CN115236680A - Positioning method, positioning device, electronic equipment and computer storage medium - Google Patents

Abstract

The embodiment of the application provides a positioning method, a positioning device, electronic equipment and a computer storage medium. The positioning method comprises the following steps: acquiring a frame relative pose for mapping a current point cloud frame acquired by a laser radar to a local map, a map absolute pose for mapping the local map to a global map, and a frame absolute pose for mapping the current point cloud frame to the global map; carrying out pose fusion on the frame absolute pose, the map absolute pose and the frame relative pose to obtain a final absolute pose corresponding to the current point cloud frame; and according to the final absolute pose and the current point cloud frame, determining abnormal global points with the position change degree larger than or equal to a screening threshold value in the global map so as to determine the frame absolute pose of the next point cloud frame based on the global map with the abnormal global points identified. The positioning method can obtain more reliable positioning output.

Description

Positioning method, device, electronic device and computer storage medium

technical field

The embodiments of the present application relate to the field of computer technologies, and in particular, to a positioning method, an apparatus, an electronic device, and a computer storage medium.

Background technique

The positioning unit is the basic unit in the intelligent driving system. The positioning unit provides real-time global pose information for the unmanned vehicle, and the lidar positioning is a very important sub-unit in the positioning unit. Most laser positioning schemes use the absolute positioning scheme. , however the absolute positioning scheme has the following disadvantages:

1) The positioning error between frames fluctuates greatly, and the pose is not smooth enough;

2) In scenes such as open space or changing environment, the positioning error of some frames is relatively large.

SUMMARY OF THE INVENTION

In view of this, the embodiments of the present application provide a positioning solution to at least partially solve the above problems.

According to a first aspect of the embodiments of the present application, a positioning method is provided, including: acquiring a frame relative pose for mapping a current point cloud frame collected by a lidar to a local map, and for mapping the local map to the map absolute pose in the global map, and the frame absolute pose for mapping the current point cloud frame to the global map; for the frame absolute pose, the map absolute pose and The relative pose of the frame is subjected to pose fusion to obtain the final absolute pose corresponding to the current point cloud frame; according to the final absolute pose and the current point cloud frame, the degree of position change in the global map is determined The abnormal global points greater than or equal to the screening threshold are used to determine the frame absolute pose of the next point cloud frame based on the global map identifying the abnormal global points.

According to a second aspect of the embodiments of the present application, a positioning device is provided, including: a first acquisition module configured to acquire a frame relative pose for mapping a current point cloud frame collected by a lidar to a local map; For mapping the local map to the map absolute pose in the global map, and for mapping the current point cloud frame to the frame absolute pose in the global map; a fusion module for the The absolute pose of the frame, the absolute pose of the map, and the relative pose of the frame are fused to obtain the final absolute pose corresponding to the current point cloud frame; an evaluation module is used to perform pose fusion according to the final absolute pose and the current point cloud frame, determine the abnormal global point in the global map whose position change degree is greater than or equal to the screening threshold, to determine the frame absolute position of the next point cloud frame based on the global map that identifies the abnormal global point posture.

According to a third aspect of the embodiments of the present application, an electronic device is provided, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface complete each other through the communication bus The memory is used for storing at least one executable instruction, and the executable instruction enables the processor to perform an operation corresponding to the positioning method described in the first aspect.

According to a fourth aspect of the embodiments of the present application, a computer storage medium is provided on which a computer program is stored, and when the program is executed by a processor, the positioning method according to the first aspect is implemented.

According to the positioning solution provided by the embodiment of the present application, when the final absolute pose is determined, the frame relative pose and the map absolute pose are fused in the frame absolute pose, so that the final fused absolute pose is more accurate and can reduce the Error fluctuation and error maximum. Moreover, based on the final absolute pose, the current point cloud frame can be mapped to the global map, so as to determine the position change degree of each global point in the global map, and to determine the abnormal global point caused by the environmental change, so as to determine the next point cloud frame. The frame absolute pose avoids the interference of abnormal global points, so as to avoid localization failure caused by environmental changes and improve robustness.

Description of drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments described in the embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings.

1A is a flowchart of steps of a positioning method according to Embodiment 1 of the present application;

FIG. 1B is a schematic diagram of an example of a scenario in the embodiment shown in FIG. 1A;

2A is a flowchart of steps of a positioning method according to Embodiment 2 of the present application;

2B is a schematic diagram of a laser positioning frame in the embodiment shown in FIG. 2A;

3 is a structural block diagram of a positioning device according to Embodiment 3 of the present application;

FIG. 4 is a schematic structural diagram of an electronic device according to Embodiment 4 of the present application.

Detailed ways

In order to make those skilled in the art better understand the technical solutions in the embodiments of the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. The embodiments described above are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments in the embodiments of the present application should fall within the protection scope of the embodiments of the present application.

The specific implementation of the embodiments of the present application is further described below with reference to the accompanying drawings of the embodiments of the present application.

Example 1

Referring to FIG. 1A , a schematic flowchart of steps of a positioning method according to Embodiment 1 of the present application is shown.

In this embodiment, the method can be applied to the field of intelligent driving to realize the positioning of the vehicle, so that navigation or driving control can be performed subsequently according to the positioning information of the vehicle. The method includes the following steps:

Step S102: Obtain the frame relative pose for mapping the current point cloud frame collected by the lidar to the local map, the map absolute pose for mapping the local map to the global map, and the current Point cloud frames are mapped to frame absolute poses in the global map.

The current point cloud frame may be the information of the reflection points in the environment where the lidar is collected at the current moment. For example, the current point cloud frame includes 3D position information of multiple reflection points in the environment at the current moment.

Figure 1B shows global space 1. The lidar is configured on the vehicle and moves from point A to point B in global space 1. During this process, the lidar collects a point cloud frame every certain time. For example, the current time is the t-th time, and the point cloud frame collected at the t-th time is the current point cloud frame.

A local map can be constructed based on at least some of the point cloud frames acquired by the lidar. The local map includes location information of multiple local points. The location information of the local point can be determined according to the location information of the reflection point in the point cloud frame used to construct the local map.

In this embodiment, the collected current point cloud frame can be input into the laser odometry, and the frame relative pose of the current point cloud frame to the local map can be determined by the laser odometry, and the current point cloud frame can be converted by the frame relative pose Mapped to the local map.

