Early warning processing method for controlling airport unmanned bus based on cloud platform

Abstract

The invention relates to an early warning processing method for controlling an airport unmanned bus based on a cloud platform, and belongs to the technical field of intelligent transportation, and the method comprises the steps: S1, sensing vehicle driving data through a sensor, transmitting the data to an automatic driving system, and synchronizing the data to the cloud platform; s2, after the cloud platform carries out noise processing on the received data, anomaly detection is carried out; s3, when an abnormal result is detected, identifying an abnormal type of the vehicle; s4, flight information is obtained, and traffic influence grade division is carried out on the airport ground according to the flight information; s5, generating a pre-scheduling scheme based on the traffic influence level, the vehicle state and the road condition for the remote scheduling abnormity; and S6, generating an exception handling personnel scheduling scheme for exceptions which cannot be remotely dispatched. According to the invention, the operation state of the unmanned bus is monitored in real time, the abnormity of the unmanned bus is handled in time, the operation efficiency of the whole airport is improved, and the delay of airplanes is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation, and particularly to a method for warning and processing of an airport driverless minibus based on cloud platform control. Background Art

[0002] With the rapid development of artificial intelligence and autonomous driving technologies, driverless vehicles have emerged in more and more scenarios. With the maturity of autonomous driving buses, in the airport aspect, the application of driverless technologies has also started to achieve unmanned pick-up and drop-off of passengers. However, with the development of the global aviation industry, the number of airport flights has increased year by year, resulting in a continuous growth of airport ground traffic flow. In such a complex and changeable environment as the airport, with a dense population and a large number of flight activities, the operation safety of driverless minibuses is particularly important. While realizing intelligent transportation at the airport by driverless minibuses, it also brings challenges to airport safety and airport operation efficiency. In order to ensure the safety of driverless minibuses during operation at the airport, a complete set of emergency warning methods needs to be established so as to be able to quickly respond when potential dangers occur and avoid accidents; at the same time, the operation efficiency of the airport should be ensured and the delay of airport aircraft should be reduced. Summary of the Invention

[0003] To solve the above-mentioned problems of the prior art, the present invention provides a method for warning and processing of an airport driverless minibus based on cloud platform control.

[0004] A method for warning and processing of an airport driverless minibus based on cloud platform control includes the following steps:

[0005] S1: Perceive vehicle driving data through sensors, send the data to the autonomous driving system, and synchronize it to the cloud platform;

[0006] S2: After performing noise processing on the received data by the cloud platform, perform anomaly detection;

[0007] S3: When an abnormal result is detected, identify the vehicle abnormal type and prompt specific warning information; the vehicle abnormal type includes two types: remotely schedulable anomaly and non-remotely schedulable.

[0008] S4: Obtain flight information and divide the influence level of ground passage at the airport according to the flight information;

[0009] S5: For remotely schedulable anomalies, generate a pre-scheduling plan based on the influence level of passage, vehicle status, and road conditions;

[0010] S6: For the exception of non-remote schedulability, determine the traffic impact level of the vehicle's location, determine the priority of emergency handling according to the traffic impact level and the vehicle status, determine the reward function according to the priority and handling time, and construct a reinforcement learning model. Generate an abnormal handling personnel scheduling plan by training the reinforcement learning model.

[0011] Further, the sensors include lidar, cameras, millimeter-wave radars, GPS, and IMU measurement units.

[0012] Further, data communication between the sensors and the cloud platform is carried out through a CPE network switch.

[0013] Further, the noise processing includes noise processing through an extended Kalman filter algorithm.

[0014] Further, the anomaly detection is: performing anomaly detection on the data through a statistics-based Z-Score anomaly detection method or a density-based anomaly detection method.

[0015] Further, the flight information includes the estimated departure time, landing time, flight type, and runway for travel.

[0016] Further, the specific traffic impact level division of the airport ground according to the flight information is as follows:

[0017] Divide the airport ground into regions according to the runway of the aircraft;

[0018] Obtain the current system time, judge the difference between the current system time and the aircraft departure time or landing time, and set the traffic impact level of the region according to the size of the time difference.

