Passenger boarding ladder automatic parking remote driving control method and system

Abstract

The invention discloses a passenger boarding ladder automatic parking remote driving control method and system, and particularly relates to the technical field of remote control, and the method comprises the steps: building an information sharing platform, and installing a real-time image collection device and a vehicle-mounted radar on a boarding ladder to collect real-time data, the information sharing platform, the real-time data and the obstacle analysis model are connected; by carrying the information sharing platform, the road information, the road traffic signals and other vehicle information are obtained through the information sharing platform, the computing power consumption can be saved, and the efficiency can be improved; in the obstacle analysis process, by directly comparing the real-time image data with the corresponding background image data, the part, which is not consistent with the background image, in the real-time image can be quickly identified, so that the existence and the position of the obstacle are accurately judged, the obstacle analysis time is shortened, and the obstacle analysis efficiency is improved. The problems that in the prior art, remote driving response time is long, and the operation risk of the airstairs is increased are effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of remote control technology, and more specifically, to a remote driving control method and system for automatic parking of a passenger boarding ladder. Background Art

[0002] The boarding ladder is a special vehicle working at the airport, which is used to build a ladder connecting the ground and the cabin door for passengers to get on and off the plane. The automatic parking remote driving control of the boarding ladder can effectively save resources and provide shuttle services for passengers.

[0003] In the prior art with application number 202310084318.1, a remote driving control method for automatic parking of a passenger boarding ladder is disclosed, which belongs to the field of intelligent driving. The proposed method adopts a new analysis technology framework that combines local image analysis with global image and voice analysis to realize the remote driving control method of the boarding ladder. It performs collaborative analysis based on the voice signals of passengers in the car in real time, which greatly improves the safety and accuracy of the remote driving control method of the boarding ladder; further improves the attention collaborative analysis model in a targeted manner, which significantly improves the analysis results and further improves the control effect of the passenger boarding ladder.

[0004] However, in actual use, the existing technology still has many shortcomings, such as the long response time of remote driving, which is specifically reflected in: over-reliance on deep learning models to identify road information of the boarding stairs, vehicle information in other lanes, and obstacle information, which increases the analysis time; the need to process real-time image information, consumes huge computing resources, and causes delays in boarding stairs operation instructions, which cannot provide smooth services for passengers. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a remote driving control method and system for automatic parking of a passenger boarding ladder. By building an information sharing platform and using the Internet of Things technology to obtain the environmental information of the boarding ladder, the response time during the remote driving of the boarding ladder is effectively shortened to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a remote driving control method for automatic parking of a passenger boarding ladder, comprising the following steps:

[0007] Step 1: Build an information sharing platform and verify the reliability of the information sharing platform; build an obstacle analysis model and verify the reliability of the obstacle analysis model;

[0008] Step 2: Install a real-time image acquisition device on the boarding ladder and collect real-time data through the vehicle-mounted radar, and connect the information sharing platform, real-time data and obstacle analysis model;

[0009] Step 3: When the vehicle-mounted radar detects an obstacle in the direction of the boarding ladder, real-time image data is collected through the vehicle-mounted camera;

[0010] Step 4: retrieve image data corresponding to the time and space of the real-time image data from the verified information sharing platform as background image data;

[0011] Step 5: The verified obstacle analysis model receives real-time data and corresponding background image data, and outputs obstacle information; the obstacle information includes: the location, size, shape and motion state of the obstacle;

[0012] Step 6: Generate and execute remote control instructions based on obstacle information and information sharing platform information, where the information sharing platform information includes environmental information and vehicle positioning data.

[0013] Preferably, building an information sharing platform includes entering environmental information in the target area into the information sharing platform to build a simulation model of the target area; obtaining real-time vehicle positioning data through a vehicle-mounted radar, and uploading the real-time vehicle positioning data and corresponding trajectory information to the target area simulation model in real time based on the Internet of Things technology; and building an information sharing platform after visualization processing.

