Intersection detection method, readable storage medium and intelligent device

Abstract

The present application relates to the technical field of autonomous driving, and specifically provides an intersection detection method, a readable storage medium and an intelligent device, which aim to solve the problem of how to effectively detect an intersection in a low-speed complex environment. For this purpose, the present application comprises: according to a static obstacle point cloud and parking space vector data of an area under test, acquiring two-dimensional point cloud data of said area; according to the two-dimensional point cloud data and a historical driving trajectory of said area, acquiring a modal fusion bird's eye view of said area; and, according to the modal fusion bird's eye view, acquiring an intersection detection result of said area. Because multiple modal data, such as historical driving trajectories, static obstacle point clouds, and parking spaces, is fused into the modal fusion bird's eye view, the obtained intersection detection result has higher accuracy. In low-speed complex environments, the present application can effectively resist trajectory noises such as environmental perception and obstacle avoidance and U-turns, achieving a better detection effect.

Description

Intersection detection method, readable storage medium and intelligent device This application claims priority to Chinese patent application CN 202311817527.9, filed on December 27, 2023, entitled “Intersection Detection Method, Readable Storage Medium and Intelligent Device”. The entire contents of the above Chinese patent application are incorporated into this application by reference. Technical Field

[0001] The present application relates to the field of autonomous driving technology, and specifically provides a road intersection detection method, a readable storage medium, and an intelligent device. Background Art

[0002] In complex, low-speed scenarios, such as underground parking lots and residential areas, identifying intersections is crucial for generating road network topology data. Traditional urban road intersection recognition methods rely heavily on vehicle trajectories. However, low-speed scenarios are often more complex, with dense and diverse intersections. Furthermore, vehicle trajectories, often due to obstacles and U-turns, make it difficult to identify a standard intersection pattern.

[0003] Accordingly, a new intersection detection solution is needed in this field to solve the above problems. Summary of the Invention

[0004] In order to overcome the above-mentioned defects, the present application is proposed to provide a solution or at least partially solve the problem of how to effectively detect intersections in low-speed complex scenarios.

[0005] In a first aspect, the present application provides a method for detecting a road intersection, the method comprising:

[0006] Acquire two-dimensional point cloud data of the area to be detected based on the static obstacle point cloud and the vector data of the parking space in the area to be detected;

[0007] Acquire a modality fusion bird's-eye view of the area to be detected based on the two-dimensional point cloud data and the historical driving trajectory of the area to be detected;

[0008] Obtain a road intersection detection result of the area to be detected based on the modal fusion bird's-eye view.

[0009] In one technical solution of the above-mentioned intersection detection method, obtaining the intersection detection result of the area to be detected based on the modal fusion bird's-eye view includes:

[0010] Performing sliding window detection based on a sliding window according to the modal fusion bird's-eye view to obtain a sliding window detection result;

[0011] The intersection detection result is obtained according to the sliding window detection result.

[0012] In one technical solution of the above-mentioned intersection detection method, the sliding window includes a first sliding window and a second sliding window, the first sliding window and the second sliding window are both centered on the same trajectory point in the historical driving trajectory, and the window size of the first sliding window is larger than the window size of the second sliding window; the sliding window detection result includes the first sliding window detection result and the second sliding window detection result;

[0013] The step of performing sliding window detection based on a sliding window according to the modal fusion bird's-eye view to obtain a sliding window detection result includes:

[0014] Performing the first sliding window detection according to the modal fusion bird's-eye view to obtain the first sliding window detection result;

[0015] The second sliding window detection is performed according to the first sliding window detection result to obtain multiple second sliding window detection results.

[0016] In one technical solution of the above-mentioned intersection detection method, obtaining the intersection detection result according to the sliding window detection result includes:

[0017] Obtaining a first sliding window feature according to the first sliding window detection result;

[0018] Obtaining a second sliding window feature according to the second sliding window detection result;

[0019] The intersection detection result is obtained according to the first sliding window feature and the second sliding window feature.

[0020] In one technical solution of the above-mentioned intersection detection method, obtaining the second sliding window feature according to the second sliding window detection result includes:

[0021] resizing the second sliding window detection result to obtain an image of the same size as the first sliding window detection result;

[0022] The second sliding window feature is obtained according to the obtained image.

