Parking space identification method, system, device, medium and program product

Abstract

The present application provides a parking space recognition method, which uses ultrasound to detect parking spaces, obtains candidate parking spaces, and determines whether the candidate parking space is a real parking space based on obstacle information learned from multi-way surround view images. If it is a real parking space, the location information of the real parking space is output. This method can prevent parking space recognition from being affected by the environment, accuracy, and obstacles on parking space recognition, obtain the location information of the real parking space, provide accurate and stable input information for automatic parking, assist in the completion of the automatic parking function, enhance the user experience, and improve the success rate of automatic parking in parking spaces.

Description

Technical Field

[0001] The present application relates to the field of automotive technology, and in particular to a method, system, device, computer-readable storage medium, and computer program product for parking space recognition. Background Art

[0002] Autonomous driving technology uses onboard sensor systems to perceive the road environment, automatically planning routes and controlling vehicle movement. In recent years, the research and development of automated parking features within autonomous driving technology has been a hot topic in academia and industry. Accurately identifying parking spaces during automated parking is a primary challenge facing researchers and engineers.

[0003] Typically, vehicles use ultrasonic radar to detect obstacles near parking spaces. The ultrasonic radar's return echo determines whether there are obstacles around the parking space and whether it can be used as a parking space. However, the obstacle information returned by ultrasonic waves is relatively simple. It can only determine the presence of an obstacle, but cannot determine its specific information, type, or category. In other words, it cannot determine whether a gap is a legal or illegal parking space. For example, a gap can be a parking space between two cars or between trees on either side of the road.

[0004] Therefore, the industry is in urgent need of an accurate parking space identification method. Summary of the Invention

[0005] This application provides a parking space recognition method that can accurately recognize parking spaces using ultrasonic waves and multi-way surround view images, thereby improving parking space recognition accuracy.

[0006] In a first aspect, the present application provides a parking space identification method, the method comprising:

[0007] Use ultrasonic waves to detect parking spaces and obtain candidate parking spaces within the area. Candidate parking spaces are represented by parking reflection points.

[0008] Determine whether a candidate parking space is a real parking space based on the obstacle information in the area learned from the multi-way surround view images;

[0009] When the candidate parking space is a real parking space, the location information of the real parking space is output.

[0010] In some possible implementations, determining whether a candidate parking space is a real parking space based on obstacle information within the area learned from the multi-way surround view images includes:

[0011] When the obstacle in the area learned from the multi-way surround view images is a stationary vehicle, the candidate parking space is determined to be a real parking space.

[0012] In some possible implementations, the method further includes:

[0013] Based on the multi-way surround view images, deep learning is used to determine that obstacles in the area are vehicles;

[0014] According to the vehicle's motion parameters, the vehicle's position sequence is obtained by performing dead reckoning through the vehicle's motion model;

[0015] According to the position sequence of the vehicle, it is determined whether the vehicle is stationary.

[0016] In some possible implementations, determining that an obstacle in an area is a vehicle through deep learning based on multiple surround view images includes:

[0017] Determine the specific location of vehicles in the area through deep learning based on multi-way surround view images;

[0018] When the specific position of the vehicle is consistent with the reflection point of the parking space, it is determined that the obstacle in the area is a vehicle.

[0019] In some possible implementations, the multi-way surround view image includes multiple frames of multi-way surround view images acquired at different locations.

[0020] In some possible implementations, the multi-path surround view images are obtained by a vehicle-mounted surround view camera.

[0021] In a second aspect, the present application provides a parking space recognition device, which includes:

[0022] The candidate parking space determination module is used to detect parking spaces using ultrasound to obtain candidate parking spaces within the area. The candidate parking spaces are represented by parking space reflection points.

[0023] The real parking space identification module is used to determine whether the candidate parking space is a real parking space based on the obstacle information in the area learned from the multi-way surround view images;

[0024] The parking space information output module is used to output the location information of the real parking space when the candidate parking space is a real parking space.

[0025] In some possible implementations, the real parking space identification module is specifically used to:

[0026] When the obstacle in the area learned from the multi-way surround view images is a stationary vehicle, the candidate parking space is determined to be a real parking space.

