Information acquisition method, device, robot and storage medium

By processing the parameter values ​​of multi-frame raster images, the robot generates obstacle perception results, solving the problem of insufficient accuracy of single-frame point cloud data and achieving higher obstacle perception accuracy and speed.

CN114387514BActive Publication Date: 2024-11-22YUANLI JUHE (CHONGQING) ROBOTICS TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111659766.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-30
Publication Date
2024-11-22
Estimated Expiration
2041-12-30

AI Technical Summary

Technical Problem

In existing technologies, robots mainly rely on single-frame point cloud data to generate obstacle perception results, which leads to insufficient accuracy of the perception results.

Method used

By using a multi-frame raster image approach, obstacle perception results are generated by determining the parameter value of each raster to be the number of target rasteres or the sum of the number and 1. Multi-frame data is used to improve perception accuracy.

Benefits of technology

By generating obstacle perception results from multiple frames of data, the accuracy and speed of obstacle perception are improved, and the impact of noise is reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114387514B_ABST
    Figure CN114387514B_ABST
Patent Text Reader

Abstract

An embodiment of the present application provides an information acquisition method, device, robot and storage medium, the method comprising: acquiring a multi-frame raster image, the multi-frame raster image comprising: a k-th frame raster image and at least one other frame raster image, wherein the multi-frame raster image is continuous, the acquisition time corresponding to the k-th frame raster image is later than the acquisition time corresponding to the other raster images, and the acquisition time corresponding to the raster image is the acquisition time of the point cloud to which the raster image belongs; for each first grid in the k-th frame raster image, determining a parameter value of the first grid, the target second grid comes from at least one other frame raster image, and the position of the center point of the target second grid in the coordinate system of the k-th frame raster image is located in the first grid; based on the parameter value of each first grid, determining a first grid related to an obstacle in the k-th frame raster image, and generating an obstacle perception result based on the first grid related to the obstacle.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of computer vision, and specifically to information acquisition methods, devices, robots and storage media. Background Art

[0002] During the operation of the robot, the robot needs to perceive obstacles in the environment in real time and generate obstacle perception results in real time. At present, the commonly used method for generating obstacle perception results is to generate obstacle perception results using a frame of point cloud currently collected. Summary of the invention

[0003] The present application provides an information acquisition method, device, robot and storage medium.

[0004] According to a first aspect of an embodiment of the present application, there is provided an information acquisition method, comprising:

[0005] Acquire multiple frames of raster images, the multiple frames of raster images comprising: a k-th frame of raster images and at least one other frame of raster images, wherein the multiple frames of raster images are continuous, the acquisition time corresponding to the k-th frame of raster images is later than the acquisition time corresponding to the other raster images, and the acquisition time corresponding to the raster images is the acquisition time of the point cloud to which the raster images belong;

[0006] For each first grid in the k-th frame grid image, determine a parameter value of the first grid, wherein the parameter value of the first grid is the number of target second grids corresponding to the first grid or the sum of the number of target second grids corresponding to the first grid and 1, the target second grid is from the at least one frame of other grid images, and the position of the center point of the target second grid in the coordinate system of the k-th frame grid image is located in the first grid;

[0007] Based on the parameter value of each first grid, a first grid related to the obstacle in the k-th frame grid image is determined, and based on the first grid related to the obstacle, an obstacle perception result is generated.

[0008] According to a second aspect of an embodiment of the present application, there is provided an information acquisition device, including:

[0009] An acquisition unit is configured to acquire a plurality of raster image frames, wherein the plurality of raster image frames include: a k-th raster image frame and at least one other raster image frame, wherein the plurality of raster image frames are continuous, an acquisition time corresponding to the k-th raster image frame is later than an acquisition time corresponding to the other raster images, and the acquisition time corresponding to the raster image is an acquisition time of a point cloud to which the raster image belongs;

[0010] A first determining unit is configured to determine, for each first grid in the k-th frame grid image, a parameter value of the first grid, wherein the parameter value of the first grid is the number of target second grids corresponding to the first grid or the sum of the number of target second grids corresponding to the first grid and 1, the target second grid is from the at least one frame of other grid images, and the position of the center point of the target second grid in the coordinate system of the k-th frame grid image is located in the first grid;

[0011] The perception unit is configured to determine the first grid related to the obstacle in the k-th frame grid image based on the parameter value of each first grid, and generate an obstacle perception result based on the first grid related to the obstacle.