[0014] Preferably, the traffic signal information and road information are image data.

[0015] Preferably, the method of verifying the reliability of the information sharing platform is:

[0016] Step 101: Collect environmental information, and record the collected environmental information as Ha; retrieve the environmental information corresponding to time and space from the information sharing platform, and record it as Hb;

[0017] Step 102: Output the shared information consistency index Yz by analyzing the similarity between the environmental information Ha and the environmental information Hb;

[0018] Step 103, by judging the relationship between the shared information consistency index Yz and the preset value, the reliability of the information sharing platform is judged. When the shared information consistency index Yz is lower than the preset value, it indicates that the information sharing platform is abnormal and in an unreliable state. Otherwise, it indicates that the information sharing platform is reliable.

[0019] Preferably, when the environmental information only includes image data, the image similarity is used to represent the shared information consistency index, the image is divided into several areas based on the location of the focus point, a weight coefficient is matched for each area, and the consistency analysis model is obtained by the weighted sum of the image similarity and the weight coefficient of each area.

[0020] Preferably, when the environmental information only includes vehicle positioning data, the vehicle positioning data similarity is used to represent the shared information consistency index;

[0021] Preferably, when the environmental information includes image data and vehicle positioning data, the shared information consistency index is represented by the sum of the image similarity and the vehicle positioning data similarity.

[0022] Preferably, the obstacle analysis model includes an obstacle static feature analysis model, an obstacle dynamic feature analysis model and a feature fusion unit;

[0023] The obstacle static feature analysis model is used to analyze the image data in the real-time data, predict the static features of the obstacle, and present the obstacle position frame in the form of a bounding box and a confidence score;

[0024] The obstacle dynamic feature analysis model obtains an obstacle position frame based on the obstacle static feature analysis model, and predicts the obstacle motion feature by analyzing the positioning data of the obstacle position frame;

[0025] The feature fusion unit is used to output obstacle information and fuse the static features of obstacles and the motion features of obstacles.

[0026] Preferably, the obstacle static feature analysis model includes an input layer, a feature extraction layer, and a classification and regression layer. The input layer is used to receive real-time image data and corresponding background image data as input; the feature extraction layer uses a convolutional neural network algorithm to extract features of the input image data, including edge, texture, color underlying features and semantic features (such as the shape and category of the object); the classification and regression layer classifies and regresses each pixel or area in the real-time image according to the extracted features. The classification operation is used to determine whether the pixel or area belongs to the obstacle category, and the regression operation is used to estimate the position and size parameters of the obstacle.

[0027] Preferably, the method for verifying the reliability of the obstacle analysis model is:

[0028] Step 201, test the obstacle analysis model several times and record the test results;

[0029] Step 202, analyzing the test results to obtain the average value of processing time, the average value of real-time data feature information entropy and obstacle accuracy parameters, which are denoted as Za, Zb and Zc respectively;

[0030] Step 204: By formula The obstacle recognition performance index Xb is calculated, where the Form(·) function is used to normalize the content in the brackets so that the value range in the brackets is between 0 and 1; X1, X2, and X3 are used to correct the obstacle accuracy parameters, processing time loss rate, and feature information entropy to adjust the degree of influence on the obstacle recognition performance index.

[0031] Preferably, the obstacle accuracy parameter is obtained in the following manner:

[0032] Analyze the difference between the predicted obstacle static characteristics and the actual obstacle static characteristics to obtain the static characteristic loss parameter La;

[0033] Analyze the difference between the predicted obstacle motion characteristics and the actual obstacle motion characteristics to obtain the motion characteristic loss parameter Lb;

[0034] Combine the static feature loss parameter La and the motion feature loss parameter Lb, through the formula The obstacle accuracy parameter Za is calculated, where La_max represents the maximum value of the static feature loss parameter and Lb_max represents the maximum value of the motion feature loss parameter.

