Safety performance quantitative indicator determination method for lane line detection module, safety performance quantitative evaluation method for autonomous driving system, and related product

Abstract

A safety performance quantitative indicator determination method for a lane line detection module, a safety performance quantitative evaluation method for an autonomous driving system, and a related product. The safety performance quantitative indicator determination method comprises: acquiring lane line detection uncertainties of a plurality of algorithm modules of a lane line detection module, hazard events of the lane line detection module being decomposed into being caused by the lane line detection uncertainties of the different algorithm modules of the lane line detection module; performing scenario modeling on the probability of the lane line detection uncertainties of the lane line detection module causing the hazard events, so as to obtain driver models in different hazard events; and, at least on the basis of the driver models in the different hazard events, obtaining quantitative objectives corresponding to the different algorithm modules of the lane line detection module.

Description

Methods for determining the safety performance quantification indicators of lane detection modules and methods for safety energy assessment of autonomous driving systems, and related products. Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a method for quantitatively evaluating the safety performance of a lane detection module and a method for quantitatively evaluating the safety of an autonomous driving system, as well as related products. Background Technology

[0002] Among related technologies, one of the key systems in autonomous vehicles that ensures a safe environment for drivers and passengers is the Advanced Driver Assistance System (ADAS). Adaptive cruise control, automatic braking / steering departure, lane keeping assist, blind spot assist, lane departure warning, and lane detection are representative of ADAS. Lane detection displays specific information about the geometric features of the lane line structure to the vehicle's intelligent system, indicating the location of the lane markings. Common lane recognition algorithms use image processing techniques, such as edge detection, color filtering, and Hough transform, to extract the position and shape information of the lane lines. Model-based lane detection algorithms typically establish a geometric model of the lane line based on its geometric characteristics, and then use methods such as Random Sample Consensus (RANSAC), least squares, and Hough transform to obtain the geometric model parameters of the lane line, thereby fitting the corresponding lane line curve.

[0003] Currently, the safety design of autonomous driving systems either sets an overall quantitative safety target at the vehicle level, or sets some experience-based quantitative safety targets at the system level, or sets some safety requirements based on experience or theoretical analysis. However, these approaches lack specific quantitative indicators and have weak theoretical basis. For example, the safety design of lane detection modules in autonomous driving systems is mostly based on qualitative design, lacking quantitative design references. A small portion of quantitative safety designs for autonomous driving systems rely on estimates based on engineering or expert experience, without theoretical basis. Current performance development of autonomous driving systems, due to the lack of safety targets, can only guarantee the availability of performance, without clarifying the extent to which performance needs to be developed and perfected. Summary of the Invention

[0004] This application provides a method for quantitatively evaluating the safety performance of a lane detection module, an electronic device, and a storage medium, which are used to at least solve one of the above-mentioned technical problems.

[0005] In a first aspect, embodiments of this application provide a method for determining the safety performance quantification index of a lane detection module, comprising: obtaining lane detection uncertainty of multiple algorithm modules of the lane detection module, wherein a hazardous event of the lane detection module is decomposed into events caused by lane detection uncertainty of different algorithm modules of the lane detection module; performing scenario modeling on the probability of hazardous events caused by lane detection uncertainty of the lane detection module to obtain driver models under different hazardous events; and obtaining quantification targets corresponding to different algorithm modules of the lane detection module based at least on the driver models under different hazardous events.

[0006] Secondly, embodiments of this application provide a method for quantitatively assessing the safety of an autonomous driving system, comprising: evaluating the safety performance of the autonomous driving system using the quantitative targets in the method described in the first aspect.

[0007] Thirdly, embodiments of this application provide an electronic device comprising: at least one processor and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the method for determining the safety performance quantification index of the lane detection module described above and the method for safety quantification assessment of an autonomous driving system.

[0008] Fourthly, embodiments of this application provide a storage medium storing one or more programs including execution instructions. The execution instructions can be read and executed by electronic devices (including but not limited to computers, servers, or network devices) to perform the safety performance quantification index determination method for any of the lane detection modules described above and the safety quantification assessment method for autonomous driving systems.

[0009] Fifthly, embodiments of this application also provide a computer program product, the computer program product including a computer program stored on a storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to execute any of the above-mentioned methods for determining the safety performance quantification index of the lane detection module and the safety quantification assessment method for an autonomous driving system.

