Obstacle detection method, device and computer equipment

Abstract

The present application relates to an obstacle detection method, device and computer equipment, wherein the method comprises: obtaining point cloud data of a laser radar; inputting the point cloud data into an obstacle detection model to obtain an obstacle detection result; inputting the point cloud data into a noise segmentation model to obtain a noise segmentation result; and verifying the obstacle detection result with the noise segmentation result under a unified view to determine whether each obstacle in the obstacle detection result is a false detection. Through the present application, the problems of large computational complexity and poor obstacle detection effect in the related technology are solved, and the filtering of false obstacle detection is achieved, thereby improving the obstacle detection effect.

Description

Technical Field

[0001] The present application relates to the technical field of point cloud data processing, and in particular to an obstacle detection method, device and computer equipment. Background Art

[0002] In autonomous driving, the environmental perception system is the biggest guarantee for vehicle driving safety, and provides basic support for high-level tasks such as subsequent vehicle path planning and behavior decision-making. Among them, point cloud data based on lidar has point cloud noise. The main sources of point cloud noise are two aspects: on the one hand, the noise caused by the surface of the target object. For example, the properties of the surface material of the target object (the point cloud reflection energy caused by the high reflectivity surface material is too strong, making the target larger than the actual size), the roughness (the uneven surface causes the emission angle of the point cloud to change), etc.; on the other hand, the noise caused by the external scanning environment, such as rain, snow, fog, dust and other particles blocking the point cloud and unable to reflect back to the point cloud. The traditional solution is: use a filtering algorithm to remove the point cloud noise in the point cloud data, and then detect obstacles based on the denoised point cloud data to improve the detection accuracy. However, the calculation amount of this solution is very large, and its final detection effect still depends on the technical level of the technicians, resulting in poor obstacle detection effect.

[0003] There is no effective solution to the problems of large amount of calculation and poor obstacle detection effect in related technologies. Summary of the invention

[0004] In this embodiment, an obstacle detection method, apparatus and computer device are provided to solve the problems of large amount of calculation and poor obstacle detection effect in the related art.

[0005] In a first aspect, an obstacle detection method is provided in this embodiment, including:

[0006] Get the point cloud data of LiDAR;

[0007] Inputting the point cloud data into an obstacle detection model to obtain an obstacle detection result;

[0008] Inputting the point cloud data into a noise segmentation model to obtain a noise segmentation result;

[0009] In a unified view, the obstacle detection result is verified with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection.

[0010] In some of the embodiments, verifying the obstacle detection result with the noise segmentation result in a unified view to determine whether each obstacle in the obstacle detection result is a false detection includes:

[0011] When the noise segmentation result is in a feature view and the obstacle detection result is in a bird's-eye view;

[0012] Converting the noise segmentation result into a grid map under a bird's-eye view to obtain a noise segmentation map;

[0013] Under the bird's-eye view, the obstacle detection result is verified with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection.

[0014] In some of the embodiments, verifying the obstacle detection result with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection includes:

[0015] Verifying the obstacle detection result with the noise segmentation map to obtain the noise point cloud ratio of each obstacle in the obstacle detection result;

[0016] The noise point cloud ratio of the obstacle is compared with a preset ratio threshold, and whether the obstacle is a false detection is determined according to the comparison result.

[0017] In some of the embodiments, verifying the obstacle detection result with the noise segmentation result in a unified view to determine whether each obstacle in the obstacle detection result is a false detection includes:

[0018] When the noise segmentation result is in a feature view and the obstacle detection result is in a bird's-eye view;

[0019] Converting the obstacle detection result into a grid map under the feature view to obtain an obstacle detection map;

[0020] Under the feature view, the obstacle detection map is verified with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection.

[0021] In some embodiments, the obstacle detection result is converted into a grid map under a feature view to obtain an obstacle detection map, including:

[0022] Based on preset projection parameters, the point cloud data corresponding to all two-dimensional points in each grid in the obstacle detection result is projected onto the grid map under the feature view to obtain the obstacle detection map.

[0023] In some of the embodiments, verifying the obstacle detection map with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection includes:

[0024] Verifying the obstacle detection map and the noise segmentation result to obtain the noise point cloud ratio of each obstacle in the obstacle detection result;

[0025] The noise point cloud ratio of the obstacle is compared with a preset ratio threshold, and whether the obstacle is a false detection is determined according to the comparison result.

