De-dynamic point method for autonomous mobile platforms and systems and devices therefor

Abstract

A dynamic point removal method for an autonomous mobile platform, and a system and device thereof. The dynamic point removal method for an autonomous mobile platform comprises the steps of: obtaining image information and point cloud information collected simultaneously at a current frame timestamp and a previous frame timestamp via a camera and a radar respectively configured on the autonomous mobile platform, to obtain current frame image data, previous frame image data and current frame point cloud data correspondingly; performing feature point processing on the current frame image data and the previous frame image data according to a relative pose relationship of the autonomous mobile platform between the current frame timestamp and the previous frame timestamp, to determine a dynamic region in the current frame image data; projecting the current frame point cloud data to an image plane of the camera, to take point clouds projected into the dynamic region as potential dynamic points; and removing the potential dynamic points in a point cloud cluster close to the camera by performing clustering processing on the potential dynamic points.

Description

Technical Field

[0001] This invention relates to the field of autonomous mobile platform technology, and more specifically to a method, system, and apparatus for removing dynamic points for autonomous mobile platforms. Background Technology

[0002] Currently, with the development of artificial intelligence technology, more and more companies and university teams are focusing on the research of autonomous mobile platforms. Autonomous mobile platforms typically refer to vehicles equipped with numerous sensors, enabling them to autonomously perceive and move, and then complete corresponding tasks through onboard task modules, such as inspection vehicles or robotic vacuum cleaners. Although the application scenarios of autonomous mobile platforms cover almost every aspect of people's lives, and autonomous mobile platforms have made considerable progress, there is currently no universal solution applicable to all scenarios. In particular, autonomous mobile platforms need to use a map of the current environment to determine their pose during autonomous navigation. Furthermore, since the positioning algorithm of autonomous mobile platforms is highly dependent on the map, the accuracy of map construction determines the accuracy of positioning. The presence of dynamic points (or dynamic obstacles) on the map not only affects the accuracy of positioning but also the accuracy of map construction. Therefore, the removal of dynamic points (or dynamic obstacles) is essential.

[0003] Existing dynamic point (or dynamic obstacle) detection schemes are typically based on rasterized maps and perform filtering in the time domain. For example, Cai Zixing et al., in "Real-time Detection of Dynamic Obstacles Based on LiDAR," proposed dividing the environment into a raster map. If an obstacle appears three times consecutively in the same map raster (i.e., the raster is occupied for three consecutive frames), then the obstacles in these raster grids are all static obstacles. If an obstacle appears twice consecutively in the same map raster (i.e., the raster is occupied for two consecutive frames), then the obstacles in these raster grids are all potential dynamic obstacles. The status of the eight surrounding raster grids is then used to determine whether the potential dynamic obstacle is static. If an obstacle appears only once in the same map raster (i.e., the raster is occupied for only one frame), then the obstacles in these raster grids are dynamic obstacles.

[0004] However, the shortcomings of existing dynamic obstacle detection schemes are very obvious: on the one hand, the existing scheme requires at least three consecutive frames of point cloud data to determine whether it is a dynamic obstacle, and it also requires three consecutive frames of data to remove dynamic points in the middle frames, which cannot truly meet the real-time requirements of mapping; on the other hand, in the mapping process, any two frames of point cloud need to be registered. If there are dynamic points in the registered point cloud, the registration accuracy will be significantly reduced, which is unavoidable during registration. Therefore, the application of existing dynamic obstacle detection methods in mapping cannot completely eliminate the influence of dynamic points. Summary of the Invention

[0005] One advantage of this invention is that it provides a method, system, and apparatus for removing dynamic points for autonomous mobile platforms, which can eliminate the influence of dynamic points on point cloud registration, help to accurately remove dynamic points from point cloud information, and facilitate the improvement of mapping accuracy.

[0006] Another advantage of the present invention is that it provides a method, system and device for removing dynamic points for autonomous mobile platforms. In one embodiment of the present invention, the method for removing dynamic points for autonomous mobile platforms can use a camera to compensate for some inherent defects of radar, such as discretized sampling and inability to accurately match.

[0007] Another advantage of the present invention is that it provides a method, system and device for removing dynamic points for autonomous mobile platforms, wherein, in one embodiment of the invention, the method for removing dynamic points for autonomous mobile platforms can be combined with cameras and radar to achieve better dynamic point removal results.

[0008] Another advantage of the present invention is that it provides a method, system and device for removing dynamic points for autonomous mobile platforms, wherein, in one embodiment of the present invention, the method for removing dynamic points for autonomous mobile platforms can remove dynamic points in the current frame data in real time using only two consecutive frames of data, so as to meet the real-time requirements of mapping.

[0009] Another advantage of the present invention is that it provides a method, system and device for removing dynamic points for autonomous mobile platforms, wherein, in one embodiment of the invention, the method for removing dynamic points for autonomous mobile platforms can utilize continuous dynamic regions in image information to accurately distinguish the boundary between dynamic points and static points, so as to remove dynamic points more accurately.

[0010] Another object of the present invention is to provide a method, system, and apparatus for removing dynamic points for autonomous mobile platforms, wherein complex structures or algorithms are not required to achieve the above-mentioned object. Therefore, the present invention successfully and effectively provides a solution that not only provides a simple method, system, and apparatus for removing dynamic points for autonomous mobile platforms, but also increases the practicality and reliability of the method, system, and apparatus for removing dynamic points for autonomous mobile platforms.

[0011] To achieve at least one of the above-mentioned inventive objectives or other objectives and advantages, the present invention provides a method for removing dynamic points for an autonomous mobile platform, comprising the steps of:

[0012] S100: Obtain image information and point cloud information simultaneously collected by the camera and radar configured on the autonomous mobile platform at the current frame timestamp and the previous frame timestamp, respectively, so as to obtain the current frame image data, the previous frame image data and the current frame point cloud data accordingly.

[0013] S200: Based on the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform, feature point processing is performed on the current frame image data and the previous frame image data to determine the dynamic region in the current frame image data;

[0014] S300: Project the current frame point cloud data onto the image plane of the camera, so as to use the point cloud within the dynamic region projected onto the current frame image data as potential dynamic points; and

[0015] S400: By clustering the potential dynamic point, remove the potential dynamic point from the point cloud clusters that are close to the camera.

[0016] According to one embodiment of this application, the sampling frequency of the camera is the same as and synchronized with the detection frequency of the radar, and the current frame timestamp and the previous frame timestamp are both timestamps of the radar.