[0012] The information acquisition method provided in the embodiment of the present application obtains a multi-frame grid map, determines the first grid related to the obstacle in the k-th frame grid map based on the parameter value of each first grid in the k-th frame grid map, generates an obstacle perception result based on the first grid related to the obstacle, and the parameter value of the first grid is the number of target second grids corresponding to the first grid or the sum of the number of target second grids corresponding to the first grid and 1, and the target second grid comes from at least one other grid map. Compared with using a single frame of data, i.e., a currently collected frame of point cloud, to generate obstacle perception results, multi-frame data, i.e., multi-frame grid maps, participate in generating obstacle perception results, and the data involved in generating obstacle perception results is richer, thereby improving the accuracy of the generated obstacle perception results. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0014] Figure 1 A flow chart showing the information acquisition method provided by an embodiment of the present application is shown;

[0015] Figure 2 A schematic diagram showing the effect of a frame of raster image is shown;

[0016] Figure 3 The structure block diagram of the information acquisition device provided by the embodiment of the present application is shown;

[0017] Figure 4 A structural block diagram of a robot provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0018] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0019] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0020] Figure 1 A flowchart of an information acquisition method provided in an embodiment of the present application is shown, and the method includes:

[0021] Step 101, obtaining multiple frames of raster images.

[0022] In the present application, a multi-frame raster image includes: a k-th frame raster image and at least one other frame raster image, and the multi-frame raster images are continuous. The acquisition time corresponding to the k-th frame raster image is later than the acquisition time corresponding to the other raster images. For each frame raster image in the multi-frame raster image, the acquisition time corresponding to the raster image is the acquisition time of the point cloud to which the raster image belongs.

[0023] The point cloud to which the raster image belongs can be collected by a device for collecting point clouds, such as a laser radar. The point cloud to which the raster image belongs can also be obtained by converting the depth image corresponding to the point cloud to which the raster image belongs. For a point cloud to which a frame of the raster image belongs, if the point cloud is obtained by converting the depth image corresponding to the point cloud, the acquisition time of the point cloud is the time when the point cloud is obtained.

[0024] The range of each frame of the grid image is the same. The boundary value of the X axis of the grid image in the coordinate system and the boundary value of the Y axis in the coordinate system can be preset. The boundary value of the X axis of the grid image in the coordinate system and the boundary value of the Y axis in the coordinate system constitute the range of the grid image. The shape of the grid in the grid image can be rectangular, and the width and height of the grid can be preset.

[0025] Please refer to Figure 2 , which shows a schematic diagram of the effect of a frame of raster image.

[0026] exist Figure 2 In FIG. 1 , a grid image is shown. The left boundary value of the grid image on the x-axis in the coordinate system is xmin, the right boundary value of the grid image on the x-axis in the coordinate system is xmax, the left boundary value of the grid image on the y-axis in the coordinate system is ymin, and the right boundary value of the grid image on the y-axis in the coordinate system is ymax. The width of the grid, i.e., the length of the grid in the y-axis direction, is y_delta, and the height of the grid, i.e., the length of the grid in the x-axis direction, is x_delta.

[0027] Step 102: for each first grid in the k-th frame grid image, determine the parameter value of the first grid.

[0028] In the present application, the grid in the grid image of the kth frame may be referred to as the first grid, and the grids in other grid images of each frame may be collectively referred to as the second grid.

[0029] In the present application, for any frame of other grid images, the point cloud to which the other grid image belongs can be projected into the other grid image, and the number of points in each second grid in the other grid image is determined. The number of points in the second grid is the number of points located in the second grid when the point cloud to which the other grid image belongs is projected into the other grid image. For each second grid in the other grid image, if the number of points in the second grid is greater than a quantity threshold, the second grid can be used as a target second grid. The quantity threshold is a custom value, for example, the quantity threshold can be the median of the number of points in all second grids in the other grid image.

[0030] For each frame of the raster image, if the number of points in a grid is less than the number threshold, the points in the grid can be considered as noise points and can be filtered out; correspondingly, if the number of points in a grid is greater than or equal to the number threshold, the points in the grid can be considered as valid points. After filtering the noise points in a frame of the raster image, the points used to represent the main obstacles can be retained in the raster image.

[0031] For each first grid, the parameter value of the first grid is the number of target second grids corresponding to the first grid or the sum of the number of target second grids corresponding to the first grid and 1.

[0032] In the present application, for a first grid, a target second grid corresponding to the first grid means that the second grid is converted to the coordinate system of the k-th frame grid image, and the center point of the second grid is located in the first grid.

[0033] In other words, for a first grid and a second grid, if the position of the center point of the second grid in the coordinate system of the k-th frame grid image is located in the first grid, the second grid serves as a target second grid corresponding to the first grid.

[0034] For example, a multi-frame grid image is composed of a 1st frame grid image, a 2nd frame grid image, ... a kth frame grid image. For a first grid i in the kth frame grid image, if the center point of a second grid j in the 1st frame grid image is located in the first grid i in the coordinate system of the kth frame grid image, then the second grid j serves as a target second grid corresponding to the first grid i, and if the center point of a second grid m in the 2nd frame grid image is located in the first grid i in the coordinate system of the kth frame grid image, then the second grid m serves as another target second grid corresponding to the first grid i.

[0035] In the present application, for a frame of a raster image, if the point cloud to which the raster image belongs is collected by a device for collection on a robot, such as a laser radar, the coordinate system of the raster image can be: a coordinate system with the position of the robot as the origin when the robot collects the point cloud to which the raster image belongs. If the point cloud to which the raster image belongs is obtained by converting the depth image corresponding to the point cloud, the coordinate system of the raster image can be: a coordinate system with the position of the robot as the origin when the robot collects the depth image corresponding to the point cloud.