[0026] In some embodiments, the point cloud data is input into a noise segmentation model to obtain a noise segmentation result, including:

[0027] Converting the point cloud data into a feature map under a feature view;

[0028] The feature map with the point cloud data is input into a noise segmentation model to obtain a noise segmentation result under a feature view.

[0029] In some of the embodiments, converting the point cloud data into a feature map under a feature view includes:

[0030] Determine whether multiple points in the point cloud data are converted to the same pixel point of the feature map;

[0031] If multiple points in the point cloud data are converted to the same pixel point of the feature map, a target point is selected from the points and converted to the feature map; the target point is the point among the points that is closest to the original location of the vehicle body coordinate system; the feature map is in the vehicle body coordinate system.

[0032] In some of the embodiments, verifying the obstacle detection result with the noise segmentation result in a unified view to determine whether each obstacle in the obstacle detection result is a false detection includes:

[0033] When both the noise segmentation result and the obstacle detection result are in a feature view, verifying the obstacle detection result with the noise segmentation result in the feature view to determine whether each obstacle in the obstacle detection result is a false detection;

[0034] Or, when the noise segmentation result and the obstacle detection result are both in a bird's-eye view, the obstacle detection result is verified with the noise segmentation result in the bird's-eye view to determine whether each obstacle in the obstacle detection result is a false detection.

[0035] In a second aspect, in this embodiment, an obstacle detection device is provided, including:

[0036] Acquisition module, used to obtain point cloud data of laser radar;

[0037] A first processing module, used for inputting the point cloud data into an obstacle detection model to obtain an obstacle detection result;

[0038] A second processing module, used for inputting the point cloud data into a noise segmentation model to obtain a noise segmentation result;

[0039] The verification module is used to verify the obstacle detection result and the noise segmentation result in a unified view to determine whether each obstacle in the obstacle detection result is a false detection.

[0040] In a third aspect, a computer device is provided in this embodiment, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the obstacle detection method described in the first aspect when executing the computer program.

[0041] Compared with the related art, the obstacle detection method, device and computer equipment provided in this embodiment obtain point cloud data of the laser radar; input the point cloud data into the obstacle detection model to obtain the obstacle detection result; input the point cloud data into the noise segmentation model to obtain the noise segmentation result; under the unified view, the obstacle detection result and the noise segmentation result are verified to determine whether each obstacle in the obstacle detection result is a false detection, thereby solving the problems of large calculation amount and poor obstacle detection effect in the related art, realizing the filtering of false obstacle detections, and improving the obstacle detection effect.

[0042] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0044] Figure 1 It is a hardware structure block diagram of a terminal device of an obstacle detection method provided in an embodiment of the present application;

[0045] Figure 2 is a flow chart of an obstacle detection method provided by an embodiment of the present application;

[0046] Figure 3 is a flow chart of an embodiment of step S240;

[0047] Figure 4 is a flow chart of another embodiment of step S240;

[0048] Figure 5 is a flowchart of step S243;

[0049] Figure 6 is a flow chart of step S245;

[0050] Figure 7 is a flowchart of step S230;

[0051] Figure 8 It is a structural block diagram of an obstacle detection device provided in one embodiment of the present application.

[0052] In the figure: 102, processor; 104, memory; 106, transmission device; 108, input and output device; 210, acquisition module; 220, first processing module; 230, second processing module; 240, verification module. DETAILED DESCRIPTION

[0053] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0054] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with general skills in the technical field to which this application belongs. The words "one", "a", "the", "these" and the like in this application do not indicate a quantitative limitation, and they may be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" may mean: A exists alone, A and B exist at the same time, and B exists alone. Generally, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0055] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1FIG. 1 is a hardware structure diagram of a terminal of the obstacle detection method of this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 and a memory 104 for storing data, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0056] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the obstacle detection method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0057] The transmission device 106 is used to receive or send data via a network. The above network includes a wireless network provided by the communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0058] In this embodiment, an obstacle detection method is provided. Figure 2 is a flow chart of the obstacle detection method of this embodiment. Figure 2 As shown, the process includes the following steps:

[0059] Step S210, obtaining point cloud data of the laser radar;

[0060] Step S220, inputting the point cloud data into the obstacle detection model to obtain the obstacle detection result;

[0061] Step S230, inputting the point cloud data into the noise segmentation model to obtain a noise segmentation result;

[0062] Step S240: Under the unified view, the obstacle detection result and the noise segmentation result are verified to determine whether each obstacle in the obstacle detection result is a false detection.

