Unmanned AVP vehicle detection system and method

Abstract

The invention belongs to the technical field of unmanned driving, and particularly relates to an unmanned AVP vehicle detection system and method, and the system comprises a user operation module which is used for enabling a user to send a parking request to a cloud platform through a mobile phone; the parking space acquisition module is used for acquiring parking spaces without parking; the cloud processing module is used for distributing an optimal parking space for the vehicle according to the parking space acquired by the parking space acquisition module after the cloud platform receives a parking request of a user, and generating a parking path plan; and the environment sensing module is used for acquiring the surrounding environment of the vehicle. According to the unmanned AVP vehicle detection system, the image optimization module is arranged, so that the image can be optimized when the phenomena of rain, snow, fog or insufficient light exist in the image, and the detection accuracy of the unmanned AVP vehicle detection system on environment information can be ensured when a complex environment is encountered.

Description

Technical Field

[0001] The present invention relates to the technical field of driverless technology, and particularly to a driverless AVP vehicle detection system and method. Background Art

[0002] AVP is a low-speed L4-level autonomous driving technology that can enable a vehicle to automatically perform actions such as passenger shuttling, overtaking, reversing, and pedestrian avoidance without driver operation, especially performing well in parking scenarios. The driverless AVP vehicle detection system is a complex system integrating multiple sensors, controllers, and algorithms, aiming to achieve the autonomous parking function of the vehicle in environments such as parking lots.

[0003] However, when the existing driverless AVP vehicle detection system is running, if it encounters a complex environment, such as bad weather (such as rain, snow, fog, etc.) or insufficient light, the performance of its sensors may be affected, resulting in inaccurate detection results, which will in turn affect the operation of the driverless AVP vehicle detection system. Therefore, a driverless AVP vehicle detection system and method are invented. Summary of the Invention

[0004] To solve the above technical problems, according to one aspect of the present invention, the following technical solutions are provided:

[0005] A driverless AVP vehicle detection system, which includes:

[0006] A user operation module for enabling a user to send a parking request to the cloud platform through a mobile phone;

[0007] A parking space acquisition module for acquiring unoccupied parking spaces;

[0008] A cloud processing module for, after receiving the user's parking request on the cloud platform, being able to allocate the optimal parking space for the vehicle according to the parking spaces acquired by the parking space acquisition module and generate a parking path plan;

[0009] An environment perception module for acquiring the surrounding environment of the vehicle;

[0010] A vehicle-end execution module for, after receiving the parking path plan sent by the cloud platform, being able to generate a control instruction for the vehicle according to the environmental information perceived by the environment perception module to control the vehicle to drive along the parking path and achieve automatic parking;

[0011] A parking completion module for, after the vehicle completes automatic parking, being able to send a parking completion message to the cloud platform so that the cloud platform updates the parking space status information of the parking lot and sends a parking completion notice to the user;

[0012] An analysis module for analyzing the images acquired by the parking space acquisition module and the environment perception module to analyze whether there are phenomena such as rain, snow, fog or insufficient light in the images;

[0013] An image optimization module for optimizing the image when the analysis module analyzes that there are phenomena such as rain, snow, fog or insufficient light in the image to improve the clarity of the image;

[0014] The image optimization module includes:

[0015] An image de-raining module for removing rain in the image when there is rain in the image;

[0016] An image de-snowing module for removing snow in the image when there is snow in the image;

[0017] An image de-fogging module for removing fog in the image when there is fog in the image;

[0018] An image light module for removing insufficient light in the image when there is insufficient light in the image.

[0019] As a preferred solution of an unmanned AVP vehicle detection system according to the present invention, wherein: the parking space acquisition module includes:

[0020] A field-end image acquisition module for acquiring images of the parking lot;

[0021] A parking space detection module for detecting whether there are empty parking spaces in the images acquired by the field-end image acquisition module;

[0022] A temporary parking space planning module for forming temporary parking spaces in other places of the parking lot when the parking space detection module does not detect empty parking spaces.

[0023] As a preferred solution of an unmanned AVP vehicle detection system according to the present invention, wherein: the parking space detection module includes:

[0024] A first storage module for storing images of various unoccupied parking spaces;

[0025] A first comparison module for comparing the images acquired by the field-end image acquisition module with the images stored in the first storage module;

[0026] A first judgment module for judging whether the images acquired by the field-end image acquisition module are similar in the first storage module. If so, it means that there are unoccupied parking spaces in the parking lot.