[0017] According to an embodiment of this application, step S200 includes the following steps:

[0018] S210: Perform trajectory extrapolation on the data collected by the inertial measurement unit and wheel speedometer configured on the autonomous mobile platform to obtain the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform;

[0019] S220: Based on the relative pose relationship of the autonomous mobile platform, feature point matching is performed on the current frame image data and the previous frame image data to obtain the transformation matrix of corresponding feature points between the current frame image data and the previous frame image data; and

[0020] S230: Based on the transformation matrix of the corresponding feature point, determine the dynamic region in the current frame image data through transmission transformation.

[0021] According to an embodiment of this application, step S210 includes the following steps:

[0022] Obtain the pose of the autonomous mobile platform at the timestamp of the previous frame, and use it as the pose of the autonomous mobile platform in the previous frame.

[0023] Linear interpolation is performed on the inertial data acquired by the inertial measurement unit and the wheel speed data acquired by the wheel speed meter to obtain the wheel speed and angular velocity at the current frame timestamp; and

[0024] Based on the pose of the autonomous mobile platform in the previous frame, the wheel speed and angular velocity between the timestamp of the previous frame and the current timestamp are integrated to obtain the pose of the autonomous mobile platform in the current frame, thereby obtaining the relative pose relationship of the autonomous mobile platform.

[0025] According to an embodiment of this application, step S220 includes the following steps:

[0026] Filter the current frame image data and the previous frame image data respectively;

[0027] Feature point extraction is performed on the filtered current frame image data and the previous frame image data to obtain each feature point in the filtered current frame image data and the previous frame image data; and

[0028] Using the relative pose relationship of the autonomous mobile platform as the initial value, the transformation matrix of the corresponding feature points between the current frame image data and the previous frame image data is calculated.

[0029] According to one embodiment of this application, while calculating the transformation matrix using the RANSC matching model, dynamic feature points in the current frame image data are also removed.

[0030] According to an embodiment of this application, step S230 includes the following steps:

[0031] Based on the transformation matrix of the corresponding feature point, a transmission transformation is performed on the previous frame image data and the current frame image data to obtain the light intensity change of each pixel in the current frame image data after the transmission transformation.

[0032] Compare whether the change in light intensity of each pixel in the current frame image data after transmission transformation is greater than a light intensity threshold; and

[0033] The pixel region where the change in light intensity is greater than the light intensity threshold is determined as the dynamic region of the current frame image data.

[0034] According to an embodiment of this application, step S300 includes the following steps:

[0035] Project all laser points in the current frame point cloud data onto the image plane of the camera to obtain the pixel points in the current frame image data corresponding to all laser points;

[0036] Determine whether the pixel corresponding to each laser point is within the dynamic region; and

[0037] In response to a pixel corresponding to a certain laser point being located within the dynamic region, that laser point is taken as the potential dynamic point.

[0038] According to an embodiment of this application, step S400 includes the following steps:

[0039] Clustering is performed on the potential dynamic points corresponding to different dynamic regions to obtain different point cloud clusters;

[0040] Based on the depth values ​​in the current frame point cloud data, calculate the average distance between each point cloud cluster and the camera; and

[0041] Remove the potential dynamic point from the point cloud cluster that is close to the camera from the current frame point cloud data.

[0042] According to another aspect of this application, one embodiment of this application further provides a de-dynamic point system for an autonomous mobile platform, wherein the autonomous mobile platform is equipped with a camera and radar, wherein the de-dynamic point system for the autonomous mobile platform includes components that are communicatively connected to each other:

[0043] A data acquisition module is used to acquire image information and point cloud information simultaneously collected by the camera and the radar at the current frame timestamp and the previous frame timestamp, respectively, so as to obtain the current frame image data, the previous frame image data and the current frame point cloud data accordingly.

[0044] A feature point processing module is used to perform feature point processing on the current frame image data and the previous frame image data based on the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform, so as to determine the dynamic region in the current frame image data.

[0045] A point cloud projection module is used to project the current frame point cloud data onto the image plane of the camera, so as to use the point cloud within the dynamic region projected onto the current frame image data as potential dynamic points; and

[0046] A clustering processing module is used to remove potential dynamic points in point cloud clusters that are close to the camera by performing clustering processing on the potential dynamic points.

[0047] According to one embodiment of this application, the feature point processing module includes a trajectory extrapolation module, a feature point matching module, and a region determination module that are communicatively connected to each other. The trajectory extrapolation module performs trajectory extrapolation on data collected via an inertial measurement unit and wheel speedometer configured on the autonomous mobile platform to obtain the relative pose relationship between the current frame timestamp and the previous frame timestamp. The feature point matching module performs feature point matching on the current frame image data and the previous frame image data based on the relative pose relationship of the autonomous mobile platform to obtain a transformation matrix of corresponding feature points between the current frame image data and the previous frame image data. The region determination module determines the dynamic region in the current frame image data through transmission transformation based on the transformation matrix of the corresponding feature points.

[0048] According to one embodiment of this application, the trajectory extrapolation module includes a pose acquisition module, a linear interpolation module, and a velocity integration module that are communicatively connected to each other. The pose acquisition module is used to acquire the pose of the autonomous mobile platform at the previous frame timestamp as the previous frame pose of the autonomous mobile platform. The linear interpolation module is used to perform linear interpolation on the inertial data collected by the inertial measurement unit and the wheel speed data collected by the wheel speed meter to obtain the wheel speed and angular velocity at the current frame timestamp. The velocity integration module is used to integrate the wheel speed and angular velocity between the previous frame timestamp and the current timestamp based on the previous frame pose of the autonomous mobile platform to obtain the current frame pose of the autonomous mobile platform, thereby obtaining the relative pose relationship of the autonomous mobile platform.

[0049] According to one embodiment of this application, the feature point matching module includes a data filtering module, a feature point extraction module, and a matrix calculation module that are communicatively connected to each other. The data filtering module is used to filter the current frame image data and the previous frame image data respectively. The feature point extraction module is used to extract feature points from the filtered current frame image data and the previous frame image data to obtain each feature point in the filtered current frame image data and the previous frame image data. The matrix calculation module is used to calculate the transformation matrix of the corresponding feature points between the current frame image data and the previous frame image data, using the relative pose relationship of the autonomous mobile platform as the initial value.