[0063] It should be noted that both the obstacle detection model and the noise segmentation model are pre-trained. The networks of the obstacle detection model and the noise segmentation model include but are not limited to convolutional neural network (CNN), fully connected neural network (FCN), generative adversarial network (GAN), residual network (ResNet), etc. The obstacle detection model and the noise segmentation model can be obtained by training the corresponding network using relevant training samples and training methods. For example, for the noise segmentation model: if the training sample is point cloud data during training, the output is the noise segmentation result under the feature view (Range View, RV); RangeView can also be called range view, interval view or field of view range view, etc. Then when the noise segmentation model is applied, the input is point cloud data, and the output is the noise segmentation result under the feature view. If the training sample is a feature map under the feature view during training, the output is the noise segmentation result under the feature view. Then when the noise segmentation model is applied, the input is the feature map under the feature view (the point cloud data needs to be converted into the feature map under the feature view), and the output is the noise segmentation result under the feature view. In this embodiment, the view may also be a bird's eye view (BEV), a point view (PV), etc. There is no limitation on the training samples and the training method.

[0064] Among them, the laser radar is set on the vehicle, and the point cloud data of the laser radar can be obtained in real time during the use of the vehicle. Then the point cloud data is input into the obstacle detection model and the noise segmentation model respectively to obtain the corresponding obstacle detection results and noise segmentation results; wherein, the obstacle detection result includes the detected obstacle and the point cloud information corresponding to the obstacle; if the obstacle detection result contains multiple obstacles, each obstacle has its corresponding point cloud information. Finally, the obstacle detection result and the noise segmentation result are verified under the unified view; at this time, the view where the obstacle detection result and the noise segmentation result are located can be determined first; if the obstacle detection result and the noise segmentation result are in the same view, then the obstacle detection result and the noise segmentation result can be directly verified to determine whether each obstacle in the obstacle detection result is a false detection. If the obstacle detection result and the noise segmentation result are in different views, the obstacle detection result can be converted to the view of the noise segmentation result, or the noise segmentation result can be converted to the view of the obstacle detection result, or the obstacle detection result and the noise segmentation result can be converted to a third-party view, and then the obstacle detection result and the noise segmentation result are verified to determine whether each obstacle in the obstacle detection result is a false detection.

[0065] Since the point cloud data collected by the laser radar contains point cloud noise, it will affect the accuracy of obstacle detection. Therefore, the traditional solution is: use a filtering algorithm to remove the point cloud noise in the point cloud data, and then detect obstacles based on the denoised point cloud data to improve the detection accuracy. However, the calculation amount of this solution is very large, and its final detection effect also depends on the technical level of the technicians, resulting in poor obstacle detection effect. In this embodiment. By obtaining the point cloud data of the laser radar; inputting the point cloud data into the obstacle detection model to obtain the obstacle detection result; inputting the point cloud data into the noise segmentation model to obtain the noise segmentation result; in a unified view, the obstacle detection result is verified with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection, which solves the problem of large calculation amount and poor obstacle detection effect in the related technology, and uses the output result of the noise segmentation model to verify the output result of the obstacle detection model, so as to filter out the false detection of obstacles and improve the detection effect of obstacles.

[0066] The above steps are described in detail below:

[0067] It should be noted that the noise segmentation result and the obstacle detection result are located in different views, and the specific implementation of step S240 is also different.

[0068] For the noise segmentation results in the feature view and the obstacle detection results in the bird's eye view:

[0069] In some of the embodiments, Figure 3 As shown, in step S240, in a unified view, the obstacle detection result and the noise segmentation result are verified to determine whether each obstacle in the obstacle detection result is a false detection, including the following steps:

[0070] Step S241, when the noise segmentation result is in the feature view and the obstacle detection result is in the bird's eye view;

[0071] Step S242, converting the noise segmentation result into a grid map under the bird's-eye view to obtain a noise segmentation map;

[0072] Step S243, in the bird's-eye view, verify the obstacle detection result with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection;

[0073] Among them, the steps from step S241 to step S243 can be considered as a unified conversion to the bird's-eye view for verification to determine whether each obstacle in the obstacle detection result is a false detection.

