Method and system for valet parking with local autonomous planning and decision making

Abstract

The application provides a valet parking method with local autonomous planning decision function, comprising: acquiring multi-source perception information of a current target vehicle according to at least one information acquisition device arranged around the target vehicle; obtaining complete obstacle information around the current target vehicle based on a multi-source perception data fusion algorithm; constructing a kinematic model and a single-track model of the target vehicle, and calculating an optimal driving local path of the target vehicle when safely passing a road with obstacles according to the obstacle information; and controlling the target vehicle to return to the path for continuous driving. By adopting a combination of a high-precision map global path and a local planning path, the entire process of valet parking is guaranteed to be efficient and reliable, the compatibility of the system is improved, and the application meets the advantages of having the ability of local autonomous planning in a complex scene, being adaptive to a complex scene with lack of coverage network and dim light, and handling a sudden situation of an intruding vehicle.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent driving of vehicles, in particular a valet parking method and system with local autonomous planning and decision-making functions. BACKGROUND

[0002] At present, the intelligent networking development of the domestic automobile market has initially taken shape. In order to adapt to the trend of intelligent development, automobile enterprises have continuously increased the research and development efforts on vehicle intelligent technology, and automobile enterprises have actively explored the use of 5G communication, edge computing, artificial intelligence and other new technologies to explore the development path of intelligent vehicles.

[0003] High-speed automatic driving is limited by legal regulations, technical maturity, safety and other factors, and cannot be truly oriented to consumers in a short time. In contrast, low-speed automatic driving and automatic parking have become a breakthrough point in the entire industry. In July 2021, Bosch and Daimler jointly announced that the automatic valet parking (AVP) developed by the two parties has been approved by the relevant departments of Baden-Württemberg, Germany, allowing it to be used in the Mercedes-Benz Museum parking lot in Stuttgart, Germany. This is the first L4 level fully automatic driving suitable for parking functions to be approved for use in daily life. Domestic traditional automobile enterprises and new energy automobile enterprises are actively exploring the development of their own valet parking technology.

[0004] At present, the automatic parking assistance system that has been widely used is based on ultrasonic or camera, which realizes the parking by stopping the vehicle near the parking space, enabling the automatic parking assistance function, detecting the environment around the parking space by the sensors, planning the parking trajectory according to the environmental information, and finally controlling the vehicle to enter the parking space. In addition, the existing autonomous valet parking system adds the function of autonomous navigation driving from the entrance of the parking lot to the vicinity of the parking space on the basis of the automatic parking assistance system. The implementation mainly includes a scheme of pure vehicle modification, that is, configuring laser radar, cameras and other sensors for the ordinary passenger vehicle to perceive environmental information and obtain the vehicle's pose, and combining the roaming method or the parking space issuing method to complete the autonomous trajectory movement and parking action in the parking lot.

[0005] However, the current automatic parking assistance system has a low success rate. Due to the limitation of sensor capability, this system has strict requirements for the parking environment, and the vehicle needs to be parked near the parking space manually, and the automatic parking assistance function needs to be started, which cannot truly realize autonomous valet parking.

[0006] The main shortcomings of the existing autonomous valet parking scheme are summarized as follows: the autonomous valet parking function is only applicable to intelligent vehicles that have been modified, the cost of a single vehicle is high, and if a laser radar is used, the cost will be further increased, and if only a camera is used, the dependence on external light is high, and the applicable parking lot will be limited. SUMMARY

[0007] In view of the deficiencies in the prior art, the present application aims to provide a valet parking method and system with local autonomous planning and decision-making functions, which fuses ultrasonic radar with visual sensor and AI Camera sensing information, so that the application can provide accurate and reliable environmental sensing information and is suitable for a wider range of application scenarios. At the same time, by combining high-precision map global paths with local planning paths, the application ensures efficient and reliable valet parking throughout the entire process, improves system compatibility, and finally provides real-time monitoring and operation for users through a mobile terminal and a cloud server, ensuring that the parking process is monitored by both the system and the user, thereby meeting the application's ability to autonomously plan locally in complex scenarios, adapt to complex scenarios with a lack of network coverage and dim lighting, and handle sudden situations where a vehicle intrudes. The above problems in the background art are solved.