[0010] Sixthly, embodiments of this application also provide a mobile platform, including the electronic device described in the third aspect.

[0011] This application provides a theoretically sound and complete method for determining and evaluating the safety quantification indicators of autonomous driving systems. Using the method described in this application, quantified safety design indicators for each algorithm module within the autonomous driving system and its lane detection module can be obtained, providing clear quantitative references for the safety design, development, and testing phases of autonomous driving systems. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 is a flowchart of a method for determining the safety performance quantitative indicators of a lane line detection module according to an embodiment of this application;

[0014] Figure 2 shows a vehicle failure decomposition scheme provided in an embodiment of this application;

[0015] Figure 3 shows a lane line error deviation model established according to an embodiment of this application when the lane line detection value is outward relative to the true value;

[0016] Figure 4 shows a lane line error deviation model established according to an embodiment of this application when the lane line detection value is inward relative to the true value;

[0017] Figure 5 shows the simulation results provided by an embodiment of this application;

[0018] Figures 6a-6d illustrate a method for calculating the quantitative index of the corresponding lane centerline based on the lateral deviation threshold of the lane line, according to an embodiment of this application.

[0019] Figure 7 is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] This invention designs a method for quantitatively evaluating the safety performance of lane detection algorithm modules in autonomous driving systems. It employs mathematical modeling, theoretical analysis, and natural driving data analysis to design a quantification scheme for the safety performance indicators of lane detection modules, thereby obtaining the safety quantitative targets of lane detection modules and providing a reference for the safety design and evaluation of lane detection modules in autonomous driving systems.

[0022] Mathematical modeling mainly involves kinematic model analysis and modeling of hazardous scenarios. Theoretical analysis primarily examines the hazards generated by each algorithm module and the rationality of the mathematical modeling. Natural driving data analysis mainly involves obtaining driver behavior data through statistical analysis of natural driving data.

[0023] The safety design of autonomous driving systems in related technologies cannot provide reasonable theoretical derivations, resulting in the lack of quantified safety targets for lane detection modules. This invention, based on reasonable mathematical modeling, obtains a quantification scheme for the safety performance targets of lane detection modules, which can be used for the safety design and evaluation of autonomous driving systems.

[0024] Please refer to Figure 1, which shows a flowchart of a method for determining the safety performance quantification index of a lane detection module according to an embodiment of this application. This method is used in autonomous driving systems and can be applied to the fields of safety design and evaluation of lane detection modules in autonomous driving systems or the testing of lane detection modules in autonomous driving systems. As part of autonomous driving safety, it can be applied to the safety design and evaluation of lane detection modules in various levels of autonomous driving systems or the testing of lane detection modules in autonomous driving systems. The execution subject of this embodiment can be an electronic device or a safety quantification index determination device installed in an electronic device. The safety quantification index determination device can be implemented by software or by a combination of software and hardware. The safety quantification index determination device can be a processor in an electronic device. The electronic device can be a device equipped with an autonomous driving system or other devices connected to an autonomous driving system.

[0025] As shown in Figure 1, in step 101, the lane line detection uncertainty of multiple algorithm modules of the lane line detection module is obtained;

[0026] In step 102, scenario modeling is performed on the probability of hazardous events caused by the lane line detection uncertainty of the lane line detection module to obtain driver models under different hazardous events;

[0027] In step 103, the quantization targets corresponding to different algorithm modules of the lane detection module are obtained based at least on the driver model under different hazardous events.

[0028] In this embodiment, for step 101, the hazardous events of the lane line detection module are decomposed into lane line detection uncertainties caused by different algorithm modules of the lane line detection module. Thus, the uncertainty of the lane line detection module can be decomposed into the uncertainties of its corresponding multiple algorithm modules. The uncertainty of the lane line detection module can be obtained by acquiring the lane line detection uncertainties of multiple algorithm modules. These multiple algorithm modules may include a lane line position detection module and a lane center line position detection module, and may also include other currently used detection modules or detection modules that will be used in the future; this application does not limit this. Furthermore, the lane line detection uncertainties of the multiple algorithm modules may include lane line position detection errors and lane center line position detection errors.

