Lane line anomaly detection method, device, equipment, medium, product and vehicle

Abstract

The invention relates to a lane line anomaly detection method, device and equipment, a medium, a product and a vehicle, and the method comprises the steps: building a state space model of a target lane line according to a lane line signal of the target lane line collected in real time after the vehicle starts an intelligent driving function or an automatic driving function, and then obtaining a state space model of the target lane line; generating a state estimation value and a first covariance of the target lane line at the first moment according to the state space model and a Kalman filter, and then generating a state recursion value and a second covariance of the target lane line at the first moment through a state recursion device; and finally, according to the state estimation value, the state recursion value, the first covariance and the second covariance corresponding to the target lane line at the first moment, performing anomaly detection on the target lane line, and generating an anomaly detection result. Wherein the target lane line is the lane line of the lane where the vehicle is located, and the anomaly detection result is used for indicating whether the target lane line is abnormal. According to the invention, the accuracy and robustness of anomaly detection are improved.

Description

Technical Field

[0001] The present invention relates to the field of vehicle technology, and in particular to a lane line anomaly detection method, device, equipment, medium, product and vehicle. Background Art

[0002] During the process of intelligent driving or autonomous driving, in addition to real-time detection of lane lines, lane line abnormality detection is also required to prevent problems such as continuing to control the vehicle's driving when the lane line is abnormal, causing the vehicle to deviate from the actual lane and collide with other vehicles or violate traffic regulations.

[0003] At present, lane line anomaly detection is mainly carried out by collecting lane line signals through on-board cameras or radars, determining the two lane lines of the current lane where the vehicle is located based on the lane line signals, and determining the position points of the two lane lines closest to the vehicle, and then determining the distance between each position point and the vehicle, comparing the sum of these two distances with the preset distance, and finally determining whether there is an abnormality in the lane line.

[0004] However, the existing technology has low accuracy in detecting lane line anomalies and has problems of missed alarms and false alarms. Summary of the invention

[0005] The purpose of the present invention is to provide a lane line anomaly detection method, device, equipment, medium, product and vehicle to solve the problem of low accuracy of lane line anomaly detection in the prior art.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A lane line anomaly detection method, comprising:

[0008] After the vehicle turns on the intelligent driving function or the automatic driving function, a state space model of the target lane line is established according to the lane line signal of the target lane line collected in real time, and the target lane line is the lane line of the lane where the vehicle is located;

[0009] Generate a state estimation value and a first covariance of the target lane line at a first moment according to the state space model and the Kalman filter;

[0010] Generate a state recursion value and a second covariance of the target lane line at the first moment through a state recursion device;

[0011] According to the state estimation value, the state recursive value, the first covariance and the second covariance corresponding to the target lane line at the first moment, the target lane line is checked for abnormality to generate the abnormality detection result, and the abnormality detection result is used to indicate whether there is an abnormality in the target lane line.

[0012] According to the above technical means, the outputs of the Kalman filter and the state filter are combined to detect anomalies of the target lane line by judging whether the two are consistent. Since there is no need to manually set parameters, the accuracy and robustness of anomaly detection can be effectively improved.

[0013] Further, the state recursion value and the second covariance of the target lane line at the first moment are generated by the state recursor, including:

[0014] Determine, according to a preset duration and a duration for which each sub-state recurser has been reset, a first sub-state recurser that needs to be reset at the first moment, the first sub-state recurser belonging to the state recurser;

[0015] Resetting the first sub-state recursor according to the state estimation value and the first covariance of the target lane line at the first moment;

[0016] According to the state recursion value and the second covariance generated by the second sub-state recurser at the second moment, the state recursion value and the second covariance of the target lane line at the first moment are generated by the second sub-state recurser, the second sub-state recurser belongs to the state recurser and is not reset at the first moment, and the second moment is earlier than the first moment.

[0017] According to the above technical means, multiple sub-state recursors are reset alternately through the Kalman filter, and the sub-state recursors that have not been reset continue to perform state recursion according to the previously reset values, which effectively reduces the accumulated error of the state recursor and improves the accuracy of subsequent abnormality detection of the target lane line.

[0018] Further, resetting the first sub-state recursor according to the state estimation value and the first covariance of the target lane line at the first moment includes:

[0019] The state recursion value generated by the first sub-state recurser at the first moment is reset to the state estimation value at the first moment, and the second covariance generated by the first sub-state recurser at the first moment is reset to the first covariance at the first moment.

[0020] According to the above technical means, the output of the Kalman filter is assigned to the first sub-state recurser, and the first sub-state recurser is initialized, thereby eliminating the accumulated error of the first sub-state recurser due to time reasons, and effectively improving the accuracy of subsequent lane line anomaly detection based on the output of the Kalman filter and the output of the state recurser.

[0021] Further, the state space model includes a state equation and an observation equation. The state space model of the target lane line is established according to the lane line signal of the target lane line collected in real time, including:

[0022] According to the lane line signal of the target lane line collected in real time, discrete system modeling is performed on the target lane line to generate the state equation;

[0023] The observation equation of the target lane line is generated according to the lane line signal of the target lane line collected in real time.

[0024] According to the above technical means, the state equation describes the change of lane line state over time, that is, how the target lane line transitions from the current state to the next state. The observation equation describes how to obtain observable output from the lane line state, so that the lane line state can be obtained from actual observations. By establishing a lane line containing state equations and observation equations for the target lane line, a theoretical basis is provided for the subsequent Kalman filter processing process.

[0025] Furthermore, the state space model includes at least one of the following states: the position, relative heading, relative curvature and relative curvature differential of the target lane line relative to the vehicle.