[0008] To achieve the above-mentioned purpose, the present application is implemented by the following technical solution: a valet parking method with local autonomous planning and decision-making functions, comprising

[0009] obtaining multi-source sensing information of a target vehicle from at least one information acquisition device arranged around the target vehicle;

[0010] extracting feature information from the multi-source sensing information based on a multi-source sensing data fusion algorithm, and performing feature matching processing on the extracted feature information to obtain complete obstacle information around the target vehicle;

[0011] constructing a kinematic model and a single-track model of the target vehicle to calculate an optimal driving local path of the target vehicle when safely passing through an obstacle road based on the obstacle information;

[0012] obtaining the minimum steering wheel angle of the target vehicle in the optimal driving local path, and controlling the target vehicle to return to the path for continuous driving.

[0013] As an improvement to the valet parking method with local autonomous planning and decision-making functions described in the present application, the obtained multi-source sensing information at least includes the contour information, category information, coordinate information of the obstacle, and the parking space information and surrounding environment situation information of the target vehicle.

[0014] As an improvement to the valet parking method with local autonomous planning and decision-making functions described in the present application, after the target vehicle safely passes through the obstacle road, the target vehicle needs to be controlled to return to the path for continuous driving based on a PID control method.

[0015] As a second aspect of the application, a valet parking system with local autonomous planning decision function is proposed, a vehicle end ultrasonic wave radar acquisition module, which feeds back reliable obstacle distance information to the target vehicle in the night, thick fog environment;

[0016] A vehicle vision sensor acquisition module, which identifies the contour information, category information and coordinate information of the current obstacle based on the received obstacle distance information;

[0017] An AI Camera perception information acquisition module, which obtains the surrounding environment situation information of the target vehicle;

[0018] A model module, which obtains the kinematic model and single-track model of the target vehicle through import or construction, and performs multi-source perception data fusion preprocessing on the obstacle distance information, the contour information, the category information, the coordinate information of the obstacle and the surrounding environment situation information of the target vehicle;

[0019] An optimal driving local path planning module, which obtains the optimal driving path of the target vehicle through the obstacle road through the target vehicle dynamics model or mathematical model simulation;

[0020] A global path regression module, which controls the path of the target vehicle returning to continue driving after passing through the obstacle road based on the PID control method.

[0021] Compared with the prior art, the application has the following advantages:

[0022] 1、The application fuses ultrasonic wave radar with vision sensor, AI Camera perception information, so that the application can provide accurate and reliable environmental perception information and is suitable for a wider range of application scenarios;

[0023] 2、At the same time, the application adopts a combination of high-precision map global path and local planning path to ensure efficient and reliable valet parking throughout the process, improve system compatibility, and provide super-long distance real-time monitoring operation for users through the mobile terminal and cloud terminal, ensuring that the parking process is monitored by the system and the user, thereby meeting the application's ability to autonomously plan locally in complex scenarios, adapt to complex scenarios with lack of coverage network and dim light, and handle situations when the vehicle suddenly intrudes. BRIEF DESCRIPTION OF DRAWINGS

[0024] The disclosure of the application will be described with reference to the accompanying drawings. It should be understood that the drawings are only for illustrative purposes, and are not intended to limit the scope of protection of the application. In the drawings, the same reference numerals are used to refer to the same parts. Among them:

[0025] Figure 1The overall timing structure schematic diagram of the information interaction of the vehicle end, the cloud end and the user mobile phone end when the target vehicle proposed in an embodiment of the present application carries out local autonomous planning decision function;

[0026] Figure 2 The schematic diagram of collision when the target vehicle proposed in an embodiment of the present application continues to drive according to the global route provided by the high-precision map and encounters an obstacle;