[0027] As a preferred solution of an unmanned AVP vehicle detection system according to the present invention, wherein: the temporary parking space planning module includes:

[0028] A free space acquisition module for acquiring images of free spaces in a parking lot;

[0029] A second storage module for storing environmental information of various vehicles not parked in parking spaces.

[0030] As a preferred solution of an unmanned AVP vehicle detection system according to the present invention, wherein: the parking space detection module further includes:

[0031] A second comparison module for comparing the images acquired by the free space acquisition module with the images stored in the second storage module;

[0032] A second judgment module for judging whether there are similar images in the second storage module for the images acquired by the free space acquisition module. If there are similar images, a temporary parking space will be acquired.

[0033] As a preferred solution of an unmanned AVP vehicle detection system according to the present invention, wherein: the parking space detection module further includes:

[0034] A size acquisition module for acquiring the size of a vehicle;

[0035] A marking module for marking in the temporary parking space according to the vehicle size acquired by the size acquisition module to mark the space to be parked in the temporary parking space;

[0036] A parking space formation module for forming a temporary parking space from the space to be parked marked by the marking module.

[0037] As a preferred solution of an unmanned AVP vehicle detection system according to the present invention, wherein: the environment perception module includes:

[0038] A vehicle-end image acquisition module for acquiring the surrounding environment of the vehicle;

[0039] An image analysis module for analyzing the images acquired by the vehicle-end image acquisition module to analyze the obstacles existing in the images and their positions.

[0040] An unmanned AVP vehicle detection method includes the following specific steps:

[0041] Step 1: The user sends a parking request to the cloud platform through the mobile phone by operating the user operation module;

[0042] Step 2: The field-end image acquisition module collects images of the parking lot. After collection, the analysis module analyzes the acquired images to check for the presence of rain, snow, fog, or insufficient light. If any of these exist, the image optimization module optimizes the images to improve their clarity. Then, the first comparison module compares the images collected by the field-end image acquisition module with those stored in the first storage module. After comparison, the first judgment module determines whether there are similar images in the first storage module. If there are, it indicates that there are unoccupied parking spaces in the parking lot. If not, the free space acquisition module obtains images of the free spaces in the parking lot. After acquisition, the second comparison module compares the images obtained by the free space acquisition module with those stored in the second storage module. After comparison, the second judgment module determines whether there are similar images in the second storage module. If there are, a temporary parking space is obtained. After that, the size acquisition module obtains the size of the vehicle. Then, the marking module marks the vehicle size obtained by the size acquisition module in the temporary parking space to mark the space for the vehicle to park, i.e., the space to be parked. After marking, the parking space formation module forms the marked space to be parked into a temporary parking space;

[0043] Step 3: After the cloud processing module receives the user's parking request on the cloud platform, it can allocate the optimal parking space for the vehicle based on the parking spaces obtained by the parking space acquisition module and generate a parking path plan;

[0044] Step 4: The vehicle-end image acquisition module collects the surrounding environment of the vehicle. After collection, the analysis module analyzes the acquired images to check for the presence of rain, snow, fog, or insufficient light. If any of these exist, the image optimization module optimizes the images to improve their clarity. Then, the image parsing module parses the images collected by the vehicle-end image acquisition module to analyze the obstacles and their positions in the images;

[0045] Step 5: After the vehicle-end execution module receives the parking path plan sent by the cloud platform, it can generate control instructions for the vehicle based on the environmental information sensed by the environmental perception module to control the vehicle to drive along the parking path and achieve automatic parking;

[0046] Step 6: After the vehicle completes automatic parking, the parking completion module can send a parking completion message to the cloud platform to enable the cloud platform to update the parking space status information of the parking lot and send a parking completion notice to the user.

[0047] Compared with the prior art:

[0048] By setting up an image optimization module, when there is rain, snow, fog or insufficient light in the image, the image can be optimized, and further, when encountering a complex environment, the detection accuracy of the environmental information by the driverless AVP vehicle detection system can be ensured; in addition, by setting up a parking space acquisition module, when there is no empty parking space in the parking lot, a temporary parking space can be generated in other places in the parking lot, and further, it can be ensured that the vehicle has a place to park. Brief Description of the Drawings

[0049] Figure 1 It is a schematic flowchart of the present invention. Detailed Embodiments

[0050] To make the objectives, technical solutions and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the drawings.