[0027] In some possible implementations, the real parking space identification module is further configured to:

[0028] Based on the multi-way surround view images, deep learning is used to determine that obstacles in the area are vehicles;

[0029] According to the vehicle's motion parameters, the vehicle's position sequence is obtained by performing dead reckoning through the vehicle's motion model;

[0030] According to the position sequence of the vehicle, it is determined whether the vehicle is stationary.

[0031] In some possible implementations, the real parking space identification module is specifically used to:

[0032] When the obstacle in the area learned from the multi-way surround view image is a stationary vehicle, determine the specific position of the vehicle;

[0033] When the specific position of the vehicle is consistent with the parking space reflection point, the candidate parking space is determined to be the real parking space.

[0034] In some possible implementations, the multi-way surround view image includes multiple frames of multi-way surround view images acquired at different locations.

[0035] In some possible implementations, the multi-path surround view images are obtained by a vehicle-mounted surround view camera.

[0036] In a third aspect, the present application provides a device comprising a processor and a memory. The processor and the memory communicate with each other. The processor is configured to execute instructions stored in the memory to cause the device to perform the parking space recognition method according to the first aspect or any implementation of the first aspect.

[0037] In a fourth aspect, the present application provides a computer-readable storage medium, in which instructions are stored, and the instructions instruct a device to execute the parking space recognition method described in the first aspect or any implementation of the first aspect.

[0038] In a fifth aspect, the present application provides a computer program product comprising instructions, which, when executed on a device, enables the device to execute the parking space identification method described in the first aspect or any one of the implementations of the first aspect.

[0039] Based on the implementation methods provided in the above aspects, this application can also be further combined to provide more implementation methods.

[0040] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0041] The present application provides a parking space recognition method that utilizes ultrasound to detect parking spaces, obtains candidate parking spaces, and then, based on obstacle information learned from multi-way surround view images, determines whether the candidate parking space is a real parking space. If so, the location information of the real parking space is output. This method prevents parking space recognition from being affected by environmental factors, accuracy, and obstacles, obtains the location information of the real parking space, and provides accurate and stable input information for automatic parking. This assists in the completion of the automatic parking function, enhances the user experience, and improves the success rate of automatic parking. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical methods of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0043] Figure 1 A flowchart of a parking space identification method provided in an embodiment of the present application;

[0044] Figure 2 A flowchart of a parking space identification method provided in an embodiment of the present application;

[0045] Figure 3 A schematic diagram of the architecture of a parking space recognition device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The following will describe the solutions in the embodiments provided in this application in conjunction with the drawings in this application.

[0047] The terms "first" and "second" in the embodiments of this application are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Therefore, features specified as "first" or "second" may explicitly or implicitly include one or more of the features.

[0048] First, some technical terms involved in the embodiments of this application are introduced.

[0049] Ultrasonic waves are sound waves with frequencies above 20,000 Hz, characterized by good directionality, adaptability, and strong penetration. Ultrasonic radar uses ultrasonic waves to transmit and measure distance by the time difference between reflection and reception of the echo. It is often used in self-driving cars to detect surrounding obstacles.

[0050] Typically, ultrasonic radar detects parking spaces by detecting obstacles within a specific area. Ideally, ultrasonic radar uses the ultrasonic signals it receives to identify empty spaces between two vehicles and determine whether they are available. However, the information returned by ultrasonic radar is often limited and cannot determine the specific details of the obstacle, specifically whether the gap between obstacles is a legal or illegal space. Consequently, parking attempts based on this information may fail, impacting user experience and hindering the development of autonomous driving technology.

[0051] In view of this, embodiments of the present application provide a method for parking space identification, which can be performed by a processing device. A processing device refers to a device with data processing capabilities, such as an electronic control unit on a vehicle. An electronic control unit is a control device composed of integrated circuits that performs a series of functions such as analyzing, processing, and transmitting data. It typically includes multiple components such as input circuits, an A / D (analog / digital) converter, a microcomputer, and an output circuit.

[0052] Specifically, the processing equipment uses ultrasound to detect parking spaces and obtain candidate parking spaces within the area. The candidate parking spaces are represented by parking reflection points. Then, based on the obstacle information in the area learned from the multi-way surround view images, it is determined whether the candidate parking space is a real parking space. When the candidate parking space is a real parking space, the location information of the real parking space is output, thereby obtaining the location information of the real parking space.