[0029] Next, in step 102, scenario modeling is performed to assess the probability of hazardous events caused by lane detection uncertainty. Through scenario analysis and modeling, driver models are obtained under different hazardous events. Specifically, due to lane detection uncertainty, vehicles may deviate from their current lane, potentially colliding with vehicles in adjacent lanes, nearby obstacles, or pedestrians, resulting in hazardous events. For these hazardous events, analysis and modeling can be performed to obtain a relationship model (driver model) between the probability of hazardous events and lane detection uncertainty.

[0030] Finally, regarding step 103, after establishing driver models under different hazardous events, at least the quantitative targets corresponding to multiple algorithm modules of the lane detection module can be obtained based on these driver models. Specifically, based on the premise that the safety of the autonomous driving system is no less than a safety reference benchmark, such as the safety of human drivers or the safety of other well-performing autonomous driving systems, the following values ​​can be obtained through scenario simulation: the critical value for preventing the vehicle from deviating from the current lane when a lane detection error is detected. If this critical value is exceeded, it is impossible to prevent the vehicle from deviating from the current lane, and a hazardous event is likely to occur in this situation. This critical value can then be used as the quantitative target, thereby obtaining the quantitative targets corresponding to multiple algorithm modules. Of course, the quantitative target can also be set smaller than this critical value; this application does not impose any restrictions here. For example, the maximum value of the lane line position detection error and the maximum value of the lane centerline position detection error.

[0031] In this embodiment of the application, the hazardous events of the lane line detection module are decomposed into lane line detection uncertainties caused by different algorithm modules of the lane line detection module. Then, scene modeling is performed on different algorithm modules to obtain the corresponding driver model. Then, based on the driver model, the quantitative targets of different algorithm modules can be obtained.

[0032] Please refer to Figure 2, which shows the breakdown scheme of an embodiment of this application.

[0033] As shown in Figure 2, this embodiment of the application adopts a layer-by-layer decomposition method, breaking down the vehicle failure problem layer by layer into various algorithm modules. The core idea is that the safety of an autonomous driving system must not be lower than human safety. First, the causes of vehicle failure are analyzed. Vehicle failure is caused by both hazardous behaviors and related scenarios. For related scenarios, scenario modeling can be performed to obtain regions of interest for different scenarios. Then, hazardous behaviors are analyzed, listing specific hazardous behaviors. Next, the causes of hazardous behaviors are analyzed to identify the performance limitations that lead to hazardous behaviors, including:

[0034] Perception-related performance limitations: Lane line state uncertainty. This mainly includes performance limitations such as positional errors in lane line detection and positional errors in lane centerline detection.

[0035] All performance limitations are caused by different algorithm modules. By analyzing and modeling these limitations, and by using natural driving datasets and driver modeling, quantitative indicators of the performance limitations of different algorithm modules can be obtained, which can then serve as quantitative targets for the safety performance of the corresponding algorithm modules.

[0036] Regarding the performance limitations related to perception: state uncertainty, which mainly includes lane line position detection error, lane center line position detection error, etc.

[0037] In some optional embodiments, obtaining the lane detection uncertainty of multiple algorithm modules of the lane detection module includes: when different algorithm modules of the lane detection module experience lane detection uncertainty, obtaining the lane detection uncertainty that leads to the exposure of a hazardous behavior, wherein the hazardous behavior is uncontrollable. Here, exposure of a hazardous behavior refers to the occurrence of a hazardous behavior. Uncontrollable hazardous behavior means that the hazardous behavior cannot be controlled (e.g., the driver's specific actions in response to the hazardous behavior cannot be predicted).

[0038] In some optional embodiments, the lane detection uncertainty is the lane position detection error. The scenario modeling of the probability of hazardous events caused by the lane detection uncertainty of the lane detection module, to obtain driver models under different hazardous events, includes: when there is a vehicle in the adjacent lane, modeling the scenario of deviation from the vehicle's lane caused by the lane position detection error, to obtain a driver model corresponding to the lane position detection error; wherein, the scenario of deviation from the vehicle's lane is: the vehicle experiences a lane position detection error, and the lane position detection error is such that even using lateral acceleration in the opposite direction of the lane position detection error, it is impossible to correct the vehicle back to its lane; the quantization target corresponding to different algorithm modules of the lane detection module, at least based on the driver models under different hazardous events, includes: at least based on the driver model corresponding to the lane position detection error, obtaining the maximum offset that allows the vehicle to remain on its lane using lateral acceleration in the opposite direction, to obtain the quantization target corresponding to the lane position detection error of the autonomous driving system.