[0026] According to the above technical means, the target lane line is described from multiple aspects, so that the lane line can be subsequently detected for abnormalities from multiple aspects, which effectively improves the accuracy of detection.

[0027] Further, performing an abnormality check on the target lane line according to the state estimation value, the state recursive value, the first covariance, and the second covariance corresponding to the target lane line at the first moment to generate the abnormality detection result includes:

[0028] Determine a state difference according to a first preset anomaly detection function and the state estimation value and the state recursive value corresponding to the target lane line at the first moment, wherein the first preset anomaly detection function obeys a Gaussian distribution;

[0029] Calculating the variance of the first preset anomaly detection function according to the first covariance and the second covariance corresponding to the target lane line at the first moment;

[0030] Determining a target chi-square value according to the state difference, the variance, and a second preset anomaly detection function corresponding to the target lane line at the first moment, wherein the second preset anomaly detection function satisfies a chi-square distribution;

[0031] According to the target chi-square value and the preset chi-square value, a chi-square test is performed on the target lane line to generate the anomaly detection result.

[0032] According to the above technical means, by performing a chi-square test on the output of the Kalman filter and the output of the state recurser, the process is simple and the result is intuitive, which can effectively improve the efficiency of lane line anomaly detection.

[0033] Further, determining a target chi-square value according to the state difference, the variance, and a second preset anomaly detection function corresponding to the target lane line at the first moment includes:

[0034] According to β (k) = β T (k)B -1 (k)β(k), calculate the target chi-square value;

[0035] Among them, ξ β (k) is the target chi-square value, β(k) is the state difference, B(k) is the variance, and n is the dimension of β(k).

[0036] A lane line abnormality detection device, comprising:

[0037] An establishment module is used to establish a state space model of a target lane line according to a lane line signal of a target lane line collected in real time after the vehicle turns on the intelligent driving function or the automatic driving function, wherein the target lane line is the lane line of the lane where the vehicle is located;

[0038] A first generating module, used for generating a state estimation value and a first covariance of the target lane line at a first moment according to the state space model and the Kalman filter;

[0039] A second generating module, configured to generate a state recursive value and a second covariance of the target lane line at the first moment through a state recursive device;

[0040] A verification module is used to perform an abnormality check on the target lane line according to the state estimation value, the state recursive value, the first covariance and the second covariance corresponding to the target lane line at the first moment, and generate the abnormality detection result, wherein the abnormality detection result is used to indicate whether there is an abnormality in the target lane line.

[0041] Furthermore, the second generating module is specifically used for:

[0042] Determine, according to a preset duration and a duration for which each sub-state recurser has been reset, a first sub-state recurser that needs to be reset at the first moment, the first sub-state recurser belonging to the state recurser;

[0043] Resetting the first sub-state recursor according to the state estimation value and the first covariance of the target lane line at the first moment;

[0044] According to the state recursion value and the second covariance generated by the second sub-state recurser at the second moment, the state recursion value and the second covariance of the target lane line at the first moment are generated by the second sub-state recurser, the second sub-state recurser belongs to the state recurser and is not reset at the first moment, and the second moment is earlier than the first moment.

[0045] Furthermore, the state space model includes a state equation and an observation equation, and the establishment module is specifically used for:

[0046] According to the lane line signal of the target lane line collected in real time, discrete system modeling is performed on the target lane line to generate the state equation;

[0047] The observation equation of the target lane line is generated according to the lane line signal of the target lane line collected in real time.

[0048] Furthermore, the state space model includes at least one of the following states: the position, relative heading, relative curvature and relative curvature differential of the target lane line relative to the vehicle.

[0049] Furthermore, the inspection module is specifically used for:

[0050] Determine a state difference according to a first preset anomaly detection function and the state estimation value and the state recursive value corresponding to the target lane line at the first moment, wherein the first preset anomaly detection function obeys a Gaussian distribution;

[0051] Calculating the variance of the first preset anomaly detection function according to the first covariance and the second covariance corresponding to the target lane line at the first moment;

[0052] Determining a target chi-square value according to the state difference, the variance, and a second preset anomaly detection function corresponding to the target lane line at the first moment, wherein the second preset anomaly detection function satisfies a chi-square distribution;

[0053] According to the target chi-square value and the preset chi-square value, a chi-square test is performed on the target lane line to generate the anomaly detection result.

[0054] Furthermore, the inspection module is specifically used for:

[0055] According to β (k) = β T (k)B -1 (k)β(k), calculate the target chi-square value;

[0056] Among them, ξ β (k) is the target chi-square value, β(k) is the state difference, B(k) is the variance, and n is the dimension of β(k).

[0057] An electronic device includes: a processor, a memory, and computer-executable instructions stored in the memory and executable on the processor, wherein the processor is used to implement the above-mentioned lane line abnormality detection method when executing the computer-executable instructions.

[0058] A computer-readable storage medium stores computer-executable instructions, which are used to implement the above-mentioned lane line anomaly detection method when executed by a processor.

[0059] A computer program product includes a computer program, and when the computer program is executed by a processor, it is used to implement the above lane line anomaly detection method.

[0060] A vehicle comprises the above electronic device.

[0061] Beneficial effects of the present invention:

[0062] (1) Since the state recursor is only related to the output of the state recursor at the previous moment, and the output of the Kalman filter is related to the actual lane line signal, the technical solution combines the outputs of the Kalman filter and the state filter, and performs anomaly detection on the target lane line by judging whether the two are consistent. Since there is no need to manually set parameters, the accuracy and robustness of anomaly detection can be effectively improved.