[0053] In order to facilitate understanding of the technical solution of the present application, the parking space recognition method provided by the present application is introduced below with reference to the accompanying drawings.

[0054] See also Figure 1 The flowchart of a parking space recognition method is shown in FIG. , and the specific steps of the method are as follows.

[0055] S102: The processing device uses ultrasound to detect parking spaces and obtains candidate parking spaces in the area.

[0056] The candidate parking spaces are represented by parking space reflection points.

[0057] In some possible implementations, the ultrasonic wave may be emitted by a vehicle-mounted ultrasonic radar, which is a common sensor and is usually used as a reversing radar.

[0058] Typically, ultrasonic parking space identification can only confirm the presence of an obstacle based on reflection points, but cannot further determine the specific details of the obstacle, resulting in misjudgment of parking spaces between obstacles. For example, ultrasonic technology may identify the gap between landscape trees on both sides of the road as a parking space.

[0059] In this embodiment, the processing device obtains candidate parking spaces within the area based on the parking space reflection point information.

[0060] S104: The processing device determines whether the candidate parking space is a real parking space based on the obstacle information in the area learned from the multi-way surround view images.

[0061] The multi-way surround view images are acquired through the on-board surround view camera, including multiple frames of multi-way surround view images acquired by the vehicle at different positions.

[0062] If only a single frame of a multi-view surround view image of the vehicle at a specific location is acquired, important information in the image may be obscured, leading to inaccurate image recognition and poor recognition accuracy. Furthermore, acquiring only a single frame of a multi-view surround view image fails to capture the vehicle's specific status. For example, an obstacle may be identified as a vehicle, but its specific state may be uncertain. The vehicle may be parking, may have completed parking, or may be in motion. This can lead to misjudgment and potentially cause accidents.

[0063] Specifically, the onboard surround view camera acquires a mosaic image of multiple surround view images of the vehicle's surroundings. When the obstacle in the area learned from the multiple surround view images is a stationary vehicle, the candidate parking space is determined to be a real parking space.

[0064] The processing device stitches the acquired multi-way surround view images into a bird's-eye view and inputs the image into a pre-trained deep learning (DL) model. The DL model outputs that the obstacles in the multi-way surround view images are vehicles.

[0065] The deep learning model is constructed through deep learning methods. Deep learning methods can solve the problem of failure of traditional machine learning vision algorithms and traditional image algorithms in image recognition under strong light, shadow and partial occlusion, and can improve the detection rate of single-frame images and the robustness of multiple scenes.

[0066] In this embodiment, by using a deep learning model to detect vehicles in the area through multi-way surround view images, it is possible to avoid the problem of inaccurate information acquisition in the area caused by obstacles such as trees, cones, and flower beds.

[0067] The processing equipment calculates the trajectory of the obstacle vehicle through the vehicle motion model based on the vehicle's wheel pulses, steering wheel angle and other vehicle motion parameters to obtain the vehicle's position sequence.

[0068] Dead reckoning is a method of obtaining the track and vehicle surrounding information based on the vehicle's turning angle and wheel speed without the aid of external navigation objects.

[0069] The processing device determines whether the vehicle is in a stationary state based on the vehicle's position sequence. When the vehicle is in a stationary state, the candidate parking space corresponding to the vehicle may be a real parking space.

[0070] In this embodiment, by combining track calculation and single-frame detection algorithm, the vehicle status of multiple detections in the time series information can be tracked and updated to obtain an accurate vehicle position sequence, solve the problem of inaccurate vehicle information due to occlusion and single-frame missed detection, improve the vehicle detection rate, and improve the detection accuracy through multi-frame images.

[0071] By tracking the targets of vehicles in the area, the processing equipment can obtain the actual motion state of the vehicle, thereby determining whether the vehicle is stationary, and avoiding the occurrence of unexpected safety accidents caused by the vehicle continuing to stop while in motion.

[0072] When the obstacle is a vehicle and is stationary, the specific location of the vehicle is obtained. The specific location of the vehicle can be obtained through a deep learning model, based on a vehicle position sequence, or other methods.

[0073] When the specific position of the vehicle is consistent with the parking space reflection point of the ultrasonic wave, the candidate parking space is determined to be the real parking space.