[0019] The beneficial effects of the present invention: The present invention monitors the running state of the driverless minibus in real time; through the centralized management method of all vehicle data, it can make a quick response in case of emergency. When potential risks or emergencies are detected, the cloud platform can remotely control the driverless minibus, and timely take safety measures such as deceleration, parking, or path re-planning, and improve the operation efficiency of the entire airport and reduce flight delays through the timely handling of the anomalies of the driverless minibus. Description of the Drawings

[0020] Figure 1 Is the method flow chart of the present invention;

[0021] Figure 2 Is the overall architecture diagram of the early warning processing of the embodiment;

[0022] Figure 3 Is the data flow chart of the extended Kalman filter for optimizing GPS / IMU / radar / camera data. Detailed Implementation Manner

[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0024] Embodiment 1, referring to Figure 1 and Figure 2 , a method for warning processing of an airport unmanned minibus based on cloud platform control, including the following steps:

[0025] S1: Sense vehicle driving data through sensors, send the data to the autonomous driving system, and synchronize it to the cloud platform;

[0026] S2: After the cloud platform processes the received data for noise, perform anomaly detection;

[0027] S3: When an abnormal result is detected, identify the vehicle abnormal type and prompt specific warning information; the vehicle abnormal types include remotely schedulable anomalies and non-remotely schedulable types;

[0028] S4: Obtain flight information and divide the influence level of ground passage at the airport according to the flight information;

[0029] S5: For remotely schedulable anomalies, generate a pre-scheduling plan based on the passage influence level, vehicle status, and road conditions;

[0030] S6: For non-remotely schedulable anomalies, judge the passage influence level of the vehicle's location, determine the priority of emergency handling according to the passage influence level and vehicle status, determine the reward function according to the priority and handling time, and construct a reinforcement learning model, and generate a dispatching plan for abnormal handling personnel by training the reinforcement learning model.

[0031] Among them, the sensors include lidar, cameras, millimeter-wave radars, GPS, and IMU measurement units. The acquired perception data is transmitted to the autonomous driving system and the CPE network switch in real time. The autonomous driving system modularly processes the perception data to generate a perception model and participates in the driving planning of the unmanned minibus. Another copy of the perception data is uploaded to the network switch, and the switch transmits the data to the cloud platform, and its architecture is as Figure 2 shown.

[0032] In this embodiment, noise processing includes noise processing through the extended Kalman filter algorithm. The cloud platform receives perception data. Since perception data usually contains noise and errors (such as multipath effects, etc.), and the extended Kalman filter can fuse perception data and other sensor data (such as accelerometers, gyroscopes, cameras, lidar, etc.) to estimate the state (position, speed, direction, etc.) of the unmanned minibus, solve the noise problem in perception data, and improve the perception prediction accuracy. The extended Kalman filter optimizes the GPS / IMU / radar / camera data process, and the architecture is as shown in Figure 3 shown; the processing process is as follows:

[0033] (1) State equation

[0034] Establish a system state equation to describe the evolution process of the system state over time. For GPS data processing, the state vector usually includes information such as position and speed:

[0035] x k+1 = f(x k , u k ) + w k ,

[0036] where f(x k , u k ) is a non-linear function, u k is the control input, and w k is the process noise.

[0037] Observation equation: Describe the relationship between the observed data and the system state. The observation equation can be expressed as:

[0038] z k = h(x k ) + v k ,

[0039] where H(x) is a non-linear function, and V K is the observation noise.

[0040] (2) Initialization

[0041] State estimation: According to the initial GPS measurement data, initialize the state vector Initial position and speed.

[0042] Covariance matrix: Initialize the covariance matrix P0 of the state estimation, which represents the uncertainty of the initial state estimation.

[0043] (3) Prediction

[0044] State prediction: Use the state equation to predict the state at the next moment:

[0045]

[0046] Covariance prediction: Update the covariance matrix of the state estimate:

[0047]

[0048] where F k is the Jacobian matrix of the state transition matrix, and Q k is the process noise covariance matrix.

[0049] (4) Update

[0050] Linearization: Linearize the observation equation around the current state estimate and calculate the Jacobian matrix H k :

[0051]

[0052] Calculate the Kalman gain:

[0053]

[0054] where R k is the observation noise covariance matrix;

[0055] State update:

[0056] Covariance update: P k+1|k+1 = (I - K k+1 H k ) P k+1|k ;

[0057] Repeat the prediction and update steps, and continuously use new GPS data to estimate and correct the state.

[0058] In this embodiment, for anomaly detection: Anomaly detection is performed on the data by using a statistical Z-Score anomaly detection method or a density-based anomaly detection method.

[0059] 1) Statistical anomaly detection method (Z-Score method), which identifies data points in the dataset that deviate significantly from the mean. It determines the anomaly points by calculating the standard deviation distance of each data point from the data mean.

[0060] Z-Score (standard score) represents the degree of deviation between a data point and the data mean, with the unit of standard deviation. Its formula is:

[0061]

[0062] Note: x: The value of the data point. μ: The mean of the data. σ: The standard deviation of the data.

[0063] Identifying abnormal points: If the absolute value of the Z-Score value of a data point is relatively large (usually greater than 3 or less than -3), then this point is considered an abnormal point.

[0064] 2) Density-based outlier detection algorithm LOF (Local Outlier Factor), which identifies points in the dataset with significantly lower local density than their neighborhood density. Different from the global outlier detection method, LOF can detect local outliers, that is, points with lower density in certain regions may be considered outliers.

