Target object detection method, device, removable device and storage medium

Abstract

The present disclosure proposes a method, device, removable device and storage medium for detecting a target object, the method comprising: obtaining a first point cloud state of a target area corresponding to the removable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the removable device is in the first state, determining the number of first target voxels among a plurality of first candidate voxels according to the first point cloud state, wherein the first target voxel is a first candidate voxel containing point cloud data, and detecting whether a target object exists in the target area according to the number of first target voxels. By implementing the method of the present disclosure, the characterization effect of the number of first target voxels obtained on whether a target object exists in the target area can be effectively improved, thereby effectively improving the detection accuracy of the target object in the target area.

Description

Technical Field

[0001] The present disclosure relates to the technical field of electronic equipment, and in particular to a target object detection method, device, removable device and storage medium. Background Art

[0002] Usually, a mobile device (such as an unmanned vehicle) has the need to detect whether a target object is close to itself. For example, a laser radar can be configured on the mobile device to detect the surrounding area of ​​the mobile device based on the laser radar to achieve zero blind spots. This will cause part of the laser to hit the mobile device and form point cloud data.

[0003] In the related art, a dilated area is usually calculated with reference to the size of the movable device, point cloud data within the dilated area is filtered, and target object detection is performed based on the remaining point cloud data.

[0004] In this way, the remaining point cloud data obtained has a poor characterization effect on whether there is a target object in the expansion area, which affects the detection accuracy of the target object in the expansion area. Summary of the invention

[0005] The present disclosure aims to solve one of the technical problems in the related art at least to some extent.

[0006] To this end, the purpose of the present disclosure is to propose a target object detection method, device, removable device and storage medium, which can effectively improve the characterization effect of the number of first target voxels obtained on whether there is a target object in the target area, thereby effectively improving the detection accuracy of the target object in the target area.

[0007] A target object detection method proposed in an embodiment of the first aspect of the present disclosure includes: acquiring a first point cloud state of a target area corresponding to a movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state; determining the number of first target voxels among a plurality of first candidate voxels based on the first point cloud state, wherein the first target voxel is a first candidate voxel containing point cloud data; and detecting whether there is a target object in the target area based on the number of the first target voxels.

[0008] In some embodiments of the present disclosure, detecting whether there is a target object in the target area according to the number of the first target voxels includes:

[0009] Determine target quantity threshold;

[0010] Whether a target object exists in the target area is detected according to the number of the first target voxels and the target number threshold.

[0011] In some embodiments of the present disclosure, detecting whether there is a target object in the target area according to the number of the first target voxels and the target number threshold includes:

[0012] determining a number of second target voxels, wherein the second target voxels are second candidate voxels containing point cloud data within the target area when the movable device is in the second state;

[0013] Whether there is a target object in the target area is detected according to the number of the first target voxels, the number of the second target voxels, and the target number threshold.

[0014] In some embodiments of the present disclosure, determining the number of second target voxels includes:

[0015] Determining first description information of the first target voxel;

[0016] Determine from a preset relationship table the second description information that is identical to the first description information, and use the number of preset mark values ​​corresponding to the identical second description information as the number of the second target voxels, wherein the second description information is the description information of the second target voxel, and the preset relationship table includes the second description information and the preset mark value corresponding to the second description information, wherein the preset mark value is used to mark that the corresponding second target voxel contains point cloud data.

[0017] In some embodiments of the present disclosure, determining the first description information of the first target voxel includes:

[0018] The position information of the center point of the first target voxel is used as the first description information.

[0019] In some embodiments of the present disclosure, before acquiring the first point cloud state of the target area corresponding to the movable device, the method further includes:

[0020] Acquire a second point cloud state of the target area corresponding to the movable device, wherein the second point cloud state describes whether each second candidate voxel in the target area contains the point cloud data when the movable device is in the second state;

[0021] Determining second description information of the second target voxel according to the second point cloud state, and determining a preset mark value corresponding to the second description information;

[0022] The preset relationship table is constructed according to the second description information and the corresponding preset tag value.

[0023] In some embodiments of the present disclosure, determining the target quantity threshold includes:

[0024] Obtain multiple reference quantity thresholds;

[0025] The maximum reference quantity threshold among the multiple reference quantity thresholds is used as the target quantity threshold.

[0026] In some embodiments of the present disclosure, the number of frames of the second point cloud state is multiple frames; wherein the obtaining of multiple reference quantity thresholds includes:

[0027] Determine the number of second target voxels obtained based on the second point cloud state of each frame;

[0028] The absolute value of a first difference between the quantities of every two second target voxels is used as the reference quantity threshold.

[0029] In some embodiments of the present disclosure, the acquiring a second point cloud state of the target area corresponding to the movable device includes:

[0030] When the movable device is in the second state, acquiring point cloud data of the target area corresponding to the movable device;

[0031] Performing voxel division on the target region to obtain a plurality of second candidate voxels in the target region;

[0032] Whether each of the second candidate voxels in the target area contains the point cloud data is determined as the second point cloud state.

[0033] In some embodiments of the present disclosure, performing voxel division on the target region to obtain a plurality of second candidate voxels in the target region includes:

[0034] Determine the voxel scale;

[0035] The target region is voxel-divided according to the set voxel scale to obtain a plurality of second candidate voxels in the target region.

[0036] In some embodiments of the present disclosure, determining the second description information of the second target voxel according to the second point cloud state includes:

[0037] determining the second target voxel from a plurality of the second candidate voxels according to the second point cloud state;

[0038] The position information of the center point of the second target voxel is used as the second description information.

[0039] In some embodiments of the present disclosure, the determining and setting the voxel scale includes:

[0040] Determine density information of point cloud data in the target area;

[0041] The set voxel scale is determined according to the density information.

[0042] In some embodiments of the present disclosure, determining the set voxel scale according to the density information includes:

[0043] Determine a plurality of first distance values ​​according to the density information, wherein the first distance values ​​are distance values ​​between adjacent point cloud data;

[0044] The maximum first distance value among the plurality of first distance values ​​is used as the set voxel scale.

[0045] In some embodiments of the present disclosure, the acquiring a first point cloud state of a target area corresponding to the movable device includes:

[0046] When the movable device is in the first state, acquiring point cloud data of the target area corresponding to the movable device;

[0047] Performing voxel division on the target area to obtain a plurality of first candidate voxels in the target area;

[0048] A condition of determining whether each of the first candidate voxels in the target area contains the point cloud data is used as the first point cloud state.

[0049] In some embodiments of the present disclosure, detecting whether there is a target object in the target area according to the number of the first target voxels, the number of the second target voxels, and the target number threshold includes:

[0050] determining a second absolute value of a difference between the number of the first target voxels and the number of the second target voxels;

[0051] Comparing the second difference absolute value with the target quantity threshold;

[0052] According to the comparison result, it is detected whether there is a target object in the target area.

[0053] In some embodiments of the present disclosure, detecting whether there is a target object in the target area according to the comparison result includes:

[0054] If the comparison result is that the absolute value of the second difference is less than or equal to the target quantity threshold, it is determined that there is no target object in the target area.

[0055] If the comparison result is that the absolute value of the second difference is greater than the target quantity threshold, it is determined that there is a target object in the target area.

[0056] In some embodiments of the present disclosure, it is characterized in that the target area is determined by first scale information and second scale information of the movable device, wherein the first scale information is used to describe the scale of the movable device, and the second scale information is used to describe the scale of the movable device to be warned.

[0057] A target object detection method proposed in an embodiment of the first aspect of the present disclosure obtains a first point cloud state of a target area corresponding to a movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state; and according to the first point cloud state, determines the number of first target voxels among multiple first candidate voxels, wherein the first target voxel is a first candidate voxel containing point cloud data; and according to the number of first target voxels, detects whether there is a target object in the target area, which can effectively improve the characterization effect of the number of first target voxels obtained on whether there is a target object in the target area, thereby effectively improving the detection accuracy of target objects in the target area.

[0058] The target object detection device proposed in the second aspect of the embodiment of the present disclosure includes: a first acquisition module, used to acquire a first point cloud state of a target area corresponding to a movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state; a first determination module, used to determine the number of first target voxels among multiple first candidate voxels according to the first point cloud state, wherein the first target voxel is a first candidate voxel containing point cloud data; and a detection module, used to detect whether there is a target object in the target area according to the number of the first target voxels.