[0039] Referring further to Figures 3 and 4, two lane line error deviation models according to embodiments of this application are shown.

[0040] In this embodiment of the application, for the lateral detection error of the lane line, the dangerous scenario considered is that there is a vehicle in the adjacent lane. The core constraint is that when the lane line has a lateral detection error, the vehicle will deviate laterally. The maximum allowable lateral deviation of the vehicle is not more than the lane itself. Therefore, a lane line lateral deviation model is established.

[0041] Please refer to Figure 3, which shows the lane line error deviation model established when the lane line detection value is outward relative to the ground truth. In the figure, the black dot chain represents the lane line ground truth, the black curve represents the lane line detection result output by the lane line detection module, the black concentric circles represent vehicles (diameter is the vehicle width), and the concentric circles of the black dashed lines represent the vehicle's position in the next frame.

[0042] Suppose that lane detection produces a large error (this error is hypothetical; the vehicle will only make unsafe movements if this error occurs), causing the vehicle to travel along the wrong lane and deviate from the true lane. If maximum lateral acceleration is applied in the opposite direction to correct the vehicle's deviation, it can ensure the vehicle stays within the lane. However, when the lane detection error is large enough, even the opposite lateral acceleration cannot correct the vehicle back into the lane. This detection error is the critical value. Therefore, in Figure 3, the intersection of the avoidance trajectory (black dashed line) perpendicular (tangential) to the true lane line point and the perpendicular deceleration direction (dashed arrow) gives the possible deviation based on dynamics. This offset is the allowable outward deviation of the lane line. The process of calculating this offset is as follows: First, calculate the slope between two points of the lane line truth value to obtain the slope in the vertical direction. Then, calculate the perpendicular line equation (represented by a solid arrow) based on the slope and one of the two points mentioned above. Next, calculate the offset trajectory based on the acceleration components. The coordinates of the intersection point (i.e., the black cross) are obtained from the offset trajectory and the perpendicular line equation. Finally, the allowable offset (the size of the vehicle can be considered) is obtained from the intersection point and the discrete point on the corresponding GT (Ground Truth).

[0043] The specific calculation process is as follows: Assume that there are two true values ​​for the lane line, a(x1, y1) and b(x2, y2). Then the slope of the tangent is c = (y2-y1) / (x2-x1), and the slope in the deceleration direction is d = 1 / -c. The deceleration a is decomposed into ax in the x direction and ay in the y direction. Then ay / ax = d, ax*ax + ay*ay = a*a. We can solve for ax and ay and then obtain their values. Using y = ay*t + 1 / 2*ay*t*t and x = ax*t + 1 / 2*ax*t*t, we can obtain the coordinates of the point and thus find the maximum offset of this point from the true value point.

[0044] Please refer to Figure 4, which shows the lane line error deviation model established when the lane line detection value is inward relative to the ground truth. In the figure, the black dot chain represents the lane line ground truth, the black curve represents the lane line detection result output by the lane line detection module, the black concentric circles represent vehicles (diameter is the vehicle width), and the concentric circles of the black dashed lines represent the vehicle's position in the next frame.

[0045] Assuming that lane detection introduces significant errors, causing vehicles to travel along incorrect lane lines and deviate from the true lane, the relative distance between the true lane line and the detected lane line is at its maximum when the vehicle crosses it. This maximum relative distance represents the maximum allowable lane line deviation. The calculation model is as follows:

[0046] lane_error_in = lane_w - car_w

[0047] In the formula, lane_error_in is the maximum allowed offset, lane_w is the lane width, and car_w is the vehicle width.

[0048] In some optional embodiments, obtaining the quantization target corresponding to the lane line position detection error of the autonomous driving system includes: performing simulation based on the driver model and actual scene parameters corresponding to the lane line position detection error to obtain the maximum allowable detection error of the lane line detection model at different detection distances; and using the maximum allowable detection error at different detection distances as the quantization target corresponding to the lane line position detection error of the autonomous driving system.