[0065] The reachability distance from data point p to point o is defined as:

[0066] Reachability-Distance k (p,o) = max(k-distance(o), Distance(p,o)),

[0067] This ensures that the distance will not be less than k-distance(o), thus avoiding too small values in density calculation;

[0068] Local Reachability Density (LRD). The local reachability density of data point p is defined as the reciprocal average of the reachability distances of all points in its k-neighborhood:

[0069]

[0070] LRD represents the local density of point p. The higher the density, the larger the LRD value;

[0071] Local Outlier Factor (LOF). The LOF value of data point p is defined as the ratio of the local reachability density of other points in its neighborhood to its own local reachability density:

[0072]

[0073] If LOFk(p) ≈ 1, then the density of point p is similar to the density of its neighborhood, and it belongs to a normal point.

[0074] If LOFk(p) >> 1, then the density of point p is significantly lower than the density of its neighborhood, and it may be an outlier.

[0075] In this embodiment, the flight information includes the estimated departure time, arrival time, flight type, and runway used.

[0076] Among them, the specific classification of the impact level on the airport ground according to flight information is as follows:

[0077] Divide the airport ground into areas according to the runway of the aircraft;

[0078] Obtain the current system time, judge the difference between the current system time and the take-off time or landing time of the aircraft, and set the traffic impact level of the area according to the size of the time difference.

[0079] In this embodiment, the cloud platform is used as the centralized control center, which can monitor the running status of the driverless minibus in real time, including key parameters such as position, speed, and battery power. This centralized management method enables quick responses in case of emergencies. When potential risks or emergencies are detected, the cloud platform can remotely control the driverless minibus to take safety measures such as decelerating and stopping, and even manual control can be performed by a remote operator to ensure safety.

[0080] In the description of the embodiments of the present invention, the terms "first", "second", "third", and "fourth" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first", "second", "third", and "fourth" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise stated, the meaning of "a plurality" is two or more.

[0081] In the description of the embodiments of the present invention, specific features, structures, materials, or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0082] In the description of the embodiments of the present invention, it should be understood that "-" and "~" represent the range between two numerical values, and this range includes the endpoints. For example: "A - B" represents a range greater than or equal to A and less than or equal to B. "A ~ B" represents a range greater than or equal to A and less than or equal to B.

[0083] In the description of the embodiments of the present invention, the term "and / or" herein is merely a description of the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0084] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A warning processing method for controlling an airport unmanned minibus based on a cloud platform, characterized in that, It includes the following steps: S1: Sense the vehicle driving data through sensors, send the data to the autonomous driving system, and synchronize it to the cloud platform; S2: After the cloud platform processes the received data for noise, perform anomaly detection; S3: When an abnormal result is detected, identify the vehicle abnormal type and prompt specific warning information; the vehicle abnormal type includes two types: remotely schedulable anomaly and non-remotely schedulable; S4: Obtain flight information and divide the ground of the airport according to the impact level of the flight information on the ground passage; S5: For remotely schedulable anomalies, generate a pre-scheduling plan based on the impact level of passage, vehicle status, and road conditions; S6: For non-remotely schedulable anomalies, judge the impact level of passage at the location where the vehicle is located, determine the priority of emergency handling according to the impact level of passage and vehicle status, determine the reward function according to the priority and processing time, and construct a reinforcement learning model, and generate a dispatching plan for anomaly handlers by training the reinforcement learning model.

2. The warning processing method for an airport unmanned minibus based on cloud platform control according to claim 1, wherein, The sensors include lidar, cameras, millimeter-wave radars, GPS, and IMU measurement units.

3. The early warning processing method for controlling an airport driverless minibus based on a cloud platform according to claim 1, characterized in that, Data communication between the sensors and the cloud platform is carried out through a CPE network switch.

4. A warning processing method for controlling an airport driverless minibus based on a cloud platform according to claim 1, characterized in that, The noise processing includes noise processing through an extended Kalman filter algorithm.

5. The early warning processing method for controlling an airport driverless minibus based on a cloud platform according to claim 1, wherein The anomaly detection is: perform anomaly detection on the data through a Z-Score anomaly detection method based on statistics or a density-based anomaly detection method.

6. The early warning processing method for controlling an unmanned airport minibus based on a cloud platform according to claim 1, wherein, The flight information includes the expected takeoff time, landing time, flight type, and runway for travel.

7. A warning processing method for controlling an airport unmanned minibus based on a cloud platform according to claim 1, characterized in that The specific division of the impact level of the ground passage of the airport according to the flight information is as follows: Divide the ground of the airport according to the runway of the aircraft; Obtain the current system time, judge the difference between the current system time and the takeoff time or landing time of the aircraft, and set the impact level of passage of the area according to the size of the time difference.

Images