maps; wherein, the local map is used to represent the map constructed from point cloud data collected within a preset time period; the relative pose information includes the initial relative pose of at least two local maps before point cloud registration processing, and the current relative pose of at least two local maps after point cloud registration processing.

[0125] For example, this step can refer to step S101 above, and will not be repeated here.

[0126] S302. Determine the relative pose error based on the relative pose information and the point cloud data in the local map; wherein, the relative pose error includes the relative pose error of the initial relative pose and the relative pose error of the current relative pose.

[0127] For example, this step can refer to step S102 above, and will not be repeated here.

[0128] S303. Subtract the relative pose error of the initial relative pose from the relative pose error of the current relative pose, and the difference is the confidence information.

[0129] For example, relative pose error can be used to measure the accuracy of relative pose, but the relative pose error after registration alone cannot determine whether point cloud registration makes the relative pose more accurate or worse. Therefore, the relative pose error of the initial relative pose and the relative pose error of the current relative pose can be calculated. Based on the relative pose errors of the initial and current relative poses, the confidence information of the point cloud registration can be determined. The confidence information can represent the accuracy of the point cloud registration.

[0130] The relative pose error of the initial relative pose can be subtracted from the relative pose error of the current relative pose. A smaller relative pose error indicates more accurate point cloud registration. Therefore, if the relative pose error of the current relative pose is less than that of the initial relative pose, the point cloud registration is considered effective. In other words, if the current relative pose obtained through registration is more accurate than the initial relative pose, the relative pose error of the current relative pose is smaller, the registration confidence is reliable, and the resulting score is positive. The better the registration effect, the higher the confidence score. Conversely, if the current relative pose obtained through registration is worse than the initial relative pose, the relative pose error of the current relative pose is larger, and the corresponding score is negative. That is, the worse the registration effect, the lower the confidence score.

[0131] By comparing the relative pose error of the initial relative pose with the relative pose error of the current relative pose, the comparison result can effectively reflect the accuracy of the point cloud registration result when the relative pose error of the current relative pose fluctuates within a certain range. It is not affected by whether the registration result is globally optimal, and is widely applicable to different scenarios, thus having higher versatility.

[0132] In this embodiment, the method further includes: if the value represented by the confidence information is greater than a preset confidence threshold, then determining the weight of the current relative pose based on the confidence information; and optimizing the mapping process for the collected point cloud data based on the weight of the current relative pose to obtain a global map.

[0133] Specifically, after obtaining the confidence information, weights can be assigned to the corresponding current relative pose. For example, the higher the confidence value, the greater the assigned weight; the lower the confidence value, the smaller the assigned weight. A graph optimization algorithm can be pre-set, using the current relative pose as an edge in the graph optimization. The weight of the current relative pose is the weight of the edge in the graph optimization. Based on the edge weights in the graph optimization, the subsequent acquired pose graph is optimized using the graph optimization algorithm, i.e., the mapping process of the subsequently acquired point cloud data is optimized to obtain a global map. In this embodiment, the graph optimization algorithm is not specifically limited.

[0134] The confidence level of the point cloud registration result is determined, which means determining the accuracy of the point cloud registration. Different noise levels are set for the registration measurement information based on the determined result. In this embodiment, the measurement information is the current relative pose, and the noise is the weight. In the subsequent graph optimization process, the information matrix of the registration measurement can be adjusted accordingly; that is, the subsequent point cloud registration process is adjusted based on the current relative pose. This allows for greater utilization of the reliable registration result for pose optimization, i.e., optimization based on the current relative pose with a larger weight, avoiding large errors in the graph optimization result caused by poor registration.

[0135] A lower weight can be assigned to the current relative pose with lower confidence information. Alternatively, a confidence threshold can be preset, and the confidence information can be compared with the threshold. If the value represented by the confidence information is less than or equal to the preset confidence threshold, the current relative pose can be deleted; that is, the current relative pose is not sent to the graph optimization module, and no weight needs to be assigned to it. If the value represented by the confidence information is greater than the preset confidence threshold, the weight of the current relative pose can be determined according to the preset weight allocation rules.