[0074] Since the noise segmentation result is used to verify whether the obstacle in the obstacle detection result is a false detection, and since the point cloud information corresponding to the obstacle is in the bird's-eye view (the bird's-eye view is built in the vehicle body coordinate system), in order to better verify the noise segmentation result in the feature view with the obstacle in the obstacle detection result in the bird's-eye view, the relevant conversion algorithm is used to project the noise segmentation information in the noise segmentation result into a grid map of a preset size under the bird's-eye view to obtain a noise segmentation map; the size of the grid map is the same as the size of the map where the obstacle is located in the obstacle detection result.

[0075] For example, the size of the grid image can be 800×320, which is in the bird's-eye view. The grid image represents the range of [-20m, 140m] in the x direction and [-32m, 32m] in the y direction in the vehicle body coordinate system, and the distance unit is 0.2m. In the conversion process, it involves the conversion of multiple pixels (RV pixels) of different noise segmentation results to pixels in the same grid image under the bird's-eye view; since noise is a strong feature, as long as one of these RV pixels is noise, the corresponding pixel in the bird's-eye view is also a noise pixel. If all of these RV pixels are non-noise pixels, the corresponding pixel in the bird's-eye view is also a non-noise pixel. For the judgment of whether it is noise, it can be: the input feature value of each pixel in the noise segmentation map corresponds to the x, y, and z position information in the vehicle body coordinate system (bird's-eye view); therefore, it can be judged which pixel in the grid image under the bird's-eye view in the obstacle detection result corresponds to. Thus, it is judged whether the point is noise. The above is just an example, and of course there is no restriction on the specific parameters of the grid image.

[0076] In this embodiment, while reducing the amount of calculation, it is possible to convert the obstacle detection results and the noise segmentation results into a bird's-eye view for verification, and determine whether each obstacle in the obstacle detection result is a false detection.

[0077] In some of the embodiments, Figure 4 As shown, in step S240, in a unified view, the obstacle detection result and the noise segmentation result are verified to determine whether each obstacle in the obstacle detection result is a false detection, including the following steps:

[0078] Step S244, when the noise segmentation result is in the feature view and the obstacle detection result is in the bird's-eye view;

[0079] Step S245, converting the obstacle detection result into a grid map under the feature view to obtain an obstacle detection map;

[0080] Step S246: Under the feature view, the obstacle detection image is verified with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection.

[0081] Among them, the steps from step S244 to step S246 can be considered as a unified conversion to the feature view for verification to determine whether each obstacle in the obstacle detection result is a false detection.

[0082] Since the noise segmentation result is used to verify whether the obstacle in the obstacle detection result is a false detection, and since the point cloud information corresponding to the obstacle is in the bird's-eye view, in order to better verify the noise segmentation result under the feature view with the obstacle in the obstacle detection result under the bird's-eye view, the obstacle detection result is converted into a grid map under the feature view of a preset size using a related conversion algorithm to obtain an obstacle detection map; the size of the grid map is the same as the size of the map where the obstacle is located in the obstacle detection result.

[0083] In this embodiment, while reducing the amount of calculation, it is possible to convert the obstacle detection result and the noise segmentation result into a feature view for verification, and determine whether each obstacle in the obstacle detection result is a false detection.

[0084] In other embodiments, a similar method to the above may be adopted to convert the obstacle detection results and the noise segmentation results into a third-party view, and then verify the obstacle detection results and the noise segmentation results to determine whether each obstacle in the obstacle detection results is a false detection, which will not be repeated.

[0085] In some of the embodiments, Figure 5As shown, step S243 of verifying the obstacle detection result with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection includes the following steps:

[0086] Step S2431, verifying the obstacle detection result with the noise segmentation map to obtain the noise point cloud ratio of each obstacle in the obstacle detection result;

[0087] Step S2432, comparing the noise point cloud ratio of the obstacle with a preset ratio threshold, and determining whether the obstacle is a false detection based on the comparison result.