[0050] According to one embodiment of this application, the region determination module includes a transmission transformation module, a change comparison module, and a region determination module that are communicatively connected to each other. The transmission transformation module is used to perform transmission transformation on the previous frame image data and the current frame image data according to the transformation matrix of the corresponding feature points to obtain the light intensity change of each pixel in the current frame image data after transmission transformation. The change comparison module is used to compare whether the light intensity change of each pixel in the current frame image data after transmission transformation is greater than a light intensity threshold. The region determination module is used to determine the pixel region whose light intensity change is greater than the light intensity threshold as the dynamic region of the current frame image data.

[0051] According to one embodiment of this application, the point cloud projection module includes a laser point projection module, a laser point judgment module, and a dynamic point screening module that are communicatively connected to each other. The laser point projection module is used to project all laser points in the current frame point cloud data onto the image plane of the camera to obtain the pixel points corresponding to all laser points in the current frame image data. The laser point judgment module is used to determine whether the pixel point corresponding to each laser point is in the dynamic region. The dynamic point screening module is used to identify a laser point as a potential dynamic point in response to the pixel point corresponding to a certain laser point being in the dynamic region.

[0052] According to one embodiment of this application, the clustering processing module includes a point cloud clustering module, a distance calculation module, and a dynamic point removal module that are communicatively connected to each other. The point cloud clustering module is used to perform clustering processing on the potential dynamic points corresponding to different dynamic regions to obtain different point cloud clusters. The distance calculation module is used to calculate the average distance between each point cloud cluster and the camera based on the depth value in the current frame point cloud data. The dynamic point removal module is used to remove the potential dynamic points in the point cloud clusters that are close to the camera from the current frame point cloud data.

[0053] According to another aspect of this application, one embodiment of this application further provides an electronic device, including:

[0054] A processor, used to execute program instructions; and

[0055] A memory, wherein the memory is configured to store program instructions executable by the processor to implement all or part of the steps in a method for removing dynamic points for an autonomous mobile platform, wherein the method for removing dynamic points for an autonomous mobile platform includes the steps of:

[0056] S100: Obtain image information and point cloud information simultaneously collected by the camera and radar configured on the autonomous mobile platform at the current frame timestamp and the previous frame timestamp, respectively, so as to obtain the current frame image data, the previous frame image data and the current frame point cloud data accordingly.

[0057] S200: Based on the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform, feature point processing is performed on the current frame image data and the previous frame image data to determine the dynamic region in the current frame image data;

[0058] S300: Project the current frame point cloud data onto the image plane of the camera, so as to use the point cloud within the dynamic region projected onto the current frame image data as potential dynamic points; and

[0059] S400: By clustering the potential dynamic point, remove the potential dynamic point from the point cloud clusters that are close to the camera.

[0060] The further objects and advantages of the invention will become fully apparent from the following description and accompanying drawings.

[0061] These and other objects, features and advantages of the present invention will be fully realized through the following detailed description, drawings and claims. Attached Figure Description

[0062] Figure 1 This is a flowchart illustrating a method for removing dynamic points for an autonomous mobile platform according to an embodiment of the present invention.

[0063] Figures 2 to 5 A flowchart illustrating the feature point processing steps in the method for removing dynamic points for an autonomous mobile platform according to the above embodiments of the present invention is shown.

[0064] Figure 6 A flowchart illustrating the point cloud projection step in the method for removing dynamic points for an autonomous mobile platform according to the above embodiments of the present invention is shown.

[0065] Figure 7 A flowchart illustrating the clustering process steps in the de-dynamic point method for an autonomous mobile platform according to the above embodiments of the present invention is shown.

[0066] Figure 8 This is a block diagram of a de-dynamic point system for an autonomous mobile platform according to an embodiment of the present invention.

[0067] Figure 9 A block diagram of a feature point processing module in a de-dynamic point system for an autonomous mobile platform according to the above embodiment of the present invention is shown.

[0068] Figure 10 An example is shown of the de-dynamic point system for an autonomous mobile platform according to the above embodiments of the present invention being applied to an autonomous mobile platform.

[0069] Figure 11 A block diagram of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation

[0070] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0071] In this invention, the term "a" in the claims and specification should be understood as "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple. Unless explicitly indicated in the disclosure of this invention that the number of the element is only one, the term "a" should not be construed as unique or single, and the term "a" should not be construed as a limitation on the quantity.

[0072] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0073] Currently, existing dynamic point (or dynamic obstacle) detection schemes are typically based on rasterized maps and perform filtering in the time domain. However, their drawbacks are significant: firstly, these schemes require at least three consecutive frames of point cloud data to determine whether an obstacle is dynamic, and also require three consecutive frames to remove dynamic points from intermediate frames, failing to truly meet the real-time requirements of mapping; secondly, during the mapping process, any two frames of point cloud data need to be registered. If dynamic points are present in the registered point cloud, the registration accuracy will be significantly reduced, which is unavoidable during registration. Therefore, existing dynamic obstacle detection methods cannot completely eliminate the influence of dynamic points in mapping. To completely eliminate the influence of dynamic points on point cloud registration, this application proposes a dynamic point removal method, system, and device for autonomous mobile platforms. This method utilizes a camera to compensate for some inherent defects of radar, such as discretized sampling and inaccurate matching.

[0074] Specifically, firstly, the camera's sampling frequency is made the same as the lidar's detection frequency, and image data and point cloud data are acquired synchronously. Next, based on the trajectory derivation results of the autonomous mobile platform, feature points on consecutive frames are matched to obtain the correspondence. Then, through transmission transformation, each pixel of one frame of image data is matched with each pixel of another frame of image data to identify areas with large changes in light intensity; objects in these areas are considered dynamic objects. Finally, by projecting the point cloud data obtained from the lidar onto the camera's image plane, the point clouds projected into the aforementioned areas are clustered, and the foreground portion is identified as dynamic points.

[0075] It is worth noting that although both radar (such as lidar) and cameras (such as monocular or binocular fisheye cameras) can be used for positioning, they are both affected by dynamic points. Therefore, radar and cameras each have their own methods for removing dynamic points. In particular, since the field of view of lidar is usually 360°, while the field of view of a camera is generally less than 180°, existing technologies do not combine the two to remove dynamic points. However, in the application scenario of autonomous mobile platforms, lidar is generally partially obscured by the platform body (such as the vehicle body), allowing the remaining field of view of the lidar to largely coincide with that of the camera. Therefore, the dynamic point removal method for autonomous mobile platforms described in this application can combine both cameras and radar sensors to achieve better dynamic point removal results.