[0063] (2) The Kalman filter is used to reset multiple sub-state recursors in turn. The sub-state recursors that have not been reset continue to perform state recursion according to the previously reset values, which effectively reduces the accumulated error of the state recursors and improves the accuracy of subsequent abnormality detection of the target lane line. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of a process flow of a first embodiment of a lane line abnormality detection method provided by an embodiment of the present invention;

[0065] Figure 2 A schematic diagram of a flow chart of a second embodiment of a lane line abnormality detection method provided by an embodiment of the present invention;

[0066] Figure 3 A schematic diagram of a process flow of a third embodiment of a lane line abnormality detection method provided by an embodiment of the present invention;

[0067] Figure 4A schematic diagram of a process flow of a fourth embodiment of a lane line abnormality detection method provided by an embodiment of the present invention;

[0068] Figure 5 A schematic diagram of the structure of a lane abnormality detection device provided by an embodiment of the present invention;

[0069] Figure 6 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0071] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards, and corresponding operation entrances shall be provided for users to choose to authorize or refuse.

[0072] Before introducing the present invention, the application background of the present invention is first explained.

[0073] With the rapid development of science and technology, intelligent driving technology and autonomous driving technology are becoming increasingly mature. Intelligent driving systems and autonomous driving systems integrate various types of sensors, complex algorithms and powerful computing power to analyze road conditions in real time and make corresponding decisions to assist users in driving vehicles, effectively improving road safety, reducing traffic congestion, and providing great convenience for users.

[0074] In practical applications, the realization of intelligent driving and autonomous driving is inseparable from the accurate perception of the road environment. Since lane lines are an important part of road traffic signs, they divide lanes by drawing lines on the road surface, indicating the direction and position of vehicles, and providing other traffic information. Therefore, lane line detection is a key link in environmental perception.

[0075] Lane line detection mainly collects lane line images through on-board cameras or radars, then identifies lane lines from the lane line images, and controls vehicle driving according to the identified lane lines to achieve purposes such as lane keeping and lane departure warning.

[0076] It can be seen that the safety of vehicle driving is closely related to the accuracy of lane detection. However, lane detection faces many challenges, such as complex road conditions, natural wear and tear, bad weather, road construction, occlusion, sensor damage, lighting changes, semantic ambiguity, etc. In addition, the lane detection algorithm needs to meet high real-time requirements and run efficiently on limited hardware resources, which greatly increases the difficulty of lane detection.

[0077] When lane line detection is abnormal, the vehicle will make wrong decisions based on the wrong detection results, and control the vehicle incorrectly, causing the vehicle to deviate from the actual lane, or even collide with other vehicles, causing traffic accidents, etc. Therefore, it is of great significance to detect lane line abnormalities and then promptly remind the driver to take over when lane line abnormalities are detected to avoid harmful events.

[0078] At present, lane line abnormality detection is mainly carried out by collecting lane line signals through on-board cameras or radars, determining the two lane lines of the current lane where the vehicle is located based on the lane line signals, and determining the position points of the two lane lines closest to the vehicle, and then determining the distance between each position point and the vehicle, and comparing the sum of these two distances with the preset distance to finally determine whether the lane line is abnormal. Specifically, if the sum of the distances is greater than the preset distance, it is considered that the lane line is abnormal; if the sum of the distances is less than or equal to the preset distance, it is considered that the lane line is not abnormal.

[0079] However, since the preset distance is manually set in advance, there is no guarantee that the preset distance can meet all scenarios. This will result in a problem of missed alarms if the preset distance is set too large; if the preset distance is set too small, when lane line abnormality detection is performed based on some lane line signals that have large fluctuations but are actually still within the normal range, the normal lane line will be mistakenly determined as an abnormal lane line, resulting in a false alarm problem. Furthermore, if the sensor finds a fault, resulting in the collected lane line signal being inconsistent with the actual situation, when processing the wrong lane line, it will also cause the actual correct lane line to be determined as an abnormal lane line, and there will also be a false alarm problem.

[0080] In summary, the existing technology has the problems of low accuracy and poor robustness when performing abnormal detection on lane lines.

[0081] Based on the above technical problems, the present invention provides a lane line anomaly detection method. When a vehicle is traveling in a current lane, it can collect the lane line signal of the target lane line of the current lane in real time, and then create a state space model based on the collected lane line signal. Considering that the output of the Kalman filter is the optimal estimate of the system state, it combines the prediction of the system's state space model and the actual lane line signal. The output of the state recursor predicts the state of the current time step based on the state estimate of the previous time step, without correction of the measurement data. If there is no abnormality in the lane line, the difference between the output of the Kalman filter and the output of the state recursor is small. Conversely, if there is an abnormality in the lane line, the difference between the output of the Kalman filter and the output of the state recursor is large. Therefore, the output of the Kalman filter and the state recursor can be compared to realize lane line anomaly detection, thereby solving the problem of low accuracy and poor robustness due to the inability to ensure the rigor of the manually set preset distance.

[0082] The technical solution of the present invention is described in detail below through specific embodiments.

[0083] It should be noted that the following specific embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0084] Figure 1 The flowchart of the lane line abnormality detection method embodiment 1 provided by the embodiment of the present invention is as follows. Figure 1 As shown, the lane line anomaly detection method may include the following steps:

[0085] S11. After the vehicle turns on the intelligent driving function or the automatic driving function, a state space model of the target lane line is established according to the lane line signal of the target lane line collected in real time.

[0086] The execution subject of the embodiment of the present invention is an electronic device, which can be a terminal device, such as a vehicle-mounted terminal set in a vehicle with an intelligent driving function or an automatic driving function, or a server. In practical applications, whether the electronic device is a terminal device or a server can be determined according to actual conditions, and there is no specific limitation on this.