[0051] The present invention provides a driverless AVP vehicle detection system. Please refer to Figure 1 ;

[0052] It includes: a user operation module for enabling a user to send a parking request to the cloud platform through a mobile phone; a parking space acquisition module for acquiring unoccupied parking spaces; a cloud processing module for, after the cloud platform receives the user's parking request, being able to allocate the optimal parking space for the vehicle according to the parking spaces acquired by the parking space acquisition module and generating a parking path plan; an environmental perception module for acquiring the surrounding environment of the vehicle; a vehicle-end execution module for, after receiving the parking path plan sent by the cloud platform, being able to generate a control instruction for the vehicle according to the environmental information perceived by the environmental perception module to control the vehicle to drive along the parking path and achieve automatic parking; a parking completion module for, after the vehicle completes automatic parking, being able to send a parking completion message to the cloud platform so that the cloud platform updates the parking space status information of the parking lot and sends a parking completion notice to the user; an analysis module for analyzing the images acquired by the parking space acquisition module and the environmental perception module to analyze whether there is rain, snow, fog or insufficient light in the images; an image optimization module for, when the analysis module analyzes that there is rain, snow, fog or insufficient light in the images, being able to optimize the images to improve the clarity of the images;

[0053] The image optimization module includes: an image de-raining module for removing rain in the image when there is rain in the image; an image de-snowing module for removing snow in the image when there is snow in the image; an image de-fogging module for removing fog in the image when there is fog in the image; an image lighting module for removing insufficient light in the image when there is insufficient light in the image;

[0054] Steps for the image de-raining module: Collect a large number of rainy images and corresponding rain-free images as the training dataset, construct the network structures of the generator and discriminator. Usually, the generator uses a convolutional neural network (CNN) to extract image features and generate rain-free images, and the discriminator also uses CNN to judge the authenticity of the input images; Use the training dataset to alternately train the generator and discriminator. During the training process, continuously adjust the parameters of the generator and discriminator to make the rain-free images generated by the generator more realistic and the discriminator can more accurately judge the authenticity of the images; After the training is completed, input the rainy image to be processed into the trained generator, and the generator outputs the image with rain removed;

[0055] Steps for the image de-snowing module: Construct a CNN model, such as the U-Net structure, and train it with a large number of paired data of snowy and snow-free images, so that the model can learn the mapping relationship between snowy images and snow-free images. During the training process, the model will automatically extract the features in the images, identify the features of the snowflakes and remove them to generate clean snow-free images;

[0056] Steps for the image de-hazing module: First, calculate the dark channel image of the input image; Then estimate the atmospheric light value according to the dark channel image; Next, estimate the transmittance; Finally, perform de-hazing processing on the image through the inversion formula of the atmospheric scattering model;

[0057] Steps for the image lighting module: By adjusting the gray histogram of the image, make its distribution uniform, so that the gray value range of the image is extended to the entire gray interval [0, 255]. This can increase the contrast of the image, improve the overall brightness, and make the dark details caused by insufficient light originally show more clearly.

[0058] The parking space acquisition module includes: a field-end image acquisition module for collecting images of the parking lot; a parking space detection module for detecting whether there are empty parking spaces in the images collected by the field-end image acquisition module; a temporary parking space planning module for forming temporary parking spaces in other places of the parking lot when the parking space detection module does not detect empty parking spaces.

[0059] The parking space detection module includes: a first storage module for storing various images of unoccupied parking spaces; a first comparison module for comparing the images collected by the field-end image acquisition module with the images stored in the first storage module; a first judgment module for judging whether there are similarities in the images collected by the field-end image acquisition module in the first storage module. If there are similarities, it means that there are unoccupied parking spaces in the parking lot.

[0060] The temporary parking space planning module includes: a vacant site acquisition module for acquiring images of vacant sites in the parking lot; a second storage module for storing various vehicle environmental information not parked in parking spaces; a second comparison module for comparing the images acquired by the vacant site acquisition module with the images stored in the second storage module; a second judgment module for judging whether the images acquired by the vacant site acquisition module are similar in the second storage module. If there is similarity, a temporary parking space will be acquired; a size acquisition module for acquiring the size of the vehicle; a marking module for marking in the temporary parking space according to the vehicle size acquired by the size acquisition module to mark the parking space to be parked in the temporary parking space; a parking space formation module for forming the parking space to be parked marked by the marking module into a temporary parking space.

[0061] The environmental perception module includes: a vehicle-end image acquisition module for acquiring the surrounding environment of the vehicle; an image analysis module for analyzing the images acquired by the vehicle-end image acquisition module to analyze the obstacles and their positions in the images.