[0036] In the present application, for each second grid in a frame of other grid images, the position of the center point of the second grid in the coordinate system of the other grid images is represented by the coordinates representing the position of the center point of the second grid in the coordinate system of the other grid images.

[0037] For each second grid in a frame of other grid images, when determining the position of the center point of the second grid in the coordinate system of the k-th grid image, the coordinates representing the position of the center point of the second grid in the coordinate system of the k-th grid image can be determined based on the coordinates representing the position of the center point of the second grid in the coordinate system of the other grid image and the coordinate conversion relationship corresponding to the coordinate system of the other grid image, thereby determining the position of the center point of the grid to be fused in the coordinate system of the k-th grid image. The coordinate conversion relationship corresponding to the coordinate system of the other grid image indicates, for a given point in the coordinate system of the other grid image, the conversion relationship between the coordinates representing the position of the point in the coordinate system of the other grid image and the coordinates representing the position of the point in the coordinate system of the k-th grid image.

[0038] Step 103: determine the first grid related to the obstacle in the k-th frame grid image based on the parameter value of each first grid, and generate an obstacle perception result based on the first grid related to the obstacle.

[0039] In the present application, the number of grid images targeted by step 101 may be k, and the number of grid images targeted for each obstacle perception result generation may be k. The parameter value of the first grid is recorded as count. For a first grid, when the parameter value of the first grid is the sum of the number of target second grids corresponding to the first grid and 1, if the count of the first grid is equal to k, it can be determined that the first grid is the first grid associated with the obstacle. When the parameter value of the first grid is the number of target second grids corresponding to the first grid, if the count of the first grid is equal to k-1, it can be determined that the first grid is the first grid associated with the obstacle.

[0040] In the present application, the obstacle perception result may include: the position of the center point of each first grid related to the obstacle in the coordinate system of the kth frame grid image.

[0041] In some embodiments, it also includes: for each frame of other grid images, obtaining the number of points in each second grid in the other grid images, and taking the second grid whose number of points is greater than a number threshold as the target second grid; wherein the number of points in each second grid in the other grid images is the number of points located in each second grid when the point cloud to which the other grid image belongs is projected into the other grid images.

[0042] For any second grid in any frame of other grid images, the number of points in the second grid can be obtained, that is, the number of points located in the second grid when the point cloud belonging to the other grid image is projected into the other grid image. If the number of points in the second grid is greater than the number threshold, the second grid can be used as the target second grid.

[0043] In the present application, considering that the second grid whose number of points is not greater than the number threshold may be a noise type second grid, for any frame of other grid images, only the second grid whose number of points is greater than the number threshold is used as the target second grid, thereby filtering the noise type second grid, and only the target second grid participates in determining the first grid related to the obstacle, thereby improving the accuracy of determining the first grid related to the obstacle, and at the same time, improving the speed of determining the first grid related to the obstacle.

[0044] In some embodiments, the target second grid is located in a connected domain in other grid graphs from which the target second grid comes, wherein the connected domain is determined by performing a connected domain analysis on other grids from which the target second grid comes by a preset connected domain analysis algorithm.

[0045] The preset connected domain analysis algorithm can be any existing connected domain analysis algorithm. The preset connected domain analysis algorithm can perform connected domain analysis on the corresponding image based on the pixel value in the corresponding image, determine the connected domain in the corresponding image, and at the same time, determine the isolated pixels in the corresponding image that are not located in the connected domain. Other grid images are equivalent to an image, the second grid in other grid images is equivalent to a pixel, and the number of points in the second grid in other grid images is equivalent to the pixel value of the second grid.

[0046] In the present application, for each frame of other grid images, a connected domain analysis may be performed on the other grid images using a preset connected domain analysis algorithm to determine the connected domains and isolated second grids in the other grid images.

[0047] For a second grid in a frame of other grid images, if the number of points in the second grid is greater than a quantity threshold and is located in a connected domain in the other grid images, the second grid may be used as a target second grid.

[0048] In the present application, considering that the second grid whose number of points is not greater than the number threshold is usually a noise type grid, the isolated second grid that is not located in the connected domain may be a noise type grid. For any frame of other grid images, the second grid whose number of points is greater than the number threshold and is located in the connected domain of the other grid image can be used as the target second grid. Thus, the noise type second grid can be filtered, and only the target second grid participates in determining the current grid related to the obstacle, thereby improving the accuracy of determining the first grid related to the obstacle, and at the same time, improving the speed of determining the first grid related to the obstacle.

[0049] In some embodiments, based on the parameter value of each first grid, determining the first grid associated with the obstacle in the kth frame grid image includes: for each first grid, when the difference between the parameter value of the first grid and k is less than a difference threshold, determining the first grid as the first grid associated with the obstacle.