[0023] In one technical solution of the above-mentioned intersection detection method, obtaining the intersection detection result according to the first sliding window feature and the second sliding window feature includes:

[0024] Fusing the first sliding window feature and the second sliding window feature to obtain a fused feature;

[0025] Obtaining a classification category corresponding to the second sliding window detection result according to the fusion feature;

[0026] According to the classification category, the intersection detection result is obtained.

[0027] In one technical solution of the above-mentioned intersection detection method, obtaining two-dimensional point cloud data of the area to be detected based on the static obstacle point cloud and the vector data of the parking space in the area to be detected includes:

[0028] Obtaining a two-dimensional point cloud of an obstacle based on the static obstacle point cloud;

[0029] Obtaining a two-dimensional point cloud of the parking space according to the parking space vector data;

[0030] The two-dimensional point cloud data of the area to be detected is obtained according to the two-dimensional point cloud of the obstacle and the two-dimensional point cloud of the parking space.

[0031] In one technical solution of the above-mentioned intersection detection method, obtaining a two-dimensional obstacle point cloud based on the static obstacle point cloud includes:

[0032] Performing ground removal processing on the static obstacle point cloud to obtain a ground-free static obstacle point cloud;

[0033] The point cloud of the static obstacle-free ground object is superimposed in the height direction to obtain the obstacle two-dimensional point cloud.

[0034] In one technical solution of the above-mentioned intersection detection method, obtaining a two-dimensional point cloud of a parking space based on the parking space vector data includes:

[0035] Acquiring two-dimensional parking space vector data according to the parking space vector data;

[0036] For the polygon corresponding to the two-dimensional parking space vector data, each edge of the polygon is sampled according to a preset sampling interval to obtain the two-dimensional point cloud of the parking space.

[0037] In one technical solution of the above-mentioned intersection detection method, obtaining a modal fusion bird's-eye view of the area to be detected based on the two-dimensional point cloud data and the historical driving trajectory of the area to be detected includes:

[0038] According to the historical driving trajectory, a trajectory polyline corresponding to the historical driving trajectory is drawn on the two-dimensional point cloud data to obtain the modal fusion bird's-eye view.

[0039] In one technical solution of the above-mentioned intersection detection method, obtaining a modal fusion bird's-eye view of the area to be detected based on the two-dimensional point cloud data and the historical driving trajectory of the area to be detected includes:

[0040] When the area to be detected is multi-layered, the modal fusion bird's-eye view of each layer is obtained based on the two-dimensional point cloud data and the historical driving trajectory of each layer.

[0041] In one technical solution of the above-mentioned intersection detection method, obtaining the intersection detection result of the area to be detected based on the modal fusion bird's-eye view includes:

[0042] Applying a preset trained deep learning model, the intersection detection result is obtained according to the modal fusion bird's-eye view.

[0043] In one technical solution of the above-mentioned intersection detection method, the method further includes training the deep learning model according to the following steps:

[0044] Label the bird's-eye view image used for training to obtain a labeled dataset;

[0045] The deep learning model is trained according to the labeled data set to obtain a trained deep learning model.

[0046] In one technical solution of the above-mentioned intersection detection method, labeling the bird's-eye view image used for training to obtain the labeled data set includes:

[0047] Marking the polygonal true value corresponding to the intersection on the bird's-eye view map used for training;

[0048] Randomly generating sliding windows of different sizes on the bird's-eye view image for training;

[0049] For each sliding window, calculating the intersection-over-union ratio between the sliding window and the true value of the polygon;

[0050] When the intersection-over-union ratio is greater than a preset threshold, marking the sliding window and other sliding windows that are concentric with the sliding window and larger in size than the sliding window as positive samples;

[0051] When the intersection-over-union ratio is less than or equal to the preset threshold, marking the sliding window as a negative sample;

[0052] The labeled data set is obtained according to the positive samples and the negative samples.

[0053] In a second aspect, a computer-readable storage medium is provided, which stores a plurality of program codes, wherein the program codes are suitable for being loaded and run by a processor to execute the intersection detection method described in any one of the technical solutions of the above-mentioned intersection detection method.