[0062] A method for detecting an unmanned AVP vehicle includes the following specific steps:

[0063] Step 1: The user sends a parking request to the cloud platform through the mobile phone using the user operation module;

[0064] Step 2: The field-end image acquisition module acquires images of the parking lot. After acquisition, the analysis module analyzes the acquired images to analyze whether there are phenomena such as rain, snow, fog, or insufficient light in the images. If so, the image optimization module optimizes the images to improve the clarity of the images. Then, the first comparison module compares the images acquired by the field-end image acquisition module with the images stored in the first storage module. After comparison, the first judgment module judges whether the images acquired by the field-end image acquisition module are similar in the first storage module. If there is similarity, it means that there are unoccupied parking spaces in the parking lot. If not, the vacant site acquisition module acquires images of the vacant sites in the parking lot. After acquisition, the second comparison module compares the images acquired by the vacant site acquisition module with the images stored in the second storage module. After comparison, the second judgment module judges whether the images acquired by the vacant site acquisition module are similar in the second storage module. If there is similarity, a temporary parking space will be acquired. After acquisition, the size acquisition module acquires the size of the vehicle. Then, the marking module marks in the temporary parking space according to the vehicle size acquired by the size acquisition module to mark the parking space to be parked in the temporary parking space. After marking, the parking space formation module forms the parking space to be parked marked by the marking module into a temporary parking space;

[0065] Step 3: After the cloud processing module receives the parking request from the user on the cloud platform, it can allocate the optimal parking space for the vehicle according to the parking spaces obtained by the parking space acquisition module, and generate a parking path plan.

[0066] Step 4: The vehicle terminal image acquisition module collects the surrounding environment of the vehicle. After the collection, the analysis module analyzes the acquired image to analyze whether there are phenomena such as rain, snow, fog or insufficient light in the image. If so, the image optimization module optimizes the image to improve the clarity of the image. Then, the image parsing module parses the image collected by the vehicle terminal image acquisition module to analyze the obstacles and their positions in the image.

[0067] Step 5: After the vehicle terminal execution module receives the parking path plan sent by the cloud platform, it can generate a control instruction for the vehicle according to the environmental information sensed by the environmental perception module to control the vehicle to drive along the parking path and achieve automatic parking.

[0068] Step 6: After the vehicle completes automatic parking through the parking completion module, it can send a parking completion message to the cloud platform, so that the cloud platform updates the parking space status information of the parking lot and sends a parking completion notice to the user.

[0069] Although the present invention has been described above with reference to the embodiments, various improvements can be made to it and components therein can be replaced with equivalents without departing from the scope of the present invention. In particular, as long as there is no structural conflict, the various features in the embodiments disclosed in the present invention can be combined with each other in any way. The exhaustive description of these combinations is not given in this specification only for the sake of saving space and resources. Therefore, the present invention is not limited to the specific embodiments disclosed in the text, but includes all technical solutions falling within the scope of the claims.

Claims

1. An unmanned AVP vehicle detection system, characterized in that: include: A user operation module, used to enable the user to send a parking request to the cloud platform via a mobile phone; A parking space acquisition module is used to acquire unparked parking spaces; The cloud processing module is used to allocate the best parking space for the vehicle and generate a parking path plan based on the parking space obtained by the parking space acquisition module after receiving the parking request from the user on the cloud platform; Environmental perception module, used to obtain the surrounding environment of the vehicle; The vehicle-side execution module is used to generate vehicle control instructions based on the environmental information sensed by the environmental perception module after receiving the parking path plan sent by the cloud platform, so as to control the vehicle to drive along the parking path and realize automatic parking; The parking completion module is used to send parking completion information to the cloud platform after the vehicle completes automatic parking, so that the cloud platform can update the parking space status information of the parking lot and send a parking completion notification to the user; An analysis module is used to analyze the images acquired by the parking space acquisition module and the environment perception module to analyze whether there is rain, snow, fog or insufficient light in the images; An image optimization module is used to optimize the image to improve the clarity of the image when the analysis module finds that there is rain, snow, fog or insufficient light in the image; The image optimization module comprises: An image rain removal module is used to remove rain from an image when there is rain in the image; An image snow removal module is used to remove snow from an image when there is snow in the image; Image defogging module, used to remove fog from an image when there is fog in the image; The image light module is used to remove the light in the image when there is insufficient light in the image.