[0050] In the present application, the number of grid maps targeted by step 101 may be k, and the number of grid maps targeted for each obstacle perception result generation may be k. For each first grid in the first grid map, the difference between the parameter value of the first grid and k is determined, and the difference between the parameter value of the first grid and k is: k minus the parameter value of the first grid. For each first grid in the first grid map, if the difference between the parameter value of the first grid and k is less than the difference threshold, the first grid can be determined as a grid associated with an obstacle.

[0051] In some embodiments, it also includes: opening up a cache area capable of storing pointer elements of k frames of raster images; when generating each frame of the raster image, storing the pointer elements of the currently generated raster image in the cache area, wherein the pointer elements of the raster image include: a pose pointer corresponding to the raster image, and a current pose transformation pointer corresponding to the raster image.

[0052] In the present application, when each frame of the raster image is generated, the pointer element of the currently generated raster image is stored in the cache area. In other words, for each frame of the raster image, after the raster image is generated, the pointer element of the raster image is stored in the cache area.

[0053] For each frame of the raster image, the pose corresponding to the raster image is the pose of the point cloud to which the other raster images belong. If the point cloud is collected by a device used for collection on the robot, such as a lidar, the pose of the point cloud is: the pose of the robot when the point cloud is collected. If the point cloud is obtained by converting the depth image corresponding to the point cloud, the pose of the point cloud is the pose of the robot when the depth image corresponding to the point cloud is collected.

[0054] For each frame of the raster image, the pose pointer corresponding to the raster image indicates the storage address of the pose corresponding to the raster image.

[0055] For each frame of other grid images, the current posture corresponding to the other grid images is transformed into a transformation between the posture corresponding to the grid image, that is, the posture of the point cloud to which the grid image belongs, and the posture corresponding to the k-th frame grid image, that is, the posture of the point cloud to which the k-th frame grid image belongs.

[0056] For each frame of other grid images, the current posture transformation pointer corresponding to the other grid images indicates the storage address of the current posture transformation corresponding to the other grid images. Through the current posture transformation pointer corresponding to the other grid images, the current posture transformation corresponding to the other grid images can be obtained.

[0057] Since there is no posture transformation between the point cloud to which the k-th frame grid image belongs and itself, the current posture transformation pointer corresponding to the k-th frame grid image may be a null pointer.

[0058] In the present application, when each frame of the raster image is generated, the pointer elements of the currently generated raster image are stored in the cache area, and the pointer elements of the raster image include: a posture pointer corresponding to the raster image, and a current posture transformation pointer corresponding to the raster image. Therefore, when it is necessary to access the posture corresponding to any frame of the raster image and the current posture transformation corresponding to any frame of the raster image in the process of generating obstacle perception results, the posture pointer corresponding to any frame of the raster image and the current posture transformation pointer corresponding to any frame of the raster image can be directly obtained from the cache area. Through the posture pointer corresponding to any frame of the raster image and the current posture transformation pointer corresponding to any frame of the raster image, the posture corresponding to any frame of the raster image and the current posture transformation pointer corresponding to any frame of the raster image can be quickly accessed.

[0059] In some embodiments, before determining the parameter value of each first grid in the k-th frame grid image, it also includes: obtaining pointer elements of each frame of other grid images from a preset cache area, the pointer elements of other grid images include: posture pointers corresponding to other grid images, and current posture transformation pointers corresponding to other grid images; obtaining the posture corresponding to each frame of other grid images through the posture pointers corresponding to each frame of other grid images, wherein the posture corresponding to other grid images is the posture of the point cloud to which the other grid images belong; obtaining the current posture transformation corresponding to each frame of other grid images through the current posture transformation pointers corresponding to each frame of other grid images, wherein the current posture transformation corresponding to other grid images is the transformation between the posture of the point cloud to which the other grid images belong and the posture of the k-th frame point cloud, and the k-th frame point cloud is the point cloud to which the k-th frame grid image belongs; for each frame of other grid images, based on the current posture transformation corresponding to the other grid images, the position of the center point of the second grid in the other grid images is transformed into the position in the coordinate system of the k-th frame grid image. In other words, the position of the center point of each second grid in the other grid image is transformed from the coordinate system of the other grid image to the coordinate system of the k-th frame grid image.

[0060] In the present application, before determining the parameter value of the first grid for each first grid in the k-th frame grid image, the pointer element of each frame of other grid images can be obtained from the preset cache area. For each frame of other grid images, the pointer element of the other grid images includes: a pose pointer corresponding to the other grid image, and a current pose transformation pointer corresponding to the other grid image. Through the pose pointer corresponding to the other grid image, the pose corresponding to the other grid image can be obtained. The pose corresponding to the other grid image is the pose of the point cloud to which the other grid image belongs. Through the current pose transformation pointer corresponding to the other grid image, the current pose transformation corresponding to the other grid image can be obtained. The current pose transformation corresponding to the other grid image is the transformation between the pose of the point cloud to which the other grid image belongs and the pose of the point cloud of the k-th frame.