[0049] Specifically, we can assume that the vehicle is in the middle of a lane with a width of 3.5m, the vehicle width is 2m (the vehicle size radius is 1m), and the vehicle speed is 130km / h. We can use the method in the above embodiment to establish a lane line index model and perform simulation. The simulation results are shown in Figure 5.

[0050] In the simulation results shown in Figure 5, the dotted line represents the avoidance trajectory of the vehicle to prevent it from deviating from the lane due to excessive lane detection error. The area 1 enclosed by the dots represents the allowable point chain with lane detection error falling within this area. In this scenario, the vehicle can avoid deviating from the lane under the avoidance trajectory. The solid black line below represents the right lane line, and the dashed black line above represents the left lane line. The table below shows the boundary values ​​of the simulation results, that is, the maximum allowable detection error of the lane detection module at different detection distances under this condition.

[0051] In a further optional embodiment, the simulation based on the driver model and actual scene parameters corresponding to the lane line position detection error includes: temporarily storing the lane line detected by the lane line detection module at the far end as the detected lane line; when the vehicle travels to the lane line at the far end, taking the detected lane line as the true lane line; and comparing the detected lane line with the true lane line to obtain the lane line position detection error.

[0052] Further optionally, after obtaining the lane line position detection error, the method further includes: determining whether the current detection error exceeds the maximum detection error of the current detection distance; and if the current detection error exceeds the maximum detection error of the current detection distance, recording the scenario cause of the excessive error.

[0053] In a specific example, the indicators in the table above can be used to determine in real time whether the lane detection error meets safety requirements, and to record scenarios where the lane error fails to meet safety standards. Specifically, the lane detection module temporarily stores the lane lines detected at a distant location (e.g., at 120m). When the vehicle reaches 120m, the detected lane line at this location is treated as the true value and compared with the temporarily stored lane line at the same location (the previous lane line result at 120m). This yields the lane detection error at the distant location, and scenarios where the error exceeds safety requirements are recorded. Furthermore, the recorded scenario data can be used to optimize the lane detection module algorithm of the autonomous driving system.

[0054] In some alternative embodiments, the lane detection uncertainty is the lane centerline position detection error. Scenario modeling of the probability of hazardous events caused by the lane detection uncertainty of the lane detection module to obtain driver models under different hazardous events includes: when there is a vehicle in the adjacent lane, scenario modeling of the deviation from the vehicle's lane caused by the lane centerline position detection error to obtain a driver model corresponding to the lane centerline position detection error; wherein, the aforementioned deviation from the vehicle's lane scenario is: the vehicle experiences a lane centerline position detection error, and the lane centerline position detection error makes it impossible to correct the vehicle back to its lane using lateral acceleration in the opposite direction of the deviation; obtaining the quantization target corresponding to different algorithm modules of the lane detection module based at least on the driver models under different hazardous events includes: at least based on the driver model corresponding to the lane centerline position detection error, obtaining the maximum offset that allows the vehicle to remain on its lane using lateral acceleration in the opposite direction, and obtaining the quantization target corresponding to the lane centerline position detection error of the autonomous driving system, wherein the lane centerline position detection error is obtained by back-calculation of the lane centerline position detection error.

[0055] Specifically, a method similar to the lane line detection error modeling and analysis described above can be used to replace the lane line detection error parameter with the lane center line position detection error. By modeling and analyzing the deviation from the vehicle's lane caused by the lane center line position detection error, the quantitative target corresponding to the lane center line position detection error of the autonomous driving system can be obtained.

[0056] Further optional, different types of lane centerline deviation include: a lane centerline deviation distance *a* resulting from two lane lines deviating by a distance *a* in the same direction; no lane centerline deviation resulting from two lane lines deviating by the same distance *a* in different directions; a lane centerline deviation of (a+b) / 2 resulting from two lane lines deviating by distances *a* and *b* in the same direction respectively; and a lane centerline deviation of |ba| / 2 resulting from one lane line deviating by distance *b* in a first direction and the other lane line deviating by distance *a* in a second direction. Therefore, after determining the direction and distance of the lane line deviation in the current scenario, the lane centerline position detection error can be obtained by inverse calculation based on the lane line position detection error.