[0027] Figure 3 The schematic diagram of the optimal driving local path through the obstacle road after switching to the local path planning when the target vehicle proposed in an embodiment of the present application continues to drive according to the global route provided by the high-precision map and encounters an obstacle;

[0028] Figure 4 The specific measurement schematic diagram of the vehicle size parameters L, C, K and M of the target vehicle proposed in an embodiment of the present application;

[0029] Figure 5 The calculation idea schematic diagram of the closest distance value B between the target vehicle proposed in an embodiment of the present application and the obstacle;

[0030] Figure 6 The path schematic diagram of the target vehicle returning to continue driving after passing through the obstacle road based on the PID control mode proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0031] It is easy to understand that, according to the technical scheme of the present application, a person skilled in the art can propose a plurality of structure modes and implementation modes which can be replaced with each other without changing the essential spirit of the present application. Therefore, the following specific embodiments and the accompanying drawings are only exemplary descriptions of the technical scheme of the present application, and should not be regarded as the whole or regarded as the limitation or restriction of the technical scheme of the present application.

[0032] The present application will be further described in detail below in combination with the accompanying drawings, but not as the limitation of the present application.

[0033] As an understanding of the technical concept and implementation principle of the present application, in order to solve the problem that the automatic parking auxiliary system in the prior art has a low parking success rate, due to the limitation of sensor capability, and also needs to manually park the vehicle near the parking space, start the automatic parking auxiliary function, and cannot truly realize autonomous valet parking, the present application first proposes a multi-source perception fusion algorithm, which fuses the ultrasonic radar at the vehicle end, visual perception information and AI Camera perception information arranged at the field end, and then constructs a target vehicle kinematics model and a single track model, solves the problem that the existing high-precision map containing complete road information cannot cope with complex road environment and real-time changing road conditions when providing global path information, thereby causing the vehicle to deviate from the preset track.

[0034] Therefore, in order to realize the above technical concept and solve the defects of the existing technical solutions.

[0035] As shown in Figure 1 As an embodiment of the present application, a valet parking method with local autonomous planning and decision-making function is proposed, which includes the following steps:

[0036] Obtain multi-source perception information of the current target vehicle based on at least one information acquisition device arranged around the target vehicle.

[0037] Based on the above technical concept, it should be noted that the obtained multi-source perception information at least includes the contour information, category information, coordinate information of the obstacle, the parking space information of the target vehicle, and the surrounding environment situation information of the target vehicle. And the information acquisition device is at least the vehicle end ultrasonic radar, vehicle visual sensor and AI Camera. It can be understood that in an embodiment of the present application, based on the ultrasonic radar, reliable distance information can be provided in the night, heavy fog and other environments without being affected by light and weather; based on the vehicle visual sensor, the contour, category and coordinate information of the obstacle can be accurately identified; based on the advantage that the AI Camera is not limited by the conditions of the vehicle itself, the surrounding environment situation information of the target vehicle can be accurately obtained, including the parking space information of the target vehicle, the surrounding charging pile information, the fee payment information, the entrance and exit gate information, the car washing service information, etc.

[0038] In an embodiment of the present application, since the above-mentioned information acquisition device (such as vehicle visual sensor) is easily affected by night, rain, snow, heavy fog and other weather, resulting in errors in the identification result, therefore, it is also necessary to extract feature information from the multi-source perception information based on a multi-source perception data fusion algorithm, and perform feature matching processing on the extracted feature information, to obtain complete obstacle information around the current target vehicle. It can be understood that through the feature extraction method, the vehicle can more accurately measure and identify the obstacle information around the vehicle, and improve the identification of the surrounding environment of the vehicle.

