Obstacle detection methods, devices, electronic equipment and computer-readable media

Abstract

This disclosure presents embodiments of an obstacle detection method, apparatus, electronic device, and computer-readable medium. One specific implementation of the method includes: acquiring an environmental point cloud dataset; cropping the environmental point cloud dataset to generate a cropped environmental point cloud dataset; downsampling the cropped environmental point cloud dataset to generate a downsampled environmental point cloud dataset; inputting the downsampled environmental point cloud dataset into an obstacle detection model to generate an obstacle information set; filtering the obstacle information in the obstacle information set to generate a filtered obstacle information set; and sending the filtered obstacle information set to a control planning terminal via an onboard communication module. This implementation improves the accuracy of obstacle information generation and reduces the risk level of autonomous vehicles during operation.

Description

Technical Field

[0001] Embodiments of this disclosure relate to the field of computer technology, and more specifically to obstacle detection methods, apparatus, electronic devices, and computer-readable media. Background Technology

[0002] Obstacle detection is a crucial step in the perception of the surrounding environment in the field of autonomous driving. Currently, common obstacle detection methods utilize appropriate data structures (e.g., K-Dimensional trees) in conjunction with clustering algorithms (e.g., Density-Based Spatial Clustering of Applications with Noise) to cluster environmental point cloud data to achieve obstacle detection.

[0003] However, when using the above method for obstacle detection, the following technical problems often arise:

[0004] First, obstacle detection results are highly dependent on the distribution of environmental point cloud data. Since environmental point cloud data is often discretely distributed, the results of obstacle detection (e.g., obstacle category, obstacle outline information) are not accurate enough. This, in turn, affects the accuracy of obstacle avoidance by autonomous vehicles, thereby increasing the risk level of autonomous vehicles during driving. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure provide obstacle detection methods, apparatuses, electronic devices, and computer-readable media to address one or more of the technical problems mentioned in the background section above.

[0007] Firstly, some embodiments of this disclosure provide an obstacle detection method, which includes: acquiring an environmental point cloud data set, wherein the environmental point cloud data is obtained by scanning the surrounding environment using a lidar installed on a target vehicle, the environmental point cloud data including: abscissa values, ordinate values, vertical coordinate values, and radar echo power values, the abscissa, ordinate, and vertical coordinate values ​​included in the environmental point cloud data being coordinate values ​​in a target vehicle coordinate system, the target vehicle coordinate system being a coordinate system with the target vehicle's direction of travel as the abscissa, the target vehicle's chassis center as the origin, a line parallel to the target vehicle's rear axle as the ordinate, and a line perpendicular to the ground as the vertical axis; cropping the environmental point cloud data set to generate a cropped environmental point cloud data set; downsampling the cropped environmental point cloud data set to generate a downsampled environmental point cloud data set; inputting the downsampled environmental point cloud data set into an obstacle detection model to generate an obstacle information set; and filtering the obstacle information in the obstacle information set to generate a filtered obstacle information set. The aforementioned set of obstacle filtering information is sent to the control planning terminal via the vehicle-mounted communication module.

[0008] Secondly, some embodiments of this disclosure provide an obstacle detection device, comprising: an acquisition unit configured to acquire an environmental point cloud data set, wherein the environmental point cloud data is obtained by scanning the surrounding environment using a lidar installed on a target vehicle; the environmental point cloud data includes: abscissa values, ordinate values, vertical coordinate values, and radar echo power values; the abscissa, ordinate, and vertical coordinate values ​​included in the environmental point cloud data are coordinate values ​​in a target vehicle coordinate system, wherein the target vehicle coordinate system is a coordinate system with the target vehicle's direction of travel as the abscissa, the target vehicle's chassis center as the origin, a line parallel to the target vehicle's rear axle as the ordinate, and a line perpendicular to the ground as the vertical axis; a clipping processing unit configured to clip the environmental point cloud data set to generate a clipped environmental point cloud data set; and a downsampling processing unit configured to downsampling the clipped environmental point cloud data set to generate a downsampled environmental point cloud data set. The input unit is configured to input the downsampled environmental point cloud data set into the obstacle detection model to generate an obstacle information set. The filtering unit is configured to filter the obstacle information in the obstacle information set to generate a filtered obstacle information set. The sending unit is configured to send the filtered obstacle information set to the control planning terminal via the vehicle communication module.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0011] The various embodiments of this disclosure have the following beneficial effects: the obstacle detection methods of some embodiments of this disclosure improve the accuracy of obstacle detection results, thereby providing more accurate data for autonomous vehicles to avoid obstacles and reducing the risk level of autonomous vehicles during operation. Specifically, the inventors have found that the reason for inaccurate obstacle detection results is that the environmental point cloud data is not preprocessed, resulting in an inaccurate clustering structure, that is, the generated obstacle information is not accurate enough. Based on this, the obstacle detection methods of some embodiments of this disclosure perform cropping, downsampling, and filtering on the environmental point cloud data, thereby making the final generated obstacle information more accurate. In addition, since environmental point cloud data is relatively sparse, this disclosure generates obstacle information based on the characteristics of environmental point cloud data using an obstacle detection model, thereby improving the accuracy of the generated obstacle information. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a schematic diagram of an application scenario of an obstacle detection method according to some embodiments of the present disclosure;