[0035] To achieve the purpose of the present invention, the following technical solution is provided: a remote driving control system for automatic parking of a passenger boarding ladder, comprising:

[0036] The information sharing platform management module is used to build the information sharing platform and verify the reliability of the information sharing platform based on the shared information consistency index;

[0037] The obstacle analysis management module is used to build an obstacle analysis model and verify the reliability of the obstacle analysis model based on the obstacle recognition performance index;

[0038] Real-time image data acquisition module: when the vehicle-mounted radar detects an obstacle in the direction of the boarding ladder, the vehicle-mounted camera acquires real-time image data;

[0039] A background image data acquisition module retrieves image data corresponding to the time and space of the real-time image data from the verified information sharing platform as background image data;

[0040] The obstacle information acquisition module receives the real-time data and the corresponding background image data after verification, and outputs the obstacle information;

[0041] The remote control module generates and executes remote control instructions based on obstacle information and information sharing platform information.

[0042] Preferably, the automatic parking remote driving control system for the passenger boarding ladder further includes:

[0043] The driving operation management module is used to obtain the reliability index of the automatic driving operation, and the automatic driving operation reliability index is obtained by comparing the actual response parameters and the expected response parameters of the boarding ladder under the operation command; the automatic driving operation reliability index is obtained in the following manner:

[0044] Get the instruction response time loss rate and instruction operation precision loss rate. Assume there are m operation instructions and j represents the sequence number of the operation instructions.

[0045] The average response time loss rate and the average operation accuracy loss rate of the jth operation instruction are recorded as st_j and sac_j respectively; the autonomous driving operation reliability index is calculated by the following formula:

[0046]

[0047] Among them, f1 and f2 represent the correction factors of time loss rate and precision loss rate respectively, which are used to adjust the influence of time loss rate and precision loss rate on the reliability index of autonomous driving operation.

[0048] Preferably, the automatic parking remote driving control system for the passenger boarding ladder also includes

[0049] The monitoring and early warning module monitors the shared information consistency index Xa, obstacle recognition performance index Xb and autonomous driving operation reliability index Xc in real time. When the monitoring data is lower than the corresponding threshold, an early warning is issued; maintenance is performed based on the fault knowledge graph to eliminate anomalies;

[0050] The remote driving control performance analysis module jointly analyzes the shared information consistency index, obstacle recognition performance index and autonomous driving operation reliability index to obtain the remote driving control coefficient, and takes measures based on the remote driving control coefficient.

[0051] Preferably, the remote driving control performance analysis module includes:

[0052] The thresholds of shared information consistency index, obstacle recognition performance index and autonomous driving operation reliability index are obtained, which are denoted as Tha, Thb and Thc respectively;

[0053] By the formula YK = e -max(Xa-Tha，0) *e -max(Xb-Thb，0) *e -max(Xc-Thc，0) The remote driving control coefficient YK is calculated;

[0054] When the remote driving control coefficient is lower than the corresponding threshold, an early warning is issued, indicating that the reliability of remote unmanned driving is low and more monitoring and maintenance resources need to be allocated;

[0055] When the remote driving control coefficient is not lower than the corresponding threshold, it indicates that the remote driving performance of the boarding ladder is normal and no measures are taken.

[0056] Technical effects and advantages of the present invention:

[0057] (1) The automatic parking remote driving control method for a passenger boarding ladder provided by the present invention can save computing power consumption and improve efficiency by carrying an information sharing platform and obtaining road information, road traffic signals and other vehicle information through the information sharing platform. In the obstacle analysis process, by directly comparing the real-time image data with the corresponding background image data, the part of the real-time image that does not match the background image can be quickly identified, thereby accurately judging the existence and position of the obstacle, reducing the obstacle analysis time, and effectively solving the problem of long remote driving response time and increased risk of boarding ladder operation in the prior art.