The global map includes position information of multiple global points, and the global points may be points obtained by data collection corresponding to the global space 1 by the lidar. In this embodiment, the global map is a priori map, that is, a priori map collected in advance by a lidar vehicle. In this embodiment, the location information of the global point in the prior map will not change. However, the position information of the global point in the prior map can be updated at regular intervals (such as half a month, one month, or half a year, etc.) as needed.

The map absolute pose is used to map the local map into the global map. In a feasible manner, the local map can be input into the global absolute positioning unit, and the absolute pose of the map can be determined by the global absolute positioning unit.

The frame absolute pose is used to map the current point cloud frame into the global map. In a feasible manner, the current point cloud frame can be input into the global absolute positioning unit, and the absolute pose of the frame can be determined by the global absolute positioning unit.

The global absolute positioning unit may adopt an appropriate unit capable of realizing corresponding functions as required.

Step S104: Perform pose fusion on the absolute pose of the frame, the absolute pose of the map, and the relative pose of the frame to obtain the final absolute pose corresponding to the current point cloud frame.

In this embodiment, the absolute pose of the frame, the absolute pose of the map, and the relative pose of the frame can be fused by using the graph optimization method, so that the absolute pose of the map and the relative pose of the frame are used as constraints, so that the final absolute pose of the fusion is obtained. The pose is more accurate. Moreover, since the frame relative pose and the map absolute pose are integrated in the final absolute pose, the maximum error value of the final absolute pose can be effectively reduced, and the error fluctuation can be reduced, so as to obtain a more accurate final absolute pose of the current point cloud frame. pose.

Step S106: According to the final absolute pose and the current point cloud frame, determine the abnormal global point in the global map whose position change degree is greater than or equal to the screening threshold, so as to identify the abnormal global point based on the global point that identifies the abnormal global point. The map determines the frame-absolute pose of the next point cloud frame.

Based on the determined final absolute pose, each reflection point in the current point cloud frame can be mapped to the global map, and for each reflection point, according to the position of the reflection point mapped to the global map, the reflection point and the nearest neighbor can be determined. The second offset distance between the global points, the second offset distance can be used as the position change degree of the global point, the screening threshold can be determined according to the position change degree of each global point, and the specific determination method of the screening threshold value can be determined as required .

According to the screening threshold, the abnormal global points whose position change degree is greater than or equal to the screening threshold can be determined from the global points, so that the environmental changes generated in the global map can be found in time, because different reflection points will appear when the environment changes, so based on the The current point cloud frame collected in real time can determine whether the real environment corresponding to the global map has changed.

In this way, the environment change can be determined in time, and then the global point corresponding to the changed environment (that is, the abnormal global point) can be determined from the global map, so that the influence of the abnormal global point can be avoided when the absolute pose of the frame corresponding to the next point cloud frame is determined. , so as to ensure that the final absolute pose can be determined accurately even in the case of environmental changes, thereby realizing the robustness of the positioning.

The implementation process of the positioning method is described below in combination with a specific usage scenario as follows:

Pre-fetch the global map in global space 1. For example, the lidar vehicle drives in the global space 1 and collects the three-dimensional position information of the reflection points. These reflection points collected by the lidar vehicle are used as the global points in the global map.

When positioning based on the global map, the information of the reflection points from the environment is collected by the lidar on the vehicle as the current point cloud frame. The current point cloud frame includes the information of the reflection points on the target object in the environment, such as position information and emission value.

On the one hand, the current point cloud frame is input into the laser odometer to obtain the frame relative pose used to map the current point cloud frame to the local map; on the other hand, the current point cloud frame and the local map can be input into the global absolute In the positioning unit, the frame absolute pose corresponding to the current point cloud frame and the map absolute pose corresponding to the local map are obtained.

Through the pose fusion unit, the absolute pose of the frame, the absolute pose of the map and the relative pose of the frame are fused to obtain a more accurate final absolute pose, and the final absolute pose is used to map the current point cloud frame to the global map .

After the current point cloud frame is mapped to the global map, the distance between the global point of the global map and the corresponding reflection point can be determined according to the mapping result, so as to determine the position change degree of the global point, thereby determining the position change in the global point. Abnormal global points whose degree is greater than or equal to the screening threshold (the screening threshold can be determined as needed), and then when determining the frame absolute pose of the next frame of point cloud, it can be based on the global map that marks the abnormal global points, thus ensuring the occurrence of abnormal global points. When the environment changes, the pose estimation can also be performed well, reducing the situation of localization failure caused by the environment change.

Through this embodiment, when the final absolute pose is determined, the frame relative pose and the map absolute pose are fused in the frame absolute pose, so that the final fused absolute pose is more accurate, and the error fluctuation and the maximum error can be reduced. value. Moreover, based on the final absolute pose, the current point cloud frame can be mapped to the global map, so as to determine the position change degree of each global point in the global map, and to determine the abnormal global point caused by the environmental change, so as to determine the next point cloud frame. The frame absolute pose avoids the interference of abnormal global points, so as to avoid localization failure caused by environmental changes and improve robustness.

The positioning method in this embodiment may be executed by any appropriate electronic device with positioning capability, including but not limited to: a server, a mobile terminal (such as a mobile phone, a PAD, etc.), a PC, and the like.

Embodiment 2

Referring to FIG. 2A , a schematic flowchart of steps of the positioning method according to the second embodiment of the present application is shown.

In this embodiment, the positioning method includes the following steps:

Step S202: Obtain a frame relative pose for mapping the current point cloud frame collected by the lidar to the local map.

In a specific implementation, the current point cloud frame collected by the lidar is input into the laser odometer, and the laser odometer matches the current point cloud frame to the local map, so as to obtain the relative pose of the frame, and then the local map can be determined. The first reference pose in the set local coordinate system and the second reference pose of the current point cloud frame in the local coordinate system. During this process, the laser odometry can also perform operations such as key point cloud frame selection and local map update.

Among them, the local map (also called sub-map) is composed of multiple key point cloud frames. For example, map the reflection points in the key point cloud frame to the local map according to the frame relative pose corresponding to each key point cloud frame. The local points are clustered together to obtain multiple clusters, and then the centroid of each cluster is calculated, and the centroid of each cluster is used as a new local point, and the position information of the centroid is used as the position information of the new local point.

Based on this, step S202 includes the following processes:

Process A1: Determine whether the current point cloud frame is a key point cloud frame.