[0087] In actual applications, the vehicle is provided with a control for controlling the switch of the intelligent driving function or the automatic driving function. When the driver wants to turn on the intelligent driving function or the automatic driving function, he can click on the control so that the vehicle responds to the driver's click operation and controls the vehicle to turn on the intelligent driving function or the automatic driving function. After the vehicle turns on the intelligent driving function or the automatic driving function, it is necessary to perform lane line detection and lane line anomaly detection in real time so that the vehicle can be controlled according to the detection results. When detecting anomalies in the lane line, it is necessary to collect the lane line signal of the target lane line in real time, and then establish a state space model of the target lane line, so that the state estimate and the first covariance at each moment can be updated through the Kalman filter according to the established state space model.

[0088] Among them, the target lane line is the lane line of the lane where the vehicle is located.

[0089] The lane line signal may be collected by a sensor (eg, radar or a vehicle-mounted camera) disposed on the vehicle, and the sensor is used to collect road information in front of the vehicle.

[0090] In a possible implementation, the state space model includes a state equation and an observation equation. According to the lane line signal of the target lane line collected in real time, a discrete system modeling of the target lane line can be performed to generate a state equation. According to the lane line signal of the target lane line collected in real time, an observation equation of the target lane line is generated.

[0091] Specifically, in this implementation, the state equation can be expressed by the following formula:

[0092]

[0093] Among them, x k (k-1) is the position state of the target lane line relative to the vehicle at the (k-1)th moment, x2(k-1) is the heading of the target lane line relative to the vehicle at the (k-1)th moment, x3(k-1) is the curvature of the lane line relative to the vehicle at the (k-1)th moment, and x4(k-1) is the differential curvature of the lane line relative to the vehicle at the (k-1)th moment.

[0094] The observation equation can be expressed by the following formula:

[0095]

[0096] Among them, z0(k) is the position state of the target lane line relative to the vehicle measured by the sensor at time k, z1(k) is the heading of the target lane line relative to the vehicle measured by the sensor at time k, z2(k) is the curvature of the target lane line relative to the vehicle measured by the sensor at time k, and z3(k) is the differential curvature of the target lane line relative to the vehicle measured by the sensor at time k.

[0097] The relationship between the state equation and the observation equation can be expressed by the following formula:

[0098] z(k)=Hx k (k)+V(k)

[0099] Among them, H is the preset observation matrix, V(k)∈M m is the preset observation noise, the mean of the observation noise is 0, the variance matrix is ​​M, and m is the dimension of the variance matrix.

[0100] Furthermore, M is determined by calibrating the parameters output by the sensor and the parameters of the actual test. For example, in the test phase before the vehicle is put into production, the state parameters of the test lane line are collected by the sensor, and the state parameters of the test lane line are actually measured using a measuring tool, and then M can be calibrated according to the state parameters collected by the sensor and the state parameters actually measured.

[0101] This implementation method describes the target lane line through state equations and observation equations. The state equation describes the change of the lane line state over time, that is, how the target lane line transitions from the current state to the next state. The observation equation describes how to obtain observable output from the lane line state, so that the lane line state can be obtained from actual observations. By establishing a lane line containing state equations and observation equations for the target lane line, a theoretical basis is provided for the subsequent Kalman filter processing process.

[0102] Optionally, the state space model includes at least one of the following states (ie, lane line state): the position, relative heading, relative curvature, and relative curvature differential of the target lane line relative to the vehicle.

[0103] Through the above states, the target lane line is described from multiple aspects, so that the lane line can be subsequently detected for abnormalities from multiple aspects, effectively improving the accuracy of detection.

[0104] S12: Generate a state estimation value and a first covariance of the target lane line at the first moment according to the state space model and the Kalman filter.

[0105] The first moment may be the current moment or a moment before the current moment. That is, the lane line signal may be collected in real time, and the target lane line may be detected for abnormality in real time according to the lane line signal; or the lane line may be detected for abnormality according to the lane line signal collected at the previous moment during the vehicle driving process.

[0106] In a possible implementation, the Kalman filter may generate a state estimation value and a first covariance of the target lane line by using the following formula.

[0107] K(k)=P k (k,k-1)H T [HP k (k,k-1)H T +M] -1

[0108]

[0109] P k (k) = [I4-K(k)H]P k (k,k-1)

[0110] Among them, K(k) is the Kalman gain at the (kth moment), P k (k,k-1) is the first covariance updated at the (k-1)th moment, P k (k) is the first covariance at the (k)th moment, is the estimated state value at the (k)th moment, is the state prediction value at the (k)th moment, and I4 is the 4X4 unit matrix.

[0111] Furthermore, It can be calculated by the following formula:

[0112]

[0113] Where A is the state transfer matrix, Γ is the noise driving matrix, W(k-1)∈R r is the process noise, R is the process noise matrix, r is the dimension of the process noise matrix, the mean of the process noise is 0, and the variance matrix is ​​Q.

[0114] It should be understood that after calculating the state estimate value and the first covariance at the (k)th moment, the state estimate value and the first covariance at the (k+1)th moment can be updated and processed based on the state estimate value and the first covariance at the (k)th moment, and the first covariance can be continuously recursively applied to estimate the optimal state estimate value.

[0115] S13, generating a state recursive value and a second covariance of the target lane line at the first moment through a state recursive device.

[0116] In a possible implementation, the state recursive value and the second covariance of the target lane line at the first moment may be determined by the following formula.

[0117]

[0118] P s (k) = AP s (k,k-1)AT +ΓQΓ T

[0119] in, is the state recursion value at the (k)th moment, is the state recursion value at the (k-1)th moment, P s (k) is the second covariance at the (k)th moment, P s (k,k-1) is the second covariance at the (k-1)th moment.