[0058] (2) The automatic parking remote driving control method for passenger boarding stairs provided by the present invention effectively ensures the quality of remote control by verifying the reliability of the information sharing platform and the obstacle analysis model, and avoids the reduction of the efficiency of obstacle recognition due to erroneous information.

[0059] (3) The automatic parking remote driving control system for the passenger boarding ladder provided by the present invention obtains the automatic driving operation reliability index through the driving operation management module, effectively quantifies the performance quality of the boarding ladder, and jointly analyzes the shared information consistency index, the obstacle recognition performance index and the automatic driving operation reliability index to obtain the quality of the remote control of the passenger boarding ladder and improve the safety of the boarding ladder. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 The present invention is a flow chart of the automatic parking remote driving control method of the boarding ladder.

[0061] Figure 2 This is a structural block diagram of the obstacle analysis model of the present invention. DETAILED DESCRIPTION

[0062] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary real-time embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the real-time embodiments set forth herein. On the contrary, these examples are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0063] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present application, its application, or uses.

[0064] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered as part of the specification.

[0065] Real-time example 1, see Figure 1The present invention provides a method for controlling the automatic parking of a passenger boarding ladder remotely, comprising the following steps:

[0066] Step 1: Build an information sharing platform and verify the reliability of the information sharing platform; build an obstacle analysis model and verify the reliability of the obstacle analysis model;

[0067] Step 2: Real-time data is collected by installing a real-time image acquisition device and a vehicle-mounted radar on the boarding ladder. The vehicle-mounted radar is used to record the real-time positioning data of the boarding ladder, and the information sharing platform, real-time data and obstacle analysis model are connected; the real-time data includes real-time positioning data and real-time image data;

[0068] Step 3: When the vehicle-mounted radar detects an obstacle in the direction of the boarding ladder, real-time image data is collected through the vehicle-mounted camera;

[0069] Step 4: retrieve image data corresponding to the time and space of the real-time image data from the verified information sharing platform as background image data;

[0070] Step 5: The verified obstacle analysis model receives real-time data and corresponding background image data, and outputs obstacle information; the obstacle information includes: the location, size, shape and motion state of the obstacle;

[0071] Step 6: Generate and execute remote control instructions based on obstacle information and information sharing platform information. Remote control instructions include avoidance, deceleration, pause and path replanning. Information sharing platform information includes environmental information and vehicle positioning data.

[0072] What needs to be further explained in the embodiments of the present invention is that building an information sharing platform includes entering environmental information in the target area, such as traffic signal information and road information, into the information sharing platform to build a simulation model of the target area; acquiring real-time vehicle positioning data through a vehicle-mounted radar, and uploading the real-time vehicle positioning data and corresponding trajectory information to the target area simulation model in real time based on the Internet of Things technology; and building an information sharing platform through visualization processing; preferably, the traffic signal information and road information are image data.

[0073] It needs to be further explained in the embodiment of the present invention that the method of verifying the reliability of the information sharing platform is:

[0074] Step 101: Collect environmental information, and record the collected environmental information as Ha; retrieve the environmental information corresponding to time and space from the information sharing platform, and record it as Hb;

[0075] Step 102: Output the shared information consistency index Yz by analyzing the similarity between the environmental information Ha and the environmental information Hb;

[0076] Step 103, by judging the relationship between the shared information consistency index Yz and the preset value, the reliability of the information sharing platform is judged. When the shared information consistency index Yz is lower than the preset value, it indicates that the information sharing platform is abnormal and in an unreliable state. Otherwise, it indicates that the information sharing platform is reliable.

[0077] In a possible real-time example, when the environmental information only includes image data, image similarity is used to represent the shared information consistency index. Based on the location of the focus point, the image is divided into several regions, and a weight coefficient is matched for each region. The consistency analysis model is obtained by weighted summation of the image similarity and weight coefficient of each region.

[0078] In a possible real-time example, when the environmental information only includes vehicle positioning data, the vehicle positioning data similarity is used to represent the shared information consistency index;

[0079] In a possible real-time example, when the environmental information includes image data and vehicle positioning data, the shared information consistency index is represented by the sum of the image similarity and the vehicle positioning data similarity.