[0059] In some embodiments of the present disclosure, the detection module includes:

[0060] A first determination submodule is used to determine a target quantity threshold;

[0061] The detection submodule is used to detect whether there is a target object in the target area according to the number of the first target voxels and the target number threshold.

[0062] In some embodiments of the present disclosure, the detection submodule is specifically used to:

[0063] determining a number of second target voxels, wherein the second target voxels are second candidate voxels containing point cloud data within the target area when the movable device is in the second state;

[0064] Whether a target object exists in the target area is detected according to the number of the first target voxels, the number of the second target voxels, and the target number threshold.

[0065] In some embodiments of the present disclosure, the detection submodule is further used to:

[0066] Determining first description information of the first target voxel;

[0067] Determine from a preset relationship table the second description information that is identical to the first description information, and use the number of preset mark values ​​corresponding to the identical second description information as the number of the second target voxels, wherein the second description information is the description information of the second target voxel, and the preset relationship table includes the second description information and the preset mark value corresponding to the second description information, wherein the preset mark value is used to mark that the corresponding second target voxel contains point cloud data.

[0068] In some embodiments of the present disclosure, the detection submodule is further used to:

[0069] The position information of the center point of the first target voxel is used as the first description information.

[0070] In some embodiments of the present disclosure, the device further includes:

[0071] A second acquisition module is used to acquire a second point cloud state of the target area corresponding to the movable device, wherein the second point cloud state describes whether each second candidate voxel in the target area contains the point cloud data when the movable device is in the second state;

[0072] A second determination module, configured to determine second description information of the second target voxel according to the second point cloud state, and determine a preset mark value corresponding to the second description information;

[0073] A processing module is used to construct the preset relationship table according to the second description information and the corresponding preset tag value.

[0074] In some embodiments of the present disclosure, the first determining submodule is specifically configured to:

[0075] Obtain multiple reference quantity thresholds;

[0076] The maximum reference quantity threshold among the multiple reference quantity thresholds is used as the target quantity threshold.

[0077] In some embodiments of the present disclosure, the number of frames of the second point cloud state is multiple frames; wherein the first determination submodule is further used to:

[0078] Determine the number of second target voxels obtained based on the second point cloud state of each frame;

[0079] The absolute value of a first difference between the quantities of every two second target voxels is used as the reference quantity threshold.

[0080] In some embodiments of the present disclosure, the second acquisition module includes:

[0081] an acquisition submodule, configured to acquire point cloud data of the target area corresponding to the movable device when the movable device is in the second state;

[0082] A processing submodule, configured to perform voxel division on the target region to obtain a plurality of second candidate voxels in the target region;

[0083] The second determination submodule is used to determine whether each of the second candidate voxels in the target area contains the point cloud data as the second point cloud state.

[0084] In some embodiments of the present disclosure, the processing submodule is specifically used to:

[0085] Determine the voxel scale;

[0086] The target region is voxel-divided according to the set voxel scale to obtain a plurality of second candidate voxels in the target region.

[0087] In some embodiments of the present disclosure, the second determining module is specifically configured to:

[0088] determining the second target voxel from a plurality of the second candidate voxels according to the second point cloud state;

[0089] The position information of the center point of the second target voxel is used as the second description information.

[0090] In some embodiments of the present disclosure, the processing submodule is further used to:

[0091] Determine density information of point cloud data in the target area;

[0092] The set voxel scale is determined according to the density information.

[0093] In some embodiments of the present disclosure, the processing submodule is further used to:

[0094] Determine a plurality of first distance values ​​according to the density information, wherein the first distance values ​​are distance values ​​between adjacent point cloud data;

[0095] The maximum first distance value among the plurality of first distance values ​​is used as the set voxel scale.

[0096] In some embodiments of the present disclosure, the first acquisition module is specifically used to:

[0097] When the movable device is in the first state, acquiring point cloud data of the target area corresponding to the movable device;

[0098] Performing voxel division on the target area to obtain a plurality of first candidate voxels in the target area;

[0099] A condition of determining whether each of the first candidate voxels in the target area contains the point cloud data is used as the first point cloud state.

[0100] In some embodiments of the present disclosure, the detection submodule is further used to:

[0101] determining a second absolute value of a difference between the number of the first target voxels and the number of the second target voxels;

[0102] Comparing the second difference absolute value with the target quantity threshold;

[0103] According to the comparison result, it is detected whether there is a target object in the target area.

[0104] In some embodiments of the present disclosure, the detection submodule is further used to:

[0105] If the comparison result is that the absolute value of the second difference is less than or equal to the target quantity threshold, it is determined that there is no target object in the target area;

[0106] If the comparison result is that the absolute value of the second difference is greater than the target quantity threshold, it is determined that there is a target object in the target area.

[0107] In some embodiments of the present disclosure, it is characterized in that the target area is determined by first scale information and second scale information of the movable device, wherein the first scale information is used to describe the scale of the movable device, and the second scale information is used to describe the scale of the movable device to be warned.

[0108] The target object detection device proposed in the second aspect of the embodiment of the present disclosure obtains a first point cloud state of a target area corresponding to a movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state, determines the number of first target voxels among multiple first candidate voxels according to the first point cloud state, wherein the first target voxel is a first candidate voxel containing point cloud data, and detects whether there is a target object in the target area according to the number of first target voxels, thereby effectively improving the characterization effect of the number of first target voxels on whether there is a target object in the target area, thereby effectively improving the detection accuracy of target objects in the target area.

[0109] The mobile device proposed in the third aspect embodiment of the present disclosure includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the target object detection method proposed in the first aspect embodiment of the present disclosure is implemented.

[0110] The fourth aspect embodiment of the present disclosure proposes a non-temporary computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the target object detection method proposed in the first aspect embodiment of the present disclosure is implemented.

[0111] The fifth aspect embodiment of the present disclosure proposes a computer program product. When the instructions in the computer program product are executed by a processor, the target object detection method proposed in the first aspect embodiment of the present disclosure is executed.

[0112] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0113] The above and / or additional aspects and advantages of the present disclosure will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0114] Figure 1 is a flowchart of a target object detection method proposed in an embodiment of the present disclosure;

[0115] Figure 2 is a flowchart of a target object detection method proposed in another embodiment of the present disclosure;

[0116] Figure 3 is a flowchart of a target object detection method proposed in another embodiment of the present disclosure;

[0117] Figure 4 is a flowchart of a target object detection method proposed in another embodiment of the present disclosure;

[0118] Figure 5 is a flowchart of a target object detection method proposed in another embodiment of the present disclosure;

[0119] Figure 6 is a schematic diagram of a target object detection process proposed in an embodiment of the present disclosure;

[0120] Figure 7 is a schematic diagram of the structure of a target object detection device proposed in an embodiment of the present disclosure;

[0121] Figure 8 is a structural schematic diagram of a target object detection device proposed in another embodiment of the present disclosure;

[0122] Fig. 9 A block diagram of an exemplary mobile device suitable for implementing embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0123] Embodiments of the present disclosure are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present disclosure, and are not to be construed as limitations of the present disclosure. On the contrary, the embodiments of the present disclosure include all changes, modifications, and equivalents that fall within the spirit and connotation of the appended claims.

[0124] Figure 1 It is a flowchart of a target object detection method proposed in an embodiment of the present disclosure.

[0125] This embodiment takes the target object detection method being configured in a target object detection device as an example. In this embodiment, the target object detection method can be configured in the target object detection device, and the target object detection device can be set in a server, or can also be set in a removable device. The embodiment of the present disclosure does not limit this.

[0126] This embodiment takes the case where the target object detection method is configured in a mobile device as an example, wherein the mobile device is a hardware device such as a smart car, a smart bicycle, etc. having various operating systems.

[0127] It should be noted that the execution subject of the embodiment of the present disclosure may be, for example, a central processing unit (CPU) in a server or a removable device in terms of hardware, or may be, for example, a related background service in a server or a removable device in terms of software, without limitation.