[0061] After obtaining the posture corresponding to each frame of other grid images and the current posture transformation corresponding to each frame of other grid images, for each frame of other grid images, based on the current posture transformation corresponding to the other grid images, the position of the center point of the second grid in the other grid images can be transformed into the position in the coordinate system of the kth frame of the grid image.

[0062] In the present application, a preset cache area stores pointer elements of each frame of other grid images. Before determining the parameter value of each first grid in the kth frame of the grid image, the pointer elements of each frame of other grid images can be directly obtained from the preset cache area, and the posture corresponding to each frame of other grid images and the current posture transformation corresponding to each frame of other grid images can be quickly obtained. Then, the position of the center point of any second grid in any frame of other grid images in the coordinate system of the kth frame of the grid image can be determined. The time consumption for obtaining the data required when determining the parameter value of each first grid is short, that is, the time consumption for obtaining the position of the center point of the second grid in the coordinate system of the kth frame of the grid image is short, thereby improving the speed of generating obstacle perception results.

[0063] In some embodiments, after generating the obstacle perception result based on the first grid related to the obstacle, it also includes: in response to the acquisition of the k+1th frame point cloud, determining the latest pose transformation corresponding to the kth frame grid image and the latest pose transformation corresponding to each target grid image, wherein the target grid image is a grid image other than the other grid image with the earliest acquisition time in at least one frame of other grid images, the latest pose transformation corresponding to the kth frame grid image is the transformation between the pose of the kth frame point cloud and the pose of the k+1th frame point cloud, and the latest pose transformation corresponding to the target grid image is the transformation between the pose of the point cloud to which the target grid image belongs and the pose of the k+1th frame point cloud; the current pose transformation pointer corresponding to the kth frame grid image in the preset cache area is replaced by the latest pose transformation pointer corresponding to the kth frame grid image, and the latest pose transformation pointer corresponding to the kth frame grid image indicates the latest pose transformation pointer corresponding to the kth frame grid image. The storage position of the pose transformation, the current pose transformation pointer corresponding to the k-th frame grid image indicates the storage position of the current pose transformation corresponding to the k-th frame grid image; for each target grid image, the current pose transformation pointer corresponding to the target grid image in the preset cache area is replaced by the latest pose transformation pointer corresponding to the target grid image, the latest pose transformation pointer corresponding to the target grid image indicates the storage position of the latest pose transformation corresponding to the target grid image, and the current pose transformation pointer corresponding to the target grid image indicates the storage position of the current pose transformation corresponding to the target grid image; based on the sequence number and k of the k+1-th frame point cloud, the target position is determined; a new pointer element is generated, the new pointer element includes: the pose pointer corresponding to the k+1-th frame grid image belonging to the k+1-th frame point cloud, and the current pose transformation pointer corresponding to the k+1-th frame grid image; the pointer element currently located at the target position in the preset cache area is replaced by the new pointer element.

[0064] In this application, the point cloud to which the k-th frame grid image belongs is referred to as the k-th frame point cloud. The k+1-th frame point cloud is the next frame point cloud of the k-th frame point cloud.

[0065] In the present application, the other raster images in at least one frame of other raster images except the one frame of other raster images with the earliest acquisition time are referred to as target raster images.

[0066] In the present application, in response to acquiring the k+1th frame of point cloud, the latest pose transformation corresponding to the kth frame grid image and the latest pose transformation corresponding to each target grid image are determined.

[0067] The latest pose transformation corresponding to the k-th frame grid image is the transformation between the pose of the k-th frame point cloud and the pose of the k+1-th frame point cloud. For each target grid image, the latest pose transformation corresponding to the target grid image is the transformation between the pose of the point cloud to which the target grid image belongs and the pose of the k+1-th frame point cloud.

[0068] The current pose transformation pointer corresponding to the k-th frame grid image in the preset cache area is replaced by the latest pose transformation pointer corresponding to the k-th frame grid image. The latest pose transformation pointer corresponding to the k-th frame grid image indicates the storage location of the latest pose transformation corresponding to the k-th frame grid image.

[0069] For each target grid image, the current pose transformation pointer corresponding to the target grid image in the preset cache area is replaced by the latest pose transformation pointer corresponding to the target grid image, and the latest pose transformation pointer corresponding to the target grid image indicates the storage location of the latest pose transformation corresponding to the target grid image, and the current pose transformation pointer corresponding to the target grid image indicates the storage location of the current pose transformation corresponding to the target grid image.

[0070] In the present application, the number of grid images targeted by step 101 can be k, and the number of grid images targeted for each obstacle perception result generation can be k. After the obstacle perception result is generated based on the first grid related to the obstacle, the target position can be determined based on the sequence number and k of the k+1th frame point cloud. The sequence number of the point cloud indicates which frame of the point cloud it is. The sequence number of the k+1th frame point cloud is k+1. The sequence number and k of the k+1th frame point cloud can be modulo calculated to obtain the target position, and the determined target position is actually the current position of the current pose transformation pointer corresponding to the other point clouds with the earliest acquisition time. The modulo calculation of the sequence number and k of the k+1th frame point cloud can be expressed as: id=frame_cout%k, id represents the position of the pose transformation pointer corresponding to the k+1th frame grid image, and frame_cout represents the sequence number of the k+1th frame point cloud.