[0080] See also Figure 2 A structural block diagram of an obstacle analysis model, wherein the obstacle analysis model includes an obstacle static feature analysis model, an obstacle dynamic feature analysis model and a feature fusion unit;

[0081] The obstacle static feature analysis model is used to analyze the image data in the real-time data and predict the static features of the obstacle, such as the obstacle position, obstacle size, and obstacle shape;

[0082] Explanation: The obstacle static feature analysis model includes an input layer, a feature extraction layer, and a classification and regression layer. The input layer is used to receive real-time image data and corresponding background image data as input; the feature extraction layer uses a convolutional neural network algorithm to extract features from the input image data, including edge, texture, color underlying features and semantic features (such as the shape and category of the object); the classification and regression layer classifies and regresses each pixel or area in the real-time image based on the extracted features. The classification operation is used to determine whether the pixel or area belongs to the obstacle category, and the regression operation is used to estimate the position and size parameters of the obstacle.

[0083] The obstacle dynamic feature analysis model obtains the obstacle position frame based on the obstacle static feature analysis model, and predicts the obstacle motion characteristics, such as the obstacle moving speed and moving direction, by analyzing the positioning data of the obstacle position frame;

[0084] The feature fusion unit is used to output obstacle information and fuse the static features of obstacles and the motion features of obstacles.

[0085] It needs to be further explained in the embodiment of the present invention that the method of verifying the reliability of the obstacle analysis model is:

[0086] Step 201, test the obstacle analysis model several times and record the test results;

[0087] Step 202, analyzing the test results to obtain the average value of processing time, the average value of real-time data feature information entropy and obstacle accuracy parameters, which are denoted as Za, Zb and Zc respectively;

[0088] Explanation: Processing time refers to the time required for the obstacle analysis model to complete recognition and respond from receiving image data. The processing time of multiple tests is calculated and averaged. The average processing time reflects the response speed of the obstacle analysis model. Information entropy is an indicator to measure data complexity and uncertainty. In obstacle analysis, the information entropy of real-time data features can reflect the complexity of the data processed by the obstacle analysis model. The information entropy of real-time data features is calculated and averaged for multiple tests to obtain the average information entropy of real-time data features. The obstacle accuracy parameter is a key indicator to measure the accuracy of the model in identifying obstacles. It is calculated by comparing the differences between the obstacles identified by the model and the actual obstacles.

[0089] Step 204: By formula The obstacle recognition performance index Xb is calculated, where the Form(·) function is used to normalize the content in the brackets so that the value range in the brackets is between 0 and 1; X1, X2, and X3 are used to correct the obstacle accuracy parameters, processing time loss rate, and feature information entropy to adjust the degree of influence on the obstacle recognition performance index.

[0090] It needs to be further explained in the embodiment of the present invention that the obstacle accuracy parameter is obtained in the following manner:

[0091] The difference between the predicted static features of the obstacle and the actual static features of the obstacle is analyzed to obtain the static feature loss parameter La; in a possible example, the static feature loss parameter La is represented by the cumulative value of the coordinate error mean square value, the IOU error mean square value and the classification error mean square value between the predicted data and the actual data;

[0092] Analyze the difference between the predicted obstacle motion characteristics and the actual obstacle motion characteristics to obtain the motion characteristic loss parameter Lb; in a possible real-time example, the motion characteristic loss parameter Lb is represented by the mean square value of the motion vector error between the predicted data and the actual data;

[0093] Combine the static feature loss parameter La and the motion feature loss parameter Lb, through the formula The obstacle accuracy parameter Za is calculated, where La_max represents the maximum value of the static feature loss parameter and Lb_max represents the maximum value of the motion feature loss parameter.