[0054] In a third aspect, a smart device includes:

[0055] at least one processor;

[0056] and, a memory communicatively coupled to the at least one processor;

[0057] Wherein, a computer program is stored in the memory, and when the computer program is executed by the at least one processor, the intersection detection method described in any one of the technical solutions of the above-mentioned intersection detection method is implemented.

[0058] The above one or more technical solutions of this application have at least one or more of the following beneficial effects:

[0059] In the technical solution for implementing the present application, the present application obtains two-dimensional point cloud data of the area to be detected based on the static obstacle point cloud and the vector data of the parking space in the area to be detected, obtains a modal fusion bird's-eye view of the area to be detected based on the two-dimensional point cloud data and the historical driving trajectory of the area to be detected, and obtains the intersection detection result of the area to be detected based on the modal fusion bird's-eye view. Through the above configuration, the modal fusion bird's-eye view in the present application integrates multiple modal data such as historical driving trajectory, static obstacle point cloud and parking space, so the intersection detection result obtained based on the modal fusion bird's-eye view has higher accuracy. In low-speed complex scenarios (such as underground parking lots, residential areas, etc.), it can effectively resist environmental perception and trajectory noise such as obstacle avoidance and U-turns, and achieve better detection effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The disclosure of this application will become more easily understood with reference to the accompanying drawings. Those skilled in the art will readily appreciate that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this application. Among them:

[0061] FIG1 is a schematic flow chart of the main steps of a method for detecting an intersection according to an embodiment of the present application;

[0062] FIG2 is a flow chart showing the main steps of a method for detecting an intersection according to an embodiment of the present application;

[0063] FIG3 is a schematic diagram of a bird's-eye view of modality fusion according to an embodiment of the present application;

[0064] FIG4 is a schematic diagram of the main network architecture of a model for obtaining intersection detection results by applying a deep learning model according to an implementation of an embodiment of the present application;

[0065] FIG5 is a schematic diagram of a connection relationship between a memory and a processor of a smart device according to an embodiment of the present application. DETAILED DESCRIPTION

[0066] Some embodiments of the present application are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and are not intended to limit the scope of protection of the present application.

[0067] In the description of this application, "module" and "processor" may include hardware, software, or a combination of both. A module may include hardware circuitry, various suitable sensors, communication ports, and memory. It may also include software components, such as program code, or a combination of software and hardware. A processor may be a central processing unit, a microprocessor, an image processor, a digital signal processor, or any other suitable processor. A processor has data and / or signal processing capabilities. A processor may be implemented in software, hardware, or a combination of both. Non-transitory computer-readable storage media include any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" refers to all possible combinations of A and B, such as only A, only B, or both A and B. The terms "at least one of A or B" or "at least one of A and B" have similar meanings to "A and / or B" and may include only A, only B, or both A and B. The singular forms "one" and "the" may also include the plural forms.

[0068] Definition of terms:

[0069] An automated driving system (ADS) is a system that continuously performs all dynamic driving tasks (DDT) within its operational domain design (ODD). Specifically, the system is only allowed to fully assume the task of autonomous vehicle control under specified appropriate driving scenarios. When the vehicle meets the ODD conditions, the system is activated, replacing the human driver as the vehicle's primary driver. The DDT refers to the continuous lateral (left and right steering) and longitudinal motion control (acceleration, deceleration, and constant speed) of the vehicle, as well as the detection and response to objects and events in the vehicle's driving environment. The ODD refers to the conditions under which the automated driving system can operate safely. These conditions can include geographic location, road type, speed range, weather, time of day, and national and local traffic laws and regulations.

[0070] Referring to FIG1 , FIG1 is a flow chart showing the main steps of a road intersection detection method according to an embodiment of the present application. As shown in FIG1 , the road intersection detection method in the embodiment of the present application mainly includes the following steps S101 to S103 .

[0071] Step S101: Acquire two-dimensional point cloud data of the area to be detected based on the static obstacle point cloud and the vector data of the parking space in the area to be detected.

[0072] In this embodiment, two-dimensional point cloud data of the area to be inspected can be generated by combining the static obstacle point cloud and parking space vector data. The static obstacle point cloud can include point cloud data of walls, fences, etc. The parking space vector data refers to parking space data generated based on the calculated composite image.

[0073] In one embodiment, a static obstacle point cloud may be obtained through an occupancy layer of a grid map.