[0071] After determining the target position based on the sequence number and k of the k+1th frame point cloud, a new pointer element can be generated, which includes: a pose pointer corresponding to the k+1th frame grid image belonging to the k+1th frame point cloud, and a current pose transformation pointer corresponding to the k+1th frame grid image. The pose corresponding to the k+1th frame grid image is the pose of the k+1th frame point cloud, and the pose pointer corresponding to the k+1th frame grid image indicates the storage location of the pose corresponding to the k+1th frame grid image. Since there is no pose transformation between the k+1th frame point cloud and itself, the current pose transformation pointer corresponding to the k+1th frame grid image can be a null pointer.

[0072] The pointer element currently located at the target position in the preset cache area is replaced by the new pointer element. After the pointer element currently located at the target position in the preset cache area is replaced by the new pointer element, the obstacle perception result can be generated again. The process of regenerating the obstacle perception result is the same as steps 101-103. The k+1 frame raster image is equivalent to the k frame raster image, and the k-1 frame raster images before the k+1 frame raster image, namely the 2nd frame raster image, the 3rd frame raster image...the kth frame raster image are all equivalent to other frame raster images.

[0073] During the operation of the robot, it is necessary to iteratively generate obstacle perception results to perceive obstacles in the robot's environment in real time. After generating an obstacle perception result once, the next generation of the obstacle perception result requires the latest pose transformation pointer corresponding to the k-th frame grid image, the latest pose transformation pointer corresponding to each frame of the target grid image, and the pose pointer corresponding to the k+1-th frame grid image, so as to access the latest pose transformation corresponding to the k-th frame grid image, the latest pose transformation corresponding to each frame of the target grid image, and the pose corresponding to the k+1-th frame grid image when generating the obstacle perception result next time.

[0074] In the present application, after the obstacle perception result is generated, the current posture transformation pointer corresponding to the k-th frame grid image in the preset cache area is replaced by the latest posture transformation pointer corresponding to the k-th frame grid image. For each target grid image, the current posture transformation pointer corresponding to the target grid image in the preset cache area is replaced by the latest posture transformation pointer corresponding to the target grid image. The pointer element currently located at the target position in the preset cache area is replaced by the new pointer element. There is no need to allocate storage space for storing the current posture transformation pointer corresponding to the k-th frame grid image, for storing the current posture transformation pointer corresponding to the target grid image, and for storing new pointer elements, saving storage space. At the same time, it can avoid the situation where the current posture transformation pointer corresponding to the k-th frame grid image, the current posture transformation pointer corresponding to the target grid image, and the pointer element currently located at the target position no longer participate in the next generation of obstacle perception results but occupy storage space, wasting storage space.

[0075] Please refer to Figure 3 , which shows a structural block diagram of an information acquisition device provided in an embodiment of the present application. The information acquisition device includes: an acquisition unit 301, a first determination unit 302, and a perception unit 303.

[0076] The acquisition unit 301 is configured to acquire multiple frames of raster images, the multiple frames of raster images including: a k-th frame of raster images and at least one other frame of raster images, wherein the multiple frames of raster images are continuous, the acquisition time corresponding to the k-th frame of raster images is later than the acquisition time corresponding to the other raster images, and the acquisition time corresponding to the raster images is the acquisition time of the point cloud to which the raster images belong;

[0077] The first determining unit 302 is configured to determine, for each first grid in the k-th frame grid image, a parameter value of the first grid, wherein the parameter value of the first grid is the number of target second grids corresponding to the first grid or the sum of the number of target second grids corresponding to the first grid and 1, the target second grid is from the at least one frame of other grid images, and the position of the center point of the target second grid in the coordinate system of the k-th frame grid image is located in the first grid;

[0078] The perception unit 303 is configured to determine the first grid related to the obstacle in the k-th frame grid image based on the parameter value of each first grid, and generate an obstacle perception result based on the first grid related to the obstacle.

[0079] In some embodiments, the information acquisition device further includes:

[0080] The second determination unit is configured to obtain the number of points in each second grid in the other grid image for each frame of the other grid image, and take the second grid whose number of points is greater than the number threshold as the target second grid; wherein the number of points in each second grid is the number of points located in each second grid when the point cloud belonging to the other grid image is projected into the other grid image.

[0081] In some embodiments, the target second grid is located in a connected domain in other grid graphs from which the target second grid comes, wherein the connected domain is determined by performing a connected domain analysis on the other grid graphs using a preset connected domain analysis algorithm.

[0082] In some embodiments, the perception unit 303 is further configured to, for each of the first grids, determine the first grid as a first grid associated with an obstacle when the difference between the parameter value of the first grid and k is less than a difference threshold.