[0088] Specifically, the obstacle detection result includes obstacle and point cloud information; each obstacle has corresponding point cloud information. The above implementation process is as follows:

[0089] First, determine whether each point cloud information in the detection result and the corresponding pixel point in the noise segmentation map are noise pixels; if they are noise pixels, mark the point cloud information as noise point cloud; secondly, traverse all the point cloud information in the obstacle detection result, and use the number of marked noise point clouds as the evaluation condition to obtain the noise point cloud ratio of each obstacle in the obstacle detection result; thirdly, traverse each obstacle to determine whether the noise point cloud ratio of the obstacle exceeds the preset ratio threshold; if the noise point cloud ratio of the obstacle exceeds the preset ratio threshold, remove the corresponding obstacle; if the noise point cloud ratio of the obstacle does not exceed the preset ratio threshold, retain the corresponding obstacle. In the above judgment process, the ratio threshold can also be adjusted according to the size of the obstacle; if the obstacle is within a range of less than ten meters, the ratio threshold can be 70%. If the obstacle is within a range of more than ten meters, the ratio threshold can be 30%.

[0090] Through this embodiment, the filtering effect of false obstacle detection is improved, and the accuracy of obstacle detection is guaranteed.

[0091] In other embodiments, in order to improve the filtering effect of false obstacle detection and ensure the accuracy of obstacle detection, multiple tests can be performed and then comprehensively evaluated. For example, if steps S2431 to S2432 are the detection results of one frame of point cloud data, the detection results of multiple frames of point cloud data can be continuously detected. If two consecutive detection results determine that the obstacle is a false detection, then it will be finally confirmed.

[0092] In some of the embodiments, Figure 6 As shown, the step S245 of converting the obstacle detection result into a grid map under the feature view to obtain an obstacle detection map includes the following steps:

[0093] Step S2451, based on preset projection parameters, the point cloud data corresponding to all two-dimensional points in each grid in the obstacle detection result is projected onto the grid map under the feature view to obtain an obstacle detection map.

[0094] Specifically, a related conversion algorithm is used to convert the obstacle detection results into a grid map under the feature view to obtain a detected grid map. Among them, the projection parameters are pre-set, which are the internal and external parameters of the point cloud data acquisition device. Since the bird's-eye view is a view from the top of an object or scene, it has no height information; while the feature view is a view from a horizontal perspective, it has height information. Therefore, the point cloud data (three-dimensional points) corresponding to all two-dimensional points in the grid are used to provide the height information required by the feature view, thereby completing the conversion of all two-dimensional points in the obstacle detection results. The implementation process can be: the point cloud data (three-dimensional points) corresponding to all two-dimensional points in each grid in the obstacle detection result are combined with the projection parameters and projected into the grid map under the feature view; in this way, each grid in each obstacle detection result is traversed to obtain an obstacle detection map.

[0095] Through this embodiment, the missing height information is provided by using point cloud data, thereby improving the accuracy of the data in the obstacle detection map.

[0096] In some embodiments, the step S246 of verifying the obstacle detection map with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection includes the following steps:

[0097] Step S2461, verifying the obstacle detection map and the noise segmentation result to obtain the noise point cloud ratio of each obstacle in the obstacle detection result;

[0098] Step S2462, comparing the noise point cloud ratio of the obstacle with a preset ratio threshold, and determining whether the obstacle is a false detection based on the comparison result.

[0099] The above process is: determine whether the point cloud information corresponding to each pixel in the obstacle detection map and the pixel corresponding to the noise segmentation result are noise pixels; if they are noise pixels, mark the point cloud information as noise point cloud; secondly, traverse all point cloud information, take the number of marked noise point clouds as the evaluation condition, and obtain the noise point cloud ratio of each obstacle in the obstacle detection result; thirdly, traverse each obstacle to determine whether the noise point cloud ratio of the obstacle exceeds the preset ratio threshold; if the noise point cloud ratio of the obstacle exceeds the preset ratio threshold, remove the corresponding obstacle; if the noise point cloud ratio of the obstacle does not exceed the preset ratio threshold, retain the corresponding obstacle. In the above judgment process, the ratio threshold can also be adjusted according to the size of the obstacle; if the obstacle is within the range of ten meters or less, the ratio threshold can be 70%. If the obstacle is within the range of ten meters or more, the ratio threshold can be 30%.

[0100] Through this embodiment, the characteristics of the feature view are utilized to identify obstacles at different distances, so as to further improve the filtering effect of obstacle false detection and ensure the accuracy of obstacle detection.