[0128] like Figure 1 As shown, the target object detection method includes:

[0129] S101: Acquire a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in a first state.

[0130] The target area may refer to an area to be detected to see whether a target object exists.

[0131] Among them, point cloud data refers to a set of point data on the surface of an object detected by a measuring instrument.

[0132] The first state may refer to a state in which the movable device is ready to detect a target object. In the first state, the target object may or may not exist in the target area, and there is no limitation on this.

[0133] Among them, voxel is the abbreviation of volume element (Volume pixel) obtained by voxelization of point cloud data. It is the smallest unit of data located on a regular grid in three-dimensional space. Its physical meaning is similar to the promotion of two-dimensional image pixels in three-dimensional space. It is a set of evenly distributed cubes located at the center of an orthogonal grid.

[0134] The first candidate voxel refers to a voxel in the target area when the movable device is in the first state.

[0135] In the embodiment of the present disclosure, the target area may be determined based on user instruction information, or may be determined by a third-party device, and there is no limitation on this.

[0136] Optionally, in some embodiments, the target area is determined by first scale information and second scale information of the movable device, wherein the first scale information is used to describe the scale of the movable device, and the second scale information is used to describe the scale of the movable device to be warned. Thus, the target area can be adapted to the size of the movable device and the warning requirements at the same time, thereby effectively improving the targeting during the target object detection process.

[0137] The first dimension information may refer to dimension information such as length, width, and height of the movable device.

[0138] The second scale information may refer to the scale to be warned of the movable device. In the embodiment of the present disclosure, the second scale information of the movable device in different directions may be the same or different, and there is no limitation on this.

[0139] For example, the movable device is an unmanned vehicle, which can be regarded as a rectangular parallelepiped with a length of 3 meters, a width of 1.5 meters, and a height of 1 meter. Assuming that the warning scale of the unmanned vehicle in all directions (up and down, front and back, left and right) is 0.1 meters, then the target area corresponds to the spatial area obtained by expanding the outermost side of the unmanned vehicle outward by 0.1 meters. The target area uses the roof and bottom of the unmanned vehicle as the top and bottom boundaries of the expansion, and the four sides 10 cm outward as the side boundaries.

[0140] In the embodiment of the present disclosure, when obtaining the first point cloud state of the target area corresponding to the movable device, it can be obtained by a third-party point cloud state acquisition device when the movable device is in the first state, or the point cloud data corresponding to the movable device in the first state can be input into a pre-trained machine learning model to obtain the first point cloud state and transmit it to the execution entity of the embodiment of the present disclosure, without limitation.

[0141] In the embodiment of the present disclosure, when the first point cloud state of the target area corresponding to the movable device is obtained, the obtained first point cloud state effectively represents whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state, thereby providing reliable data support for the subsequent determination of the number of first target voxels.

[0142] S102: Determine the number of first target voxels in a plurality of first candidate voxels according to a first point cloud state, wherein the first target voxels are first candidate voxels containing point cloud data.

[0143] It can be understood that point cloud data refers to a set of point data on the appearance surface of an object detected by a measuring instrument, and among the multiple first candidate voxels in the target area, there may be first candidate voxels that do not contain point cloud data, and the number of first target voxels has a high correlation with the target object.

[0144] Therefore, in the embodiment of the present disclosure, when the number of first target voxels in a plurality of first candidate voxels is determined according to the first point cloud state, a reliable detection basis can be provided for the target object detection process from the dimension of the number of first target voxels.

[0145] S103: Detect whether there is a target object in the target area according to the number of the first target voxels.

[0146] The target object may refer to other objects in the target area except the movable device. For example, when the movable device is an unmanned vehicle, the target object may refer to obstacles in the target area, such as other vehicles, trees, signboards, etc.

[0147] In the embodiment of the present disclosure, when detecting whether there is a target object in the target area based on the number of first target voxels, the confidence that there is a target object in the target area can be determined based on the number of first target voxels, and then whether there is a target object in the target area can be determined based on the obtained confidence. Alternatively, the change value of the number of first target voxels within an indicated time range can be obtained, and then whether there is a target object in the target area can be determined based on the obtained change value. There is no limitation to this.

[0148] In the embodiment of the present disclosure, when detecting whether there is a target object in the target area, it can also be based on feature information of other dimensions of the first target voxel, such as coordinates, center of gravity, etc., and there is no limitation to this.

[0149] In this embodiment, by acquiring a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state, the number of first target voxels among multiple first candidate voxels is determined according to the first point cloud state, wherein the first target voxel is a first candidate voxel containing point cloud data, and based on the number of first target voxels, detecting whether a target object exists in the target area, the characterization effect of the number of first target voxels obtained on whether a target object exists in the target area can be effectively improved, thereby effectively improving the detection accuracy of target objects in the target area.

[0150] Figure 2 It is a flowchart of a target object detection method proposed in another embodiment of the present disclosure.

[0151] like Figure 2 As shown, the target object detection method includes:

[0152] S201: Acquire a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in a first state.

[0153] S202: Determine the number of first target voxels in a plurality of first candidate voxels according to a first point cloud state, wherein the first target voxels are first candidate voxels containing point cloud data.

[0154] The description of S201 and S202 can be specifically referred to the above embodiment, which will not be repeated here.

[0155] S203: Determine a target quantity threshold.

[0156] The target quantity threshold may be a threshold value set for relevant information on the quantity of first target voxels during the target object detection process.

[0157] In the embodiment of the present disclosure, when determining the target number threshold, the number of first target voxels in multiple frames may be obtained, and then the target number threshold may be determined based on the number of multiple first target voxels. Alternatively, the target number threshold may be determined using a mathematical method, and there is no limitation to this.

[0158] Optionally, in some embodiments, when determining the target quantity threshold, multiple reference quantity thresholds may be obtained, and the maximum reference quantity threshold among the multiple reference quantity thresholds may be used as the target quantity threshold. Thus, the impact of interference factors on the target object detection process can be reduced to a great extent, thereby effectively improving the practicality of the obtained target quantity threshold.

[0159] The reference quantity threshold may be a quantity threshold that may be used as a target quantity threshold.

[0160] In the embodiment of the present disclosure, when obtaining multiple reference quantity thresholds, the number of second target voxels obtained based on the second point cloud state of each frame can be determined, and the number of the multiple second target voxels obtained can be used as the reference quantity threshold. Alternatively, a communication link between the execution entity of the embodiment of the present disclosure and the big data server can be established in advance, and then multiple reference quantity thresholds can be obtained from the big data server. There is no limitation on this.

[0161] Optionally, in some embodiments, when obtaining multiple reference quantity thresholds, the number of second target voxels obtained based on the second point cloud state of each frame can be determined, and the absolute value of the first difference between the numbers of every two second target voxels can be used as the reference quantity threshold. Thus, the obtained reference quantity threshold can accurately characterize the impact that interference factors may have on the number of second target voxels, thereby providing practical and reliable reference data for determining the target quantity threshold.

[0162] The second point cloud state is used to describe whether each second candidate voxel in the target area contains point cloud data when the movable device is in the second state.

[0163] The second target voxel refers to a voxel containing point cloud data in the target area when the movable device is in the second state.

[0164] The first absolute value of the difference refers to the absolute value of the difference between the numbers of every two second target voxels.

[0165] It can be understood that when the movable device is in the second state, there are no obstacles in the target area, and the number of corresponding second target voxels is only related to the movable device and interference factors (such as noise interference). Therefore, when determining the number of second target voxels obtained based on the second point cloud state of each frame, the absolute value of the first difference between the numbers of every two second target voxels is used as the reference number threshold, the obtained reference number threshold can effectively characterize the fluctuation value of the number of second target voxels caused by interference factors.

[0166] For example, the target quantity threshold may refer to a noise interference threshold, which is related to the laser radar noise fluctuation. It may capture point cloud data for a sufficiently long period of time and ensure that there are no obstacles in the expansion area during this period. Then, according to the method for obtaining the absolute value of the second difference in the above embodiment, multiple reference quantity thresholds are obtained, and the maximum reference quantity threshold among the multiple reference quantity thresholds is used as the target quantity threshold.