Since the local map is constructed from key point cloud frames, it is necessary to determine whether the current point cloud frame collected is a key point cloud frame.

A way to determine whether it is a key point cloud frame may be: summing the frame relative pose of the previous point cloud frame of the current point cloud frame and a preset adjustment amount, and using the summation result as the initial frame of the current point cloud frame relative pose. If the position difference or the pose difference between the initial frame relative pose of the current point cloud frame and the frame relative pose corresponding to the last key point cloud frame in the local map is greater than or equal to the set pose difference threshold (which can be based on experience or It needs to be determined, which is not limited in this embodiment), then the current point cloud frame is determined as the key point cloud frame, otherwise, the current point cloud frame is not determined as the key point cloud frame.

If the current point cloud frame is a key point cloud frame, execute process B1; otherwise, execute C1.

Process B1: Use the current point cloud frame as the key point cloud frame to update the local map.

Case 1, if the number of key point cloud frames in the currently used local map is not greater than the split value, the current point cloud frame is directly added to the currently used local map, and the local points in the local map and Location information of local points.

Case 2, if the number of key point cloud frames in the currently used local map is greater than the split value, but less than the total accommodation value (the value can be determined according to the positioning capability, storage capability, etc., such as 10 frames, 20 frames, 50 frames, etc.) , the current point cloud frame can be added to the currently used local map, and the local points in the local map and the location information of the local points can be calculated in the aforementioned manner. Also add the current point cloud frame to the alternate local map.

Case 3, if the number of key point cloud frames in the currently used local map is equal to the total accommodation value, then use the spare local map as the new currently used local map, and determine the situation to which the new currently used local map belongs is Case 1 Or 2, if the new currently used local map belongs to case 1, refer to the aforementioned case 1, add the current point cloud frame to the new currently used local map, and then perform subsequent operations.

Process C1 may be performed after updating the local map.

Process C1: Determine the relative pose of the frame according to the local map corresponding to the current point cloud frame.

In a feasible manner, according to the relative pose of the initial frame of the current point cloud frame, a plurality of reflection points of the current point cloud frame are mapped to the local map, and the mapping position of each reflection point is obtained. Determine the k nearest local points of each reflection point in the local map according to the mapping position, construct the reference surface of the reflection point with the k nearest local points, and calculate the distance between the reflection point and the corresponding reference surface.

Calculate the square sum of the distance from each reflection point to the corresponding reference surface, and according to the square sum, according to the Gauss-Newton iteration method, optimize the relative pose of the initial frame, and then use the optimized frame relative pose to convert the current point cloud frame. Map to the local map, and then calculate the sum of squares of the distances from each reflection point to the corresponding reference surface. According to the sum of squares, the relative pose of the frame is optimized according to the Gauss-Newton iteration method. This iterative optimization is performed until convergence. Frame-relative pose as the frame-relative pose of the current point cloud frame.

The condition for convergence can be that the sum of two adjacent squares is equal or the difference is less than the set value (the set value can be determined as required).

Optionally, in addition to obtaining the frame relative pose of the current point cloud frame in step S202, the first reference pose of the local map in the set local coordinate system and the current point cloud frame in the local coordinate system can also be obtained. The second reference pose, the first reference pose and the second reference pose can make the determination of the subsequent absolute pose more accurate.

The local coordinate system can be a coordinate system selected according to needs, and the local coordinate system corresponds to the laser odometer (or lidar). For example, the relative pose of the frame corresponding to the first point cloud frame collected by the lidar can be used as the local coordinate system. The corresponding pose is either the pose of the global coordinate system corresponding to the global map as the pose of the local coordinate system, or any one of the poses can be selected as the pose of the local coordinate system.

After the pose of the local coordinate system is determined, the transformation relationship between the local coordinate system and the global coordinate system can be determined. Of course, since the pose of the local coordinate system is the frame relative pose of the first point cloud frame of the lidar, and there may be errors in the frame relative pose, the pose of the local coordinate system can be optimized, so the local The transformation relationship between the coordinate system and the global coordinate system will also change with optimization.

After the pose of the local coordinate system is determined, the first reference pose of the local map in the local coordinate system can be determined through the process D1, and the second reference pose of the current point cloud frame in the local coordinate system can be determined through the process E1.

Process D1: Determine a first reference pose of the local map in the local coordinate system according to the frame relative pose of at least one key point cloud frame in the local map.

In a feasible manner, the first reference pose is determined according to the frame relative pose of the first key point cloud frame in the local map and the pose of the local coordinate system.

Process E1: Determine the second reference pose of the current point cloud frame in the local coordinate system.

In a feasible manner, the product of the frame relative pose of the current point cloud frame and the first reference pose is used as the second reference pose.

Step S204: Obtain the absolute map pose for mapping the local map to the global map.

In a feasible manner, step S204 includes the following sub-steps:

Sub-step S2041: Determine the map predicted absolute pose for mapping the local map into the global map.

For example, sub-step S2041 includes the following processes:

Process A2: Obtain the frame absolute pose corresponding to the previous point cloud frame, the second reference pose of the previous point cloud frame in the local coordinate system, and the first reference pose for mapping the local map to the local coordinate system. Reference pose.

其是已经计算得到的位姿，故不再赘述。前一点云帧在所述局部坐标系下的第二参考位姿可以记作

其是已经计算出的值，故不再赘述。用于将局部地图映射到局部坐标系中的第一参考位姿可以记作

其可以通过前述的过程D1确定，故不再赘述。The absolute pose of the frame corresponding to the previous cloud frame can be recorded as

It is the pose that has been calculated, so it is not repeated here. The second reference pose of the previous point cloud frame in the local coordinate system can be written as

It is an already calculated value, so it will not be repeated here. The first reference pose used to map the local map into the local coordinate system can be written as

It can be determined through the aforementioned process D1, so it is not repeated here.

Process B2: Based on the first reference pose, the frame absolute pose corresponding to the previous point cloud frame, and the second reference pose of the previous point cloud frame, determine the map prediction absolute pose corresponding to the local map .

In a feasible manner, the product of the first reference pose, the frame absolute pose corresponding to the previous point cloud frame, and the second reference pose of the previous point cloud frame may be used as the map prediction absolute pose. The absolute pose of the map prediction is expressed as:

Sub-step S2042: Predict the absolute pose according to the map, and match the local map with the global map to determine an absolute map pose for mapping the local map to the global map.