[0101] For the case where both the noise segmentation results and the obstacle detection results are in the feature view, or the noise segmentation results and the obstacle detection results are in the bird's-eye view:

[0102] In some embodiments, in step S240, in a unified view, verifying the obstacle detection result with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection includes the following steps:

[0103] When both the noise segmentation result and the obstacle detection result are in the feature view, the obstacle detection result is verified with the noise segmentation result in the feature view to determine whether each obstacle in the obstacle detection result is a false detection;

[0104] Alternatively, when both the noise segmentation result and the obstacle detection result are in the bird's-eye view, the obstacle detection result is verified with the noise segmentation result in the bird's-eye view to determine whether each obstacle in the obstacle detection result is a false detection.

[0105] The specific implementation of this embodiment can refer to the above-mentioned related processes, which will not be repeated here.

[0106] In some embodiments, the step S220 of inputting the point cloud data into the obstacle detection model to obtain the obstacle detection result includes the following steps:

[0107] Step S221, preprocessing the point cloud data;

[0108] Step S222: input the preprocessed point cloud data into the obstacle detection model to obtain the obstacle detection result.

[0109] Specifically, preprocessing includes but is not limited to filtering, denoising, etc. Among them, filtering is applicable to the scene of vehicle driving on the road. The point cloud data on both sides of the road are filtered using the road edge detection algorithm, which can effectively reduce the number of irrelevant point clouds, thereby obtaining effective point cloud data. Since the collection of point cloud data is affected by factors such as the accuracy of the laser radar, it is necessary to denoise the point cloud data.

[0110] Through this embodiment, the preprocessed point cloud data is input into the obstacle detection model to obtain the obstacle detection result, which can effectively improve the recognition accuracy of the obstacle detection model.

[0111] In some of the embodiments, Figure 7 As shown, step S230 of inputting point cloud data into the noise segmentation model to obtain a noise segmentation result includes the following steps:

[0112] Step S231, converting the point cloud data into a feature map under the feature view;

[0113] Step S232: input the feature map with point cloud data into the noise segmentation model to obtain the noise segmentation result under the feature view.

[0114] Specifically, all point cloud data are converted into a feature map under the RV view (RV feature map). The size of the feature map is 640×128, its upper and lower sizes are [-64°, 64°], its left and right sizes are [-12.8°, 12.8°], and the resolution is 0.2. The above is just an example, and of course there is no restriction on the specific parameters of the feature map.

[0115] The above feature map is input into the noise segmentation model to obtain the noise segmentation result under the feature view; the value of each pixel in the noise segmentation result is a value in the range of [0, 1], which represents the possible row of the pixel being noise, 1 is the most likely, and 0 is the least likely. If the predicted value of the pixel is greater than the preset noise threshold, the pixel is a noise pixel, otherwise, the pixel is a non-noise pixel.

[0116] The view where the feature map is located is related to the view of the training sample of the noise segmentation model; for example, if the training sample of the noise segmentation model is in the RV view, then all the point cloud data are converted to a feature map in the RV view. In other embodiments, if the training sample of the noise segmentation model is in other views, then the point cloud data will be converted to the corresponding view accordingly, and there is no limitation on this.

[0117] The training sample may be an RV feature map, and each feature point of the RV feature map includes at least three feature information (x, y, z). Of course, the more feature information a feature point has, the better the noise segmentation effect of the noise segmentation model will be. In other embodiments, the feature information of the feature point may also include reflection intensity information and depth information, etc. The depth information is If each feature point includes the above five feature information, the RV feature map can be expressed as 640×128×5. Because the feature map with point cloud data corresponds to the RV feature map of the training sample, the feature map with point cloud data can also be expressed as 640×128×5.

[0118] Through this embodiment, the point cloud data is converted into a feature map under a feature view, and effective and stable data is retained, so that the output of the noise segmentation model is more generalized.

[0119] In some embodiments, the step S231 of converting the point cloud data into a feature map under the feature view includes the following steps:

[0120] Determine whether multiple points in the point cloud data are converted to the same pixel point in the feature map;

[0121] If multiple points in the point cloud data are converted to the same pixel point of the feature map, the target point is selected from the points and converted to the feature map; the target point is the point in the point that is closest to the original location of the vehicle body coordinate system; the feature map is in the vehicle body coordinate system.

[0122] Specifically, in the process of converting point cloud data to feature maps, multiple different points will be converted to the same pixel in the feature map. This is due to geometric properties. In order to give priority to detecting obstacles closest to the vehicle body, the point closest to the original location of the vehicle body coordinate system is selected as the target point and converted to the feature map. In other embodiments, these points can also be fused into one target point and then converted to the feature map, which can improve the robustness of detection.