[0167] S204: Detect whether there is a target object in the target area according to the number of the first target voxels and the target number threshold.

[0168] In some embodiments, the target number threshold can be determined based on the number of multiple first target voxels. When detecting whether a target object exists in the target area based on the number of first target voxels and the target number threshold, a comparison result between the number of first target voxels and the target number threshold can be determined, and then whether a target object exists in the target area is determined based on the comparison result.

[0169] In other embodiments, when detecting whether a target object exists in a target area based on the number of first target voxels and the target number threshold, the number of first target voxels and the target number threshold can also be input into a pre-trained target object detection model to detect whether a target object exists in the target area.

[0170] Of course, in some embodiments, any other possible method may be used to detect whether there is a target object in the target area according to the number of first target voxels and the target number threshold, such as drawing method, combination of numbers and shapes, etc., without limitation.

[0171] Optionally, in some embodiments, when detecting whether there is a target object in the target area based on the number of first target voxels and a target number threshold, the number of second target voxels may be determined, wherein the second target voxel is a second candidate voxel containing point cloud data in the target area when the movable device is in the second state. Based on the number of first target voxels, the number of second target voxels, and the target number threshold, it is detected whether there is a target object in the target area. Thus, the number of second target voxels can be effectively combined in the target object detection process, thereby providing accurate reference data for the number of first target voxels, which can effectively improve the reliability of the target object detection process.

[0172] The second candidate voxel refers to a voxel obtained by voxelizing the point cloud data of the target area when the movable device is in the second state.

[0173] That is to say, in the embodiment of the present disclosure, after determining the number of first target voxels in multiple first candidate voxels according to the first point cloud state, the target number threshold can be determined, and whether there is a target object in the target area can be detected based on the number of first target voxels and the target number threshold. Since the target number threshold can be flexibly configured according to the specific application scenario, it provides a reliable reference basis for the target object detection process, which can effectively improve the adaptability between the target detection process and the application scenario.

[0174] In this embodiment, by determining the target quantity threshold, according to the number of first target voxels and the target quantity threshold, whether there is a target object in the target area is detected. Since the target quantity threshold can be flexibly configured according to the specific application scenario, it provides a reliable reference basis for the target object detection process, which can effectively improve the adaptability between the target detection process and the application scenario. By determining the number of second target voxels, wherein the second target voxel is a second candidate voxel containing point cloud data in the target area when the movable device is in the second state, according to the number of first target voxels, the number of second target voxels, and the target quantity threshold, whether there is a target object in the target area is detected. Thus, the number of second target voxels can be effectively combined in the target object detection process, thereby providing accurate reference data for the number of first target voxels, which can effectively improve the reliability of the target object detection process. By obtaining multiple reference quantity thresholds, the maximum reference quantity threshold among the multiple reference quantity thresholds is used as the target quantity threshold, thereby, the influence of interference factors on the target object detection process can be greatly reduced, thereby effectively improving the practicality of the obtained target quantity threshold. By determining the number of second target voxels obtained based on the second point cloud state of each frame, and taking the absolute value of the first difference between the numbers of every two second target voxels as a reference number threshold, the obtained reference number threshold can accurately characterize the impact that interference factors may have on the number of second target voxels, thereby providing practical and reliable reference data for determining the target number threshold.

[0175] Figure 3 It is a flowchart of a target object detection method proposed in another embodiment of the present disclosure.

[0176] like Figure 3 As shown, the target object detection method includes:

[0177] S301: Acquire a second point cloud state of a target area corresponding to the movable device, wherein the second point cloud state describes whether each second candidate voxel in the target area contains point cloud data when the movable device is in the second state.

[0178] In the embodiment of the present disclosure, when obtaining the second point cloud state of the target area corresponding to the movable device, it is possible to pre-establish a voxel framework corresponding to the target area, determine the position information of each second candidate voxel in the voxel framework, and then determine the second point cloud state based on the position information of the point cloud data and the position information of the second candidate voxel. There is no limitation to this.

[0179] In the disclosed embodiment, when the second point cloud state of the target area corresponding to the movable device is obtained, the corresponding voxel features can be obtained when there are no obstacles in the target area, thereby providing reliable data support for constructing a preset relationship table.

[0180] S302: Determine second description information of a second target voxel according to a second point cloud state, and determine a preset mark value corresponding to the second description information.

[0181] The preset marking value may be used to mark in the preset relationship table that the voxel corresponding to the second description information contains point cloud data.

[0182] In the embodiment of the present disclosure, when determining the second description information of the second target voxel based on the second point cloud state, the second target voxel can be determined from multiple second candidate voxels based on the second point cloud state, and then the position information of any point of the second target voxel is used as the second description information, for example, it can be any vertex of the second target voxel, and there is no limitation on this.

[0183] In the disclosed embodiment, when the second description information of the second target voxel is determined according to the second point cloud state, and the preset mark value corresponding to the second description information is determined, the amount of data in the process of constructing the preset relationship table can be effectively reduced while ensuring the practicality of the preset relationship table.

[0184] S303: Construct a preset relationship table according to the second description information and the corresponding preset tag value.

[0185] The preset relationship table refers to a data relationship table pre-constructed based on the second description information and the corresponding preset tag value. There may be multiple second description information in the preset relationship table, and each second description information has a corresponding preset tag value.

[0186] For example, the second description information and the corresponding preset tag value can be stored in the storage medium of the execution body of the embodiment of the present disclosure, and then loaded and mapped into a hash data table structure during initialization before the unmanned vehicle runs, wherein the key is the second description information and the value is the corresponding preset tag value. The method of mapping into a hash table can be implemented based on hash_map in the STL library of c++.

[0187] That is to say, before obtaining the first point cloud state of the target area corresponding to the movable device, the embodiment of the present disclosure can obtain the second point cloud state of the target area corresponding to the movable device, wherein the second point cloud state describes whether each second candidate voxel in the target area contains point cloud data when the movable device is in the second state, and according to the second point cloud state, the second description information of the second target voxel is determined, and the preset mark value corresponding to the second description information is determined, and according to the second description information and the corresponding preset mark value, a preset relationship table is constructed, whereby the obtained preset relationship table can accurately describe the second description information of the second target voxel in the target area in the second state and the preset mark value corresponding to the second description information, thereby providing a reliable execution basis for the process of determining the number of second target voxels.

[0188] S304: Acquire a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state.

[0189] S305: Determine the number of first target voxels in a plurality of first candidate voxels according to the first point cloud state, wherein the first target voxels are first candidate voxels containing point cloud data.

[0190] S306: Determine a target quantity threshold.

[0191] The description of S304-S306 can be specifically referred to the above embodiment, which will not be repeated here.

[0192] S307: Determine first description information of the first target voxel.

[0193] The first description information refers to the description information corresponding to the first target voxel.

[0194] Optionally, in some embodiments, when determining the first description information of the first target voxel, the position information of the center point of the first target voxel can be used as the first description information. Thus, the correlation between the first description information and the second description information can be guaranteed, thereby effectively improving the reliability of the process of retrieving the second description information from the preset relationship table based on the first description information.

[0195] S308: Determine the second description information that is identical to the first description information from the preset relationship table, and use the number of preset mark values ​​corresponding to the identical second description information as the number of second target voxels, wherein the second description information is the description information of the second target voxel, and the preset relationship table includes the second description information and the preset mark value corresponding to the second description information, wherein the preset mark value is used to mark that the corresponding second target voxel contains point cloud data.

[0196] In the embodiment of the present disclosure, since the first description information and the second description information both describe the position information of the voxel center in the target area, the number of second target voxels can be determined from a preset relationship table based on the second description information that is the same as the first description information, so as to facilitate subsequent analysis and comparison in combination with the first target data.

[0197] That is to say, after determining the target quantity threshold, the embodiment of the present disclosure can determine the first description information of the first target voxel, determine the second description information that is identical to the first description information from the preset relationship table, and use the number of preset mark values ​​corresponding to the identical second description information as the number of second target voxels, wherein the second description information is the description information of the second target voxel, the preset relationship table includes the second description information, and the preset mark value corresponding to the second description information, wherein the preset mark value is used to mark that the corresponding second target voxel contains point cloud data, thereby, the number of the two target voxels can be quickly and accurately determined based on the preset relationship table, which can effectively improve the efficiency of target object detection.