[0014] Figure 2 This is a flowchart of some embodiments of the obstacle detection method according to the present disclosure;

[0015] Figure 3 This is a schematic diagram of a non-trimmed region in some embodiments of the obstacle detection method according to this disclosure;

[0016] Figure 4 This is a schematic diagram of an obstacle detection model in some embodiments of the obstacle detection method according to the present disclosure;

[0017] Figure 5 These are schematic diagrams illustrating the structure of some embodiments of the obstacle detection device according to this disclosure;

[0018] Figure 6 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0019] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0020] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0021] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0022] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0023] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0024] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] Figure 1 This is a schematic diagram illustrating an application scenario of the obstacle detection method according to some embodiments of this disclosure.

[0026] exist Figure 1In the application scenario, firstly, the computing device 101 can acquire an environmental point cloud data set 102. This environmental point cloud data is obtained by scanning the surrounding environment using a lidar installed on the target vehicle. The environmental point cloud data includes: horizontal coordinate values, vertical coordinate values, and radar echo power values. The horizontal, vertical, and radar echo values ​​are coordinates in the target vehicle coordinate system, which is a coordinate system with the target vehicle's direction of travel as the horizontal axis, the target vehicle's chassis center as the origin, a line parallel to the target vehicle's rear axle as the vertical axis, and a line perpendicular to the ground as the center axis. Secondly, the computing device 101 can crop the environmental point cloud data set 102 to generate a cropped environmental point cloud data set 103. Then, the computing device 101 can downsample the cropped environmental point cloud data set 103 to generate a downsampled environmental point cloud data set 104. Then, the computing device 101 can input the downsampled environmental point cloud data set 104 into the obstacle detection model 105 to generate an obstacle information set 106. Next, the computing device 101 can filter the obstacle information in the obstacle information set 106 to generate a filtered obstacle information set 107. Finally, the computing device 101 can send the filtered obstacle information set 107 to the control planning terminal 109 via the vehicle communication module 108.

[0027] It should be noted that the aforementioned computing device 101 can be either hardware or software. When the computing device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. When the computing device is software, it can be installed within the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here.

[0028] It should be understood that Figure 1 The number of computing devices shown is merely illustrative. Any number of computing devices can be used depending on implementation needs.

[0029] Continue to refer to Figure 2 The diagram illustrates a flow 200 of some embodiments of an obstacle detection method according to the present disclosure. The obstacle detection method includes the following steps:

[0030] Step 201: Obtain the environmental point cloud data set.

[0031] In some embodiments, the entity performing the obstacle detection method (e.g. Figure 1The aforementioned computing device 101) can acquire the aforementioned environmental point cloud data set via a wired or wireless connection. The environmental point cloud data can be obtained by scanning the surrounding environment using a lidar installed on the target vehicle. The environmental point cloud data can include: horizontal coordinate values, vertical coordinate values, and radar echo power values. The horizontal, vertical, and radar echo values ​​included in the environmental point cloud data are coordinate values ​​in the target vehicle coordinate system. The target vehicle coordinate system can be a coordinate system with the target vehicle's direction of travel as the horizontal axis, the target vehicle's chassis center as the origin, a line parallel to the target vehicle's rear axle as the vertical axis, and a line perpendicular to the ground as the vertical axis.

[0032] Step 202: Prune the environmental point cloud dataset to generate a pruned environmental point cloud dataset.

[0033] In some embodiments, the aforementioned execution entity may perform various methods to prune the environmental point cloud dataset to generate a pruned environmental point cloud dataset.