[0094] Explanation: When using the obstacle analysis model to predict obstacles, the predicted static features of the obstacles constitute a static feature set. There is an inconsistency between the predicted static features and the actually observed static features. The difference between the predicted static features and the actual static features of the obstacles is obtained by calculating a certain distance (such as Euclidean distance, Manhattan distance, etc.) or similarity (such as cosine similarity, Pearson correlation coefficient, etc.) between the predicted features and the actual features; the motion feature loss parameter Lb is used to quantify the degree of inconsistency between the predicted motion features and the actually observed motion features. The calculation method may be similar to that of the static feature loss parameter La, but it will be performed for motion features. The present invention does not specifically describe or limit this.

[0095] Real-time Example 2, the real-time example of the present invention provides a remote driving control system for automatic parking of a passenger boarding ladder, comprising:

[0096] The information sharing platform management module is used to build the information sharing platform and verify the reliability of the information sharing platform based on the shared information consistency index;

[0097] The obstacle analysis management module is used to build an obstacle analysis model and verify the reliability of the obstacle analysis model based on the obstacle recognition performance index;

[0098] Real-time image data acquisition module: when the vehicle-mounted radar detects an obstacle in the direction of the boarding ladder, the vehicle-mounted camera acquires real-time image data;

[0099] A background image data acquisition module retrieves image data corresponding to the time and space of the real-time image data from the verified information sharing platform as background image data;

[0100] The obstacle information acquisition module receives the real-time data and the corresponding background image data after verification, and outputs the obstacle information;

[0101] The remote control module generates and executes remote control instructions based on obstacle information and information sharing platform information.

[0102] It needs to be further explained in the embodiment of the present invention that the automatic parking remote driving control system for the passenger boarding ladder further includes:

[0103] The driving operation management module is used to obtain the reliability index of the automatic driving operation, and the automatic driving operation reliability index is obtained by comparing the actual response parameters and the expected response parameters of the boarding ladder under the operation command; the automatic driving operation reliability index is obtained in the following manner:

[0104] Get the instruction response time loss rate and instruction operation precision loss rate. Assume there are m operation instructions and j represents the sequence number of the operation instructions.

[0105] The average response time loss rate and the average operation accuracy loss rate of the jth operation instruction are recorded as st_j and sac_j respectively; the autonomous driving operation reliability index is calculated by the following formula:

[0106]

[0107] Among them, f1 and f2 represent the correction factors of time loss rate and precision loss rate respectively, which are used to adjust the influence of time loss rate and precision loss rate on the reliability index of autonomous driving operation.

[0108] It is necessary to further explain in the embodiment of the present invention that the automatic parking remote driving control system for the passenger boarding ladder also includes:

[0109] The monitoring and early warning module monitors the shared information consistency index Xa, obstacle recognition performance index Xb and autonomous driving operation reliability index Xc in real time. When the monitoring data is lower than the corresponding threshold, an early warning is issued; maintenance is performed based on the fault knowledge graph to eliminate anomalies;

[0110] The remote driving control performance analysis module jointly analyzes the shared information consistency index, obstacle recognition performance index and autonomous driving operation reliability index to obtain the remote driving control coefficient, and takes measures based on the remote driving control coefficient.

[0111] It needs to be further explained in the embodiment of the present invention that the remote driving control performance analysis module includes:

[0112] The thresholds of shared information consistency index, obstacle recognition performance index and autonomous driving operation reliability index are obtained, which are denoted as Tha, Thb and Thc respectively;

[0113] By the formula YK = e -max(Xa-Tha，0) *e -max(Xb-Thb，0) *e -max(Xc-Thc，0) The remote driving control coefficient YK is calculated;

[0114] When the remote driving control coefficient is lower than the corresponding threshold, an early warning is issued, indicating that the reliability of remote unmanned driving is low and more monitoring and maintenance resources need to be allocated;

[0115] When the remote driving control coefficient is not lower than the corresponding threshold, it indicates that the remote driving performance of the boarding ladder is normal and no measures are taken.