[0123] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0124] In this embodiment, an obstacle detection device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. The terms "module", "unit", "subunit", etc. used below can be a combination of software and / or hardware that implements the predetermined functions. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0125] Figure 8 is a structural block diagram of the obstacle detection device of this embodiment, such as Figure 8 As shown, the device comprises:

[0126] An acquisition module 210 is used to acquire point cloud data of a laser radar;

[0127] The first processing module 220 is used to input the point cloud data into the obstacle detection model to obtain the obstacle detection result;

[0128] The second processing module 230 is used to input the point cloud data into the noise segmentation model to obtain a noise segmentation result;

[0129] The verification module 240 is used to verify the obstacle detection result and the noise segmentation result in a unified view, and determine whether each obstacle in the obstacle detection result is a false detection.

[0130] The above device solves the problems of large amount of calculation and poor obstacle detection effect in related technologies, achieves filtering of false obstacle detection, and improves the obstacle detection effect.

[0131] In some of the embodiments, the noise segmentation result is in the feature view, and the obstacle detection result is in the bird's-eye view; the verification module 240 is further used to convert the noise segmentation result into a grid map in the bird's-eye view to obtain a noise segmentation map;

[0132] Verify the obstacle detection result with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection;

[0133] Or, the obstacle detection result is converted into a grid map under the feature view to obtain an obstacle detection map;

[0134] The obstacle detection map is verified with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection.

[0135] In some of the embodiments, the verification module 240 is further used to verify the obstacle detection result with the noise segmentation map to obtain the noise point cloud ratio of each obstacle in the obstacle detection result;

[0136] The noise point cloud ratio of the obstacle is compared with the preset ratio threshold, and whether the obstacle is a false detection is determined based on the comparison result.

[0137] In some of the embodiments, the verification module 240 is further used to project the point cloud data corresponding to all two-dimensional points in each grid in the obstacle detection result into the grid map under the feature view based on preset projection parameters to obtain an obstacle detection map.

[0138] In some of the embodiments, the verification module 240 is further used to verify the obstacle detection map and the noise segmentation result to obtain the noise point cloud ratio of each obstacle in the obstacle detection result;

[0139] The noise point cloud ratio of the obstacle is compared with the preset ratio threshold, and whether the obstacle is a false detection is determined based on the comparison result.

[0140] In some of the embodiments, the second processing module 230 is further used to convert the point cloud data into a feature map under the feature view;

[0141] The feature map with point cloud data is input into the noise segmentation model to obtain the noise segmentation result under the feature view.

[0142] In some of the embodiments, the second processing module 230 is further used to determine whether multiple points in the point cloud data are converted to the same pixel point of the feature map;

[0143] If multiple points in the point cloud data are converted to the same pixel point of the feature map, the target point is selected from the points and converted to the feature map; the target point is the point in the point that is closest to the original location of the vehicle body coordinate system; the feature map is in the vehicle body coordinate system.

[0144] In some of the embodiments, the noise segmentation result and the obstacle detection result are both in a feature view, or the noise segmentation result and the obstacle detection result are both in a bird's eye view;

[0145] The verification module 240 is further used to verify the obstacle detection result and the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection.

[0146] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0147] In this embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0148] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0149] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0150] S1, obtain the point cloud data of the laser radar;

[0151] S2, input the point cloud data into the obstacle detection model to obtain the obstacle detection result;

[0152] S3, inputting the point cloud data into the noise segmentation model to obtain the noise segmentation result;

[0153] S4, in a unified view, verify the obstacle detection result with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection.

[0154] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0155] In addition, in combination with the obstacle detection method provided in the above embodiments, a storage medium may be provided in this embodiment to implement the obstacle detection method. The storage medium stores a computer program, and when the computer program is executed by a processor, any obstacle detection method in the above embodiments is implemented.