[0034] In some optional implementations of the embodiments, the execution entity performs pruning processing on the environmental point cloud dataset to generate a pruned environmental point cloud dataset, which may include the following steps:

[0035] The first step is to obtain the lateral sensing distance and the sensing radius of the aforementioned LiDAR. The lateral sensing distance can be the maximum sensing distance to the left or right of the target vehicle. The sensing radius of the aforementioned LiDAR can be the maximum sensing distance of the aforementioned LiDAR.

[0036] The second step is to determine the non-clipping area based on the lateral sensing distance and sensing radius.

[0037] Optionally, the aforementioned execution entity can determine the non-clipping area (e.g., based on the lateral sensing distance and sensing radius) using the following formula. Figure 3 (The shaded area shown in the image):

[0038]

[0039] Where TR represents the aforementioned lateral sensing distance. x represents the horizontal coordinate of the environmental point cloud data in the aforementioned environmental point cloud dataset. y represents the vertical coordinate of the environmental point cloud data in the aforementioned environmental point cloud dataset. R represents the aforementioned sensing radius. CH represents the aforementioned vehicle body length.

[0040] The third step is to select environmental point cloud data that fall within the non-clipping region from the environmental point cloud data set as the clipped environmental point cloud data, thus obtaining the clipped environmental point cloud data set.

[0041] As an example, environmental point cloud data whose horizontal and vertical coordinate values ​​both fall within the non-clipping region can be selected from the above environmental point cloud data set as the clipped environmental point cloud data set, thus obtaining the clipped environmental point cloud data set.

[0042] Step 203: Downsample the cropped environmental point cloud dataset to generate a downsampled environmental point cloud dataset.

[0043] In some embodiments, the execution entity performs downsampling processing on the cropped environmental point cloud dataset to generate a downsampled environmental point cloud dataset, which may include the following steps:

[0044] The first step is to obtain ground information from a high-precision map.

[0045] The second step is to construct a fitting plane based on the above ground information.

[0046] The third step is to remove the environmental point cloud data that falls into the above-mentioned fitting plane from the above environmental point cloud data set to generate a downsampled environmental point cloud data set.

[0047] In some optional implementations of certain embodiments, the execution entity performs downsampling processing on the clipped environmental point cloud data set to generate a downsampled environmental point cloud data set, which may include the following steps:

[0048] The first step is to construct an octree based on the cropped environmental point cloud dataset and the preset maximum recursion depth.

[0049] The second step is to identify the environmental point cloud data contained in the above octree as downsampled environmental point cloud data, and obtain the set of downsampled environmental point cloud data.

[0050] Step 204: Input the downsampled environmental point cloud data set into the obstacle detection model to generate an obstacle information set.

[0051] In some embodiments, the execution entity can input a downsampled environmental point cloud dataset into an obstacle detection model to generate an obstacle information set. The obstacle detection model may include convolutional layers, pooling layers, and fully connected layers. The convolutional layers are used for feature extraction, the pooling layers for feature compression, and the fully connected layers for feature-based classification.

[0052] In some optional implementations of certain embodiments, the aforementioned execution entity inputs the downsampled environmental point cloud data set into an obstacle detection model to generate an obstacle information set. The obstacle detection model may include: a first feature extraction layer, a voxel segmentation and feature stitching layer, a second feature extraction layer, a unit feature tiling layer, a third feature extraction layer, and an obstacle attribute regression layer. It may include the following steps:

[0053] The first step involves performing sparse convolution on the environmental point cloud dataset using the first feature extraction layer 401 in the obstacle detection model to generate the first feature. The environmental point cloud dataset is an n×4 vector set. The first feature is an n×m vector set.

[0054] The second step involves dividing the first feature into voxels and stitching features together using the voxel segmentation and feature stitching layer 402 in the obstacle detection model to generate the second feature.

[0055] The third step involves further extracting features from the second feature extraction layer 403 in the obstacle detection model to generate the third feature. This third feature is an n×s vector group.

[0056] The fourth step involves tiling the third feature onto the corresponding voxel grid using the unit feature tiling layer 404 in the obstacle detection model to generate the fourth feature.

[0057] The fifth step involves using the third feature extraction layer 405 in the obstacle detection model to perform two-dimensional convolutional feature extraction on the fourth feature to generate the fifth feature.

[0058] The sixth step involves using the fifth feature to perform regression processing on the obstacle attributes through the obstacle attribute regression layer 406 in the obstacle detection model to generate an obstacle information set.