[0057] In a specific example, the lateral position error of the lane centerline can be inferred from the lateral deviation threshold of the lane line. There are four ways in which the lane centerline deviates due to the deviation of the lane line. For the method of calculating the corresponding quantitative index of the lane centerline when the lateral deviation threshold of the lane line is known, please refer to Figures 6a-6d.

[0058] In Figures 6a-6d, the solid black lines represent the true values ​​of lane lines, the dashed black lines represent the true values ​​of lane center lines, the solid gray lines represent the lane line detection results, and the dashed gray lines represent the lane center line detection results.

[0059] As shown in Figure 6a, when both lane lines deviate to the left by the same distance 'a', the deviation of the lane centerline is the same as the lane line deviation, which is 'a'. As shown in Figure 6b, when one lane line deviates to the left and the other to the right by the same distance 'a', the lane centerline has no deviation. As shown in Figure 6c, when both lane lines deviate to the left by different distances 'a' and 'b', the deviation of the lane centerline is (a+b) / 2. As shown in Figure 6d, when one lane line deviates to the left and the other to the right by distances 'b' and 'a' respectively, the deviation of the lane centerline is |ba| / 2.

[0060] This application also provides a method for quantitatively assessing the safety of an autonomous driving system, comprising: evaluating the safety performance of the autonomous driving system using the quantitative targets described in any of the preceding methods. Thus, the safety performance of the autonomous driving system can be assessed through quantitative targets. The executing entity in this application embodiment can be an electronic device or a safety quantitative assessment device installed in an electronic device. The safety quantitative assessment device can be implemented through software or a combination of software and hardware. The safety quantitative assessment device can be a processor in an electronic device. The electronic device can be a device equipped with an autonomous driving system or other devices connected to the autonomous driving system.

[0061] In some optional embodiments, the above-described evaluation of the safety performance of the autonomous driving system includes: determining the lane line detection error based on the detected lane line positions; comparing the lane line detection error with a quantized target corresponding to the lane line position detection error; and evaluating the safety performance of the autonomous driving system based on the comparison result. Thus, the lane line detection error can be determined using the above method, and this error can be used to evaluate the safety performance of the autonomous driving system.

[0062] In some optional embodiments, the above-mentioned evaluation of the safety performance of the autonomous driving system includes: determining the lane centerline detection error by detecting the lane centerline position; comparing the lane centerline detection error with a quantized target corresponding to the lane centerline position detection error; and evaluating the safety performance of the autonomous driving system based on the comparison result. Thus, the lane centerline detection error can be determined using the above method, and then the lane centerline error can be used to evaluate the safety performance of the autonomous driving system.

[0063] Optionally, the above method further includes: if the safety performance assessment does not meet the preset requirements, then optimizing the corresponding algorithm and conducting another test and assessment. Furthermore, if the safety performance assessment of lane line detection or lane center line detection does not meet the preset requirements, the algorithm modules related to lane line detection or lane center line detection can be optimized, and then tested and assessed again until the assessment performance meets the preset requirements. The preset requirements are based on the assessment results of the aforementioned evaluation of the safety performance of the autonomous driving system, indicating that the system meets safety requirements. In a specific example, when the maximum allowable detection error of the lane line detection module at a certain detection distance is an outward lateral error of 1.01m and an inward lateral error of 1.5m, if the lane line detection error at that detection distance exceeds the outward lateral error of 1.01m and the inward lateral error of 1.5m, for example, an outward offset of 1.6m, then according to the comparison results, this error exceeds the safety range, and the system is assessed as not meeting the safety requirements. That is, the safety performance assessment of this module does not meet the preset requirements, and the lane line detection algorithm needs to be optimized to meet the preset requirements.

[0064] Optionally, the above method further includes: if the safety performance assessment meets preset requirements, deploying the corresponding algorithm to the autonomous driving system. Thus, when the performance assessment meets preset requirements, the corresponding algorithm or module can be deployed to the autonomous driving system. In a specific example, when the maximum allowable detection error of the lane detection module at a certain detection distance is an outward lateral error of 0.99m and an inward lateral error of 1.5m, if the lane detection error at that detection distance does not exceed the outward lateral error of 0.99m and the inward lateral error of 1.5m (e.g., an outward lateral error of 0.8m and an inward lateral error of 1.2m), then according to the comparison results, this error is within the safe range, and the system is assessed as meeting the safety requirements. That is, the module's safety performance assessment meets the preset requirements, and the lane detection algorithm or module can be deployed to the autonomous driving system.