[0076] illustrative method

[0077] Refer to the attached diagram in the instruction manual. Figures 1 to 7 As shown, a method for removing dynamic points for an autonomous mobile platform according to an embodiment of the present invention is illustrated. Specifically, as Figure 1 As shown, the method for removing dynamic points for an autonomous mobile platform may include the following steps:

[0078] S100: Obtain image information and point cloud information simultaneously collected by the camera and radar configured on the autonomous mobile platform at the current frame timestamp and the previous frame timestamp, respectively, so as to obtain the current frame image data, the previous frame image data and the current frame point cloud data accordingly.

[0079] S200: Based on the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform, feature point processing is performed on the current frame image data and the previous frame image data to determine the dynamic region in the current frame image data;

[0080] S300: Project the current frame point cloud data onto the image plane of the camera, so that the point cloud within the dynamic region projected onto the current frame image data is used as potential dynamic points; and

[0081] S400: By clustering the potential dynamic points, the potential dynamic points in the point cloud clusters that are close to the camera are removed.

[0082] It is worth noting that the camera configured on this autonomous mobile platform is implemented as a fisheye camera so that its field of view coincides as closely as possible with the effective field of view of the lidar, which helps to more comprehensively remove dynamic points for better mapping. It is understood that the fisheye camera described in this application can be a monocular fisheye camera or a binocular fisheye camera.

[0083] Furthermore, according to the above embodiments of this application, the timestamp in step S100 of the method for removing dynamic points for an autonomous mobile platform can, but is not limited to, be implemented as the radar's timestamp, simply by adjusting the camera's sampling frequency to be the same as and synchronized with the radar's detection frequency. Of course, in other examples of this application, the timestamp in step S100 can also be implemented as the camera's timestamp; in this case, it is necessary to adjust the radar's detection frequency to be the same as and synchronized with the camera's sampling frequency.

[0084] It is worth mentioning that, since step S200 of the method for removing dynamic points for an autonomous mobile platform described in this application requires the relative pose relationship between the current frame timestamp and the previous frame timestamp, the method for removing dynamic points for an autonomous mobile platform described in this application can obtain the phase pose relationship between different frame timestamps through trajectory extrapolation. In this case, it is necessary to configure an inertial measurement unit and wheel speedometer on the autonomous mobile platform. Of course, in other examples of this application, the method for removing dynamic points for an autonomous mobile platform can also obtain the phase pose relationship between different frame timestamps through visual odometry. In this case, only the image information acquired by the camera is needed to obtain the phase pose relationship between different frame timestamps, without the need for additional sensors such as inertial measurement units and wheel speedometers.

[0085] For example, the autonomous mobile platform can typically be further configured with an inertial measurement unit and wheel speedometers to perform trajectory extrapolation using data collected via the inertial measurement unit and wheel speedometers, thereby accurately obtaining the phase pose relationship of the autonomous mobile platform between different frame timestamps. Specifically, as... Figure 2 As shown, step S200 of the method for removing dynamic points for an autonomous mobile platform according to this application may include the following steps:

[0086] S210: Perform trajectory extrapolation on the data collected by the inertial measurement unit and wheel speedometer configured on the autonomous mobile platform to obtain the relative pose relationship of the autonomous mobile platform between the current frame timestamp and the previous frame timestamp;

[0087] S220: Based on the relative pose relationship of the autonomous mobile platform, feature point matching is performed on the current frame image data and the previous frame image data to obtain the transformation matrix of corresponding feature points between the current frame image data and the previous frame image data; and

[0088] S230: Based on the transformation matrix of the corresponding feature points, determine the dynamic region in the current frame image data through transmission transformation.

[0089] More specifically, such as Figure 3 As shown, step S210 of the method for removing dynamic points for an autonomous mobile platform according to this application may include the following steps:

[0090] S211: Obtain the pose of the autonomous mobile platform at the timestamp of the previous frame, and use it as the pose of the autonomous mobile platform in the previous frame.

[0091] S212: Perform linear interpolation on the inertial data acquired by the inertial measurement unit and the wheel speed data acquired by the wheel speed meter to obtain the wheel speed and angular velocity at the current frame timestamp; and

[0092] S213: Based on the previous frame pose of the autonomous mobile platform, integrate the wheel speed and angular velocity between the previous frame timestamp and the current frame timestamp to obtain the current frame pose of the autonomous mobile platform, thereby obtaining the relative pose relationship of the autonomous mobile platform.

[0093] It is understood that when the previous frame timestamp is the first frame timestamp, the pose (i.e., position and orientation) of the autonomous mobile platform is zero. Furthermore, the current frame pose of the autonomous mobile platform in this application is implemented as the pose of the autonomous mobile platform at that current frame timestamp.

[0094] It is worth noting that in the above embodiments of this application, since the timestamps of the inertial measurement unit and the wheel speedometer are usually not the same as or synchronized with the timestamp of the radar, step S212 of the method for removing dynamic points for an autonomous mobile platform in this application uses linear interpolation to obtain the wheel speed and angular velocity of the current frame. Of course, in other examples of this application, when the autonomous mobile platform can synchronize the timestamps of the inertial measurement unit, the wheel speedometer, the radar, and the camera, step S212 of the method for removing dynamic points for an autonomous mobile platform in this application can be omitted, and the relative pose relationship between the current frame timestamp and the previous frame timestamp can still be obtained.

[0095] According to the above embodiments of this application, as Figure 4 As shown, step S220 of the method for removing dynamic points for an autonomous mobile platform may include the following steps:

[0096] S221: Filter the current frame image data and the previous frame image data respectively;

[0097] S222: Perform feature point extraction processing on the filtered current frame image data and the previous frame image data to obtain each feature point in the filtered current frame image data and the previous frame image data; and

[0098] S223: Using the relative pose relationship of the autonomous mobile platform as the initial value, calculate the transformation matrix of the corresponding feature points between the current frame image data and the previous frame image data.

[0099] For example, in step S221 of the method for removing dynamic points for an autonomous mobile platform described in this application: the current frame image data and the previous frame image data are filtered using a Gaussian smoothing filter model.

[0100] In step S222 of the method for removing dynamic points for an autonomous mobile platform described in this application: feature points in the filtered current frame image data and the previous frame image data are extracted using the FAST corner extraction model to calculate the ORB descriptor for each feature point.

[0101] In step S223 of the method for removing dynamic points for an autonomous mobile platform described in this application: using a RANSC matching model, the correspondence between feature points between the current frame image data and the previous frame image data is found using the relative pose relationship of the autonomous mobile platform as the initial value, so as to obtain the transformation matrix of the corresponding feature points.