[0059] Step 205: Filter the obstacle information in the obstacle information set to generate a filtered obstacle information set.

[0060] In some embodiments, the execution entity can filter out obstacle information whose corresponding confidence values ​​are not within a preset range from the obstacle information set to generate a filtered obstacle information set. The preset range can be [0, 0.2].

[0061] Step 206: Send the set of filtered obstacle information to the control planning terminal via the vehicle communication module.

[0062] In some embodiments, the aforementioned execution entity can transmit the set of filtered obstacle information to the control planning terminal via a wired or wireless connection through an onboard communication module.

[0063] The various embodiments of this disclosure have the following beneficial effects: the obstacle detection methods of some embodiments of this disclosure improve the accuracy of obstacle detection results, thereby providing more accurate data for autonomous vehicles to avoid obstacles and reducing the risk level of autonomous vehicles during operation. Specifically, the inventors have found that the reason for inaccurate obstacle detection results is that the environmental point cloud data is not preprocessed, resulting in an inaccurate clustering structure, that is, the generated obstacle information is not accurate enough. Based on this, the obstacle detection methods of some embodiments of this disclosure perform cropping, downsampling, and filtering on the environmental point cloud data, thereby making the final generated obstacle information more accurate. In addition, since environmental point cloud data is relatively sparse, this disclosure generates obstacle information based on the characteristics of environmental point cloud data using an obstacle detection model, thereby improving the accuracy of the generated obstacle information.

[0064] Further reference Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of an obstacle detection device, which are similar to... Figure 2 Corresponding to the method embodiments shown, the device can be specifically applied to various electronic devices.

[0065] like Figure 5As shown, the obstacle detection device 500 in some embodiments includes: an acquisition unit 501, a cropping processing unit 502, a downsampling processing unit 503, an input unit 504, a filtering processing unit 505, and a transmission unit 506. The acquisition unit 501 is configured to acquire an environmental point cloud data set, wherein the environmental point cloud data is obtained by scanning the surrounding environment using a lidar installed on a target vehicle. The environmental point cloud data includes: abscissa values, ordinate values, vertical coordinate values, and radar echo power values. The abscissa, ordinate, and vertical coordinate values ​​included in the environmental point cloud data are coordinate values ​​in the target vehicle coordinate system. The target vehicle coordinate system is a coordinate system with the target vehicle's direction of travel as the abscissa, the target vehicle's chassis center as the origin, a line parallel to the target vehicle's rear axle as the ordinate, and a line perpendicular to the ground as the vertical axis. The cropping processing unit 502 is configured to crop the environmental point cloud data set to generate a cropped environmental point cloud data set. Downsampling processing unit 503 is configured to downsampling the cropped environmental point cloud data set to generate a downsampled environmental point cloud data set. Input unit 504 is configured to input the downsampled environmental point cloud data set into an obstacle detection model to generate an obstacle information set. Filtering processing unit 505 is configured to filter the obstacle information in the obstacle information set to generate a filtered obstacle information set. Transmission unit 506 is configured to transmit the filtered obstacle information set to a control planning terminal via an onboard communication module.

[0066] It is understandable that the units described in the device 500 are related to the reference. Figure 2 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method also apply to the device 500 and the units contained therein, and will not be repeated here.

[0067] The following is for reference. Figure 6 It illustrates electronic devices suitable for implementing some embodiments of this disclosure (e.g., Figure 1 A schematic diagram of the structure of the computing device 101)600. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0068] like Figure 6As shown, electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 601, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 602 or a program loaded from storage device 608 into random access memory (RAM) 603. RAM 603 also stores various programs and data required for the operation of electronic device 600. Processing device 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0069] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic device 600 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 An electronic device 600 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 6 Each box shown can represent a device or multiple devices as needed.

[0070] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a storage device 608, or installed from a ROM 602. When the computer program is executed by the processing device 601, it performs the functions defined above in the methods of some embodiments of this disclosure.