[0198] S309: Determine a second absolute value of a difference between the number of the first target voxels and the number of the second target voxels.

[0199] The second absolute value of the difference refers to the absolute value of the difference between the number of the first target voxels and the number of the second target voxels.

[0200] It is understandable that the second absolute value of the difference may be generated based on the target object and / or interference factors.

[0201] In the embodiment of the present disclosure, when determining the second absolute value of the difference between the number of first target voxels and the number of second target voxels, the obtained second absolute value of the difference can accurately characterize the characteristic information of the number of target voxels in the target area in the first state and the second state, so as to facilitate subsequent comparison with the target number threshold.

[0202] S310: Compare the second difference absolute value with the target quantity threshold.

[0203] In the disclosed embodiment, the target quantity threshold can effectively characterize the maximum degree of influence of interference factors on the target voxel quantity. When compared based on the second difference absolute value and the target quantity threshold, it can provide a reliable basis for determining whether there is a target object in the target area.

[0204] S311: Detect whether there is a target object in the target area according to the comparison result.

[0205] In the embodiment of the present disclosure, when detecting whether there is a target object in the target area based on the comparison results, the confidence level of the existence of the target object in the target area can be determined based on the comparison results, and whether there is a target object in the target area can be determined based on the confidence level. Alternatively, the comparison results can be input into a pre-trained machine learning model to determine whether there is a target object in the target area, and there is no limitation on this.

[0206] Optionally, in some embodiments, when detecting whether there is a target object in the target area based on the comparison results, if the comparison result is that the absolute value of the second difference is less than or equal to the target number threshold, then it is determined that there is no target object in the target area; if the comparison result is that the absolute value of the second difference is greater than the target number threshold, then it is determined that there is a target object in the target area. Thus, accurate judgment of whether there is a target object in the target area can be achieved based on the comparison results, which can effectively improve the applicability of the target object detection process.

[0207] That is to say, in the embodiment of the present disclosure, after determining the number of the first target voxels and the number of the second target voxels, the second absolute value of the difference between the number of the first target voxels and the number of the second target voxels can be determined, and the second absolute value of the difference can be compared with the target number threshold value. Based on the comparison result, it is detected whether there is a target object in the target area. Thus, the obtained second absolute difference can effectively characterize the influence of the target object and / or interference factors on the number of the first target voxels in the second state, and the target number threshold value can characterize the influence of the interference factors on the number of the first target voxels, thereby effectively improving the rationality of the judgment logic based on the comparison result.

[0208] In this embodiment, by obtaining the second point cloud state of the target area corresponding to the movable device, wherein the second point cloud state describes whether each second candidate voxel in the target area contains point cloud data when the movable device is in the second state, the second description information of the second target voxel is determined according to the second point cloud state, and the preset mark value corresponding to the second description information is determined, and a preset relationship table is constructed according to the second description information and the corresponding preset mark value. Thus, the obtained preset relationship table can accurately describe the second description information of the second target voxel in the target area in the second state and the preset mark value corresponding to the second description information, thereby providing a reliable execution basis for the process of determining the number of second target voxels. By determining the first description information of the first target voxel, determining the second description information identical to the first description information from the preset relationship table, and taking the number of preset mark values ​​corresponding to the identical second description information as the number of the second target voxels, wherein the second description information is the description information of the second target voxel, the preset relationship table includes the second description information, and the preset mark value corresponding to the second description information, wherein the preset mark value is used to mark the corresponding second target voxel as containing point cloud data, thereby, the number of the two target voxels can be quickly and accurately determined based on the preset relationship table, which can effectively improve the efficiency of target object detection. By determining the second difference absolute value between the number of the first target voxels and the number of the second target voxels, the second difference absolute value is compared with the target number threshold, and according to the comparison result, whether there is a target object in the target area is detected, thereby, the obtained second absolute difference can effectively characterize the influence of the target object and / or interference factors on the number of the first target voxels in the second state, and the target number threshold can characterize the influence of the interference factors on the number of the first target voxels, thereby effectively improving the rationality of the judgment logic based on the comparison result. If the comparison result is that the absolute value of the second difference is less than or equal to the target quantity threshold, it is determined that there is no target object in the target area. If the comparison result is that the absolute value of the second difference is greater than the target quantity threshold, it is determined that there is a target object in the target area. Therefore, it is possible to accurately determine whether there is a target object in the target area based on the comparison result, which can effectively improve the applicability of the target object detection process. By using the position information of the center point of the first target voxel as the first description information, the correlation between the first description information and the second description information can be guaranteed, thereby effectively improving the reliability of the process of retrieving the second description information from the preset relationship table based on the first description information.

[0209] Figure 4 It is a flowchart of a target object detection method proposed in another embodiment of the present disclosure.

[0210] like Figure 4 As shown, the target object detection method includes:

[0211] S401: When the movable device is in the second state, acquiring point cloud data of a target area corresponding to the movable device.

[0212] In the embodiment of the present disclosure, when point cloud data of a target area corresponding to the movable device is obtained when the movable device is in the second state, point cloud features of the target area corresponding to the movable device in the second state can be determined, thereby providing reliable data support for constructing a preset relationship table.

[0213] For example, an unmanned vehicle is equipped with several laser radars and the laser radar scanning area ensures that there are no blind spots around the unmanned vehicle. For example, the unmanned vehicle can install a laser radar in the front, back, left and right to fill in the blind spots. The unmanned vehicle radar external parameters are calibrated through the calibration tool (the calibration of the external parameters of the laser radar refers to solving the relative transformation relationship between the laser radar measurement coordinate system and the measurement coordinate system of other sensors, that is, the rotation and translation transformation matrix), and the coordinates of each radar point cloud are transformed according to the calibration parameters to be unified into the unmanned vehicle coordinate system.

[0214] S402: Divide the target region into voxels to obtain a plurality of second candidate voxels in the target region.

[0215] In the embodiment of the present disclosure, when performing voxel division on the target area to obtain multiple second candidate voxels in the target area, the number of second candidate voxels can be determined in advance, and then the target area can be voxel divided according to the number of second candidate voxels. Alternatively, a third-party voxel division device can be used to perform voxel division on the target area to obtain multiple second candidate voxels in the target area, and there is no limitation on this.

[0216] Optionally, in some embodiments, when performing voxel division on the target area to obtain multiple second candidate voxels in the target area, a set voxel scale may be determined, and the target area may be voxel divided according to the set voxel scale to obtain multiple second candidate voxels in the target area. Since the set voxel size can be flexibly configured based on the application scenario, when the target area is voxel divided based on the set voxel scale, the applicability of the obtained second candidate voxels in the target object detection process can be effectively improved.

[0217] The pre-set scale refers to a pre-set voxel scale.

[0218] In the disclosed embodiment, when the target area is divided into voxels to obtain multiple second candidate voxels in the target area, the messy and unorganized point cloud data can be converted into voxel data to effectively reduce the difficulty of data processing and improve data processing efficiency.

[0219] S403: Determine whether each second candidate voxel in the target area contains point cloud data as a second point cloud state.

[0220] For example, the unmanned vehicle can be moved to a horizontal ground and there are no obstacles in the target area corresponding to the unmanned vehicle (i.e., in the second state), and a frame of point cloud is collected. The point cloud is intercepted according to the above expansion boundary parameters (10 cm), and the point cloud within the expansion boundary is retained. Since there is no blind spot around the unmanned vehicle, some point clouds may be hit on the vehicle body and retained due to the installation tolerance, and the point cloud situation of each vehicle on the vehicle body is not necessarily the same. Then, the intercepted point cloud is downsampled by the voxel center (the point cloud is divided into voxels, and then the center of the non-empty voxel is calculated to replace all points in the voxel to achieve downsampling of the point cloud). The side length of a single voxel can be confirmed according to the point cloud density of the laser radar. If the maximum distance between the two radar points in the collision area is 2cm, the side length of the voxel can be set to 2cm accordingly. In this way, it can be ensured that the amount of data is reduced in the expansion area to improve the calculation speed without losing the obstacle detection accuracy. In addition, voxelization is convenient for the subsequent steps to match the pixels to determine the obstacles. The method of downsampling the voxel center can be implemented by the Voxel Center Down Sample class of the PCL library. The process of obtaining the first point cloud state may refer to the process of obtaining the second point cloud state described above.