[0083] In some embodiments, the information acquisition device further includes:

[0084] The first preprocessing unit is configured to open up a cache area capable of storing pointer elements of k frames of raster images; when generating each frame of the raster image, the pointer of the currently generated raster image is stored in the cache area, wherein the pointer elements of the raster image include: a posture pointer corresponding to the raster image and a current posture transformation pointer corresponding to the raster image.

[0085] In some embodiments, the information acquisition device further includes:

[0086] The second preprocessing unit is configured to obtain the pointer elements of the other raster images of each frame from the preset cache area before determining the parameter value of the first raster for each first raster in the raster image of the kth frame, wherein the pointer elements of the other raster images include: a pose pointer corresponding to the other raster images and a current pose transformation pointer corresponding to the other raster images; obtain the pose corresponding to the other raster images of each frame through the pose pointer corresponding to the other raster images of each frame, wherein the pose corresponding to the other raster images is the pose of the point cloud to which the other raster images belong; obtain the pose corresponding to the other raster images of each frame through the pose pointer corresponding to the other raster images of each frame, wherein the pose corresponding to the other raster images is the pose of the point cloud to which the other raster images belong; obtain the pose corresponding to the other raster images of each frame through the pose pointer corresponding to the other raster images of each frame The current pose transformation pointer corresponding to the grid image is used to obtain the current pose transformation corresponding to each frame of the other grid images, wherein the current pose transformation corresponding to the other grid images is the transformation between the pose of the point cloud to which the other grid images belong and the pose of the point cloud of the k-th frame, and the point cloud of the k-th frame is the point cloud to which the k-th frame grid image belongs; for each frame of the other grid image, based on the current pose transformation corresponding to the other grid images, the position of the center point of the second grid in the other grid image in the coordinate system of the other grid image is transformed into the position of the center point of the second grid in the coordinate system of the k-th frame grid image.

[0087] In some embodiments, the information acquisition device further includes:

[0088] The updating unit is configured to, after generating an obstacle perception result based on a first grid related to the obstacle, determine the latest pose transformation corresponding to the k-th frame grid image and the latest pose transformation corresponding to each target grid image in response to the acquisition of the k+1-th frame point cloud, wherein the target grid image is the grid images other than the one frame of other grid images with the earliest acquisition time in the at least one frame of other grid images, the latest pose transformation corresponding to the k-th frame grid image is the transformation between the pose of the k-th frame point cloud and the pose of the k+1-th frame point cloud, and the latest pose transformation corresponding to the target grid image is the transformation between the pose of the point cloud to which the target grid image belongs and the pose of the k+1-th frame point cloud; the current pose transformation pointer corresponding to the k-th frame grid image in the preset cache area is replaced by the latest pose transformation pointer corresponding to the k-th frame grid image, and the latest pose transformation pointer corresponding to the k-th frame grid image indicates the storage of the latest pose transformation corresponding to the k-th frame grid image. a storage position, the current posture transformation pointer corresponding to the k-th frame grid image indicates the storage position of the current posture transformation corresponding to the k-th frame grid image; for each of the target grid images, the current posture transformation pointer corresponding to the target grid image in the preset cache area is replaced by the latest posture transformation pointer corresponding to the target grid image, the latest posture transformation pointer corresponding to the target grid image indicates the storage position of the latest posture transformation corresponding to the target grid image, and the current posture transformation pointer corresponding to the target grid image indicates the storage position of the current posture transformation corresponding to the target grid image; based on the sequence number of the k+1-th frame point cloud and the k, determine the target position; generate a new pointer element, the new pointer element includes: the posture pointer corresponding to the k+1-th frame grid image belonging to the k+1-th frame point cloud, and the current posture transformation pointer corresponding to the k+1-th frame grid image; replace the pointer element currently located at the target position in the preset cache area with the new pointer element.

[0089] Figure 4401 is a block diagram of a robot provided by the present application. The robot includes a processing component 401, which further includes one or more processors, and a memory resource represented by a memory 402 for storing instructions executable by the processing component 401, such as an application. The application stored in the memory 402 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 401 is configured to execute instructions to perform the above method. The robot may also include a collection component 403. The collection component 403 may be a laser radar for collecting point clouds or a camera that can collect depth images. The robot may also include a wired or wireless network interface 404, which is configured to connect the robot to a network. The robot may also include an input / output (I / O) interface 405. The robot may operate based on an operating system stored in the memory 402, such as Windows ServerTM, MacOS XTM, UnixTM, LinuxTM, FreeBSDTM or the like.

[0090] In an exemplary embodiment, a storage medium including instructions is also provided, such as a memory including instructions, and the above instructions can be executed by the robot to complete the above information acquisition method. Optionally, the storage medium can be a non-temporary computer-readable storage medium, for example, the non-temporary computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0091] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0092] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for obtaining information, It is characterized in that The method comprises: Acquire multiple frames of raster images, the multiple frames of raster images comprising: a k-th frame of raster images and at least one other frame of raster images, wherein the multiple frames of raster images are continuous, the acquisition time corresponding to the k-th frame of raster images is later than the acquisition time corresponding to the other raster images, and the acquisition time corresponding to the raster images is the acquisition time of the point cloud to which the raster images belong; For each first grid in the k-th frame grid image, determine a parameter value of the first grid, wherein the parameter value of the first grid is the number of target second grids corresponding to the first grid or the sum of the number of target second grids corresponding to the first grid and 1, the target second grid is from the at least one frame of other grid images, and the position of the center point of the target second grid in the coordinate system of the k-th frame grid image is located in the first grid; Based on the parameter value of each first grid, a first grid related to the obstacle in the k-th frame grid image is determined, and based on the first grid related to the obstacle, an obstacle perception result is generated.