[0071] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0072] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0073] The aforementioned computer-readable medium may be included in the aforementioned device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: acquire an environmental point cloud data set, wherein the environmental point cloud data is obtained by scanning the surrounding environment using a lidar installed on the target vehicle; the environmental point cloud data includes: abscissa values, ordinate values, vertical coordinate values, and radar echo power values; the abscissa, ordinate, and vertical coordinate values ​​included in the environmental point cloud data are coordinate values ​​in a target vehicle coordinate system, wherein the target vehicle coordinate system has the target vehicle's direction of travel as the abscissa, the target vehicle's chassis center as the origin, a line parallel to the target vehicle's rear axle as the ordinate, and a line perpendicular to the ground as the vertical axis; clip the aforementioned environmental point cloud data set to generate a clipped environmental point cloud data set; and downsample the clipped environmental point cloud data set to generate a downsampled environmental point cloud data set. The downsampled environmental point cloud dataset is input into the obstacle detection model to generate an obstacle information set. The obstacle information in this set is then filtered to generate a filtered obstacle information set. This filtered obstacle information set is then transmitted to the control and planning terminal via the vehicle communication module.

[0074] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0075] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0076] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including an acquisition unit, a cropping processing unit, a downsampling processing unit, an input unit, a filtering processing unit, and a transmission unit. The names of these units do not necessarily limit the specific unit; for example, the acquisition unit may also be described as a "unit for acquiring a set of environmental point cloud data."

[0077] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0078] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for obstacle detection, comprising: obtaining a set of environment point cloud data, wherein the environment point cloud data is scanned by a laser radar installed on a target vehicle, and the environment point cloud data comprises a horizontal coordinate value, a vertical coordinate value, a vertical coordinate value, and a radar echo power value, the horizontal coordinate value, the vertical coordinate value, and the vertical coordinate value included in the environment point cloud data are coordinate values in a target vehicle coordinate system, and the target vehicle coordinate system is a coordinate system with the direction of travel of the target vehicle as the horizontal axis, the center of the chassis of the target vehicle as the origin, a line parallel to the rear axle of the target vehicle as the vertical axis, and a line perpendicular to the ground as the vertical axis; performing clipping processing on the set of environment point cloud data to generate a set of clipped environment point cloud data; performing down-sampling processing on the set of clipped environment point cloud data to generate a set of down-sampled environment point cloud data; inputting the set of down-sampled environment point cloud data into an obstacle detection model to generate a set of obstacle information, wherein the obstacle detection model comprises a first feature extraction layer, a voxel segmentation and feature splicing layer, a second feature extraction layer, a unit feature tiling layer, a third feature extraction layer, and an obstacle attribute regression layer; performing filtering processing on the obstacle information in the set of obstacle information to generate a set of filtered obstacle information; sending the set of filtered obstacle information to a control planning terminal through a vehicle-mounted communication module, wherein the inputting the set of down-sampled environment point cloud data into the obstacle detection model to generate the set of obstacle information comprises: performing sparse convolution on the set of environment point cloud data through the first feature extraction layer in the obstacle detection model to generate first features, wherein the set of environment point cloud data is an n x 4 vector group, and the first features are an n x m vector group; performing voxel segmentation and feature splicing on the first features through the voxel segmentation and feature splicing layer in the obstacle detection model to generate second features; performing further feature extraction on the second features through the second feature extraction layer in the obstacle detection model to generate third features, wherein the third features are an n x s vector group; tiling the third features into corresponding voxel grids through the unit feature tiling layer in the obstacle detection model to generate fourth features; performing two-dimensional convolution feature extraction on the fourth features through the third feature extraction layer in the obstacle detection model to generate fifth features; based on the fifth features, performing regression processing on obstacle attributes through the obstacle attribute regression layer in the obstacle detection model to generate the set of obstacle information, wherein the down-sampling processing on the set of clipped environment point cloud data to generate the set of down-sampled environment point cloud data comprises: obtaining ground information in a high-precision map; constructing a fitting plane based on the ground information; The environment point cloud data in the environment point cloud data set falling into the fitted plane is removed to generate a down-sampled environment point cloud data set, or the down-sampling processing is performed on the cropped environment point cloud data set to generate a down-sampled environment point cloud data set, comprising: An octree is constructed based on the cropped environment point cloud data set and a preset maximum recursion depth; The environment point cloud data contained in the octree is determined as the down-sampled environment point cloud data to obtain a down-sampled environment point cloud data set; The environment point cloud data set is cropped to generate a cropped environment point cloud data set, comprising: An input unit is configured to input the down-sampled environment point cloud data set to an obstacle detection model to generate an obstacle information set, wherein the obstacle detection model comprises: a first feature extraction layer, a voxel segmentation and feature splicing layer, a second feature extraction layer, a unit feature tiling layer, a third feature extraction layer, and an obstacle attribute regression layer; A filtering processing unit is configured to perform filtering processing on the obstacle information in the obstacle information set to generate a filtered obstacle information set.