[0221] That is to say, in the embodiment of the present disclosure, when the movable device is in the second state, point cloud data of the target area corresponding to the movable device can be obtained, the target area can be divided into voxels to obtain multiple second candidate voxels in the target area, and whether each second candidate voxel in the target area contains point cloud data is determined as the second point cloud state. As a result, the indication effect of the second point cloud state on the point cloud data in the target area in the second state can be effectively improved, and the point cloud data in the second state can be converted into the second point cloud state during the target object detection process, thereby effectively reducing the amount of data in the target object detection process and improving data processing efficiency.

[0222] S404: Determine a second target voxel from a plurality of second candidate voxels according to the second point cloud state.

[0223] In the disclosed embodiment, when determining the second target voxel from a plurality of second candidate voxels according to the second point cloud state, the second target voxel containing the point cloud data can be accurately screened out from the plurality of second candidate voxels to effectively reduce redundant information.

[0224] S405: Use the position information of the center point of the second target voxel as the second description information.

[0225] That is to say, in the embodiment of the present disclosure, after determining whether each second candidate voxel in the target area contains point cloud data as the second point cloud state, the second target voxel can be determined from multiple second candidate voxels based on the second point cloud state, and the position information of the center point of the second target voxel can be used as the second description information. Thus, multiple second candidate voxels can be screened based on the second point cloud state to reduce redundant information in the target object detection process, which can effectively improve the practicality of the second description information and the indication effect on the second target voxel.

[0226] S406: Determine a preset mark value corresponding to the second description information according to the second point cloud state.

[0227] S407: Construct a preset relationship table according to the second description information and the corresponding preset tag value.

[0228] The description of S406 and S407 can be specifically referred to the above embodiment, which will not be repeated here.

[0229] S408: When the movable device is in the first state, obtain point cloud data of the target area corresponding to the movable device.

[0230] In the embodiment of the present disclosure, when the movable device is in the first state, when point cloud data of the target area corresponding to the movable device is obtained, a reliable detection basis can be provided for detecting whether there is a target object in the target area in the first state, and the real-time nature of the data to be detected can be ensured.

[0231] S409: Divide the target area into voxels to obtain a plurality of first candidate voxels in the target area.

[0232] It is understandable that the point cloud data in the target area may be irregularly distributed, which has a high difficulty in data processing. When the target area is divided into voxels to obtain multiple first candidate voxels in the target area, the data processing efficiency can be effectively improved.

[0233] S410: Determine whether each first candidate voxel in the target area contains point cloud data as a first point cloud state.

[0234] That is to say, in the embodiment of the present disclosure, after constructing a preset relationship table according to the second description information and the corresponding preset tag value, the point cloud data of the target area corresponding to the movable device can be obtained when the movable device is in the first state, and the target area can be divided into voxels to obtain multiple first candidate voxels in the target area, and whether each first candidate voxel in the target area contains point cloud data is determined as the first point cloud state. Therefore, the point cloud data of the target area can be converted into first candidate voxels for processing based on the voxel division of the target area, so as to effectively improve the point cloud data processing efficiency, and can effectively improve the applicability of the obtained first point cloud state in the target object detection process.

[0235] S411: Determine the number of first target voxels in a plurality of first candidate voxels according to a first point cloud state, wherein the first target voxels are first candidate voxels containing point cloud data.

[0236] S412: Detect whether there is a target object in the target area according to the number of the first target voxels.

[0237] The description of S411 and S412 can be specifically referred to the above embodiment, which will not be repeated here.

[0238] In this embodiment, by obtaining point cloud data of the target area corresponding to the movable device when the movable device is in the second state, performing voxel division on the target area to obtain a plurality of second candidate voxels in the target area, and determining whether each second candidate voxel in the target area contains point cloud data as the second point cloud state, thereby effectively improving the indication effect of the second point cloud state on the point cloud data in the target area under the second state, and being able to convert the point cloud data in the second state into the second point cloud state during the target object detection process, thereby effectively reducing the amount of data in the target object detection process and improving data processing efficiency. By determining a set voxel scale, performing voxel division on the target area according to the set voxel scale to obtain a plurality of second candidate voxels in the target area, since the set voxel size can be flexibly configured based on the application scenario, when performing voxel division on the target area based on the set voxel scale, the applicability of the obtained second candidate voxels in the target object detection process can be effectively improved. By determining the second target voxel from multiple second candidate voxels according to the second point cloud state, and using the position information of the center point of the second target voxel as the second description information, multiple second candidate voxels can be screened based on the second point cloud state to reduce redundant information in the target object detection process, which can effectively improve the practicality of the second description information and the indication effect of the second target voxel. When the movable device is in the first state, the point cloud data of the target area corresponding to the movable device is obtained, the target area is voxel-divided to obtain multiple first candidate voxels in the target area, and whether each first candidate voxel in the target area contains point cloud data is determined as the first point cloud state. Therefore, the point cloud data of the target area can be converted into first candidate voxels for processing based on the voxel division of the target area, so as to effectively improve the point cloud data processing efficiency, and can effectively improve the applicability of the obtained first point cloud state in the target object detection process.

[0239] Figure 5 It is a flowchart of a target object detection method proposed in another embodiment of the present disclosure.

[0240] like Figure 5 As shown, the target object detection method includes:

[0241] S501: When the movable device is in the second state, acquiring point cloud data of a target area corresponding to the movable device.

[0242] For the description of S501, please refer to the above embodiment, which will not be repeated here.

[0243] S502: Determine density information of point cloud data in the target area.

[0244] Among them, the density information can describe the distance information between the point cloud data in the target area.

[0245] It is understandable that the voxel scale may affect the detection accuracy of the target object. When determining the density information of the point cloud data in the target area, reliable data support can be provided for determining the set voxel scale.

[0246] S503: Determine and set the voxel scale according to the density information.

[0247] In the embodiment of the present disclosure, when determining the set voxel scale based on density information, the first distance value between adjacent point cloud data can be determined based on the density information, and then the first distance value with the largest number among multiple first distance values ​​is used as the set voxel scale, or the median value among multiple first distance values ​​can be taken as the set voxel scale, and there is no limitation on this.

[0248] Optionally, in some embodiments, when determining the set voxel scale based on density information, multiple first distance values ​​can be determined based on the density information, wherein the first distance value is the distance value between adjacent point cloud data, and the maximum first distance value among the multiple first distance values ​​is used as the set voxel scale. In this way, the accuracy of target object detection can be ensured while reducing the amount of data.

[0249] That is to say, in the embodiment of the present disclosure, when the movable device is in the second state, after acquiring the point cloud data of the target area corresponding to the movable device, the density information of the point cloud data in the target area can be determined, and the set voxel scale can be determined based on the density information. Thus, the obtained set voxel scale can be adapted to the density information of the point cloud data in the target area, thereby effectively improving the practicality of the obtained set voxel scale.

[0250] S504: performing voxel division on the target region according to the set voxel scale to obtain a plurality of second candidate voxels in the target region.

[0251] S505: Determine whether each second candidate voxel in the target area contains point cloud data as a second point cloud state.

[0252] S506: Determine second description information of a second target voxel according to the second point cloud state, and determine a preset mark value corresponding to the second description information.

[0253] S507: Construct a preset relationship table according to the second description information and the corresponding preset tag value.

[0254] S508: Acquire a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state.

[0255] S509: Determine the number of first target voxels in a plurality of first candidate voxels according to the first point cloud state, wherein the first target voxels are first candidate voxels containing point cloud data.

[0256] S510: Detect whether there is a target object in the target area according to the number of the first target voxels.

[0257] The description of S504-S510 can be specifically referred to the above embodiment, which will not be repeated here.