[0102] Preferably, while using the RANSC matching model to calculate the transformation matrix, the RANSC matching model can also remove dynamic feature points in the current frame image data to ensure the accuracy of the transformation matrix for the corresponding feature points.

[0103] It is worth noting that, according to the above embodiments of this application, such as Figure 5 As shown, step S230 of the method for removing dynamic points for an autonomous mobile platform may include the following steps:

[0104] S231: Based on the transformation matrix of the corresponding feature points, perform a transmission transformation on the previous frame image data and the current frame image data to obtain the light intensity change of each pixel in the current frame image data after the transmission transformation.

[0105] S232: Compare whether the change in light intensity of each pixel in the current frame image data after transmission transformation is greater than a light intensity threshold; and

[0106] S233: The pixel region where the change in light intensity is greater than the light intensity threshold is determined as the dynamic region of the current frame image data.

[0107] Preferably, the light intensity threshold can be set according to the ambient brightness of the environment where the autonomous mobile platform is located. For example, when the environment is bright, the light intensity threshold is larger; while when the environment is dark, the light intensity threshold is smaller.

[0108] It is worth mentioning that the dynamic point removal method for autonomous mobile platforms described in this application utilizes image information to obtain a continuous dynamic region, which allows for more accurate differentiation of the boundaries between dynamic and static points. However, since point cloud data is discrete, existing technologies that only use this point cloud data to remove dynamic points also obtain a discrete dynamic region, failing to accurately differentiate the boundaries between dynamic and static points, resulting in low accuracy in dynamic point removal.

[0109] Furthermore, after obtaining the dynamic region on the current frame image, the de-dynamic point method for the autonomous mobile platform can project all laser points in the current frame point cloud data onto the image plane of the camera to obtain the pixel corresponding to the laser point. If the pixel corresponding to the laser point is located within the dynamic region on the current frame image, then the laser point may be a dynamic point, i.e., a potential dynamic point.

[0110] For example, such as Figure 6 As shown, step S300 of the method for removing dynamic points for an autonomous mobile platform according to this application may include the following steps:

[0111] S310: Project all laser points in the current frame point cloud data onto the image plane of the camera to obtain the pixel points in the current frame image data corresponding to all the laser points;

[0112] S320: Determine whether the pixel corresponding to each laser point is within the dynamic region; and

[0113] S330: In response to a pixel corresponding to a certain laser point being located within the dynamic region, the certain laser point is taken as the potential dynamic point.

[0114] It is worth noting that since the same dynamic object moves between the previous frame timestamp and the current frame timestamp, the same dynamic object usually corresponds to two dynamic regions. In other words, the pixel region in the current frame image data where the light intensity changes may be the location of the dynamic object at the current frame timestamp, mainly because the dynamic object moves to this region and obscures the background, causing a change in light intensity; at the same time, the pixel region in the current frame image data where the light intensity changes may also be the location of the dynamic object at the previous frame timestamp, mainly because the dynamic object leaves this region and exposes the background, causing a change in light intensity. Therefore, the potential dynamic point is not necessarily a true dynamic point. Therefore, the dynamic point removal method for autonomous mobile platforms described in this application needs to further examine the potential dynamic points to determine whether the potential dynamic points are true dynamic points by judging the foreground and background, and delete the potential dynamic points that are determined to be true dynamic points.

[0115] Specifically, such as Figure 8 As shown, step S400 of the method for removing dynamic points for an autonomous mobile platform according to this application may include the following steps:

[0116] S410: Cluster the potential dynamic points corresponding to different dynamic regions to obtain different point cloud clusters;

[0117] S420: Calculate the average distance between each point cloud cluster and the camera based on the depth values ​​in the current frame point cloud data; and

[0118] S430: Remove potential dynamic points within point cloud clusters that are close to the camera from the current frame point cloud data.

[0119] For example, a dynamic object typically corresponds to two dynamic regions. The method for removing dynamic points for an autonomous mobile platform described in this application clusters the point clouds within the two dynamic regions, resulting in two point cloud clusters. One of these clusters represents the foreground (a dynamic object such as an obstacle), and the other represents the background. In this case, the application can distinguish between the foreground and background by calculating the centroids of the two point cloud clusters and delete the point clouds belonging to the foreground, thus achieving the purpose of removing dynamic points. It is understood that the centroid of the point cloud cluster can refer to the average distance between the point cloud cluster and the camera. That is, when the point cloud data is projected onto the camera coordinate system, the point cloud with a larger z-axis value (i.e., farther from the camera) is the background, and the point cloud with a smaller z-axis value (i.e., closer to the camera) is the foreground.

[0120] In summary, the dynamic point removal method for autonomous mobile platforms described in this application can remove dynamic points from the current frame point cloud data in real time using only the current frame data and the previous frame data. In contrast, existing technologies require at least three frames of data and can only remove dynamic points from intermediate frames, making real-time dynamic point removal impossible. Furthermore, since real-time performance is essential for mapping, the dynamic point removal methods in existing technologies are not suitable for real-time mapping and are difficult to apply on autonomous mobile platforms.

[0121] Indicative System

[0122] Refer to the attached diagram in the instruction manual. Figures 8 to 10 As shown, a de-dynamic point system for an autonomous mobile platform according to an embodiment of the present invention is illustrated, wherein the autonomous mobile platform 50 is equipped with a camera 51 and a radar 52. Specifically, as Figure 8 and Figure 10As shown, the de-dynamic point system 1 for the autonomous mobile platform may include a data acquisition module 10, a feature point processing module 20, a point cloud projection module 30, and a clustering processing module 40 that are communicatively connected to each other. The data acquisition module 10 is used to acquire image information and point cloud information simultaneously collected by the camera 51 and the radar 52 at the current frame timestamp and the previous frame timestamp, respectively, to correspondingly obtain the current frame image data, the previous frame image data, and the current frame point cloud data. The feature point processing module 20 is used to process the data according to the autonomous mobile platform 50... Based on the relative pose relationship between the current frame timestamp and the previous frame timestamp, feature point processing is performed on the current frame image data and the previous frame image data to determine the dynamic region in the current frame image data; wherein the point cloud projection module 30 is used to project the current frame point cloud data onto the image plane of the camera 52, so as to take the point cloud in the dynamic region projected onto the current frame image data as potential dynamic points; wherein the clustering processing module 40 is used to remove the potential dynamic points in the point cloud clusters that are close to the camera 52 by performing clustering processing on the potential dynamic points.