2. An obstacle detection device, comprising: An obtaining unit configured to obtain an environment point cloud data set, wherein the environment point cloud data is obtained by scanning the surrounding environment through a laser radar installed on a target vehicle, and the environment point cloud data comprises a horizontal coordinate value, a vertical coordinate value, a vertical coordinate value, and a radar echo power value, the horizontal coordinate value, the vertical coordinate value, and the vertical coordinate value comprised in the environment point cloud data are coordinate values in a target vehicle coordinate system, and the target vehicle coordinate system is a coordinate system with the direction of travel of the target vehicle as the horizontal axis, the center of the chassis of the target vehicle as the origin, a line parallel to the rear axle of the target vehicle as the vertical axis, and a line perpendicular to the ground as the vertical axis; A cropping processing unit configured to crop the environment point cloud data set to generate a cropped environment point cloud data set; A down-sampling processing unit configured to perform down-sampling processing on the cropped environment point cloud data set to generate a down-sampled environment point cloud data set; An input unit configured to input the down-sampled environment point cloud data set to an obstacle detection model to generate an obstacle information set, wherein the obstacle detection model comprises: a first feature extraction layer, a voxel segmentation and feature splicing layer, a second feature extraction layer, a unit feature tiling layer, a third feature extraction layer, and an obstacle attribute regression layer; A filtering processing unit configured to perform filtering processing on the obstacle information in the obstacle information set to generate a filtered obstacle information set. ​ ​ ​ The sending unit is configured to send the filtered obstacle information set to a control planning terminal through a vehicle-mounted communication module, wherein the inputting of the down-sampled environment point cloud data set into the obstacle detection model to generate the obstacle information set comprises: performing sparse convolution on the environment point cloud data set through a first feature extraction layer in the obstacle detection model to generate first features, wherein the environment point cloud data set is a group of n×4 vectors, and the first features are a group of n×m vectors; performing voxel division and feature splicing on the first features through a voxel division and feature splicing layer in the obstacle detection model to generate second features; performing further feature extraction on the second features through a second feature extraction layer in the obstacle detection model to generate third features, wherein the third features are a group of n×s vectors; tiling the third features into corresponding voxel grids through a cell feature tiling layer in the obstacle detection model to generate fourth features; performing two-dimensional convolution feature extraction on the fourth features through a third feature extraction layer in the obstacle detection model to generate fifth features; based on the fifth features, performing regression processing on obstacle attributes through an obstacle attribute regression layer in the obstacle detection model to generate the obstacle information set, wherein the down-sampling processing of the cropped environment point cloud data set to generate the down-sampled environment point cloud data set comprises: obtaining ground information in a high-precision map; constructing a fitting plane based on the ground information; removing environment point cloud data in the environment point cloud data set that falls on the fitting plane to generate the down-sampled environment point cloud data set, or the down-sampling processing of the cropped environment point cloud data set to generate the down-sampled environment point cloud data set comprises: based on the cropped environment point cloud data set and a preset maximum recursion depth, constructing an octree; determining environment point cloud data contained in the octree as down-sampled environment point cloud data to obtain the down-sampled environment point cloud data set; wherein the cropping processing of the environment point cloud data set to generate the cropped environment point cloud data set comprises: obtaining a lateral perception distance and a perception radius of the lidar, wherein the lateral perception distance is a maximum perception distance on the left or right side of the target vehicle, and the perception radius of the lidar is a maximum perception distance of the lidar; based on the lateral perception distance and the perception radius, determining a non-cropping region; selecting environment point cloud data falling in the non-cropping region from the environment point cloud data set as cropped environment point cloud data to obtain the cropped environment point cloud data set; wherein the determination of the non-cropping region based on the lateral perception distance and the perception radius comprises: determining the non-cropping region through the following formula: In the formula, TR represents the lateral perception distance, x represents a lateral coordinate included in the environmental point cloud data, y represents a longitudinal coordinate included in the environmental point cloud data, R represents the perception radius, and CH represents the body length of the target vehicle. 3.An electronic device, comprising: one or more processors; a memory device having stored thereon one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors being caused to implement the method of claim 1.

4. A computer readable medium having stored thereon a computer program, wherein, the program, when executed by a processor, implements the method of claim 1.

Images