[0074] In one embodiment, for multi-layer scenes, such as multi-layer parking lots, two-dimensional point cloud data of each layer can be obtained based on the static obstacle point cloud and parking space vector data of each layer.

[0075] Step S102: obtaining a modal fusion bird's-eye view of the area to be detected based on the two-dimensional point cloud data and the historical driving trajectory of the area to be detected.

[0076] In this embodiment, the two-dimensional point cloud data and the historical driving trajectory of the area to be detected can be fused to obtain a modal fusion bird's-eye view of the area to be detected.

[0077] In one embodiment, the two-dimensional point cloud data and the historical driving trajectory can be mapped onto the same image at a certain scale to obtain a modal fusion bird's-eye view.

[0078] In one embodiment, when the area to be inspected is multi-layered, a modal fusion bird's-eye view of each layer is obtained based on the two-dimensional point cloud data and historical driving trajectory of each layer.

[0079] Step S103: Obtaining the intersection detection result of the area to be detected based on the modal fusion bird's-eye view.

[0080] In this embodiment, the intersection detection of the area to be detected can be performed based on the modal fusion bird's-eye view, so as to obtain the intersection detection result.

[0081] In one embodiment, the modal fusion bird's-eye view image can be input into a deep learning-based neural network model for intersection detection to obtain intersection detection results.

[0082] Based on the above steps S101 to S103, the embodiment of the present application obtains two-dimensional point cloud data of the area to be detected based on the static obstacle point cloud and the vector data of the parking space in the area to be detected, obtains a modal fusion bird's-eye view of the area to be detected based on the two-dimensional point cloud data and the historical driving trajectory of the area to be detected, and obtains the intersection detection result of the area to be detected based on the modal fusion bird's-eye view. Through the above configuration, the modal fusion bird's-eye view in the embodiment of the present application integrates multiple modal data such as historical driving trajectory, static obstacle point cloud and parking space, so the intersection detection result obtained based on the modal fusion bird's-eye view has higher accuracy. In low-speed complex scenes (such as underground parking lots, residential areas, etc.), it can effectively resist environmental perception and trajectory noise such as obstacle avoidance and U-turns, and achieve better detection effect.

[0083] Step S101, step S102 and step S103 are further described below.

[0084] In one implementation of the embodiment of the present application, step S101 may further include the following steps S1011 to S1013:

[0085] Step S1011: Obtain a two-dimensional point cloud of the obstacle based on the static obstacle point cloud.

[0086] In this embodiment, step S1011 may further include the following steps S10111 and S10112:

[0087] Step S10111: performing ground removal processing on the static obstacle point cloud to obtain a ground-free static obstacle point cloud.

[0088] Step S10112: Superimpose the point cloud of the static obstacle-free ground object in the height direction to obtain a two-dimensional point cloud of the obstacle.

[0089] In this embodiment, the static obstacle point cloud is expressed in a local three-dimensional coordinate system. Therefore, the static obstacle point cloud can be first processed to remove the ground, and then the ground-free static obstacle point cloud can be superimposed in the height direction to obtain the obstacle two-dimensional point cloud.

[0090] Step S1012: Obtain a two-dimensional point cloud of the parking space based on the parking space vector data.

[0091] In this embodiment, step S1012 may further include the following steps S10121 and S10122:

[0092] Step S10121: Obtain two-dimensional parking space vector data based on the parking space vector data.

[0093] Step S10122: For the polygon corresponding to the two-dimensional parking space vector data, each edge of the polygon is sampled according to a preset sampling interval to obtain a two-dimensional point cloud of the parking space.

[0094] In this embodiment, the parking space vector data is similarly expressed in a local 3D coordinate system. Since the parking space vector data is a closed polygon, the z value can be ignored, resulting in 2D parking space vector data. Sampling is performed along each edge of the polygon corresponding to the 2D parking space vector data at a preset sampling interval to obtain a set of 2D points, which serve as the 2D parking space point cloud. Those skilled in the art can adjust the sampling interval based on actual application needs.

[0095] Step S1013: Acquire two-dimensional point cloud data of the area to be detected based on the two-dimensional point cloud of the obstacle and the two-dimensional point cloud of the parking space.

[0096] In this embodiment, the obstacle two-dimensional point cloud and the parking space two-dimensional point cloud may be used as the two-dimensional point cloud data of the area to be detected.