[0123] It is worth noting that, in one example of this application, such as Figure 9 and Figure 10 As shown, the feature point processing module 20 may include a trajectory extrapolation module 21, a feature point matching module 22, and a region determination module 23 that are communicatively connected to each other. The trajectory extrapolation module 21 is used to extrapolate the trajectory of the data collected by the inertial measurement unit and wheel speedometer configured on the autonomous mobile platform 50 to obtain the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform 50. The feature point matching module 22 is used to perform feature point matching on the current frame image data and the previous frame image data according to the relative pose relationship of the autonomous mobile platform 50 to obtain the transformation matrix of corresponding feature points between the current frame image data and the previous frame image data. The region determination module 23 is used to determine the dynamic region in the current frame image data by transmission transformation according to the transformation matrix of the corresponding feature points.

[0124] In one example of this application, such as Figure 9 and Figure 10As shown, the trajectory extrapolation module 21 includes a pose acquisition module 211, a linear interpolation module 212, and a velocity integration module 213 that are communicatively connected to each other. The pose acquisition module 211 is used to acquire the pose of the autonomous mobile platform 50 at the previous frame timestamp, as the previous frame pose of the autonomous mobile platform. The linear interpolation module 212 is used to perform linear interpolation on the inertial data collected by the inertial measurement unit and the wheel speed data collected by the wheel speed meter to obtain the wheel speed and angular velocity at the current frame timestamp. The velocity integration module 213 is used to integrate the wheel speed and angular velocity between the previous frame timestamp and the current timestamp based on the previous frame pose of the autonomous mobile platform to obtain the current frame pose of the autonomous mobile platform 50, thereby obtaining the relative pose relationship of the autonomous mobile platform 50.

[0125] In one example of this application, such as Figure 9 and Figure 10 As shown, the feature point matching module 22 includes a data filtering module 221, a feature point extraction module 222, and a matrix calculation module 223 that are communicatively connected to each other. The data filtering module 221 is used to filter the current frame image data and the previous frame image data respectively. The feature point extraction module 222 is used to extract feature points from the filtered current frame image data and the previous frame image data to obtain each feature point in the filtered current frame image data and the previous frame image data. The matrix calculation module 223 is used to calculate the transformation matrix of the corresponding feature points between the current frame image data and the previous frame image data, using the relative pose relationship of the autonomous mobile platform as the initial value.

[0126] In one example of this application, such as Figure 9 and Figure 10 As shown, the region determination module 23 includes a transmission transformation module 231, a change amount comparison module 232, and a region determination module 233 that are communicatively connected to each other. The transmission transformation module 231 is used to perform transmission transformation on the previous frame image data and the current frame image data according to the transformation matrix of the corresponding feature point, so as to obtain the light intensity change of each pixel in the current frame image data after transmission transformation. The change amount comparison module 232 is used to compare whether the light intensity change of each pixel in the current frame image data after transmission transformation is greater than a light intensity threshold. The region determination module 233 is used to determine the pixel region whose light intensity change is greater than the light intensity threshold as the dynamic region of the current frame image data.

[0127] In one example of this application, such as Figure 8 and Figure 10As shown, the point cloud projection module 30 includes a laser point projection module 31, a laser point judgment module 32, and a dynamic point screening module 33 that are communicatively connected to each other. The laser point projection module 31 is used to project all laser points in the current frame point cloud data onto the image plane of the camera to obtain the pixel points corresponding to all laser points in the current frame image data. The laser point judgment module 32 is used to determine whether the pixel point corresponding to each laser point is in the dynamic region. The dynamic point screening module 33 is used to identify a laser point as a potential dynamic point in response to the pixel point corresponding to a certain laser point being in the dynamic region.

[0128] In one example of this application, such as Figure 8 and Figure 10 As shown, the clustering processing module 40 includes a point cloud clustering module 41, a distance calculation module 42, and a dynamic point removal module 43 that are communicatively connected to each other. The point cloud clustering module 41 is used to perform clustering processing on the potential dynamic points corresponding to different dynamic regions to obtain different point cloud clusters. The distance calculation module 42 is used to calculate the average distance between each point cloud cluster and the camera based on the depth value in the current frame point cloud data. The dynamic point removal module 43 is used to remove the potential dynamic points in the point cloud clusters that are close to the camera from the current frame point cloud data.

[0129] Indicative electronic devices

[0130] Below, for reference Figure 11 To describe the electronic device according to embodiments of the present invention ( Figure 11 A block diagram of an electronic device according to an embodiment of the present invention is shown. Figure 11 As shown, the electronic device 60 includes one or more processors 61 and a memory 62.

[0131] The processor 61 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 60 to perform desired functions.

[0132] The memory 62 may include one or more computational program products, which may include various forms of computationally readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computational program instructions may be stored on the computationally readable storage medium, and the processor 61 may execute the program instructions to implement the methods of the various embodiments of the present invention described above and / or other desired functions.

[0133] In one example, such as Figure 11 As shown, the electronic device 60 may also include an input device 63 and an output device 64, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0134] For example, the input device 63 could be, for instance, a camera module for acquiring image data or video data, etc.

[0135] The output device 64 can output various information to the outside, including classification results. The output device 64 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0136] Of course, for the sake of simplicity, Figure 11 Only some of the components of the electronic device 60 relevant to the present invention are shown, omitting components such as buses, input / output interfaces, etc. In addition, the electronic device 60 may include any other suitable components depending on the specific application.

[0137] Schematic computing program products

[0138] In addition to the methods and apparatus described above, embodiments of the present invention may also be computing program products, which include computing program instructions that, when executed by a processor, cause the processor to perform the steps of the methods according to various embodiments of the present invention described in the "Exemplary Methods" section above.

[0139] The computing program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of the present invention. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0140] Furthermore, embodiments of the present invention may also be computationally readable storage media storing computation program instructions, which, when executed by a processor, cause the processor to perform the steps of the method described above in this specification.

[0141] The computationally readable storage medium may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. Readable storage media may include, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0142] The basic principles of the present invention have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in the present invention are merely examples and not limitations, and should not be considered as essential features of each embodiment of the present invention. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the present invention to the necessity of employing the aforementioned specific details.

[0143] The block diagrams of devices, apparatuses, devices, and systems involved in this invention are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0144] It should also be noted that in the apparatus, device, and method of the present invention, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of the present invention.