[0120] In another possible implementation, in order to ensure the processing accuracy of the state recursors, a Kalman filter can be used to reset multiple state recursors alternately, thereby eliminating the errors accumulated by the state recursors over time. Furthermore, the reset state recursors do not output, and the state recursors that are not reset output the state recursion value and the second covariance, so that the target lane line can be detected for abnormality based on the output of the Kalman filter and the output of the state recursors that are not reset.

[0121] Exemplarily, it is assumed that the state recurser has two sub-state recursors, namely state recursor 1 and state recursor 2. In period 1, state recursor 1 is reset, state recursor 2 works, and outputs the state recursion value and the second covariance; in period 2, state recursor 1 is reset, and state recursor 1 outputs the state recursion value and the second covariance based on the reset value.

[0122] It should be understood that the specific implementation principle and implementation process of this method will be Figure 2 The detailed description is given in the illustrated embodiment and will not be repeated here.

[0123] S14. Perform an abnormality check on the target lane line according to the state estimation value, the state recursive value, the first covariance and the second covariance corresponding to the target lane line at the first moment, and generate an abnormality detection result.

[0124] Among them, the anomaly detection result is used to indicate whether there is an anomaly in the target lane line.

[0125] In a possible implementation, a first error between a state estimation value and a state recursive value may be calculated, and a second error between a first covariance and a second covariance may be calculated, and an abnormality detection result may be determined comprehensively based on the first error and the second error.

[0126] Exemplarily, if the first error is greater than a first preset error, and / or the second error is greater than a second preset error, an abnormality detection result indicating that there is an abnormality in the target lane line is determined; if the first error does not exceed the first preset error, and the second error does not exceed the second preset error, an abnormality detection result indicating that there is no abnormality in the target lane line is determined.

[0127] In another possible implementation, a chi-square test may be performed on the target lane line according to the state estimation value, the state recursive value, the first covariance, and the second covariance corresponding to the target lane line at the first moment, thereby generating an anomaly detection result.

[0128] It should be understood that the process and principle of the chi-square test on the target lane line will be Figure 3 The illustrated embodiment is described in detail and will not be elaborated here.

[0129] It can be seen from the existing lane line anomaly detection scheme that the low accuracy of the existing technology is mainly caused by the following two aspects: 1. The preset distance is not set accurately; 2. The sensor that collects the lane line signal fails. In this technical solution, when the lane line itself is abnormal due to natural wear, occlusion, light changes, etc., since the state recursor is only related to the output of the state recursor at the previous moment, and the output of the Kalman filter is related to the lane line signal, the output of the state recursor and the output of the Kalman filter are different in this scenario. It can be judged whether the target lane line is abnormal based on the output of the state recursor and the output of the Kalman filter. In this process, there is no need to manually set parameters, which solves the problem of low anomaly detection accuracy caused by the inability to ensure the rigor of manually set parameters. For the scenario where the sensor fails, since the Kalman filter will not be able to make a correct state estimate based on the erroneous lane line signal, the output of the Kalman filter and the state recursor will be inconsistent, and then it is judged whether the target lane line is abnormal.

[0130] The lane line abnormality detection method provided by the embodiment of the present invention, after the vehicle turns on the intelligent driving function or the automatic driving function, establishes the state space model of the target lane line according to the lane line signal of the target lane line collected in real time, and then generates the state estimation value and the first covariance of the target lane line at the first moment according to the state space model and the Kalman filter, and then generates the state recursion value and the second covariance of the target lane line at the first moment through the state recursion. Finally, according to the state estimation value, the state recursion value, the first covariance and the second covariance corresponding to the target lane line at the first moment, the target lane line is tested for abnormality and an abnormality detection result is generated. Among them, the target lane line is the lane line of the lane where the vehicle is located, and the abnormality detection result is used to indicate whether the target lane line is abnormal. Since the state recursion is only related to the output of the state recursion at the previous moment, and the output of the Kalman filter is related to the actual lane line signal, the technical solution combines the output of the Kalman filter and the state filter, and performs abnormality detection on the target lane line by judging whether the two are consistent. Since there is no need to manually set parameters, the accuracy and robustness of abnormality detection can be effectively improved.

[0131] Optionally, in some embodiments, after generating an abnormality detection result indicating that an abnormality exists in the target lane line, a reminder message may also be output, where the reminder message is used to remind the user to control the driving of the vehicle.

[0132] In actual applications, the method of outputting the reminder information can be at least one of the following: displaying the reminder information on the central control screen, controlling the vibration of the cockpit seat and / or steering wheel, controlling the warning lights in the vehicle to flash, and controlling the vehicle to play a warning sound.

[0133] Figure 2 The flowchart of the second embodiment of the lane line abnormality detection method provided by the embodiment of the present invention is as follows. Figure 2 As shown, S13 may include the following steps:

[0134] S21. Determine the first sub-state recurser that needs to be reset at the first moment according to the preset duration and the duration that each sub-state recurser has been reset.

[0135] The state recurser may include at least two sub-state recursors, and the first sub-state recursor belongs to the state recursor.

[0136] In a possible implementation, the reset duration of each sub-state recurser can be preset. Exemplarily, the reset duration of each sub-state recurser can be directly set to 30 seconds, or the reset frequency of each sub-state recurser can be set, for example, reset once every 30 seconds, and then determine the reset duration (30 seconds) of each sub-state recurser. Further, for each sub-state recurser, if the current reset duration of the sub-state recurser is greater than 0 and less than the preset duration, the sub-state recurser is determined to be the first sub-state recurser; if the current reset duration of each sub-state recurser is 0, the sub-state recurser that has not been reset at the previous moment is determined as the first sub-state recurser.