In a feasible manner, the global absolute positioning unit may predict the pose according to the map, and match the local map with the global map, thereby determining the absolute map pose for mapping the local map to the global map.

For example, the local points of the local map are mapped to the global map according to the predicted pose of the map, and the k nearest global points are determined according to the mapping positions of the local points. For the k nearest global points corresponding to each local point, based on the abnormal global points in the global map corresponding to the current point cloud frame, the k nearest global points are screened out. For example, for the k nearest global points corresponding to the local point A, the abnormal global points are removed.

If the number of the remaining normal global points is greater than or equal to 3, and the reference surface can be formed, then the distance between these local points and the reference surface is calculated, and according to the sum of the squares of the distances of these local points, the map is calculated according to the Gauss-Newton iteration method. The predicted pose is optimized, and the optimized absolute map pose is obtained.

Then map the local map to the global coordinate system according to the optimized absolute pose of the map, and repeat the above process of determining the k nearest global points according to the mapping positions of the local points, so that the absolute pose of the map is iteratively optimized until meet the convergence conditions. The pose thus optimized is the absolute pose of the map.

Step S206: Obtain the frame absolute pose for mapping the current point cloud frame to the global map.

In a feasible manner, the step S206 includes the following sub-steps:

Sub-step S2061: According to the final absolute pose corresponding to the previous point cloud frame collected by the lidar, the second reference pose of the previous point cloud frame in the local coordinate system, and the current point cloud frame In the second reference pose in the local coordinate system, a frame predicted absolute pose for mapping the current point cloud frame to the global map is determined.

由于其是已经计算出的位姿，故不再进行赘述。The final absolute pose corresponding to the previous cloud frame can be recorded as

Since it is the already calculated pose, it will not be repeated here.

由于其是已经计算出的位姿，故不再赘述。The second reference pose of the previous cloud frame can be written as

Since it is the already calculated pose, it will not be repeated here.

其可以通过前述的过程E1确定，故不再赘述。The second reference pose of the current point cloud frame in the local coordinate system is recorded as

It can be determined through the aforementioned process E1, so it is not repeated here.

The frame prediction absolute pose used to map the current point cloud frame to the global map can be the product of the three, which can be expressed as:

Sub-step S2062: Determine normal reflection points and abnormal reflection points in the reflection points included in the current point cloud frame.

The abnormal reflection point is a reflection point whose first offset distance from the corresponding global point in the global map is greater than or equal to a change threshold.

The change threshold may be determined according to needs or experience, which is not limited in this embodiment.

Sub-step S2062 includes the following processes:

Process A3: Obtain the inter-frame displacement increment, and determine the predicted absolute position corresponding to the current point cloud frame according to the inter-frame displacement increment and the final absolute pose corresponding to the previous point cloud frame collected by the lidar posture.

The inter-frame displacement increment can be determined as needed or empirically.

The abnormal screening absolute pose corresponding to the current point cloud frame may be the sum of the inter-frame displacement increment and the final absolute pose corresponding to the previous point cloud frame collected by the lidar.

Process B3: According to the predicted absolute pose and the global map corresponding to the current point cloud frame, determine the first offset between the reflection point of the current point cloud frame and the corresponding global point in the global map distance.

Screen the absolute pose according to the abnormality and map the reflection points of the current point cloud frame to the global map. According to the mapping position of each reflection point, determine the nearest global point of each reflection point from the global points of the global map, and calculate each reflection point. The first offset distance between the point and the corresponding nearest-neighbor global point.

Process C3: Determine and identify abnormal reflection points in the current point cloud frame according to the first offset distance of each reflection point

If the first offset distance is greater than or equal to the change threshold, it is determined as an abnormal reflection point.

Since abnormal reflection points may be generated due to environmental changes, which will adversely affect the positioning accuracy, when determining the absolute pose of the frame, the abnormal reflection points are removed to ensure accuracy.

Sub-step S2063: Predict the absolute pose according to the frame, match the normal reflection points outside the abnormal reflection points in the current point cloud frame with the global map, and determine the corresponding position of the current point cloud frame according to the matching result. Frame absolute pose.

For example, sub-step S2063 includes the following processes:

Process A4: Predict the absolute pose according to the frame, and map each normal reflection point in the current point cloud frame to the global map to obtain a first mapping position of each normal reflection point.

For example, multiply each normal reflection point in the current point cloud frame by the yaw angle, pitch angle and roll angle in the predicted absolute pose of the frame, plus the offsets of the x-axis, y-axis and z-axis, the The result is the first mapped position in the global map.

Process B4: According to the first mapping position of each normal reflection point and the position of the global point in the global map, determine k adjacent global points corresponding to each of the normal reflection points.

According to the first mapping position, k adjacent global points are obtained from the global points included in the global map, and these k adjacent global points may not include abnormal global points, or may include abnormal global points. The abnormal global point is determined according to the global map corresponding to the current point cloud frame.

Process C4: For each of the normal reflection points, screen out the abnormal global points among the k adjacent global points corresponding to the normal reflection points, and determine the relative reference of each of the normal reflection points constructed by the remaining normal global points The face offset distance for the face.

For example, for the normal reflection point M, if the corresponding k adjacent global points include abnormal global points, the abnormal global points are screened out.

If the number of normal global points remaining after screening is greater than or equal to 3, the reference surface is determined according to the remaining normal global points, and the surface offset distance between the normal reflection point and the reference surface is calculated.

Alternatively, if the number of normal global points remaining after screening is less than 3, since the reference surface cannot be formed, the normal reflection point can be discarded.

Process D4: Determine the absolute pose of the frame corresponding to the current point cloud frame according to the offset distance of each of the surfaces.

In a feasible way, according to the square sum of the offset distances of each surface, the Gauss-Newton iteration method is used to optimize the frame prediction absolute pose to obtain the optimized frame absolute pose, and then iterate with the optimized frame absolute pose Optimization, and finally obtain the absolute pose of the frame corresponding to the current point cloud frame. The iterative optimization process may be to return to process A4, and replace the frame predicted absolute pose with the optimized frame absolute pose, and continue to execute until the converged frame absolute pose is obtained. The converged frame absolute pose may be that the deviation of the frame absolute poses of two adjacent optimizations is less than or equal to a certain value.

Step S208: Perform pose fusion on the absolute pose of the frame, the absolute pose of the map, and the relative pose of the frame to obtain the final absolute pose corresponding to the current point cloud frame.