[0145] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the invention. Therefore, the invention is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0146] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and explained in the embodiments, and any variations or modifications may be made to the implementation of the present invention without departing from the stated principles.

Claims

1. A method for removing dynamic points in an autonomous mobile platform, characterized in that, Including the following steps: S100: Obtain image information and point cloud information simultaneously collected by the camera and radar configured on the autonomous mobile platform at the current frame timestamp and the previous frame timestamp, respectively, so as to obtain the current frame image data, the previous frame image data and the current frame point cloud data accordingly; wherein, the sampling frequency of the camera is the same as and synchronized with the detection frequency of the radar, and the current frame timestamp and the previous frame timestamp are both timestamps of the radar. S200: Based on the relative pose relationship between the current frame timestamp and the previous frame timestamp, the autonomous mobile platform performs feature point processing on the current frame image data and the previous frame image data; wherein, firstly, using the relative pose relationship as an initial value, the transformation matrix of corresponding feature points between the current frame image data and the previous frame image data is calculated, and then the previous frame image data and the current frame image data are subjected to transmission transformation according to the transformation matrix to obtain the light intensity change of each pixel in the current frame image data after transmission transformation. By comparing the light intensity change with a light intensity threshold, the pixel region with the light intensity change greater than the light intensity threshold is determined as a continuous dynamic region in the current frame image data. S300: Project the current frame point cloud data onto the image plane of the camera, so as to use the point cloud within the dynamic region projected onto the current frame image data as potential dynamic points; and S400: Cluster the potential dynamic points to obtain different point cloud clusters, and calculate the average distance between each point cloud cluster and the camera based on the depth value of the different point cloud clusters in the camera coordinate system. Distinguish between point cloud clusters that are close to the camera and point cloud clusters that are far from the camera, and remove potential dynamic points in point cloud clusters that are close to the camera.

2. The method for removing dynamic points for an autonomous mobile platform as described in claim 1, wherein, Step S200 includes the following steps: S210: Perform trajectory extrapolation on the data collected by the inertial measurement unit and wheel speedometer configured on the autonomous mobile platform to obtain the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform; S220: Based on the relative pose relationship of the autonomous mobile platform, feature point matching is performed on the current frame image data and the previous frame image data to obtain the transformation matrix of corresponding feature points between the current frame image data and the previous frame image data; and S230: Based on the transformation matrix of the corresponding feature point, determine the dynamic region in the current frame image data through transmission transformation.

3. The method for removing dynamic points for an autonomous mobile platform as described in claim 2, wherein, Step S210 includes the following steps: Obtain the pose of the autonomous mobile platform at the timestamp of the previous frame, and use it as the pose of the autonomous mobile platform in the previous frame. Linear interpolation is performed on the inertial data collected by the inertial measurement unit and the wheel speed data collected by the wheel speed meter to obtain the wheel speed and angular velocity at the current frame timestamp; as well as Based on the pose of the autonomous mobile platform in the previous frame, the wheel speed and angular velocity between the timestamp of the previous frame and the timestamp of the current frame are integrated to obtain the pose of the autonomous mobile platform in the current frame, thereby obtaining the relative pose relationship of the autonomous mobile platform.

4. The method for removing dynamic points for an autonomous mobile platform as described in claim 2, wherein, Step S220 includes the following steps: Filter the current frame image data and the previous frame image data respectively; Feature point extraction is performed on the filtered current frame image data and the previous frame image data to obtain each feature point in the filtered current frame image data and the previous frame image data; and Using the relative pose relationship of the autonomous mobile platform as the initial value, the transformation matrix of the corresponding feature points between the current frame image data and the previous frame image data is calculated.

5. The method for removing dynamic points for an autonomous mobile platform as described in claim 4, wherein, The RANSC matching model calculates the transformation matrix while also removing dynamic feature points from the current frame image data.

6. The method for removing dynamic points for an autonomous mobile platform as described in claim 2, wherein, Step S230 includes the following steps: Based on the transformation matrix of the corresponding feature point, a transmission transformation is performed on the previous frame image data and the current frame image data to obtain the light intensity change of each pixel in the current frame image data after the transmission transformation. Compare whether the change in light intensity of each pixel in the current frame image data after transmission transformation is greater than a light intensity threshold. as well as The pixel region where the change in light intensity is greater than the light intensity threshold is determined as the dynamic region of the current frame image data.

7. The method for removing dynamic points for an autonomous mobile platform as described in claim 1, wherein, Step S300 includes the following steps: Project all laser points in the current frame point cloud data onto the image plane of the camera to obtain the pixel points in the current frame image data corresponding to all laser points; Determine whether the pixel corresponding to each laser point is within the dynamic region; and In response to a pixel corresponding to a certain laser point being located within the dynamic region, that laser point is taken as the potential dynamic point.

8. The method for removing dynamic points for an autonomous mobile platform as described in any one of claims 1 to 7, wherein, Step S400 includes the following steps: Clustering is performed on the potential dynamic points corresponding to different dynamic regions to obtain different point cloud clusters; Calculate the average distance between each point cloud cluster and the camera based on the depth value in the current frame point cloud data; as well as Remove the potential dynamic point from the point cloud cluster that is close to the camera from the current frame point cloud data.

9. A de-dynamic point system for an autonomous mobile platform, wherein the autonomous mobile platform is equipped with a camera and radar, characterized in that, The de-dynamic point system for the autonomous mobile platform includes interconnected components: A data acquisition module is used to acquire image information and point cloud information simultaneously collected by the camera and the radar at the current frame timestamp and the previous frame timestamp, respectively, so as to obtain the current frame image data, the previous frame image data and the current frame point cloud data accordingly; wherein, the sampling frequency of the camera is the same as and synchronized with the detection frequency of the radar, and the current frame timestamp and the previous frame timestamp are both timestamps of the radar. A feature point processing module is used to perform feature point processing on the current frame image data and the previous frame image data based on the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform, so as to determine the dynamic region in the current frame image data. Specifically, the module first calculates the transformation matrix of corresponding feature points between the current frame image data and the previous frame image data using the relative pose relationship as the initial value. Then, the module performs a transmission transformation on the previous frame image data and the current frame image data based on the transformation matrix to obtain the light intensity change of each pixel in the current frame image data after the transmission transformation. By comparing the light intensity change with a light intensity threshold, the pixel region with the light intensity change greater than the light intensity threshold is determined as a continuous dynamic region in the current frame image data. A point cloud projection module is used to project the current frame point cloud data onto the image plane of the camera, so as to use the point cloud within the dynamic region projected onto the current frame image data as potential dynamic points; and A clustering module is used to cluster the potential dynamic points to obtain different point cloud clusters, and calculate the average distance between each point cloud cluster and the camera based on the depth value of the different point cloud clusters in the camera coordinate system. The module distinguishes between point cloud clusters that are close to the camera and point cloud clusters that are far from the camera, and removes potential dynamic points in point cloud clusters that are close to the camera.