[0137] Exemplarily, it is assumed that the state deducer may include two sub-state deducers, namely state deducer 1 and state deducer 2. At the beginning, the first one of state deducer 1 and state deducer 2 may be randomly determined as the first sub-state deducer (e.g., state deducer 1), and the first sub-state deducer is reset until the reset duration of the first sub-state deducer is equal to the preset duration. Thereafter, the reset duration of the first sub-state deducer is reset to 0, and the other sub-state deducer (state deducer 2) is determined as the new first sub-state deducer, and the new first sub-state deducer is reset until the reset duration of the new first sub-state deducer is equal to the preset duration.

[0138] S22. Reset the first sub-state recurser according to the state estimation value and the first covariance of the target lane line at the first moment.

[0139] In a possible implementation, the state recursion value generated by the first sub-state recurser at the first moment is reset to the state estimation value at the first moment, and the second covariance generated by the first sub-state recurser at the first moment is reset to the first covariance at the first moment.

[0140] In this implementation, since the Kalman filter processes the collected lane line signal and then outputs it, the accuracy of the Kalman filter output is higher than that of the state recursor output. Therefore, the output of the Kalman filter can be assigned to the first sub-state recursor, and the first sub-state recursor is initialized to eliminate the error accumulated in the first sub-state recursor due to time reasons, which effectively improves the accuracy of subsequent lane line anomaly detection based on the output of the Kalman filter and the output of the state recursor.

[0141] S23, according to the state recursion value and the second covariance generated by the second sub-state recurser at the second moment, generate the state recursion value and the second covariance of the target lane line at the first moment through the second sub-state recurser.

[0142] The second sub-state recurser belongs to the state recurser and is not reset at the first moment.

[0143] Among them, the second moment is earlier than the first moment.

[0144] It should be understood that the manner in which the second sub-state recurser generates the state recursive value and the second covariance of the target lane line at the first moment can refer to the relevant formula in S13, which will not be repeated here.

[0145] In the above embodiment, multiple sub-state recursors are reset alternately through the Kalman filter, and the sub-state recursors that have not been reset continue to perform state recursion according to the previously reset values, which effectively reduces the accumulated error of the state recursor and improves the accuracy of subsequent abnormality detection of the target lane line.

[0146] Figure 3 The flowchart of the lane line abnormality detection method embodiment 3 provided by the embodiment of the present invention is as follows. Figure 3 As shown, S14 may include the following steps:

[0147] S31. Determine a state difference according to a first preset anomaly detection function and a state estimation value and a state recursive value corresponding to the target lane line at the first moment.

[0148] Among them, the first preset anomaly detection function obeys Gaussian distribution.

[0149] Optionally, the first preset anomaly detection function can be expressed by the following formula:

[0150]

[0151] Wherein, β(k) is the state difference, and the mean value of the first preset abnormality detection function is zero.

[0152] S32. Calculate the variance of the first preset anomaly detection function according to the first covariance and the second covariance corresponding to the target lane line at the first moment.

[0153] Optionally, the variance of the first preset anomaly detection function can be expressed by the following formula:

[0154] B(k)=E[β(k)β T (k)]=P s (k)-P k (k)

[0155] Wherein, B(k) is the variance of the first preset anomaly detection function, and E[] is the covariance calculation formula.

[0156] S33, determining a target chi-square value according to the state difference and variance corresponding to the target lane line at the first moment and a second preset anomaly detection function.

[0157] The second preset anomaly detection function satisfies the chi-square distribution.

[0158] Optionally, the second preset anomaly detection function can be expressed by the following formula:

[0159] ξ β (k) = β T (k)B -1 (k)β(k)

[0160] Among them, ξ β (k) is the target chi-square value, which follows a χ2 distribution with n degrees of freedom, where n is the dimension of β(k).

[0161] S34. Perform a chi-square test on the target lane line according to the target chi-square value and the preset chi-square value to generate an anomaly detection result.

[0162] In a possible implementation, the target chi-square value may be compared with a preset chi-square value to perform a chi-square test on the target lane line. Specifically, if the target chi-square value is greater than or equal to the preset chi-square value, an abnormality detection result indicating that the target lane line is abnormal is determined; if the target chi-square value is less than the preset chi-square value, an abnormality detection result indicating that the target lane line is not abnormal is determined.

[0163] Exemplarily, the anomaly detection result can be generated by the following formula:

[0164]

[0165] Among them, ε β It is a preset chi-square value, which can be preset according to actual conditions.

[0166] In the above embodiment, by performing a chi-square test on the output of the Kalman filter and the output of the state recurser, the process is simple and the result is intuitive, which can effectively improve the efficiency of lane line anomaly detection.

[0167] Next, based on the lane line anomaly detection method shown in any of the above embodiments, a specific example will be used to explain it in practice.

[0168] Figure 4 Schematic diagram of the process of the fourth embodiment of the lane line abnormality detection method provided by the embodiment of the present invention. Figure 4 As shown, the lane line anomaly detection method may include the following steps:

[0169] Step 1: Collect the lane line signal of the vehicle through the vehicle's sensor.

[0170] Step 2: Construct a state space model based on the collected lane line signals.

[0171] Step 3: Input the state space model into the Kalman filter to generate the state estimation value and the first covariance in real time.

[0172] Step 4: According to the preset reset frequency, the sub-state recursors are reset in turn according to the output of the Kalman filter.

[0173] Among them, state recurser No. 1 and state recurser No. 2 are both sub-state recursors.