[0136] The beneficial effect of this setting is that by comprehensively considering the joint differential entropy and the separate differential entropy, the confidence of the point cloud registration result can be examined by using the change in the difference between the relative poses before and after registration. During pose optimization, the confidence information can be effectively used to adjust the weight of the edges of the point cloud registration, thereby improving the effect of pose optimization and establishing a more accurate global map.

[0137] This disclosure determines the relative pose information of multiple local maps before and after registration, and determines the relative pose error of the relative pose information before and after registration based on the point cloud data in the local maps. Based on the relative pose error of the relative pose information before and after registration, confidence information is determined, that is, the accuracy of the point cloud registration process is determined. By using the relative pose before and after registration to examine the confidence of the point cloud registration result, the accuracy of confidence determination is improved, and the accuracy of the point cloud registration process is accurately determined.

[0138] Figure 4 This is a structural block diagram of a confidence determination device based on a local map, provided in an embodiment of this disclosure. For ease of explanation, only the parts relevant to the embodiments of this disclosure are shown. (Refer to...) Figure 4 The confidence determination device 400 based on local maps includes: a first determination unit 401, a second determination unit 402 and a third determination unit 403.

[0139] The first determining unit 401 is used to determine the relative pose information between at least two local maps; wherein the local map is used to represent a map constructed from point cloud data collected within a preset time period; the relative pose information includes the initial relative pose of the at least two local maps before point cloud registration processing, and the current relative pose of the at least two local maps after point cloud registration processing.

[0140] The second determining unit 402 is used to determine the relative pose error based on the relative pose information and the point cloud data in the local map; wherein the relative pose error includes the relative pose error of the initial relative pose and the relative pose error of the current relative pose.

[0141] The third determining unit 403 is used to determine confidence information based on the relative pose error of the initial relative pose and the relative pose error of the current relative pose; wherein the confidence information is used to indicate the accuracy of point cloud registration processing performed on at least two local maps.

[0142] Figure 5 A structural block diagram of a confidence determination device based on a local map provided in this disclosure embodiment is shown below. Figure 5 As shown, the confidence determination device 500 based on local maps includes a first determination unit 501, a second determination unit 502 and a third determination unit 503, wherein the second determination unit 502 includes a first determination module 5021 and a second determination module 5022.

[0143] The first determining module 5021 is used to determine the differential entropy of the point cloud data of the local map based on the point cloud data in the local map; wherein, the differential entropy of the point cloud data is used to represent the sum of the differential entropies of the point cloud data;

[0144] The second determining module 5022 is used to determine the relative pose error of the initial relative pose based on the initial relative pose and the differential entropy of the point cloud data of the local map; and to determine the relative pose error of the current relative pose based on the current relative pose and the differential entropy of the point cloud data of the local map.

[0145] In one example, the first determining module 5021 includes:

[0146] The first determining submodule is used to determine the neighboring point cloud data of the point cloud in the local map according to a preset radius range;

[0147] The second determining submodule is used to determine the differential entropy information of the point cloud data in the local map based on the coordinate information of the neighboring point cloud data in the local map; wherein, the differential entropy information represents the differential entropy corresponding to the coordinate information of the point cloud data;

[0148] The third determining submodule is used to determine the differential entropy of the point cloud data in the local map based on the differential entropy information of the point cloud data in the local map.

[0149] In one example, the second determined submodule is specifically used for:

[0150] Based on the coordinate information of the neighboring point cloud data in the local map, the covariance information of the point cloud data in the local map is determined; based on the covariance information of the point cloud data in the local map, the differential entropy information of the point cloud data in the local map is determined.

[0151] In one example, the third determining submodule is specifically used for:

[0152] The differential entropy information of the point cloud data in the local map is accumulated to obtain the differential entropy of the point cloud data in the local map.

[0153] In one example, the second determining module 5022 includes:

[0154] The fourth determining submodule is used to determine the separation differential entropy information between the at least two local maps based on the differential entropy of the point cloud data of the local map and the number of point cloud data in the local map; wherein, the separation differential entropy information represents the ratio of the sum of the differential entropy of the point cloud data of the at least two local maps to the sum of the number of point cloud data in the at least two local maps;

[0155] The first fusion submodule is used to fuse the point cloud data in the local map according to the initial relative pose to obtain the first fused point cloud data.