[0097] In one implementation of the embodiment of the present application, step S102 may be further configured as follows:

[0098] According to the historical driving trajectory, the trajectory polyline corresponding to the historical driving trajectory is drawn on the two-dimensional point cloud data to obtain a modal fusion bird's-eye view.

[0099] In one embodiment, for ease of identification, the two-dimensional point cloud data and trajectory polylines in the modality fusion bird's-eye view may be marked with different colors.

[0100] In this embodiment, please refer to Figure 3, which is a schematic diagram of a modal fusion bird's-eye view according to an embodiment of the present application. As shown in Figure 3, the two-dimensional point cloud data and the trajectory polyline of the historical driving trajectory can be fused into a single image for expression, namely, a modal fusion bird's-eye view. In order to effectively distinguish multimodal data, the two-dimensional point cloud data and the trajectory polyline can be marked with different colors (the colors are not shown in Figure 3), for example, the two-dimensional point cloud data can be expressed in black and the trajectory polyline can be expressed in red.

[0101] In one implementation of the embodiment of the present application, step S103 may further include the following steps S1031 and S1032:

[0102] Step S1031: performing sliding window detection based on a sliding window according to the modal fusion bird's-eye view to obtain a sliding window detection result.

[0103] In this embodiment, the sliding window includes a first sliding window and a second sliding window. The first sliding window and the second sliding window are both centered on the same track point in the historical driving track. The window size of the first sliding window is larger than the window size of the second sliding window. The sliding window detection results include the first sliding window detection results and the second sliding window detection results. Step S1031 can further include the following steps S10311 and S10312:

[0104] Step S10311: Perform a first sliding window detection based on the modal fusion bird's-eye view to obtain a first sliding window detection result.

[0105] In this embodiment, a first sliding window with a larger window size may be used to perform sliding window detection on the modality fusion bird's-eye view image to obtain a first sliding window detection result.

[0106] In one implementation, the window size of the first sliding window may be the number of pixels corresponding to 30m.

[0107] Step S10312: Perform a second sliding window detection based on the first sliding window detection result to obtain multiple second sliding window detection results.

[0108] In this embodiment, considering that the sizes of intersections in low-speed complex scenarios vary, sliding window detection can be performed based on the first sliding window detection result using a second sliding window with a smaller window size, thereby obtaining a second sliding window detection result.

[0109] In one embodiment, the window size of the second sliding window may be multiple, such as the number of pixels corresponding to 5m, 7m, 10m, 15m, and 20m.

[0110] Step S1032: Obtain intersection detection results based on the sliding window detection results.

[0111] In this embodiment, step S1032 may further include the following steps S10321 to S10323:

[0112] Step S10321: Obtain a first sliding window feature according to the first sliding window detection result.

[0113] In this embodiment, feature extraction may be performed on each image corresponding to the first sliding window detection result to obtain first sliding window features.

[0114] Step S10322: Obtain a second sliding window feature according to the second sliding window detection result.

[0115] In this embodiment, step S10322 may further include the following steps S103221 and S103222:

[0116] Step S103221: resize the second sliding window detection result to obtain an image with the same size as the first sliding window detection result.

[0117] Step S103222: Obtain a second sliding window feature based on the acquired image.

[0118] In this embodiment, the second sliding window detection result may be resized to an image having the same size as the first sliding window detection result, and feature extraction may be performed based on the resized image to obtain the second sliding window feature.

[0119] Step S10323: Obtain intersection detection results based on the first sliding window features and the second sliding window features.

[0120] In this embodiment, step S10323 may further include the following steps S103231 to S103233:

[0121] Step S103231: Fusing the first sliding window feature and the second sliding window feature to obtain a fused feature.

[0122] Step S103232: Obtain the classification category corresponding to the second sliding window detection result based on the fusion feature.

[0123] Step S103233: Obtain intersection detection results based on the classification category.

[0124] In this embodiment, the first sliding window features and the second sliding window features can be fused. The resulting fused features are then used to identify the classification category corresponding to the second sliding window detection result, thereby obtaining an intersection detection result. Specifically, if the classification category is an intersection, the second sliding window detection result is used as the intersection detection result, or the first sliding window detection result is used as the intersection detection result. If the classification category is not an intersection, the second sliding window detection result is ignored. Using sliding windows of varying sizes for intersection detection can accommodate the varying sizes of intersections in complex low-speed scenarios while also effectively preserving intersection contextual information.