Wherein, step S208 includes the following process:

Process A5: Obtain the first reference pose of the local map in the set local coordinate system, the second reference pose of the current point cloud frame in the local coordinate system, and the local coordinate system and The conversion relationship of the global coordinate system corresponding to the global map.

可以使用前述过程获取的位姿，故不再赘述。The first reference pose can be expressed as

The pose obtained in the foregoing process can be used, so it is not repeated here.

可以使用前述过程获取的位姿，故不再赘述。The second reference pose can be expressed as

The pose obtained in the foregoing process can be used, so it is not repeated here.

在局部坐标系的位姿确定的情况下转换关系是可知的，故不再赘述。The transformation relation can be expressed as

The transformation relationship is known when the pose of the local coordinate system is determined, so it is not repeated here.

Process B5: According to the absolute pose of the frame, the absolute pose of the map, and the relative pose of the frame, construct a fused graph optimization problem according to a graph optimization rule.

可以使用前述过程获取的位姿，故不再赘述。The frame absolute pose can be expressed as

The pose obtained in the foregoing process can be used, so it is not repeated here.

可以使用前述过程获取的位姿，故不再赘述。The absolute pose of the map can be expressed as

The pose obtained in the foregoing process can be used, so it is not repeated here.

可以使用前述过程获取的位姿，故不再赘述。The frame relative pose can be expressed as

The pose obtained in the foregoing process can be used, so it is not repeated here.

The graph optimization rule can adopt any appropriate rule, so it will not be repeated here.

Process C5: Solve the graph optimization problem according to the first reference pose, the second reference pose, and the conversion relationship to obtain the final fused absolute pose corresponding to the current point cloud frame.

In a feasible manner, the first reference pose, the second reference pose and the conversion relationship are optimized, and the optimized first reference pose, the second reference pose and the conversion relationship to determine the final frame absolute pose.

The optimization process is for example:

On the one hand, the relative pose of the intermediate frame is calculated according to the first reference pose and the second reference pose, and then the difference between the relative pose of the intermediate frame and the relative pose of the frame is calculated, and then the first reference pose and the first reference pose are calculated according to the difference. The second reference pose is optimized.

On the other hand, the absolute pose of the intermediate map is calculated according to the first reference pose and the transformation relationship, and then the difference between the absolute pose of the intermediate map and the absolute map pose is calculated, and then the first reference pose and the transformation relationship are calculated according to the difference. optimization.

On the other hand, the absolute pose of the intermediate frame is calculated according to the second reference pose and the transformation relationship, and then the difference between the absolute pose of the intermediate frame and the absolute pose of the frame is calculated, and then the second reference pose and the transformation relationship are performed according to the difference. optimization.

In this way, the three aspects are jointly optimized until the first reference pose, the second reference pose and the transformation relationship with the three smallest differences are obtained.

When the final absolute pose is determined according to the optimized first reference pose, the second reference pose and the conversion relationship, for the key point cloud frame, the final absolute pose is expressed as

The non-key point cloud frame whose final absolute pose is represented as

Step S210: Map the current point cloud frame to the global map according to the final absolute pose, so as to obtain the second mapping position of each reflection point in the current point cloud frame in the global map .

Step S212: According to the second mapping position of each of the reflection points and the nearest global point corresponding to each of the reflection points in the global map, determine the nearest neighbor corresponding to each of the reflection points in the global map. Second offset distance between global points.

Step S214: According to the second offset distance of the reflection point corresponding to each of the global points, use a normalization function to determine the confidence of the final absolute pose.

In a feasible manner, summing is performed according to the second offset distances of the reflection points to obtain the second offset distance sum, and a normalization function (such as a softmax function) is used to normalize the second offset distance sum. process, so that it is normalized to between [0, 1], so as to obtain the confidence of the final absolute pose.

The confidence can be calculated by the confidence evaluation unit, and the reliability of the positioning can be evaluated through the confidence.

Step S216 : determining, according to the final absolute pose and the current point cloud frame, an abnormal global point whose position change degree in the global map is greater than or equal to a screening threshold.

In a feasible manner, step S216 includes the following processes:

Process A6: Map the reflection point of the current point cloud frame to the global map according to the final absolute pose, so as to obtain a second mapping position of the reflection point in the global map.

Process B6: Determine a second offset distance between each of the reflection points and the corresponding closest global point in the global map according to the second mapping position of each of the reflection points.

For example, for each reflection point, a global point with the smallest distance from the reflection point is selected from the global points, and the distance between the reflection point and the selected global point is used as the second offset distance.

Process C6: According to the second offset distance of the reflection point corresponding to each of the global points, an abnormal global point whose position change degree is greater than or equal to the screening threshold is determined.

For example, the global map includes global points 1 to N, where N is greater than 1. Wherein, the P global points have corresponding reflection points, and the position change degree of the P global points may be the second offset distance of the corresponding reflection points. The degree of position change of the remaining global points can be determined to be 0.

According to the position change degree of each global point, the number of global points distributed in different position change degree intervals can be determined, and then the screening threshold is determined according to the distribution, and the global point whose position change degree is greater than or equal to the screening threshold is determined as an abnormal global point according to the screening threshold. point.

In this way, abnormal global points in the global map can be determined according to the current point cloud frame and the corresponding final absolute pose, so as to determine whether there is an environmental change in time. The abnormal global point in the global map can be applied to the next point cloud frame, for example, it is used when determining the frame absolute pose of the next point cloud frame and the map absolute pose of the local map corresponding to the next point cloud frame. The abnormal global points are filtered out during the pose, so as to avoid the localization failure caused by the environment change, and it can also reduce the error fluctuation range and the maximum error value.

As shown in Figure 2B, the present method can be performed and implemented due to a robust laser positioning framework comprising a laser odometry, a global absolute positioning unit, an environmental risk assessment unit, and a positional reliability assessment unit. This method can provide continuous stable and reliable positioning output, that is, output the final absolute pose of the current point cloud frame.

Due to the introduction of the environmental risk assessment unit and the fixed position reliability assessment unit, the environmental changes can be determined in a timely and accurate manner, so as to greatly reduce the failure of positioning due to environmental changes; the fixed position reliability assessment unit can give the current final absolute position posture reliability. In the process of determining the final absolute pose, the matching of the local map and the global map is used, that is, the absolute map pose. This matching has a positive effect in the situation of environmental changes and short-term map output, which can effectively improve the reliability of positioning. sex.