[0156] The fifth determining submodule is used to determine the first joint differential entropy information between the at least two local maps based on the number of point cloud data in the first fused point cloud data and the local map; wherein, the first joint differential entropy information characterizes the magnitude of the joint differential entropy of the at least two local maps before registration;

[0157] The sixth determining submodule is used to determine the relative pose error of the initial relative pose based on the separated differential entropy information and the first joint differential entropy information.

[0158] In one example, the first fusion submodule is specifically used for:

[0159] A target coordinate system is determined from the coordinate systems of point cloud data from at least two local maps; based on the initial relative pose, the point cloud data of the local maps are mapped to the target coordinate system to obtain the first fused point cloud data in the target coordinate system.

[0160] In one example, the fifth determining submodule is specifically used for:

[0161] Based on the first fused point cloud data, determine the point cloud data differential entropy of the first fused point cloud data; based on the number of point cloud data in the local map and the point cloud data differential entropy of the first fused point cloud data, determine the first joint differential entropy information between the at least two local maps.

[0162] In one example, the second determining module 5022 includes:

[0163] The fourth determining submodule is used to determine the separation differential entropy information between the at least two local maps based on the differential entropy of the point cloud data of the local map and the number of point cloud data in the local map;

[0164] The second fusion submodule is used to fuse the point cloud data in the local map according to the current relative pose to obtain the second fused point cloud data;

[0165] The seventh determining submodule is used to determine the second joint differential entropy information between the at least two local maps based on the number of point cloud data in the second fused point cloud data and the local map; wherein the second joint differential entropy information characterizes the magnitude of the joint differential entropy of the at least two local maps after registration;

[0166] The eighth determining submodule is used to determine the relative pose error of the current relative pose based on the separated differential entropy information and the second joint differential entropy information.

[0167] In one example, the second fusion submodule is specifically used for:

[0168] A target coordinate system is determined from the coordinate systems of point cloud data from at least two local maps; based on the current relative pose, the point cloud data of the local maps are mapped to the target coordinate system to obtain the second fused point cloud data in the target coordinate system.

[0169] In one example, the seventh determining submodule is specifically used for:

[0170] Based on the second fused point cloud data, determine the point cloud data differential entropy of the second fused point cloud data; based on the number of point cloud data in the local map and the point cloud data differential entropy of the second fused point cloud data, determine the second joint differential entropy information between the at least two local maps.

[0171] In one example, the third determining unit 503 is specifically used for:

[0172] The confidence information is obtained by subtracting the relative pose error of the initial relative pose from the relative pose error of the current relative pose.

[0173] In one example, the device also includes:

[0174] The weight determination module is used to determine the weight of the current relative pose based on the confidence information if the value represented by the confidence information is greater than a preset confidence threshold.

[0175] The graph optimization module is used to optimize the graph construction process of the collected point cloud data according to the weight of the current relative pose, so as to obtain a global map.

[0176] According to embodiments of this disclosure, this disclosure also provides an electronic device.

[0177] Figure 6 A structural block diagram of an electronic device provided in this disclosure embodiment, such as... Figure 6 As shown, the electronic device 600 includes: at least one processor 602; and a memory 601 communicatively connected to the at least one processor 602; wherein the memory stores instructions executable by the at least one processor 602, the instructions being executed by the at least one processor 602 to enable the at least one processor 602 to perform the confidence determination method based on local maps disclosed herein.

[0178] The electronic device 600 also includes a receiver 603 and a transmitter 604. The receiver 603 is used to receive instructions and data sent by other devices, and the transmitter 604 is used to send instructions and data to external devices.

[0179] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0180] According to embodiments of this disclosure, this disclosure also provides a computer program product comprising: a computer program stored in a readable storage medium, at least one processor of an electronic device being able to read the computer program from the readable storage medium, and the at least one processor executing the computer program causing the electronic device to perform the method provided in any of the above embodiments.