[0116] In summary, the embodiments of the present invention can achieve effective evaluation of the reliability of the remote unmanned driving system by setting the remote driving control coefficient and the corresponding threshold; according to the evaluation results, the system can automatically take corresponding measures to ensure the safety and reliability of the remote driving, which helps to improve the safety and efficiency of the remote unmanned driving system.

[0117] Finally: The above description is only a preferred example of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A remote driving control method for automatic parking of a passenger boarding ladder, characterized in that: The following steps are involved: Step 1: Build an information sharing platform and verify the reliability of the information sharing platform; build an obstacle analysis model and verify the reliability of the obstacle analysis model; Step 2: Install a real-time image acquisition device on the boarding ladder and collect real-time data through the vehicle-mounted radar, and connect the information sharing platform, real-time data and obstacle analysis model; Step 3: When the vehicle-mounted radar detects an obstacle in the direction of the boarding ladder, real-time image data is collected through the vehicle-mounted camera; Step 4: retrieve image data corresponding to the time and space of the real-time image data from the verified information sharing platform as background image data; Step 5: The verified obstacle analysis model receives real-time data and corresponding background image data, and outputs obstacle information; Obstacle information includes: the location, size, shape and motion state of the obstacle; Step 6: Generate and execute remote control instructions based on obstacle information and information sharing platform information.

2. The automatic parking remote driving control method for passenger boarding stairs according to claim 1 is characterized in that: Building an information sharing platform includes entering the environmental information in the target area into the information sharing platform to build a simulation model of the target area; obtaining the real-time positioning data of the vehicle through the vehicle-mounted radar, and uploading the real-time positioning data of the vehicle and the corresponding trajectory information to the simulation model of the target area in real time based on the Internet of Things technology; and building an information sharing platform after visualization processing.

3. The automatic parking remote driving control method for passenger boarding stairs according to claim 2 is characterized in that: The ways to verify the reliability of the information sharing platform are: Step 101: Collect environmental information, and record the collected environmental information as Ha; retrieve the environmental information corresponding to time and space from the information sharing platform, and record it as Hb; Step 102: Output the shared information consistency index Yz by analyzing the similarity between the environmental information Ha and the environmental information Hb; Step 103, by judging the relationship between the shared information consistency index Yz and the preset value, the reliability of the information sharing platform is judged. When the shared information consistency index Yz is lower than the preset value, it indicates that the information sharing platform is abnormal and in an unreliable state. Otherwise, it indicates that the information sharing platform is reliable.

4. The automatic parking remote driving control method for passenger boarding stairs according to claim 1 is characterized in that: The obstacle analysis model includes an obstacle static feature analysis model, an obstacle dynamic feature analysis model and a feature fusion unit; The obstacle static feature analysis model is used to analyze the image data in the real-time data, predict the static features of the obstacle, and present the obstacle position frame in the form of a bounding box and a confidence score; The obstacle dynamic feature analysis model obtains the obstacle position frame based on the obstacle static feature analysis model, and predicts the obstacle motion feature by analyzing the positioning data of the obstacle position frame. The feature fusion unit is used to output obstacle information and fuse the static features of obstacles and the motion features of obstacles.

5. The automatic parking remote driving control method for passenger boarding stairs according to claim 1 is characterized in that: The reliability of the obstacle analysis model is verified by: Step 201, test the obstacle analysis model several times and record the test results; Step 202, analyzing the test results to obtain the average value of processing time, the average value of real-time data feature information entropy and obstacle accuracy parameters, which are denoted as Za, Zb and Zc respectively; Step 204: By formula The obstacle recognition performance index Xb is calculated, where the Form(·) function is used to normalize the content in the brackets so that the value range in the brackets is between 0 and 1; X1, X2, and X3 are used to correct the obstacle accuracy parameters, processing time loss rate, and feature information entropy to adjust the degree of influence on the obstacle recognition performance index.