[0258] In this embodiment, by determining the density information of the point cloud data in the target area, the set voxel scale is determined according to the density information, thereby making the obtained set voxel scale adaptable to the density information of the point cloud data in the target area, thereby effectively improving the practicality of the obtained set voxel scale, and by determining multiple first distance values ​​according to the density information, wherein the first distance value is the distance value between adjacent point cloud data, and taking the maximum first distance value among the multiple first distance values ​​as the set voxel scale, thereby reducing the amount of data while ensuring the accuracy of target object detection.

[0259] For example, Figure 6 As shown, Figure 6 This is a schematic diagram of a target object detection process proposed in an embodiment of the present disclosure. Taking an unmanned vehicle equipped with a laser radar device as an example, the laser radar device scans an area so that there is no blind spot around the unmanned vehicle. The external parameters of the unmanned vehicle radar have been calibrated and each lightning point cloud has been integrated into the same coordinate system. The target area is expanded outward based on the shape of the unmanned vehicle. The expansion thickness can be determined based on the warning demand (for example, it can be 10 cm). The target object within the range of 10 cm can be identified as a collision danger state and a warning is issued. The target object detection process can be as follows:

[0260] Ensure that the unmanned vehicle is in the second state, that is, there are no obstacles around it, collect a frame of point cloud data, and intercept the point cloud in the target area according to the vehicle height and expansion thickness;

[0261] The intercepted point cloud is downsampled at the voxel center. The size of a single voxel can be determined based on the point cloud density. For example, it can be set to 2 cm, which can reduce the amount of data without affecting the accuracy of obstacle detection.

[0262] The voxelized data is stored. The stored data is loaded and mapped into a hash data table structure when the unmanned vehicle is initialized before running. The key is the second description information (voxel coordinates) and the value is whether the voxel contains point cloud data.

[0263] When the unmanned vehicle is running, a frame of data is intercepted according to the above interception method, and the intercepted point cloud data is down-sampled at the voxel center according to the above voxel parameters to obtain multiple first candidate voxels;

[0264] Traverse each first candidate voxel, count the number of first target voxels containing point cloud data, use the first description information of the first target voxel as the key to search in the above hash data table structure to determine the number of second target voxels in the hash data table structure, find the difference between the number of the first target voxels and the number of the second target voxels and take the absolute value (i.e., the second difference absolute value), and set a noise interference elimination threshold (target number threshold);

[0265] Determine the comparison result between the absolute value of the second difference and the target quantity threshold. If the absolute value of the second difference is less than or equal to the target quantity threshold, it is determined that there are no obstacles in the target area, and continue to analyze the first point cloud data of the next frame; if the absolute value of the second difference is greater than the target quantity threshold, it is determined that there are obstacles in the target area, and an early warning is reported.

[0266] Figure 7 It is a schematic diagram of the structure of a target object detection device proposed in one embodiment of the present disclosure.

[0267] like Figure 7 As shown, the target object detection device 70 includes:

[0268] A first acquisition module 701 is used to acquire a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in a first state;

[0269] A first determination module 702 is used to determine the number of first target voxels in a plurality of first candidate voxels according to the first point cloud state, wherein the first target voxel is a first candidate voxel containing point cloud data; and

[0270] The detection module 703 is used to detect whether there is a target object in the target area according to the number of the first target voxels.

[0271] In some embodiments of the present disclosure, Figure 8 As shown, Figure 8 703 is a schematic diagram of a target object detection device according to another embodiment of the present disclosure. The detection module 703 includes:

[0272] The first determination submodule 7031 is used to determine a target quantity threshold;

[0273] The detection submodule 7032 is used to detect whether there is a target object in the target area according to the number of the first target voxels and the target number threshold.

[0274] In some embodiments of the present disclosure, the detection submodule 7031 is specifically used to:

[0275] determining a number of second target voxels, wherein the second target voxels are second candidate voxels containing point cloud data within the target area when the movable device is in the second state;

[0276] Whether there is a target object in the target area is detected according to the number of the first target voxels, the number of the second target voxels, and the target number threshold.

[0277] In some embodiments of the present disclosure, the detection submodule 7031 is further used to:

[0278] Determining first description information of a first target voxel;

[0279] Determine second description information that is identical to the first description information from a preset relationship table, and use the number of preset mark values ​​corresponding to the identical second description information as the number of second target voxels, wherein the second description information is description information of the second target voxel, the preset relationship table includes the second description information, and a preset mark value corresponding to the second description information, wherein the preset mark value is used to mark that the corresponding second target voxel contains point cloud data.

[0280] In some embodiments of the present disclosure, the detection submodule 7031 is further used to:

[0281] The position information of the center point of the first target voxel is used as the first description information.

[0282] In some embodiments of the present disclosure, the apparatus further includes:

[0283] A second acquisition module 704 is used to acquire a second point cloud state of the target area corresponding to the movable device, wherein the second point cloud state describes whether each second candidate voxel in the target area contains point cloud data when the movable device is in the second state;

[0284] A second determination module 705 is used to determine second description information of a second target voxel according to a second point cloud state, and to determine a preset mark value corresponding to the second description information;

[0285] The processing module 706 is used to construct a preset relationship table according to the second description information and the corresponding preset tag value.

[0286] In some embodiments of the present disclosure, the first determining submodule 7031 is specifically configured to:

[0287] Obtain multiple reference quantity thresholds;

[0288] The maximum reference quantity threshold among the multiple reference quantity thresholds is used as the target quantity threshold.

[0289] In some embodiments of the present disclosure, the number of frames of the second point cloud state is multiple frames; wherein the first determination submodule 7031 is further used to:

[0290] Determine the number of second target voxels obtained based on the second point cloud state of each frame;

[0291] The absolute value of the first difference between the quantities of every two second target voxels is used as a reference quantity threshold.

[0292] In some embodiments of the present disclosure, the second acquisition module 704 includes:

[0293] An acquisition submodule 7041 is used to acquire point cloud data of a target area corresponding to the movable device when the movable device is in the second state;

[0294] The processing submodule 7042 is used to perform voxel division on the target area to obtain a plurality of second candidate voxels in the target area;

[0295] The second determination submodule 7043 is used to determine whether each second candidate voxel in the target area contains point cloud data as a second point cloud state.

[0296] In some embodiments of the present disclosure, the processing submodule 7042 is specifically configured to:

[0297] Determine the voxel scale;

[0298] The target region is voxel-divided according to the set voxel scale to obtain a plurality of second candidate voxels in the target region.

[0299] In some embodiments of the present disclosure, the second determining module 705 is specifically configured to:

[0300] determining a second target voxel from a plurality of second candidate voxels according to the second point cloud state;

[0301] The position information of the center point of the second target voxel is used as the second description information.

[0302] In some embodiments of the present disclosure, the processing submodule 7042 is further configured to:

[0303] Determine the density information of point cloud data in the target area;

[0304] According to the density information, the voxel scale is determined and set.

[0305] In some embodiments of the present disclosure, the processing submodule 7042 is further configured to:

[0306] Determine a plurality of first distance values ​​according to the density information, wherein the first distance value is a distance value between adjacent point cloud data;

[0307] The maximum first distance value among the multiple first distance values ​​is used as the set voxel scale.

[0308] In some embodiments of the present disclosure, the first acquisition module 701 is specifically configured to:

[0309] When the movable device is in the first state, acquiring point cloud data of a target area corresponding to the movable device;

[0310] Performing voxel division on the target area to obtain a plurality of first candidate voxels in the target area;

[0311] Whether each first candidate voxel in the target area contains point cloud data is determined as a first point cloud state.

[0312] In some embodiments of the present disclosure, the detection submodule 7032 is further configured to:

[0313] determining a second absolute value of a difference between the number of first target voxels and the number of second target voxels;

[0314] Comparing the absolute value of the second difference with the target quantity threshold;

[0315] According to the comparison results, it is detected whether there is a target object in the target area.

[0316] In some embodiments of the present disclosure, the detection submodule 7032 is further configured to:

[0317] If the comparison result is that the absolute value of the second difference is less than or equal to the target quantity threshold, it is determined that there is no target object in the target area;

[0318] If the comparison result is that the absolute value of the second difference is greater than the target quantity threshold, it is determined that there is a target object in the target area.

[0319] In some embodiments of the present disclosure, it is characterized in that the target area is determined by first scale information and second scale information of the movable device, wherein the first scale information is used to describe the scale of the movable device, and the second scale information is used to describe the scale of the movable device to be warned.