[0125] In one embodiment, please refer to Figure 4, which is a schematic diagram of the main network architecture of the model for applying a deep learning model to obtain intersection detection results according to an embodiment of the present application. As shown in Figure 4, the trained deep learning model can be applied to implement step S103, that is, the process of obtaining intersection recognition results based on the modal fusion bird's-eye view. Among them, the large sliding window (first sliding window) is 30×30, and the large sliding window obtains the first sliding window detection result and performs feature extraction through backbone (backbone network) 1 to obtain the first sliding window feature; the small sliding window (second sliding window) performs sliding window detection on the basis of the first sliding window detection result to obtain the second sliding window detection result, and the second sliding window detection result is subjected to feature extraction through backbone2 to obtain the second sliding window feature. The first sliding window feature and the second sliding window feature are feature fused, and finally the classification result of the small sliding window is output, thereby obtaining the intersection detection result.

[0126] In one embodiment, the bird's-eye view image used for training can be annotated to construct an annotated dataset, and the annotated dataset can be used to train the deep learning model to obtain a trained deep learning model.

[0127] In one embodiment, the labeled dataset may be obtained according to the following steps S201 to S206:

[0128] Step S201: Mark the polygonal true values ​​corresponding to the intersection on the bird's-eye view map used for training.

[0129] Step S202: Randomly generate sliding windows of different sizes on the bird's-eye view image used for training.

[0130] Step S203: For each sliding window, calculate the intersection-over-union ratio between the sliding window and the polygon true value.

[0131] Step S204: When the intersection-over-union ratio is greater than a preset threshold, the sliding window and other sliding windows that are concentric with the sliding window and larger in size than the sliding window are marked as positive samples.

[0132] Step S205: When the intersection-over-union ratio is less than or equal to a preset threshold, the sliding window is marked as a negative sample.

[0133] Step S206: Obtain a labeled data set based on the positive samples and negative samples.

[0134] In this embodiment, the training bird's-eye view image can be annotated with ground truth intersections, using polygons as polygon ground truths. Randomly generate sliding windows of varying sizes on the training bird's-eye view image. If the intersection-over-union (IOU) between the sliding window and the ground truth polygon is greater than a preset threshold, the sliding window and any other concentric sliding windows larger than the sliding window are considered positive samples and classified as "at the intersection." Conversely, if the IOU is less than or equal to the preset threshold, the sliding window is labeled as a negative sample and classified as "not at the intersection."

[0135] In one embodiment, please refer to Figure 2, which is a flow chart of the main steps of an intersection detection method according to an embodiment of the present application. As shown in Figure 2, in this embodiment, the input data of the intersection detection method is historical driving trajectories, static obstacle point clouds, and parking space vector data. Two-dimensional point cloud data is generated based on the static obstacle point cloud and parking space vector data. A modal fusion bird's-eye view is generated based on the historical driving trajectories and the two-dimensional point cloud data. A deep learning-based sliding window detection is performed based on the modal fusion bird's-eye view to obtain the intersection detection result.

[0136] It should be pointed out that although the various steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effect of the present application, different steps do not have to be performed in such an order. They can be performed simultaneously (in parallel) or in other orders. These changes are within the scope of protection of the present application.

[0137] It will be understood by those skilled in the art that all or part of the processes in the method for implementing the above embodiment of the present application can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of each of the above method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file or some intermediate form. The computer-readable storage medium can include: any entity or device, medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium that can carry the computer program code. It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media do not include electric carrier signals and telecommunication signals.

[0138] Furthermore, the present application also provides a computer-readable storage medium. In a computer-readable storage medium embodiment according to the present application, the computer-readable storage medium can be configured to store a program for executing the intersection detection method of the above-mentioned method embodiment, and the program can be loaded and run by the processor to implement the above-mentioned intersection detection method. For ease of explanation, only the parts related to the embodiment of the present application are shown. For specific technical details not disclosed, please refer to the method part of the embodiment of the present application. The computer-readable storage medium can be a storage device formed by various electronic devices. Optionally, the computer-readable storage medium in the embodiment of the present application is a non-transitory computer-readable storage medium.