[0156] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0157] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0158] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0159] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. An obstacle detection method, characterized in that: include: Get the point cloud data of LiDAR; Inputting the point cloud data into an obstacle detection model to obtain an obstacle detection result; Inputting the point cloud data into a noise segmentation model to obtain a noise segmentation result; Under a unified view, verifying the obstacle detection result with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection; The step of verifying the obstacle detection result and the noise segmentation result under the unified view to determine whether each obstacle in the obstacle detection result is a false detection includes: When the noise segmentation result is in a feature view and the obstacle detection result is in a bird's-eye view; Converting the noise segmentation result into a grid map under a bird's-eye view to obtain a noise segmentation map; Under the bird's-eye view, verifying the obstacle detection result with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection; Or, in the unified view, verifying the obstacle detection result with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection, including: When the noise segmentation result is in a feature view and the obstacle detection result is in a bird's-eye view; Converting the obstacle detection result into a grid map under the feature view to obtain an obstacle detection map; Under the feature view, verifying the obstacle detection map with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection; Or, in the unified view, verifying the obstacle detection result with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection, including: When both the noise segmentation result and the obstacle detection result are in a feature view, verifying the obstacle detection result with the noise segmentation result in the feature view to determine whether each obstacle in the obstacle detection result is a false detection; Or, when the noise segmentation result and the obstacle detection result are both in a bird's-eye view, the obstacle detection result is verified with the noise segmentation result in the bird's-eye view to determine whether each obstacle in the obstacle detection result is a false detection.

2. The obstacle detection method according to claim 1, characterized in that: Verifying the obstacle detection result with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection includes: Verifying the obstacle detection result with the noise segmentation map to obtain the noise point cloud ratio of each obstacle in the obstacle detection result; The noise point cloud ratio of the obstacle is compared with a preset ratio threshold, and whether the obstacle is a false detection is determined according to the comparison result.

3. The obstacle detection method according to claim 1, characterized in that: The obstacle detection result is converted into a grid map under the feature view to obtain an obstacle detection map, including: Based on preset projection parameters, the point cloud data corresponding to all two-dimensional points in each grid in the obstacle detection result is projected onto the grid map under the feature view to obtain the obstacle detection map.

4. The obstacle detection method according to claim 3, characterized in that: Verifying the obstacle detection map with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection includes: Verifying the obstacle detection map and the noise segmentation result to obtain the noise point cloud ratio of each obstacle in the obstacle detection result; The noise point cloud ratio of the obstacle is compared with a preset ratio threshold, and whether the obstacle is a false detection is determined according to the comparison result.

5. The obstacle detection method according to claim 1, characterized in that: The point cloud data is input into the noise segmentation model to obtain the noise segmentation result, including: Converting the point cloud data into a feature map under a feature view; The feature map with the point cloud data is input into a noise segmentation model to obtain a noise segmentation result under a feature view.

6. The obstacle detection method according to claim 5, characterized in that: Converting the point cloud data into a feature map under a feature view includes: Determine whether multiple points in the point cloud data are converted to the same pixel point of the feature map; If multiple points in the point cloud data are converted to the same pixel point of the feature map, a target point is selected from the points and converted to the feature map; the target point is the point among the points that is closest to the original location of the vehicle body coordinate system; the feature map is in the vehicle body coordinate system.

7. An obstacle detection device, characterized in that: include: Acquisition module, used to obtain point cloud data of laser radar; A first processing module, used for inputting the point cloud data into an obstacle detection model to obtain an obstacle detection result; A second processing module, used for inputting the point cloud data into a noise segmentation model to obtain a noise segmentation result; A verification module, used to verify the obstacle detection result and the noise segmentation result in a unified view, and determine whether each obstacle in the obstacle detection result is a false detection; The verification module is further used when the noise segmentation result is in a feature view and the obstacle detection result is in a bird's eye view; Converting the noise segmentation result into a grid map under a bird's-eye view to obtain a noise segmentation map; Under the bird's-eye view, verifying the obstacle detection result with the noise segmentation map to determine whether each obstacle in the obstacle detection result is a false detection; Or, the verification module is further used when the noise segmentation result is in a feature view and the obstacle detection result is in a bird's eye view; Converting the obstacle detection result into a grid map under the feature view to obtain an obstacle detection map; Under the feature view, verifying the obstacle detection map with the noise segmentation result to determine whether each obstacle in the obstacle detection result is a false detection; or, the verification module is further configured to verify the obstacle detection result and the noise segmentation result in the feature view when both the noise segmentation result and the obstacle detection result are in the feature view, so as to determine whether each obstacle in the obstacle detection result is a false detection; Or, when the noise segmentation result and the obstacle detection result are both in a bird's-eye view, the obstacle detection result is verified with the noise segmentation result in the bird's-eye view to determine whether each obstacle in the obstacle detection result is a false detection.

8. A computer device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the steps of the obstacle detection method according to any one of claims 1 to 6.

Images