2. The unmanned AVP vehicle detection system according to claim 1, characterized in that: The parking space acquisition module includes: The field-side image acquisition module is used to collect images of the parking lot; The parking space detection module is used to detect whether there is an empty parking space in the image collected by the field-side image acquisition module; The temporary parking space planning module is used to form a temporary parking space elsewhere in the parking lot when the parking space detection module does not detect an empty parking space.

3. The unmanned AVP vehicle detection system according to claim 2, characterized in that: The parking space detection module comprises: A first storage module is used to store various unparked parking space images; A first comparison module, used for comparing the image acquired by the field-side image acquisition module with the image stored in the first storage module; The first judgment module is used to judge whether the image collected by the field-side image collection module is similar to the image in the first storage module. If so, it means that there is an unparked parking space in the parking lot.

4. The unmanned AVP vehicle detection system according to claim 3, characterized in that: The temporary parking space planning module includes: A vacant space acquisition module is used to acquire images of vacant spaces in a parking lot; The second storage module is used to store various vehicle environment information that are not parked in parking spaces.

5. The unmanned AVP vehicle detection system according to claim 4, characterized in that: The parking space detection module also includes: A second comparison module, used for comparing the image acquired by the vacant space acquisition module with the image stored in the second storage module; The second judgment module is used to judge whether the image acquired by the vacant space acquisition module is similar to the image in the second storage module. If so, a temporary parking space will be acquired.

6. The unmanned AVP vehicle detection system according to claim 5, characterized in that: The parking space detection module also includes: A size acquisition module, used to acquire the size of the vehicle; A marking module, used for marking the vehicle size acquired by the size acquisition module in the temporary parking space, so as to mark the parking space to be parked in the temporary parking space; The parking space forming module is used to form a temporary parking space from the parking space marked by the marking module.

7. The unmanned AVP vehicle detection system according to claim 1, characterized in that: The environment perception module comprises: The vehicle-side image acquisition module is used to collect the surrounding environment of the vehicle; The image analysis module is used to analyze the images collected by the vehicle-side image acquisition module to analyze the obstacles in the image and their locations.

8. A method for detecting an unmanned AVP vehicle, characterized in that: The specific steps are as follows: Step 1: The user operation module enables the user to send a parking request to the cloud platform via a mobile phone; Step 2: The image of the parking lot is collected by the field-side image collection module. After the collection, the acquired image will be analyzed by the analysis module to analyze whether there is rain, snow, fog or insufficient light in the image. If so, the image will be optimized by the image optimization module to improve the clarity of the image. After that, the image collected by the field-side image collection module will be compared with the image stored in the first storage module through the first comparison module. After the comparison, the first judgment module will be used to judge whether the image collected by the field-side image collection module is similar to the image in the first storage module. If so, it means that there is an unparked parking space in the parking lot. If not, the vacant space acquisition module will be used to compare the parking space. The image of the vacant space in the parking lot is acquired. After acquisition, the image acquired by the vacant space acquisition module is compared with the image stored in the second storage module through the second comparison module. After comparison, the second judgment module is used to judge whether the image acquired by the vacant space acquisition module is similar in the second storage module. If so, a temporary parking space is acquired. After acquisition, the size of the vehicle is acquired through the size acquisition module. Then, the marking module is used to mark the temporary parking space according to the vehicle size acquired by the size acquisition module, so as to mark the space to be parked in the temporary parking space. After marking, the space to be parked marked by the marking module is formed into a temporary parking space through the parking space forming module; Step 3: After receiving the user's parking request on the cloud platform through the cloud processing module, the optimal parking space can be allocated to the vehicle according to the parking space obtained by the parking space acquisition module, and a parking path plan can be generated; Step 4: The vehicle's surrounding environment is collected through the vehicle-side image acquisition module. After the collection, the acquired image is analyzed by the analysis module to analyze whether there is rain, snow, fog or insufficient light in the image. If so, the image is optimized by the image optimization module to improve the image clarity. After that, the image collected by the vehicle-side image acquisition module is analyzed by the image analysis module to analyze the obstacles in the image and their positions; Step 5: After receiving the parking path plan sent by the cloud platform, the vehicle-side execution module can generate vehicle control instructions based on the environmental information sensed by the environmental perception module to control the vehicle to drive along the parking path and realize automatic parking; Step 6: After the vehicle completes automatic parking, the parking completion module can send parking completion information to the cloud platform so that the cloud platform can update the parking space status information of the parking lot and send a parking completion notification to the user.

Images