[0139] Furthermore, the present application also provides an intelligent device, which may include at least one processor; and a memory communicatively connected to the at least one processor; wherein a computer program is stored in the memory, and when the computer program is executed by the at least one processor, the intersection detection method described in any of the above embodiments is implemented. The intelligent device of the present application may include driving devices, smart cars, robots and other devices. Referring to Figure 5, Figure 5 is a schematic diagram of the connection relationship between the memory and the processor of the intelligent device according to an embodiment of the present application. As shown in Figure 5, the memory and the processor of the intelligent device are communicatively connected via a bus.

[0140] In some embodiments of the present application, the smart device further includes at least one sensor for sensing information. The sensor is communicatively coupled to any of the types of processors described herein. Optionally, the smart device further includes an autonomous driving system for guiding the smart device to drive autonomously or with assistance. The processor communicates with the sensor and / or autonomous driving system to perform the method described in any of the above embodiments.

[0141] Furthermore, it should be understood that since the configuration of each module is merely for the purpose of illustrating the functional units of the apparatus of the present application, the physical devices corresponding to these modules may be the processor itself, or a portion of the software in the processor, a portion of the hardware, or a combination of software and hardware. Therefore, the number of modules in the figure is merely illustrative.

[0142] Those skilled in the art will appreciate that the various modules in the device can be adaptively split or merged. Such splitting or merging of specific modules will not cause the technical solution to deviate from the principles of this application. Therefore, the technical solutions after splitting or merging will fall within the scope of protection of this application.

[0143] The relevant user personal information that may be involved in the various embodiments of this application is strictly in accordance with the requirements of laws and regulations, following the principles of legality, legitimacy and necessity, and based on the reasonable purposes of business scenarios, to process the personal information that users actively provide during the use of products / services or generated due to the use of products / services, as well as the personal information obtained with the user's authorization.

[0144] The user personal information processed in this application will vary depending on the specific product / service scenario and will be based on the specific scenario in which the user uses the product / service. This may involve the user's account information, device information, driving information, vehicle information, or other related information. The applicant will treat the user's personal information and its processing with a high degree of diligence.

[0145] This application attaches great importance to the security of user personal information and has taken reasonable and feasible security protection measures that comply with industry standards to protect user information and prevent personal information from being accessed, disclosed, used, modified, damaged or lost without authorization.

[0146] Thus far, the technical solutions of the present application have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is readily understood by those skilled in the art that the scope of protection of the present application is obviously not limited to these specific embodiments. Without departing from the principles of the present application, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present application.

Claims

1. A road intersection detection method, characterized in that, The method includes: Obtaining two-dimensional point cloud data of the area to be detected according to the static obstacle point cloud of the area to be detected and the vector data of the parking spaces; Obtaining a modal fusion bird's-eye view of the area to be detected according to the two-dimensional point cloud data and the historical driving trajectory of the area to be detected; Obtaining an intersection detection result of the area to be detected according to the modal fusion bird's-eye view.

2. The intersection detection method according to claim 1, wherein The obtaining the intersection detection result of the area to be detected according to the modal fusion bird's-eye view includes: Performing a sliding window detection based on the sliding window on the modal fusion bird's-eye view to obtain a sliding window detection result; Obtaining the intersection detection result according to the sliding window detection result.

3. The intersection detection method according to claim 2, wherein The sliding window includes a first sliding window and a second sliding window. Both the first sliding window and the second sliding window are centered on the same trajectory point in the historical driving trajectory. The window size of the first sliding window is larger than the window size of the second sliding window; the sliding window detection result includes a first sliding window detection result and a second sliding window detection result; The performing a sliding window detection based on the sliding window on the modal fusion bird's-eye view to obtain a sliding window detection result includes: Performing the first sliding window detection on the modal fusion bird's-eye view to obtain the first sliding window detection result; Performing the second sliding window detection according to the first sliding window detection result to obtain a plurality of second sliding window detection results.

4. The intersection detection method according to claim 3, wherein The obtaining the intersection detection result according to the sliding window detection result includes: Obtaining a first sliding window feature according to the first sliding window detection result; Obtaining a second sliding window feature according to the second sliding window detection result; Obtaining the intersection detection result according to the first sliding window feature and the second sliding window feature.