[0074] Specifically, the specific position of the vehicle can be converted into coordinates in the world coordinate system, and the parking space reflection point in the ultrasound can also be converted into coordinates in the world coordinate system. The two coordinates are compared to determine whether the parking space reflection point in the ultrasound is the position of a stationary vehicle. The specific position of the vehicle and the parking space reflection point in the ultrasound can be converted into coordinates in the world coordinate system or into other coordinates, thereby accurately comparing the specific position of the vehicle and the position of the parking space reflection point in the ultrasound.

[0075] In some possible implementations, the number of obstacles in the area is greater than one, the number of obstacles is multiple, the number of parking space reflection points is multiple, and the number of vehicles is multiple.

[0076] When the specific position of the vehicle is consistent with the parking space reflection point, the candidate parking space can be determined to be a real parking space. Specifically, when the specific positions of all vehicles are consistent with all parking space reflection points, the candidate parking space can be determined to be a real parking space. When any one of them is inconsistent, the candidate parking space is an invalid parking space.

[0077] S106: When the candidate parking space is a real parking space, the processing device outputs location information of the real parking space.

[0078] In this way, parking space recognition can be prevented from being affected by the environment, accuracy and obstacles, the location information of the actual parking space can be obtained, accurate and stable input information can be provided for automatic parking, the completion of the automatic parking function can be assisted, the user experience can be enhanced, and the success rate of automatic parking can be increased.

[0079] In summary, this application provides a parking space recognition method that uses ultrasound to detect parking spaces, obtains candidate parking spaces, and then determines whether the candidate parking space is a real parking space based on obstacle information learned from multi-way surround view images. If it is a real parking space, the location information of the real parking space is output. In this way, it is possible to integrate the detection of the surrounding environment, the tracking of obstacle information, and the matching of the parking space reflection point of the ultrasound, avoiding problems such as identifying obstacles as non-vehicles or vehicles in a non-stationary state, which may result in the identification of illegal parking spaces, thereby improving the accuracy of parking space recognition.

[0080] This application provides another embodiment of a parking space identification method, such as Figure 2 shown.

[0081] Specifically, parking space recognition is started. On the one hand, the visual deep learning vehicle detection module is started to obtain multi-way surround view images, detect vehicles around the environment, track and update the vehicle position, and convert the stationary vehicle information into world coordinate system coordinates.

[0082] Meanwhile, the ultrasonic parking space detection module is activated to detect parking spaces in the surrounding environment and convert the detected parking space reflection point information into world coordinates. Based on the vehicle's world coordinates and the world coordinates of the parking space reflection point, the processing device determines whether the parking space reflection point represents a stationary vehicle, that is, whether the candidate parking space is a real parking space. If the parking space reflection point represents a stationary vehicle, the processing device outputs the location information of the real parking space.

[0083] This can accurately identify parking spaces, provide accurate and stable input information for automatic parking, assist in the completion of the automatic parking function, enhance the user experience, and increase the success rate of automatic parking.

[0084] Corresponding to the above method embodiment, the present application also provides a parking space recognition device, see Figure 3 The device 300 includes: a candidate parking space determination module 302, a real parking space determination module 304 and a parking space information output module 306.

[0085] The candidate parking space determination module 302 is configured to detect parking spaces using ultrasound to obtain candidate parking spaces within the area, where the candidate parking spaces are represented by parking space reflection points.

[0086] A real parking space determination module 304 is configured to determine whether a candidate parking space is a real parking space based on obstacle information within the area learned from the multi-way surround view images;

[0087] The parking space information output module 306 is configured to output the position information of the real parking space when the candidate parking space is a real parking space.

[0088] In some possible implementations, the real parking space identification module 304 is specifically configured to:

[0089] When the obstacle in the area learned from the multi-way surround view images is a stationary vehicle, the candidate parking space is determined to be a real parking space.

[0090] In some possible implementations, the real parking space identification module 304 is further configured to:

[0091] Based on the multi-way surround view images, deep learning is used to determine that obstacles in the area are vehicles;

[0092] According to the vehicle's motion parameters, the vehicle's position sequence is obtained by performing dead reckoning through the vehicle's motion model;

[0093] According to the position sequence of the vehicle, it is determined whether the vehicle is stationary.