[0181] Figure 7 A schematic structural block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0182] like Figure 7 As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded into random access memory (RAM) 703 from storage unit 708. The RAM 703 may also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0183] Multiple components in device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 707, such as various types of monitors, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0184] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as the confidence determination method based on a local map. For example, in some embodiments, the confidence determination method based on a local map can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the confidence determination method based on a local map described above can be performed. Alternatively, in other embodiments, the computing unit 701 may be configured, by any other suitable means (e.g., by means of firmware), to perform a confidence determination method based on a local map.

[0185] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0186] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0187] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0188] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0189] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0190] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service system that addresses the shortcomings of traditional physical hosts and VPS (Virtual Private Server) services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0191] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0192] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for determining confidence based on local maps, comprising: determining relative pose information between at least two local maps; wherein the local map is used to represent a map constructed by point cloud data collected in a preset time period; the relative pose information comprises an initial relative pose of the at least two local maps before point cloud registration processing and a current relative pose of the at least two local maps after point cloud registration processing; determining relative pose errors according to the relative pose information and point cloud data in the local map; wherein the relative pose errors comprise a relative pose error of the initial relative pose and a relative pose error of the current relative pose; determining confidence information according to the relative pose error of the initial relative pose and the relative pose error of the current relative pose; wherein the confidence information is used to indicate accuracy of point cloud registration processing performed on the at least two local maps; wherein the determining relative pose errors according to the relative pose information and point cloud data in the local map comprises: determining point cloud data differential entropy of the local map according to point cloud data in the local map; wherein the point cloud data differential entropy is used to represent a sum of differential entropy of point cloud data; determining the relative pose error of the initial relative pose according to the initial relative pose and the point cloud data differential entropy of the local map, and determining the relative pose error of the current relative pose according to the current relative pose and the point cloud data differential entropy of the local map; wherein the determining the relative pose error of the initial relative pose according to the initial relative pose and the point cloud data differential entropy of the local map comprises: determining separation differential entropy information between the at least two local maps according to the point cloud data differential entropy of the local map and a quantity of point cloud data in the local map; wherein the separation differential entropy information represents a ratio of a sum of the point cloud data differential entropy of the at least two local maps to a sum of the quantity of point cloud data in the at least two local maps; fusing point cloud data in the local map according to the initial relative pose to obtain first fused point cloud data; determining first joint differential entropy information between the at least two local maps according to the first fused point cloud data and the quantity of point cloud data in the local map; wherein the first joint differential entropy information represents a size of joint differential entropy of the at least two local maps before registration; determining the relative pose error of the initial relative pose according to the separation differential entropy information and the first joint differential entropy information; wherein the determining point cloud data differential entropy of the local map according to point cloud data in the local map comprises: determining neighboring point cloud data of point cloud data in the local map according to a preset radius range; determining differential entropy information of point cloud data in the local map according to coordinate information of the neighboring point cloud data in the local map; wherein the differential entropy information represents differential entropy corresponding to coordinate information of point cloud data. ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ 2. The method of claim 1, wherein, ​ ​ ​ According to the differential entropy information of the point cloud data in the local map, the differential entropy of the point cloud data of the local map is determined.

3. The method of claim 2, wherein, The differential entropy information of the point cloud data in the local map is determined according to the coordinate information of the adjacent point cloud data in the local map. According to the coordinate information of the adjacent point cloud data in the local map, the covariance information of the point cloud data in the local map is determined. According to the covariance information of the point cloud data in the local map, the differential entropy information of the point cloud data in the local map is determined.

4. The method of claim 2 or 3, wherein, The differential entropy of the point cloud data of the local map is determined according to the differential entropy information of the point cloud data in the local map. The differential entropy of the point cloud data of the local map is determined according to the differential entropy information of the point cloud data in the local map.

5. The method of claim 4, wherein, The first fused point cloud data is obtained by fusing the point cloud data in the local map according to the initial relative pose. A target coordinate system is determined from the coordinate systems of the point cloud data of at least two local maps. The point cloud data of the local map is mapped into the target coordinate system according to the initial relative pose, and the first fused point cloud data in the target coordinate system is obtained.