10. The de-dynamic point system for an autonomous mobile platform as described in claim 9, wherein, The feature point processing module includes a trajectory extrapolation module, a feature point matching module, and a region determination module that are communicatively connected to each other. The trajectory extrapolation module is used to extrapolate the data collected by the inertial measurement unit and wheel speedometer configured on the autonomous mobile platform to obtain the relative pose relationship between the current frame timestamp and the previous frame timestamp of the autonomous mobile platform. The feature point matching module is used to perform feature point matching on the current frame image data and the previous frame image data according to the relative pose relationship of the autonomous mobile platform, so as to obtain the transformation matrix of corresponding feature points between the current frame image data and the previous frame image data; The region determination module is used to determine the dynamic region in the current frame image data by means of transmission transformation based on the transformation matrix of the corresponding feature point.

11. The de-dynamic point system for an autonomous mobile platform as described in claim 10, wherein, The trajectory extrapolation module includes a pose acquisition module, a linear interpolation module, and a velocity integration module that are communicatively connected to each other. The pose acquisition module is used to acquire the pose of the autonomous mobile platform at the previous frame timestamp, as the previous frame pose of the autonomous mobile platform. The linear interpolation module is used to perform linear interpolation on the inertial data collected by the inertial measurement unit and the wheel speed data collected by the wheel speed meter to obtain the wheel speed and angular velocity at the current frame timestamp. The velocity integration module is used to integrate the wheel speed and angular velocity between the previous frame timestamp and the current frame timestamp based on the previous frame pose of the autonomous mobile platform, so as to obtain the current frame pose of the autonomous mobile platform and thus obtain the relative pose relationship of the autonomous mobile platform.

12. The de-dynamic point system for an autonomous mobile platform as described in claim 10, wherein, The feature point matching module includes a data filtering module, a feature point extraction module, and a matrix calculation module that are communicatively connected to each other. The data filtering module is used to filter the current frame image data and the previous frame image data respectively. The feature point extraction module is used to perform feature point extraction processing on the filtered current frame image data and the previous frame image data to obtain each feature point in the filtered current frame image data and the previous frame image data; The matrix calculation module is used to calculate the transformation matrix of the corresponding feature points between the current frame image data and the previous frame image data, using the relative pose relationship of the autonomous mobile platform as the initial value.

13. The de-dynamic point system for an autonomous mobile platform as described in claim 10, wherein, The region determination module includes a transmission transformation module, a change comparison module, and a region determination module that are communicatively connected to each other. The transmission transformation module is used to perform transmission transformation on the previous frame image data and the current frame image data according to the transformation matrix of the corresponding feature point, so as to obtain the light intensity change of each pixel in the current frame image data after transmission transformation. The change comparison module is used to compare whether the light intensity change of each pixel in the current frame image data after transmission transformation is greater than a light intensity threshold. The region determination module is used to determine the pixel region whose light intensity change is greater than the light intensity threshold as the dynamic region of the current frame image data.

14. The de-dynamic point system for an autonomous mobile platform as described in any one of claims 10 to 13, wherein, The point cloud projection module includes a laser point projection module, a laser point judgment module, and a dynamic point screening module that are communicatively connected to each other. The laser point projection module is used to project all laser points in the current frame point cloud data onto the image plane of the camera to obtain the pixels in the current frame image data corresponding to all the laser points. The laser point judgment module is used to determine whether the pixel corresponding to each laser point is in the dynamic region. The dynamic point screening module is used to identify a laser point as a potential dynamic point in response to the pixel corresponding to a certain laser point being in the dynamic region.

15. The de-dynamic point system for an autonomous mobile platform as described in claim 14, wherein, The clustering processing module includes a point cloud clustering module, a distance calculation module, and a dynamic point removal module that are communicatively connected to each other. The point cloud clustering module is used to perform clustering processing on the potential dynamic points corresponding to different dynamic regions to obtain different point cloud clusters. The distance calculation module is used to calculate the average distance between each point cloud cluster and the camera based on the depth value in the current frame point cloud data. The dynamic point removal module is used to remove the potential dynamic points in the point cloud clusters that are close to the camera from the current frame point cloud data.

16. An electronic device, characterized in that, include: A processor is used to execute program instructions; and A memory, wherein the memory is configured to store program instructions executable by the processor to implement all or part of the steps in a method for removing dynamic points for an autonomous mobile platform, wherein the method for removing dynamic points for an autonomous mobile platform includes the steps of: S100: Obtain image information and point cloud information simultaneously collected by the camera and radar configured on the autonomous mobile platform at the current frame timestamp and the previous frame timestamp, respectively, so as to obtain the current frame image data, the previous frame image data and the current frame point cloud data accordingly; wherein, the sampling frequency of the camera is the same as and synchronized with the detection frequency of the radar, and the current frame timestamp and the previous frame timestamp are both timestamps of the radar. S200: Based on the relative pose relationship between the current frame timestamp and the previous frame timestamp, the autonomous mobile platform performs feature point processing on the current frame image data and the previous frame image data to determine the dynamic region in the current frame image data; wherein, firstly, using the relative pose relationship as the initial value, the transformation matrix of the corresponding feature points between the current frame image data and the previous frame image data is calculated, and then the previous frame image data and the current frame image data are subjected to transmission transformation according to the transformation matrix to obtain the light intensity change of each pixel in the current frame image data after transmission transformation. By comparing the light intensity change with the light intensity threshold, the pixel region with the light intensity change greater than the light intensity threshold is determined as a continuous dynamic region in the current frame image data; S300: Project the current frame point cloud data onto the image plane of the camera, so as to use the point cloud within the dynamic region projected onto the current frame image data as potential dynamic points; and S400: Cluster the potential dynamic points to obtain different point cloud clusters, and calculate the average distance between each point cloud cluster and the camera based on the depth value of the different point cloud clusters in the camera coordinate system. Distinguish between point cloud clusters that are close to the camera and point cloud clusters that are far from the camera, and remove potential dynamic points in point cloud clusters that are close to the camera.

Images