Among them, the laser odometer provides the frame relative pose information of the current point cloud frame relative to the local map; the global absolute positioning unit provides the frame absolute pose of the current point cloud frame relative to the global map, and the map absolute position of the local map relative to the global map. pose.

The pose fusion unit is used to fuse the frame relative pose output by the laser odometer with the frame absolute pose and map absolute pose output by the global absolute positioning unit to output the final absolute pose.

The risk assessment unit can detect the abnormal reflection points of the current point cloud frame relative to the prior global map and the position change degree of each global point in the prior global map, and then determine the abnormal global point.

The global absolute positioning unit will apply these abnormal reflection points and abnormal global points to the frame absolute pose of the current point cloud frame and the map absolute pose of the local map by setting lower weights for the abnormal reflection points and abnormal global points (such as Screening out these outliers) to reduce the negative impact of environmental changes on global positioning, thereby improving reliability.

This method can solve the problem that there is no final confidence output and it is impossible to judge whether the positioning result is good or bad, and the change information can be fed back to the absolute pose determination after environmental change detection, and better positioning can be performed for large-scale environmental changes. .

In summary, this method can provide stable and reliable positioning output, and has better robustness, which can greatly reduce the situation of positioning failure caused by environmental changes; it can give the reliability of laser positioning, and when the positioning error is large The problem can be reported in time; the fusion of relative pose and global absolute pose can reduce the fluctuation of positioning error and the maximum error value.

The method can be applied to autonomous driving scenarios, and the vehicle can be positioned continuously, stably and reliably through a laser positioning framework including absolute positioning, relative positioning, environmental risk assessment and positional reliability assessment. This method can be applied to different types of autonomous vehicles. For example, for logistics vehicles, public service vehicles, medical service vehicles, terminal service vehicles, etc., as long as the vehicle carries a laser odometer, this method can be used for positioning.

For logistics vehicles, it can transport goods to the appropriate location according to the set route during the autonomous driving process. In the process of automatic driving, the frame relative pose that maps the current point cloud frame to the local map, the map absolute pose that maps the local map to the global map, and the frame absolute pose that maps the current point cloud frame to the global map are fused The final absolute pose is obtained, and the abnormal global point is determined according to the final absolute pose, and the abnormal global point is considered when determining the frame absolute pose of the next point cloud frame, so that the frame absolute pose determination is more accurate.

Similarly, for public service vehicles, such as fire trucks, de-icers, sprinklers, snow plows, garbage disposal vehicles, traffic command vehicles, and the like. In order to accurately determine the current position of these public service vehicles in the process of automatic driving, they need to accurately determine their own poses. These public service vehicles can also be configured with this method, using the current point cloud frame collected by the onboard laser odometer. , a priori global map and local map, determine the relative pose of the current point cloud frame mapped to the frame in the local map, the local map is mapped to the absolute map pose in the global map, and the current point cloud frame is mapped to the The absolute pose of the frame in the global map, according to the relative pose of the frame, the absolute pose of the frame and the absolute pose of the map, the final absolute pose is determined, and then the abnormal global point is determined based on the final absolute pose and the current point cloud frame. The global map identifying the abnormal global point determines the frame absolute pose of the next point cloud frame, thereby improving the accuracy of the final absolute pose. The process for other vehicles with different functions is similar, so it is not repeated here.

Through this embodiment, when the final absolute pose is determined, the frame relative pose and the map absolute pose are fused in the frame absolute pose, so that the final fused absolute pose is more accurate, and the error fluctuation and the maximum error can be reduced. value. Moreover, based on the final absolute pose, the current point cloud frame can be mapped to the global map, so as to determine the position change degree of each global point in the global map, and to determine the abnormal global point caused by the environmental change, so as to determine the next point cloud frame. The frame absolute pose avoids the interference of abnormal global points, so as to avoid localization failure caused by environmental changes and improve robustness.

The positioning method in this embodiment may be executed by any appropriate electronic device with positioning capability, including but not limited to: a server, a mobile terminal (such as a mobile phone, a PAD, etc.), a PC, and the like.

Embodiment 3

Referring to FIG. 3 , a structural block diagram of a positioning apparatus according to Embodiment 3 of the present application is shown.

In this embodiment, the positioning device includes:

The first obtaining module 302 is used to obtain the frame relative pose for mapping the current point cloud frame collected by the lidar to the local map, and the map absolute pose for mapping the local map to the global map , and the frame absolute pose for mapping the current point cloud frame to the global map;

A fusion module 304, configured to perform pose fusion on the absolute pose of the frame, the absolute pose of the map, and the relative pose of the frame to obtain the final absolute pose corresponding to the current point cloud frame;

The evaluation module 306 is configured to determine, according to the final absolute pose and the current point cloud frame, an abnormal global point whose position change degree in the global map is greater than or equal to the screening threshold, so as to identify the abnormal global point based on the identification of the abnormal global point. The global map determines the frame absolute pose of the next point cloud frame.

Optionally, the first obtaining module 302 is configured to determine a map used for mapping the local map to the global map when the absolute pose of the map used for mapping the local map to the global map is obtained. Predict the absolute pose according to the map in posture.

Optionally, the first obtaining module 302 is configured to obtain the frame absolute pose corresponding to the previous point cloud frame, the The second reference pose of the point cloud frame in the local coordinate system, and the first reference pose for mapping the local map into the local coordinate system; based on the first reference pose, the previous point cloud The absolute pose of the frame corresponding to the frame and the second reference pose of the cloud frame at the previous point are used to determine the absolute pose of the map prediction corresponding to the local map.

Optionally, the first obtaining module 302 is configured to obtain the final absolute pose corresponding to the previous point cloud frame collected by the lidar when obtaining the absolute pose of the frame mapped from the current point cloud frame to the global map. , and the second reference pose of the previous point cloud frame under the local coordinate system, and the second reference pose of the current point cloud frame under the local coordinate system, determine the current The point cloud frame is mapped to the frame prediction absolute pose in the global map; the normal reflection point and the abnormal reflection point in the reflection points included in the current point cloud frame are determined, wherein the abnormal reflection point is the same as the global reflection point. The first offset distance between the corresponding global points in the map is greater than or equal to the reflection point of the change threshold; the absolute pose is predicted according to the frame, and the normal reflection point outside the abnormal reflection point in the current point cloud frame is compared with all the reflection points. The global map is matched, and the absolute pose of the frame corresponding to the current point cloud frame is determined according to the matching result.