[0174] Step 5: Perform a chi-square test based on the output of the sub-state recursor that has not been reset and the output of the Kalman filter to generate an anomaly detection result.

[0175] Among them, the sub-state recursor that has not been reset recursively recurses the output at the current moment based on the output at the previous moment, and the state transfer matrix used in the recursive process is obtained from the state space model.

[0176] The following are embodiments of the device of the present invention, which can be used to implement the method embodiments of the present invention. For details not disclosed in the device embodiments of the present invention, please refer to the method embodiments of the present invention.

[0177] Figure 5 Schematic diagram of the structure of the lane line abnormality detection device provided by an embodiment of the present invention. Figure 5 As shown, the lane line abnormality detection device 50 includes:

[0178] Establishing module 51, which is used to establish a state space model of the target lane line according to the lane line signal of the target lane line collected in real time after the vehicle turns on the intelligent driving function or the automatic driving function, and the target lane line is the lane line of the lane where the vehicle is located.

[0179] The first generating module 52 is used to generate a state estimation value and a first covariance of the target lane line at a first moment according to the state space model and the Kalman filter.

[0180] The second generating module 53 is used to generate the state recursion value and the second covariance of the target lane line at the first moment through a state recursive device.

[0181] The inspection module 54 is used to perform an abnormality inspection on the target lane line according to the state estimation value, state recursive value, first covariance and second covariance corresponding to the target lane line at the first moment, and generate an abnormality detection result, and the abnormality detection result is used to indicate whether there is an abnormality in the target lane line.

[0182] Further, the second generating module 53 is specifically used for:

[0183] According to the preset duration and the duration for which each sub-state recurser has been reset, a first sub-state recurser that needs to be reset at the first moment is determined, and the first sub-state recurser belongs to the state recurser.

[0184] The first sub-state recurser is reset according to the state estimation value and the first covariance of the target lane line at the first moment.

[0185] According to the state recursion value and second covariance generated by the second sub-state recurser at the second moment, the state recursion value and second covariance of the target lane line at the first moment are generated by the second sub-state recurser. The second sub-state recurser belongs to the state recurser and is not reset at the first moment. The second moment is earlier than the first moment.

[0186] Further, the state space model includes a state equation and an observation equation, and a module 51 is established, which is specifically used for:

[0187] According to the lane line signal of the target lane line collected in real time, discrete system modeling is performed on the target lane line to generate a state equation.

[0188] According to the lane line signal of the target lane line collected in real time, an observation equation of the target lane line is generated.

[0189] Furthermore, the state space model includes at least one of the following states: a position of the target lane line relative to the vehicle, a relative heading, a relative curvature, and a relative curvature differential.

[0190] Furthermore, the inspection module 54 is specifically used for:

[0191] The state difference is determined according to a first preset anomaly detection function and a state estimation value and a state recursive value corresponding to the target lane line at the first moment, and the first preset anomaly detection function obeys a Gaussian distribution.

[0192] The variance of the first preset anomaly detection function is calculated according to the first covariance and the second covariance corresponding to the target lane line at the first moment.

[0193] A target chi-square value is determined based on the state difference, variance and a second preset anomaly detection function corresponding to the target lane line at the first moment, and the second preset anomaly detection function satisfies the chi-square distribution.

[0194] According to the target chi-square value and the preset chi-square value, a chi-square test is performed on the target lane line to generate anomaly detection results.

[0195] Furthermore, the inspection module 54 is specifically used for:

[0196] According to β (k) = β T (k)B -1 (k)β(k), calculate the target chi-square value;

[0197] Among them, ξ β (k) is the target chi-square value, β(k) is the state difference, B(k) is the variance, and n is the dimension of β(k).

[0198] The lane line anomaly detection device provided in an embodiment of the present invention can be used to execute the lane line anomaly detection method in any of the above embodiments. Its implementation principle and technical effects are similar and will not be repeated here.

[0199] It should be noted that it should be understood that the division of the various modules of the above device is only a division of logical functions. In actual implementation, all or part of them can be integrated into one physical entity, or they can be physically separated. And these modules can all be implemented in the form of software called by processing elements. They can also be implemented in the form of hardware. Some modules can also be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. In addition, all or part of these modules can be integrated together or implemented independently. The processing element here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each module above can be completed by an integrated logic circuit of hardware in the processor element or instructions in the form of software.

[0200] Figure 6 Schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 6As shown, the electronic device 60 may include: a processor 61, a memory 62, and computer execution instructions stored in the memory 62 and executable on the processor 61. When the processor 61 executes the computer execution instructions, the lane line abnormality detection method provided in any of the aforementioned embodiments is implemented.

[0201] Optionally, the above-mentioned components of the electronic device 60 may be connected via a system bus.

[0202] The memory 62 may be a separate storage unit or a storage unit integrated in the processor. The number of processors may be one or more.

[0203] Optionally, the electronic device 60 may also include a communication interface for interacting with other devices.

[0204] It should be understood that the processor 61 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the present invention can be directly embodied as being executed by a hardware processor, or can be executed by a combination of hardware and software modules in the processor.

[0205] The system bus may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The system bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage.

[0206] All or part of the steps of the above-mentioned method embodiments can be completed by hardware related to program instructions. The above-mentioned program can be stored in a readable memory. When the program is executed, the steps of the above-mentioned method embodiments are executed. The above-mentioned memory (storage medium) includes: read-only memory (ROM), RAM, flash memory, hard disk, solid state drive, magnetic tape, floppy disk, optical disc and any combination thereof.

[0207] The electronic device provided in an embodiment of the present invention can be used to execute the lane line anomaly detection method provided in any of the above method embodiments. Its implementation principle and technical effects are similar and will not be repeated here.