[0320] It should be noted that the above explanation of the target object detection method is also applicable to the target object detection device of this embodiment, and will not be repeated here.

[0321] In this embodiment, by acquiring a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state, the number of first target voxels among multiple first candidate voxels is determined according to the first point cloud state, wherein the first target voxel is a first candidate voxel containing point cloud data, and based on the number of first target voxels, detecting whether a target object exists in the target area, the characterization effect of the number of first target voxels obtained on whether a target object exists in the target area can be effectively improved, thereby effectively improving the detection accuracy of target objects in the target area.

[0322] Fig. 9 A block diagram of an exemplary mobile device suitable for implementing embodiments of the present disclosure is shown. Fig. 9 The shown removable device 12 is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0323] like Fig. 9 As shown, the removable device 12 is in the form of a general purpose computing device. The components of the removable device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 connecting various system components (including the system memory 28 and the processing unit 16).

[0324] The bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor or a local bus using any of a variety of bus structures. For example, these architectures include but are not limited to Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus and Peripheral Component Interconnection (PCI) bus.

[0325] The removable device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the removable device 12, including volatile and non-volatile media, removable and non-removable media.

[0326] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The removable device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Fig. 9 Not shown, often called a "hard drive").

[0327] although Fig. 9 Not shown in the figure, a disk drive for reading and writing a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing a removable non-volatile optical disk (e.g., a compact disc read only memory (hereinafter referred to as: CD-ROM), a digital versatile disc read only memory (hereinafter referred to as: DVD-ROM) or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present disclosure.

[0328] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described in the present disclosure.

[0329] The removable device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), may also communicate with one or more devices that enable the human body to interact with the removable device 12, and / or may communicate with any device that enables the removable device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). Such communication may be performed through an input / output (I / O) interface 22. Furthermore, the removable device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 20. As shown, the network adapter 20 communicates with other modules of the removable device 12 through a bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the removable device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0330] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28 , such as implementing the target object detection method mentioned in the above embodiment.

[0331] In order to implement the above embodiments, the present disclosure further proposes a non-temporary computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the target object detection method proposed in the above embodiments of the present disclosure is implemented.

[0332] In order to implement the above embodiments, the present disclosure further proposes a computer program product. When an instruction processor in the computer program product executes, the target object detection method proposed in the above embodiments of the present disclosure is executed.

[0333] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0334] It should be understood that the present disclosure is not limited to the exact 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 disclosure is limited only by the appended claims.

[0335] It should be noted that, in the description of the present disclosure, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present disclosure, unless otherwise specified, the meaning of "plurality" is two or more.

[0336] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present disclosure includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present disclosure belong.

[0337] It should be understood that the various parts of the present disclosure can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0338] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0339] In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0340] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0341] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0342] Although the embodiments of the present disclosure have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present disclosure. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present disclosure.

Claims

1. A target object detection method, characterized in that: include: Acquire a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in the first state; Determining the number of first target voxels among the plurality of first candidate voxels according to the first point cloud state, wherein the first target voxels are first candidate voxels containing point cloud data; and Determine target quantity threshold; determining a number of second target voxels, wherein the second target voxels are second candidate voxels containing point cloud data within the target area when the movable device is in the second state; determining a second absolute value of a difference between the number of the first target voxels and the number of the second target voxels; Comparing the second difference absolute value with the target quantity threshold; According to the comparison result, it is detected whether there is a target object in the target area.

2. The method according to claim 1, characterized in that The determining the number of the second target voxels comprises: Determining first description information of the first target voxel; Determine from a preset relationship table the second description information that is identical to the first description information, and use the number of preset mark values ​​corresponding to the identical second description information as the number of the second target voxels, wherein the second description information is the description information of the second target voxel, and the preset relationship table includes the second description information and the preset mark value corresponding to the second description information, wherein the preset mark value is used to mark that the corresponding second target voxel contains point cloud data.

3. The method according to claim 2, characterized in that The determining the first description information of the first target voxel includes: The position information of the center point of the first target voxel is used as the first description information.

4. The method according to claim 2, characterized in that Before acquiring the first point cloud state of the target area corresponding to the movable device, the method further includes: Acquire a second point cloud state of the target area corresponding to the movable device, wherein the second point cloud state describes whether each second candidate voxel in the target area contains the point cloud data when the movable device is in the second state; Determining second description information of the second target voxel according to the second point cloud state, and determining a preset mark value corresponding to the second description information; The preset relationship table is constructed according to the second description information and the corresponding preset tag value.

5. The method according to claim 4, characterized in that Determining the target quantity threshold comprises: Obtain multiple reference quantity thresholds; The maximum reference quantity threshold among the multiple reference quantity thresholds is used as the target quantity threshold.

6. The method according to claim 5, characterized in that The number of frames of the second point cloud state is multiple frames; wherein the obtaining of multiple reference quantity thresholds includes: Determine the number of second target voxels obtained based on the second point cloud state of each frame; The absolute value of a first difference between the quantities of every two second target voxels is used as the reference quantity threshold.

7. The method according to claim 4, characterized in that The acquiring a second point cloud state of the target area corresponding to the movable device includes: When the movable device is in the second state, acquiring point cloud data of the target area corresponding to the movable device; Performing voxel division on the target region to obtain a plurality of second candidate voxels in the target region; Whether each of the second candidate voxels in the target area contains the point cloud data is determined as the second point cloud state.

8. The method according to claim 7, characterized in that The voxel division of the target region to obtain a plurality of second candidate voxels in the target region includes: Determine the voxel scale; The target region is voxel-divided according to the set voxel scale to obtain a plurality of second candidate voxels in the target region.

9. The method according to claim 8, characterized in that The determining, according to the second point cloud state, second description information of the second target voxel includes: determining the second target voxel from a plurality of the second candidate voxels according to the second point cloud state; The position information of the center point of the second target voxel is used as the second description information.

10. The method according to claim 8, characterized in that in, The determining and setting the voxel scale includes: Determine density information of point cloud data in the target area; The set voxel scale is determined according to the density information.

11. The method according to claim 10, characterized in that The step of determining the set voxel scale according to the density information includes: Determine a plurality of first distance values ​​according to the density information, wherein the first distance values ​​are distance values ​​between adjacent point cloud data; The maximum first distance value among the plurality of first distance values ​​is used as the set voxel scale.

12. The method according to claim 7, characterized in that The obtaining of a first point cloud state of a target area corresponding to the movable device includes: When the movable device is in the first state, acquiring point cloud data of the target area corresponding to the movable device; Performing voxel division on the target area to obtain a plurality of first candidate voxels in the target area; A condition of determining whether each of the first candidate voxels in the target area contains the point cloud data is used as the first point cloud state.

13. The method according to claim 1, characterized in that The detecting whether there is a target object in the target area according to the comparison result includes: If the comparison result is that the absolute value of the second difference is less than or equal to the target quantity threshold, it is determined that there is no target object in the target area; If the comparison result is that the absolute value of the second difference is greater than the target quantity threshold, it is determined that there is a target object in the target area.

14. The method according to any one of claims 1 to 13, characterized in that: The target area is determined by first scale information and second scale information of the movable device, wherein the first scale information is used to describe the scale of the movable device, and the second scale information is used to describe the scale of the movable device to be warned.

15. A target object detection device, characterized in that: include: A first acquisition module is used to acquire a first point cloud state of a target area corresponding to the movable device, wherein the first point cloud state describes whether each first candidate voxel in the target area contains point cloud data when the movable device is in a first state; A first determination module is configured to determine the number of first target voxels in the plurality of first candidate voxels according to the first point cloud state, wherein the first target voxels are first candidate voxels containing point cloud data; and A detection module is used to determine a target quantity threshold; determine the number of second target voxels, wherein the second target voxels are second candidate voxels containing point cloud data in the target area when the movable device is in a second state; determine a second absolute value of a difference between the number of the first target voxels and the number of the second target voxels; compare the second absolute value of the difference with the target quantity threshold; and detect whether a target object exists in the target area based on the comparison result.

16. A movable device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 14.

17. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: in, The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 14.

18. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 14.

Images