6. The method of claim 5, wherein, The first joint differential entropy information between the at least two local maps is determined according to the number of point cloud data in the local map and the point cloud data differential entropy of the first fused point cloud data. The point cloud data differential entropy of the first fused point cloud data is determined according to the first fused point cloud data. The first joint differential entropy information between the at least two local maps is determined according to the number of point cloud data in the local map and the point cloud data differential entropy of the first fused point cloud data.

7. The method of any one of claims 1-3, 5-6, wherein, The relative pose error of the current relative pose is determined according to the current relative pose and the point cloud data differential entropy of the local map. The separation differential entropy information between the at least two local maps is determined according to the point cloud data differential entropy of the local map and the number of point cloud data in the local map. The second fused point cloud data is obtained by fusing the point cloud data in the local map according to the current relative pose. The second joint differential entropy information between the at least two local maps is determined according to the second fused point cloud data and the number of point cloud data in the local map, wherein the second joint differential entropy information represents the size of the joint differential entropy of the at least two local maps after registration. The relative pose error of the current relative pose is determined according to the separation differential entropy information and the second joint differential entropy information.

8. The method of claim 7, wherein, The second fused point cloud data is obtained by fusing the point cloud data in the local map according to the initial relative pose. A target coordinate system is determined from the coordinate systems of the point cloud data of at least two local maps. The point cloud data of the local map is mapped into the target coordinate system according to the current relative pose, and the second fused point cloud data in the target coordinate system is obtained.

9. The method of claim 8, wherein, The second joint differential entropy information between the at least two local maps is determined according to the second fused point cloud data and the quantity of point cloud data in the local map, and the second joint differential entropy information between the at least two local maps comprises: The point cloud data differential entropy of the second fused point cloud data is determined according to the second fused point cloud data. The second joint differential entropy information between the at least two local maps is determined according to the quantity of point cloud data in the local map and the point cloud data differential entropy of the second fused point cloud data.

10. The method of any one of claims 1-3, 5-6, and 8-9, wherein, The confidence information is determined according to the relative pose error of the initial relative pose and the relative pose error of the current relative pose, and the confidence information comprises: The relative pose error of the initial relative pose is subtracted from the relative pose error of the current relative pose, and a difference value obtained is the confidence information.

11. The method of any one of claims 1-3, 5-6, and 8-9, further comprising: If a value represented by the confidence information is greater than a preset confidence threshold, then a weight of the current relative pose is determined according to the confidence information; The mapping processing for the collected point cloud data is optimized according to the weight of the current relative pose, and a global map is obtained.

12. A confidence determination apparatus based on a local map, comprising: A first determination unit configured to determine relative pose information between at least two local maps; wherein the local map is configured to represent a map constructed by point cloud data collected in a preset time period; the relative pose information comprises an initial relative pose of the at least two local maps before point cloud registration processing and a current relative pose of the at least two local maps after point cloud registration processing; A second determination unit configured to determine relative pose errors according to the relative pose information and point cloud data in the local map; wherein the relative pose errors comprise a relative pose error of the initial relative pose and a relative pose error of the current relative pose; A third determination unit configured to determine confidence information according to the relative pose error of the initial relative pose and the relative pose error of the current relative pose; wherein the confidence information is configured to indicate accuracy of point cloud registration processing performed on the at least two local maps; The second determination unit comprises: A first determination module configured to determine point cloud data differential entropy of the local map according to point cloud data in the local map; wherein the point cloud data differential entropy is configured to represent a sum of differential entropies of point cloud data; A second determination module configured to determine the relative pose error of the initial relative pose according to the initial relative pose and the point cloud data differential entropy of the local map, and determine the relative pose error of the current relative pose according to the current relative pose and the point cloud data differential entropy of the local map; The second determination module comprises: a fourth determining sub-module, configured to determine separation differential entropy information between the at least two local maps according to the point cloud data differential entropy of the local map and the quantity of the point cloud data in the local map; wherein the separation differential entropy information represents a ratio of a sum of the point cloud data differential entropy of the at least two local maps to a sum of the quantity of the point cloud data in the at least two local maps; a first fusing sub-module, configured to fuse the point cloud data in the local map according to the initial relative pose, to obtain first fused point cloud data; a fifth determining sub-module, configured to determine first joint differential entropy information between the at least two local maps according to the first fused point cloud data and the quantity of the point cloud data in the local map; wherein the first joint differential entropy information represents a size of joint differential entropy of the at least two local maps before registration; a sixth determining sub-module, configured to determine a relative pose error of the initial relative pose according to the separation differential entropy information and the first joint differential entropy information.