Optionally, the first obtaining module 302 is configured to obtain an inter-frame displacement increment when determining a normal reflection point and an abnormal reflection point in the reflection points included in the current point cloud frame, and according to the inter-frame displacement Increment and the final absolute pose corresponding to the last point cloud frame collected by the lidar, determine the predicted absolute pose corresponding to the current point cloud frame; according to the predicted absolute pose and the current point cloud frame The corresponding global map is to determine the first offset distance between the reflection point of the current point cloud frame and the corresponding global point in the global map; according to the first offset distance of each of the reflection points, determine and identify Abnormal reflection points in the current point cloud frame.

Optionally, the first acquisition module 302 is used to predict the absolute pose according to the frame, match the normal reflection points outside the abnormal reflection points in the current point cloud frame with the global map, and match As a result, when the absolute pose of the frame corresponding to the current point cloud frame is determined, the absolute pose is predicted according to the frame, and each normal reflection point in the current point cloud frame is mapped to the global map to obtain each the first mapping position of the normal reflection point; according to the first mapping position of each normal reflection point and the position of the global point in the global map, determine k adjacent global points corresponding to each of the normal reflection points; For each of the normal reflection points, screen out the abnormal global points among the k adjacent global points corresponding to the normal reflection points, and determine the surface of each of the normal reflection points relative to the reference surface constructed by the remaining normal global points Offset distance; according to the offset distance of each surface, determine the absolute pose of the frame corresponding to the current point cloud frame.

Optionally, the fusion module 304 is configured to obtain the first reference pose of the local map in the set local coordinate system, and the second reference pose of the current point cloud frame in the local coordinate system. , and the transformation relationship of the global coordinate system corresponding to the local coordinate system and the global map; according to the absolute pose of the frame, the absolute pose of the map and the relative pose of the frame, build a fusion Graph optimization problem; solve the graph optimization problem according to the first reference pose, the second reference pose and the conversion relationship to obtain the final absolute pose of fusion corresponding to the current point cloud frame .

Optionally, the device further includes:

The second obtaining module 308 is configured to map the current point cloud frame to the global map according to the final absolute pose, so as to obtain the reflection points in the current point cloud frame in the global map the second mapping position of ;

A determination module 310 is configured to determine, according to the second mapping position of each of the reflection points and the nearest global point corresponding to each of the reflection points in the global map, that each of the reflection points corresponds to that in the global map The second offset distance between the nearest global points of ;

The normalization module 312 is configured to use a normalization function to determine the confidence of the final absolute pose according to the second offset distance of the reflection point corresponding to each of the global points.

Optionally, the evaluation module 306 is configured to determine, according to the final absolute pose and the current point cloud frame, an abnormal global point whose position change degree in the global map is greater than or equal to the screening threshold, according to the In the final absolute pose, the reflection point of the current point cloud frame is mapped to the global map to obtain the second mapping position of the reflection point in the global map; according to the second mapping position of each reflection point mapping the position, determining the second offset distance between each of the reflection points and the corresponding nearest global point in the global map; determining the position change according to the second offset distance of the reflection point corresponding to each of the global points Anomalous global points whose degree is greater than or equal to the screening threshold.

The positioning apparatus in this embodiment is used to implement the corresponding positioning methods in the foregoing multiple method embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here. In addition, for the function implementation of each module in the positioning apparatus of this embodiment, reference may be made to the descriptions of the corresponding parts in the foregoing method embodiments, and details are not repeated here.

Embodiment 4

Referring to FIG. 4 , a schematic structural diagram of an electronic device according to Embodiment 4 of the present application is shown. The specific embodiments of the present application do not limit the specific implementation of the electronic device.

As shown in FIG. 4 , the electronic device may include: a processor (processor) 402 , a communication interface (Communications Interface) 404 , a memory (memory) 406 , and a communication bus 408 .

in:

The processor 402 , the communication interface 404 , and the memory 406 communicate with each other through the communication bus 408 .

The communication interface 404 is used to communicate with other electronic devices or servers.

The processor 402 is configured to execute the program 410, and specifically may execute the relevant steps in the foregoing positioning method embodiments.

Specifically, the program 410 may include program code including computer operation instructions.

The processor 402 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. One or more processors included in the smart device may be the same type of processors, such as one or more CPUs; or may be different types of processors, such as one or more CPUs and one or more ASICs.

The memory 406 is used to store the program 410 . Memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk memory.

The program 410 can be specifically used to cause the processor 402 to execute the steps corresponding to the foregoing positioning method.

For the specific implementation of the steps in the program 410, reference may be made to the corresponding descriptions in the corresponding steps and units in the foregoing positioning method embodiments, which are not repeated here. Those skilled in the art can clearly understand that, for the convenience and brevity of description, for the specific working process of the above-described devices and modules, reference may be made to the corresponding process descriptions in the foregoing method embodiments, which will not be repeated here.

It should be pointed out that, according to the needs of implementation, each component/step described in the embodiments of the present application may be split into more components/steps, or two or more components/steps or part of operations of components/steps may be combined into New components/steps to achieve the purpose of the embodiments of the present application.

The above-described methods according to the embodiments of the present application may be implemented in hardware, firmware, or as software or computer codes that may be stored in a recording medium (such as CD ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or implemented by Network downloaded computer code originally stored in a remote recording medium or non-transitory machine-readable medium and will be stored in a local recording medium so that the methods described herein can be stored on a computer using a general purpose computer, special purpose processor or programmable or such software processing on a recording medium of dedicated hardware such as ASIC or FPGA. It will be understood that a computer, processor, microprocessor controller or programmable hardware includes storage components (eg, RAM, ROM, flash memory, etc.) that can store or receive software or computer code, when the software or computer code is executed by a computer, When accessed and executed by a processor or hardware, the positioning methods described herein are implemented. Furthermore, when a general purpose computer accesses code for implementing the positioning methods shown herein, execution of the code converts the general purpose computer into a special purpose computer for executing the positioning methods shown herein.

Those of ordinary skill in the art can realize that the units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Experts may use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of the embodiments of the present application.