6. The automatic parking remote driving control method for passenger boarding stairs according to claim 5 is characterized in that: The obstacle accuracy parameter is obtained as follows: Analyze the difference between the predicted obstacle static characteristics and the actual obstacle static characteristics to obtain the static characteristic loss parameter La; Analyze the difference between the predicted obstacle motion characteristics and the actual obstacle motion characteristics to obtain the motion characteristic loss parameter Lb; Combine the static feature loss parameter La and the motion feature loss parameter Lb, through the formula The obstacle accuracy parameter Za is calculated, where La_max represents the maximum value of the static feature loss parameter and Lb_max represents the maximum value of the motion feature loss parameter.

7. A remote control system for automatic parking of a passenger boarding ladder, used to implement the remote control method for automatic parking of a passenger boarding ladder as described in claim 1, characterized in that: include: The information sharing platform management module is used to build the information sharing platform and verify the reliability of the information sharing platform based on the shared information consistency index; The obstacle analysis management module is used to build an obstacle analysis model and verify the reliability of the obstacle analysis model based on the obstacle recognition performance index; Real-time image data acquisition module: when the vehicle-mounted radar detects an obstacle in the direction of the boarding ladder, the vehicle-mounted camera acquires real-time image data; A background image data acquisition module retrieves image data corresponding to the time and space of the real-time image data from the verified information sharing platform as background image data; The obstacle information acquisition module receives the real-time data and the corresponding background image data after verification, and outputs the obstacle information; The remote control module generates and executes remote control instructions based on obstacle information and information sharing platform information.

8. The automatic parking remote driving control system for passenger boarding stairs according to claim 7 is characterized in that: The automatic parking remote driving control system for the passenger boarding ladder also includes: The driving operation management module is used to obtain the reliability index of the automatic driving operation, and the automatic driving operation reliability index is obtained by comparing the actual response parameters and the expected response parameters of the boarding ladder under the operation command; the automatic driving operation reliability index is obtained in the following manner: Get the instruction response time loss rate and instruction operation precision loss rate. Assume there are m operation instructions and j represents the sequence number of the operation instructions. The average response time loss rate and the average operation accuracy loss rate of the jth operation instruction are recorded as st_j and sac_j respectively; the autonomous driving operation reliability index is calculated by the following formula: Among them, f1 and f2 represent the correction factors of time loss rate and precision loss rate respectively, which are used to adjust the influence of time loss rate and precision loss rate on the reliability index of autonomous driving operation.

9. The automatic parking remote driving control system for passenger boarding stairs according to claim 8, characterized in that: The automatic parking remote driving control system for the passenger boarding ladder also includes The monitoring and early warning module monitors the shared information consistency index Xa, obstacle recognition performance index Xb and autonomous driving operation reliability index Xc in real time, and issues an early warning when the monitoring data is lower than the corresponding threshold; Perform maintenance based on the fault knowledge graph to eliminate anomalies; The remote driving control performance analysis module jointly analyzes the shared information consistency index, obstacle recognition performance index and autonomous driving operation reliability index to obtain the remote driving control coefficient, and takes measures based on the remote driving control coefficient.

10. The automatic parking remote driving control system for passenger boarding stairs according to claim 9, characterized in that: The remote driving control performance analysis module includes: The thresholds of shared information consistency index, obstacle recognition performance index and autonomous driving operation reliability index are obtained, which are denoted as Tha, Thb and Thc respectively; By the formula YK = e -max(Xa-Tha，0) *e -max(Xb-Thb，0) *e -max(Xc-Thc，0) The remote driving control coefficient YK is calculated; When the remote driving control coefficient is lower than the corresponding threshold, an early warning is issued, indicating that the reliability of remote unmanned driving is low and more monitoring and maintenance resources need to be allocated; when the remote driving control coefficient is not lower than the corresponding threshold, it indicates that the remote driving performance of the boarding ladder is normal and no measures are taken.

Images