13. The apparatus of claim 12, wherein, The first determining module comprises: a first determining sub-module, configured to determine adjacent point cloud data of the point cloud in the local map according to a preset radius range; a second determining sub-module, configured to determine differential entropy information of the point cloud data in the local map according to coordinate information of the adjacent point cloud data in the local map; wherein the differential entropy information represents differential entropy corresponding to the coordinate information of the point cloud data; a third determining sub-module, configured to determine point cloud data differential entropy of the local map according to the differential entropy information of the point cloud data in the local map.

14. The apparatus of claim 13, wherein, The second determining sub-module is specifically configured to: determine covariance information of the point cloud data in the local map according to the coordinate information of the adjacent point cloud data in the local map; and determine the differential entropy information of the point cloud data in the local map according to the covariance information of the point cloud data in the local map.

15. The apparatus of claim 13 or 14, wherein, The third determining sub-module is specifically configured to: perform accumulation processing on the differential entropy information of the point cloud data in the local map, to obtain the point cloud data differential entropy of the local map.

16. The apparatus of claim 15, wherein, The first fusing sub-module is specifically configured to: determine a target coordinate system from coordinate systems of point cloud data of at least two local maps; and map the point cloud data of the local map into the target coordinate system according to the initial relative pose, to obtain first fused point cloud data in the target coordinate system.

17. The apparatus of claim 16, wherein, The fifth determining sub-module is specifically configured to: determine point cloud data differential entropy of the first fused point cloud data according to the first fused point cloud data; and determine first joint differential entropy information between the at least two local maps according to the quantity of the point cloud data in the local map and the point cloud data differential entropy of the first fused point cloud data.

18. The apparatus of any one of claims 12-14, 16-17, wherein, The second determining module comprises: a fourth determining sub-module, configured to determine separation differential entropy information between the at least two local maps according to the point cloud data differential entropy of the local map and the quantity of the point cloud data in the local map; a second fusion submodule configured to fuse the point cloud data in the local map according to the current relative pose to obtain second fused point cloud data; a seventh determination submodule configured to determine second joint differential entropy information between the at least two local maps according to the second fused point cloud data and a quantity of the point cloud data in the local map, wherein the second joint differential entropy information represents a size of joint differential entropy of the at least two local maps after registration; an eighth determination submodule configured to determine a relative pose error of the current relative pose according to the separated differential entropy information and the second joint differential entropy information.

19. The apparatus of claim 18, wherein, The second fusion submodule is specifically configured to: determine a target coordinate system from coordinate systems of the point cloud data of the at least two local maps, and map the point cloud data of the local map into the target coordinate system according to the current relative pose to obtain second fused point cloud data in the target coordinate system.

20. The apparatus of claim 19, wherein, The seventh determination submodule is specifically configured to: determine point cloud data differential entropy of the second fused point cloud data according to the second fused point cloud data, and determine the second joint differential entropy information between the at least two local maps according to the quantity of the point cloud data in the local map and the point cloud data differential entropy of the second fused point cloud data.

21. The apparatus of any of claims 12-14, 16-17, and 19-20, wherein, The third determination unit is specifically configured to: subtract the relative pose error of the initial relative pose from the relative pose error of the current relative pose, and obtain a difference value as the confidence information.

22. The apparatus of any one of claims 12-14, 16-17, and 19-20, further comprising: a weight determination module configured to determine a weight of the current relative pose according to the confidence information if a value represented by the confidence information is greater than a preset confidence threshold; a graph optimization module configured to optimize mapping processing for the collected point cloud data according to the weight of the current relative pose to obtain a global map.

23. An electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the local map based confidence determination method of any one of claims 1-11.

24. A non-transitory computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to enable the computer to perform the local map based confidence determination method of any one of claims 1-11.

25. A computer program product comprising a computer program which, when executed by a processor, implements the steps of the local map based confidence determination method of any one of claims 1-11.

Images