[0208] An embodiment of the present invention provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed on a computer, the computer executes the above-mentioned lane line anomaly detection method.

[0209] The computer-readable storage medium mentioned above can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory, electrically erasable programmable read-only memory, erasable programmable read-only memory, programmable read-only memory, read-only memory, magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0210] Optionally, a readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0211] An embodiment of the present invention also provides a computer program product, which includes a computer program. The computer program is stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium. When at least one processor executes the computer program, the above-mentioned lane line anomaly detection method can be implemented.

[0212] An embodiment of the present invention further provides a vehicle, which includes the above-mentioned electronic device 60.

[0213] It should be understood that the present invention is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A lane line anomaly detection method, It is characterized in that include: After the vehicle turns on the intelligent driving function or the automatic driving function, a state space model of the target lane line is established according to the lane line signal of the target lane line collected in real time, and the target lane line is the lane line of the lane where the vehicle is located; Generate a state estimation value and a first covariance of the target lane line at a first moment according to the state space model and the Kalman filter; Generate a state recursion value and a second covariance of the target lane line at the first moment through a state recursion device; According to the state estimation value, the state recursive value, the first covariance and the second covariance corresponding to the target lane line at the first moment, the target lane line is checked for abnormality to generate the abnormality detection result, and the abnormality detection result is used to indicate whether there is an abnormality in the target lane line.

2. The method according to claim 1, It is characterized in that The state recursion value and the second covariance of the target lane line at the first moment are generated by the state recursion device, including: Determine, according to a preset duration and a duration for which each sub-state recurser has been reset, a first sub-state recurser that needs to be reset at the first moment, the first sub-state recurser belonging to the state recurser; Resetting the first sub-state recursor according to the state estimation value and the first covariance of the target lane line at the first moment; According to the state recursion value and the second covariance generated by the second sub-state recurser at the second moment, the state recursion value and the second covariance of the target lane line at the first moment are generated by the second sub-state recurser, the second sub-state recurser belongs to the state recurser and is not reset at the first moment, and the second moment is earlier than the first moment.

3. The method according to claim 2, It is characterized in that The resetting the first sub-state recursor according to the state estimation value and the first covariance of the target lane line at the first moment includes: The state recursion value generated by the first sub-state recurser at the first moment is reset to the state estimation value at the first moment, and the second covariance generated by the first sub-state recurser at the first moment is reset to the first covariance at the first moment.

4. The method according to any one of claims 1 to 3, It is characterized in that The state space model includes a state equation and an observation equation. The state space model of the target lane line is established according to the lane line signal of the target lane line collected in real time, including: According to the lane line signal of the target lane line collected in real time, discrete system modeling is performed on the target lane line to generate the state equation; The observation equation of the target lane line is generated according to the lane line signal of the target lane line collected in real time.

5. The method according to any one of claims 1 to 3, It is characterized in that The state space model includes at least one of the following states: a position, a relative heading, a relative curvature, and a relative curvature differential of the target lane line relative to the vehicle.

6. The method according to any one of claims 1 to 3, It is characterized in that The step of performing an abnormality check on the target lane line according to the state estimation value, the state recursive value, the first covariance, and the second covariance corresponding to the target lane line at the first moment to generate an abnormality detection result includes: Determine a state difference according to a first preset anomaly detection function and the state estimation value and the state recursive value corresponding to the target lane line at the first moment, wherein the first preset anomaly detection function obeys a Gaussian distribution; Calculating the variance of the first preset anomaly detection function according to the first covariance and the second covariance corresponding to the target lane line at the first moment; Determining a target chi-square value according to the state difference, the variance, and a second preset anomaly detection function corresponding to the target lane line at the first moment, wherein the second preset anomaly detection function satisfies a chi-square distribution; According to the target chi-square value and the preset chi-square value, a chi-square test is performed on the target lane line to generate the anomaly detection result.

7. The method according to claim 6, It is characterized in that The determining a target chi-square value according to the state difference, the variance and a second preset anomaly detection function corresponding to the target lane line at the first moment includes: According to β (k) = β T (k)B -1 (k)β(k), calculate the target chi-square value; Among them, ξ β (k) is the target chi-square value, β(k) is the state difference, B(k) is the variance, and n is the dimension of β(k).

8. A lane line abnormality detection device, It is characterized in that include: An establishment module is used to establish a state space model of a target lane line according to a lane line signal of a target lane line collected in real time after the vehicle turns on the intelligent driving function or the automatic driving function, wherein the target lane line is the lane line of the lane where the vehicle is located; A first generating module, used for generating a state estimation value and a first covariance of the target lane line at a first moment according to the state space model and the Kalman filter; A second generating module, configured to generate a state recursive value and a second covariance of the target lane line at the first moment through a state recursive device; A verification module is used to perform an abnormality check on the target lane line according to the state estimation value, the state recursive value, the first covariance and the second covariance corresponding to the target lane line at the first moment, and generate the abnormality detection result, wherein the abnormality detection result is used to indicate whether there is an abnormality in the target lane line.

9. An electronic device, include: A processor, a memory, and a computer-executable instruction stored in the memory and executable on the processor, wherein the processor is used to implement the lane line anomaly detection method as described in any one of claims 1 to 7 when executing the computer-executable instruction.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the lane line anomaly detection method as described in any one of claims 1 to 7.

11. A computer program product comprising a computer program, It is characterized in that When the computer program is executed by a processor, it is used to implement the lane line anomaly detection method as described in any one of claims 1 to 7.

12. A vehicle, It is characterized in that Comprising the electronic device as claimed in claim 9.

Images