The above embodiments are only used to illustrate the embodiments of the present application, but are not intended to limit the embodiments of the present application. Those of ordinary skill in the relevant technical field can also make various Therefore, all equivalent technical solutions also belong to the scope of the embodiments of the present application, and the patent protection scope of the embodiments of the present application should be defined by the claims.

Claims (12)

1. A method of positioning, comprising:

acquiring a frame relative pose for mapping a current point cloud frame acquired by a laser radar into a local map, a map absolute pose for mapping the local map into the global map, and a frame absolute pose for mapping the current point cloud frame into the global map;

performing pose fusion on the frame absolute pose, the map absolute pose and the frame relative pose to obtain a final absolute pose corresponding to the current point cloud frame;

and determining abnormal global points with the position change degree larger than or equal to a screening threshold value in the global map according to the final absolute pose and the current point cloud frame so as to determine the frame absolute pose of the next point cloud frame based on the global map identifying the abnormal global points.

2. The method of claim 1, wherein the obtaining a map absolute pose for mapping the local map into the global map comprises:

determining a map predicted absolute pose for mapping the local map into the global map;

and according to the map prediction absolute pose, matching the local map with the global map to determine a map absolute pose for mapping the local map into the global map.

3. The method of claim 2, wherein the determining a map predicted absolute pose for mapping the local map into the global map comprises:

acquiring a frame absolute pose corresponding to a previous point cloud frame, a second reference pose of the previous point cloud frame in the local coordinate system, and a first reference pose for mapping a local map to the local coordinate system;

and determining a map prediction absolute pose corresponding to the local map based on the first reference pose, the frame absolute pose corresponding to the previous point cloud frame and the second reference pose of the previous point cloud frame.

4. The method of claim 1, wherein the obtaining a frame absolute pose of the current point cloud frame mapping into the global map comprises:

determining a frame prediction absolute pose for mapping the current point cloud frame into the global map according to a final absolute pose corresponding to a previous point cloud frame acquired by the laser radar, a second reference pose of the previous point cloud frame in the local coordinate system, and a second reference pose of the current point cloud frame in the local coordinate system;

determining normal reflection points and abnormal reflection points in the reflection points contained in the current point cloud frame, wherein the abnormal reflection points are reflection points of which the first offset distance with the corresponding global points in the global map is greater than or equal to a variation threshold value;

and according to the frame prediction absolute pose, matching normal reflection points outside the abnormal reflection points in the current point cloud frame with the global map, and determining the frame absolute pose corresponding to the current point cloud frame according to a matching result.

5. The method of claim 4, wherein said determining normal and abnormal reflection points of the reflection points contained in the current point cloud frame comprises:

acquiring an inter-frame displacement increment, and determining a predicted absolute pose corresponding to the current point cloud frame according to the inter-frame displacement increment and a final absolute pose corresponding to a previous point cloud frame acquired by the laser radar;

determining a first offset distance between a reflection point of the current point cloud frame and a corresponding global point in a global map according to the predicted absolute pose and the global map corresponding to the current point cloud frame;

and determining and identifying abnormal reflection points in the current point cloud frame according to the first offset distance of each reflection point.

6. The method of claim 5, wherein the predicting absolute pose from the frame, matching normal reflection points outside abnormal reflection points in the current point cloud frame with the global map, and determining a frame absolute pose corresponding to the current point cloud frame from the matching result comprises:

according to the frame prediction absolute pose, mapping each normal reflection point in the current point cloud frame to the global map to obtain a first mapping position of each normal reflection point;

determining k adjacent global points corresponding to each normal reflection point according to the first mapping position of each normal reflection point and the position of the global point in the global map;

for each normal reflection point, screening out abnormal global points in k adjacent global points corresponding to the normal reflection point, and determining the plane offset distance of each normal reflection point relative to a reference plane constructed by the rest normal global points;

and determining the frame absolute pose corresponding to the current point cloud frame according to the surface offset distance.

7. The method of claim 1, wherein the pose fusing the frame absolute pose, the map absolute pose, and the frame relative pose to obtain a final absolute pose corresponding to the current point cloud frame comprises:

acquiring a first reference pose of the local map in a set local coordinate system, a second reference pose of the current point cloud frame in the local coordinate system and a conversion relation of a global coordinate system corresponding to the local coordinate system and the global map;

constructing a fused graph optimization problem according to the frame absolute pose, the map absolute pose and the frame relative pose and a graph optimization rule;

and solving the graph optimization problem according to the first reference pose, the second reference pose and the conversion relation so as to obtain a fused final absolute pose corresponding to the current point cloud frame.

8. The method of claim 1, wherein the method further comprises:

mapping the current point cloud frame into the global map according to the final absolute pose so as to obtain a second mapping position of each reflection point in the current point cloud frame in the global map;

determining a second offset distance between each reflection point and a corresponding nearest global point in the global map according to the second mapping position of each reflection point and the nearest global point corresponding to each reflection point in the global map;

and determining the confidence coefficient of the final absolute pose by using a normalization function according to the second offset distance of the reflection point corresponding to each global point.

9. The method of claim 1, wherein the determining abnormal global points in the global map having a position change degree greater than or equal to a filtering threshold according to the final absolute pose and the current point cloud frame comprises:

mapping the reflection point of the current point cloud frame into the global map according to the final absolute pose so as to obtain a second mapping position of the reflection point in the global map;

determining a second offset distance between each reflection point and the corresponding nearest global point in the global map according to the second mapping position of each reflection point;

and determining abnormal global points with the position change degree larger than or equal to a screening threshold value according to the second offset distance of the reflection points corresponding to the global points.

10. A positioning device, comprising:

the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring a frame relative pose for mapping a current point cloud frame acquired by a laser radar into a local map, a map absolute pose for mapping the local map into a global map and a frame absolute pose for mapping the current point cloud frame into the global map;

the fusion module is used for carrying out pose fusion on the frame absolute pose, the map absolute pose and the frame relative pose so as to obtain a final absolute pose corresponding to the current point cloud frame;

and the evaluation module is used for determining abnormal global points with the position change degree larger than or equal to a screening threshold value in the global map according to the final absolute pose and the current point cloud frame so as to determine the frame absolute pose of the next point cloud frame based on the global map identifying the abnormal global points.

11. An electronic device, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;

the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the positioning method according to any one of claims 1-9.

12. A computer storage medium, on which a computer program is stored which, when being executed by a processor, carries out the positioning method according to any one of claims 1-9.

Images