[0039] As Figures 2-6 shown, as an embodiment of the present application, the present application also includes:

[0040] The target vehicle kinematics model and monorail model are constructed to calculate the optimal driving local path of the target vehicle when passing through the road with obstacles, and it is noted that the purpose of this step is to plan a local optimal path by the local path optimization algorithm in combination with the perception fusion information when the target vehicle cannot continue to drive according to the path information provided by the high-precision map or cannot drive into the target parking space. The adaptability of the vehicle in the automatic parking process is improved, and the success rate of the valet parking is ensured. Therefore,

[0041] Based on the above technical concept, in the embodiment of the present application, the specific implementation of controlling the optimal driving local path of the target vehicle when passing through the road with obstacles is:

[0042] S1, according to the obtained obstacle information, the minimum wheel rotation angle of the target vehicle when passing through the road with obstacles is calculated

[0043]

[0044] In the formula, a, b, c represent the calibrated constant values related to the target vehicle EPS; and δ represents the current target vehicle steering wheel rotation angle;

[0045] S2, the maximum turning radius R of the target vehicle on the road with obstacles is calculated:

[0046]

[0047] In the formula, L, C, K, M represent vehicle size parameters;

[0048] S3, based on the obtained maximum turning radius R and the target vehicle coordinates (x2, y2), the coordinates (x3, y3) of the center M of the target vehicle during turning are calculated;

[0049] S4, according to the obtained center M coordinates, the maximum radius R1 of the obstacle from the center M during the turning of the target vehicle is calculated:

[0050]

[0051] In the formula, x1, y1 represent the coordinate information of the obstacle, which is obtained by the acquisition device based on the target vehicle information;

[0052] S5, the closest distance value B between the target vehicle and the obstacle is calculated:

[0053] B = R1 - R0 (4)

[0054] In the formula, R0 represents the maximum turning radius R of the target vehicle on the road with obstacles, and it can be understood that, in a specific implementation, in order to ensure that the target vehicle can safely bypass the obstacle, it is necessary to ensure that the distance B between the target vehicle and the obstacle is greater than a certain value, and therefore, the current steering wheel steering angle δ needs to be greater than a certain value, that is, after the angle δ of the vehicle steering wheel is calculated according to the requirement of the safety distance, the optimal driving local path of the target vehicle can be obtained.

[0055] Based on the above technical concept, it can be understood that the local path planning algorithm proposed in the present application is mainly divided into two parts of obstacle avoidance and returning to the global path, after the minimum steering wheel steering angle of the target vehicle in the optimal driving local path is obtained, the target vehicle can safely pass through the road with obstacles, and it is also necessary to control the target vehicle to return to the path for continuing driving based on the PID control mode.

[0056] In an embodiment of the present application, by fusing the ultrasonic radar with the visual sensor, AICamera sensing information, the present application can provide accurate and reliable environmental sensing information, and is suitable for a wider range of application scenarios. At the same time, by adopting the combination of high-precision map global path and local planning path, the entire process of valet parking is ensured to be efficient and reliable, and the compatibility of the system is improved. Through the mobile terminal and cloud mode, real-time monitoring operation is provided for the user, and the parking process is ensured to be carried out under the double monitoring of the system and the user, thereby meeting the advantages of the present application, such as the ability of local autonomous planning in complex scenes, the adaptability to complex scenes with lack of network coverage and dim light, and the handling of the situation when the vehicle suddenly intrudes.

[0057] As a second aspect of the present application, a valet parking system with local autonomous planning and decision-making function is proposed, comprising

[0058] The vehicle end ultrasonic wave radar collection module feeds back reliable obstacle distance information to the target vehicle in the night and thick fog environment; the vehicle vision sensor collection module identifies the contour information, category information and coordinate information of the current obstacle based on the received obstacle distance information; the AI camera sensing information collection module acquires the surrounding environment situation information of the target vehicle; the model module obtains the kinematics model and single-track model of the target vehicle through import or construction, and carries out multi-source sensing data fusion preprocessing on the obstacle distance information, the contour information, the category information, the coordinate information of the obstacle and the surrounding environment situation information of the target vehicle; the optimal driving local path planning module obtains the optimal driving path of the target vehicle when passing through the road with obstacles through the target vehicle dynamics model or mathematical model simulation; the global path regression module controls the path of the target vehicle after passing through the road with obstacles to continue driving based on the PID control mode.