5. The intersection detection method according to claim 4, wherein The obtaining the second sliding window feature according to the second sliding window detection result includes: Adjusting the size of the second sliding window detection result to obtain an image with the same size as the first sliding window detection result; Obtaining the second sliding window feature according to the obtained image.

6. The intersection detection method according to claim 4, wherein The obtaining the intersection detection result according to the first sliding window feature and the second sliding window feature includes: Performing feature fusion on the first sliding window feature and the second sliding window feature to obtain a fusion feature; Obtaining a classification category corresponding to the second sliding window detection result according to the fusion feature; Obtaining the intersection detection result according to the classification category.

7. The intersection detection method according to claim 1, wherein The obtaining two-dimensional point cloud data of the area to be detected according to the static obstacle point cloud of the area to be detected and the vector data of the parking spaces includes: Obtaining an obstacle two-dimensional point cloud according to the static obstacle point cloud; Obtain the two-dimensional point cloud of the parking space according to the vector data of the parking space; Obtain the two-dimensional point cloud data of the area to be detected according to the two-dimensional point cloud of the obstacle and the two-dimensional point cloud of the parking space.

8. The intersection detection method according to claim 7, wherein The obtaining the two-dimensional point cloud of the obstacle according to the static obstacle point cloud includes: Perform ground removal processing on the static obstacle point cloud to obtain a static obstacle point cloud without ground; Superimpose the static obstacle point cloud without ground in the height direction to obtain the two-dimensional point cloud of the obstacle.

9. The intersection detection method according to claim 7, wherein The obtaining the two-dimensional point cloud of the parking space according to the vector data of the parking space includes: Obtain two-dimensional parking space vector data according to the vector data of the parking space; For the polygon corresponding to the two-dimensional parking space vector data, collect each side of the polygon at a preset sampling interval to obtain the two-dimensional point cloud of the parking space.

10. The intersection detection method according to claim 1, wherein The obtaining the modality fusion bird's-eye view of the area to be detected according to the two-dimensional point cloud data and the historical driving trajectory of the area to be detected includes: Draw a trajectory polyline corresponding to the historical driving trajectory in the two-dimensional point cloud data according to the historical driving trajectory to obtain the modality fusion bird's-eye view.

11. The intersection detection method according to claim 1, wherein The obtaining the modality fusion bird's-eye view of the area to be detected according to the two-dimensional point cloud data and the historical driving trajectory of the area to be detected includes: When the area to be detected is multi-layered, obtain the modality fusion bird's-eye view of each layer according to the two-dimensional point cloud data and the historical driving trajectory of each layer.

12. The intersection detection method according to any one of claims 1 to 11, wherein The obtaining the intersection detection result of the area to be detected according to the modality fusion bird's-eye view includes: Apply a preset trained deep learning model to obtain the intersection detection result according to the modality fusion bird's-eye view.

13. The intersection detection method according to claim 12, wherein The method further includes training the deep learning model according to the following steps: Annotate the bird's-eye view for training to obtain an annotation data set; Train the deep learning model according to the annotation data set to obtain a trained deep learning model.

14. The intersection detection method according to claim 13, wherein The annotating the bird's-eye view for training to obtain an annotation data set includes: Annotate the true polygon corresponding to the intersection on the bird's-eye view for training; Randomly generate sliding windows of different sizes on the bird's-eye view for training; For each sliding window, calculate the intersection over union between the sliding window and the true polygon; When the intersection over union is greater than a preset threshold value, label the sliding window and other sliding windows that are concentric with the sliding window and have a size larger than the sliding window as positive samples; When the intersection over union is less than or equal to the preset threshold, label the sliding window as a negative sample; Obtain the labeled data set according to the positive samples and the negative samples.

15. A computer-readable storage medium storing multiple program codes, characterized in that, The program code is adapted to be loaded and run by a processor to execute the intersection detection method according to any one of claims 1 to 14.

16. An intelligent device, characterized in that, Comprising: At least one processor; And a memory communicatively connected to the at least one processor; Wherein, a computer program is stored in the memory, and when the computer program is executed by the at least one processor, the intersection detection method according to any one of claims 1 to 14 is implemented.

Images