[0094] In some possible implementations, the real parking space identification module 304 is specifically configured to:

[0095] When the obstacle in the area learned from the multi-way surround view image is a stationary vehicle, determine the specific position of the vehicle;

[0096] When the specific position of the vehicle is consistent with the parking space reflection point, the candidate parking space is determined to be the real parking space.

[0097] In some possible implementations, the multi-way surround view image includes multiple frames of multi-way surround view images acquired at different locations.

[0098] In some possible implementations, the multi-path surround view images are obtained by a vehicle-mounted surround view camera.

[0099] The present application provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is run on a device, the device executes the above-mentioned parking space recognition method.

[0100] The present application provides a computer program product comprising instructions, which, when executed on a device, enables the device to perform the above-mentioned parking space identification method.

[0101] It should also be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided in this application, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines.

[0102] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course can also be implemented by special hardware including application-specific integrated circuits, special CPUs, special memories, special components, etc. In general, all functions performed by computer programs can be easily implemented with corresponding hardware, and the specific hardware structures used to implement the same function can also be diverse, such as analog circuits, digital circuits or special circuits, etc. However, for the present application, software program implementation is a better implementation method in most cases. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer's floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk or optical disk, etc., and includes a number of instructions to enable a computer device (which can be a personal computer, training equipment, or network equipment, etc.) to execute the methods described in each embodiment of the present application.

[0103] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0104] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, a computer, a training device or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website, a computer, a training device or a data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium can be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

Claims

1. A parking space recognition method, characterized in that: The method comprises: Using ultrasound to detect parking spaces, obtaining candidate parking spaces within the area, wherein the candidate parking spaces are represented by parking space reflection points; Determining whether the candidate parking space is a real parking space based on obstacle information within the area learned from the multi-way surround view images; When the candidate parking space is a real parking space, outputting the location information of the real parking space; The determining whether the candidate parking space is a real parking space based on the obstacle information in the area learned from the multi-way surround view images includes: Determining, based on the multi-way surround view images, that obstacles in the area are obstacle vehicles through a deep learning model; According to the motion parameters of the vehicle, dead reckoning is performed on the obstacle vehicle using a vehicle motion model to obtain a position sequence of the obstacle vehicle; Determining whether the obstacle vehicle is in a stationary state based on the position sequence of the obstacle vehicle, and determining the specific position of the vehicle when the obstacle vehicle is in a stationary state; When the specific position of the vehicle is consistent with the parking space reflection point, the candidate parking space corresponding to the obstacle vehicle is determined to be a real parking space.

2. The method according to claim 1, characterized in that The multi-way surround view image includes multiple frames of multi-way surround view images acquired at different positions.

3. The method according to claim 1, characterized in that The multi-path surround view images are obtained through a vehicle-mounted surround view camera.

4. A parking space recognition device, characterized in that: The device comprises: A candidate parking space determination module is used to detect parking spaces using ultrasound to obtain candidate parking spaces within the area, where the candidate parking spaces are represented by parking space reflection points; a real parking space identification module, configured to identify whether the candidate parking space is a real parking space based on obstacle information within the area learned from the multi-way surround view images; The real parking space identification module is specifically configured to determine, based on the multi-way surround view images, that an obstacle in the area is an obstructing vehicle through a deep learning model; According to the motion parameters of the vehicle, dead reckoning is performed on the obstacle vehicle using a vehicle motion model to obtain a position sequence of the obstacle vehicle; Determining whether the obstacle vehicle is in a stationary state based on the position sequence of the obstacle vehicle, and determining the specific position of the vehicle when the obstacle vehicle is in a stationary state; When the specific position of the vehicle is consistent with the parking space reflection point, determining that the candidate parking space corresponding to the obstacle vehicle is the real parking space; The parking space information output module is used to output the position information of the real parking space when the candidate parking space is a real parking space.

5. A device, characterized in that The device includes a processor and a memory; The processor is configured to execute instructions stored in the memory, so that the device performs the method according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that The method comprises instructions for instructing a device to execute the method according to any one of claims 1 to 3.

7. A computer program product, characterized in that When the computer program product is run on a computer, the computer is caused to perform the method according to any one of claims 1 to 3.

Images