[0059] The technical scope of the present application is not limited to the above description, and those skilled in the art can make various modifications and changes to the above embodiments without departing from the technical idea of the present application, and these modifications and changes should be within the protection scope of the present application.

Claims

1. A valet parking method with local autonomous planning and decision-making capabilities, characterized in that: Includes the following steps: Multi-source perception information of the target vehicle is obtained by using at least one information collection device located around the target vehicle. Based on the multi-source perception data fusion algorithm, feature information is extracted from the multi-source perception information, and the extracted feature information is subjected to feature matching processing to obtain complete obstacle information around the current target vehicle. A kinematic model and a single-track model of the target vehicle are constructed to calculate the optimal local driving path for the target vehicle to safely pass through a road with obstacles, based on the obstacle information. The implementation method is as follows: S1, Based on the obtained obstacle information, calculate the minimum wheel angle Ø for the target vehicle to safely pass through the road with obstacles: ； In the formula, a, b, and c represent calibration constants related to the EPS of the target vehicle; δ represents the current steering wheel angle of the target vehicle. S2, calculate the maximum turning radius R of the target vehicle on the road with obstacles: ； In the formula, L, C, K, and M all represent vehicle size parameters; S3, based on the obtained maximum turning radius R and the target vehicle coordinates (x2, y2), calculate the coordinates (x3, y3) of the center M of the target vehicle during the turning process; S4, based on the obtained coordinates of the center M, calculate the maximum radius R1 of the obstacle's distance from the center M during the target vehicle's turning process: ； In the formula, x1 and y1 represent the coordinate information of the obstacle, which is obtained based on the information collection equipment set up on the target vehicle; S5, Calculate the closest distance B between the target vehicle and the obstacle: ； In the formula, R0 represents the maximum turning radius R of the target vehicle on a road with obstacles; For a target vehicle to safely bypass an obstacle, the closest distance B between the target vehicle and the obstacle must be greater than a certain value. Based on the safety distance requirement, the steering wheel angle δ of the vehicle is calculated to obtain the optimal local driving path of the target vehicle. Obtain the minimum steering wheel angle of the target vehicle in the optimal local driving path, and control the target vehicle to return to the path to continue driving.

2. The valet parking method with local autonomous planning and decision-making function according to claim 1, characterized in that: The acquired multi-source perception information includes at least the outline information, category information, and coordinate information of obstacles identified by the vehicle's visual sensors, the surrounding environment situation information of the target vehicle acquired by the AI ​​Camera, and the obstacle distance information acquired by the ultrasonic radar.

3. The valet parking method with local autonomous planning and decision-making function according to claim 1 or 2, characterized in that: After the target vehicle safely passes through the road with obstacles, it is also necessary to control the target vehicle to return to the path of continued travel based on PID control.

4. A valet parking system with local autonomous planning and decision-making function, based on the valet parking method with local autonomous planning and decision-making function according to any one of claims 1-3, characterized in that: include The vehicle-mounted ultrasonic radar acquisition module provides reliable obstacle distance information to the target vehicle at night and in dense fog. The vehicle vision sensor acquisition module identifies the outline, category, and coordinate information of the current obstacle based on the received obstacle distance information; The AI ​​Camera perception information acquisition module obtains information about the surrounding environment of the target vehicle. The model module obtains the kinematic model and monorail model of the target vehicle through import or construction, and performs multi-source perception data fusion preprocessing on the obstacle distance information, the outline information, category information, coordinate information of the obstacle, and the surrounding environment situation information of the target vehicle. The optimal driving local path planning module obtains the optimal driving path of the target vehicle when it passes through a road with obstacles by simulating the target vehicle's mathematical model. The global path regression module uses PID control to guide a target vehicle back to its original path after passing through a road with obstacles.

Images