2. The method according to claim 1, It is characterized in that Also includes: For each frame of the other grid image, the number of points in each second grid in the other grid image is obtained, and the second grid whose number of points is greater than a number threshold is used as the target second grid; wherein the number of points in each second grid is the number of points located in each second grid when the point cloud belonging to the other grid image is projected into the other grid image.

3. The method according to claim 2, It is characterized in that The target second grid is located in a connected domain in other grid graphs from which the target second grid comes, wherein the connected domain is determined by performing a connected domain analysis on the other grid graphs using a preset connected domain analysis algorithm.

4. The method according to any one of claims 1 to 3, It is characterized in that Determining the first grid related to the obstacle in the k-th frame grid image based on the parameter value of each first grid comprises: For each of the first grids, when the difference between the parameter value of the first grid and k is less than a difference threshold, the first grid is determined as a first grid associated with an obstacle.

5. The method according to any one of claims 1 to 3, It is characterized in that Also includes: Open up a cache area capable of storing pointer elements of k-frame raster images; When generating each frame of the raster image, the pointer element of the currently generated raster image is stored in the cache area, wherein the pointer element of the raster image includes: a posture pointer corresponding to the raster image and a current posture transformation pointer corresponding to the raster image.

6. The method according to claim 5, It is characterized in that Before determining the parameter value of each first grid in the k-th frame grid image, the method further includes: Obtaining pointer elements of the other raster images of each frame from a preset cache area, wherein the pointer elements of the other raster images include: a posture pointer corresponding to the other raster images and a current posture transformation pointer corresponding to the other raster images; Obtaining the pose corresponding to the other raster images in each frame through the pose pointer corresponding to the other raster images in each frame, wherein the pose corresponding to the other raster images is the pose of the point cloud to which the other raster images belong; Obtain the current pose transformation corresponding to each other grid image in each frame through the current pose transformation pointer corresponding to the other grid images in each frame, wherein the current pose transformation corresponding to the other grid images is the transformation between the pose of the point cloud to which the other grid images belong and the pose of the point cloud of the k-th frame, and the point cloud of the k-th frame is the point cloud to which the k-th grid image belongs; For each frame of the other grid images, based on the current posture transformation corresponding to the other grid images, the position of the center point of the second grid in the other grid images is transformed into a position in the coordinate system of the kth frame grid image.

7. The method according to claim 6, It is characterized in that After generating an obstacle perception result based on the first grid related to the obstacle, the method further includes: In response to the acquisition of the k+1th frame of point cloud, determine the latest pose transformation corresponding to the kth frame of grid image and the latest pose transformation corresponding to each target grid image, wherein the target grid image is the grid images other than the one frame of other grid images with the earliest acquisition time in the at least one frame of other grid images, the latest pose transformation corresponding to the kth frame of grid image is the transformation between the pose of the kth frame of point cloud and the pose of the k+1th frame of point cloud, and the latest pose transformation corresponding to the target grid image is the transformation between the pose of the point cloud to which the target grid image belongs and the pose of the k+1th frame of point cloud; The current pose transformation pointer corresponding to the k-th frame grid image in the preset cache area is replaced by the latest pose transformation pointer corresponding to the k-th frame grid image, the latest pose transformation pointer corresponding to the k-th frame grid image indicates the storage location of the latest pose transformation corresponding to the k-th frame grid image, and the current pose transformation pointer corresponding to the k-th frame grid image indicates the storage location of the current pose transformation corresponding to the k-th frame grid image; For each of the target grid images, the current pose transformation pointer corresponding to the target grid image in the preset cache area is replaced by the latest pose transformation pointer corresponding to the target grid image, the latest pose transformation pointer corresponding to the target grid image indicates the storage location of the latest pose transformation corresponding to the target grid image, and the current pose transformation pointer corresponding to the target grid image indicates the storage location of the current pose transformation corresponding to the target grid image; Determine the target position based on the sequence number of the k+1th frame point cloud and k; Generate a new pointer element, the new pointer element comprising: a pose pointer corresponding to the k+1th frame grid image belonging to the k+1th frame point cloud, and a current pose transformation pointer corresponding to the k+1th frame grid image; The pointer element in the preset cache area that is currently located at the target position is replaced by the new pointer element.

8. A robot, include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method according to any one of claims 1 to 7.

9. A storage medium, when the instructions in the storage medium are executed by a processor of a robot, the robot is enabled to perform the method according to any one of claims 1 to 7.

10. A computer program product, It is characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.