[0065] The method for evaluating the safety performance of an autonomous driving system based on lane centerline detection error is similar to the example above, and will not be repeated here.

[0066] In other embodiments, this application also provides a non-volatile computer storage medium storing computer-executable instructions that can execute the method for determining the safety performance quantification index of the lane detection module in any of the above method embodiments.

[0067] As one implementation, the non-volatile computer storage medium of this application stores computer-executable instructions, which are configured as follows:

[0068] The lane detection uncertainty of multiple algorithm modules of the lane detection module is obtained, wherein the hazardous events of the lane detection module are decomposed into lane detection uncertainties caused by different algorithm modules of the lane detection module;

[0069] The probability of hazardous events caused by the lane line detection uncertainty of the lane line detection module is modeled in a scenario to obtain driver models under different hazardous events;

[0070] The quantization targets corresponding to different algorithm modules of the lane detection module are obtained based on driver models under different hazardous events.

[0071] As another implementation, for an autonomous driving system, the non-volatile computer storage medium of this application stores computer-executable instructions, which can execute the safety quantification assessment method for an autonomous driving system in any of the above method embodiments. The computer-executable instructions are set as follows:

[0072] The safety performance of the autonomous driving system is evaluated using the quantitative targets described in the above implementation method.

[0073] The non-volatile computer-readable storage medium may include a program storage area and a data storage area. The program storage area may store an operating system and an application program required for at least one function. The data storage area may store data created based on the use of the lane detection module's safety performance quantification indicator determination device or the safety quantification assessment device for an autonomous driving system. Furthermore, the non-volatile computer-readable storage medium may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the non-volatile computer-readable storage medium may optionally include memory remotely configured relative to a processor. This remote memory can be connected via a network to the lane detection module's safety performance quantification indicator determination device or the safety quantification assessment device for an autonomous driving system. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0074] This application also provides a computer program product, which includes a computer program stored on a non-volatile computer-readable storage medium. The computer program includes program instructions, which, when executed by a computer, cause the computer to execute any of the above-mentioned methods for determining the safety performance quantification index of the lane detection module and the safety quantification assessment method for an autonomous driving system.

[0075] Figure 7 is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. As shown in Figure 7, the device includes one or more processors 710 and a memory 720. Figure 7 shows an example of one processor 710. The device for determining the safety performance quantification index of the lane detection module or for the safety quantification assessment method of an autonomous driving system may further include an input device 730 and an output device 740. The processor 710, memory 720, input device 730, and output device 740 can be connected via a bus or other means. Figure 7 shows an example of connection via a bus. The memory 720 is the aforementioned non-volatile computer-readable storage medium. The processor 710 executes various functional applications and data processing of the server by running non-volatile software programs, instructions, and modules stored in the memory 720, thereby implementing the method for determining the safety performance quantification index of the lane detection module or the safety quantification assessment method of an autonomous driving system as described in the above method embodiment. The input device 730 can receive input digital or character information and generate key signal inputs related to user settings and function control of the safety quantification assessment device for an autonomous driving system. The output device 740 may include a display screen or other display device.

[0076] This application also provides a mobile platform, which includes: a vehicle body, a power system, and electronic devices as described in the above embodiments. The power system is installed on the vehicle body and provides power; the principle and implementation of the electronic devices are consistent with those described in the above embodiments, and will not be repeated here. The electronic devices may be a controller installed on the mobile platform, or other computing devices connected to the controller. Optionally, the mobile platform may include at least one of the following: a vehicle, a mobile robot, or an unmanned vehicle.

[0077] The above-described product can perform the methods provided in the embodiments of this application, and has the corresponding functional modules and beneficial effects for performing the methods. Technical details not described in detail in this embodiment can be found in the methods provided in the embodiments of this application.

[0078] In one embodiment, the above-described electronic device is applied in a safety performance quantification index determination device for a lane detection module, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0079] The lane detection uncertainty of multiple algorithm modules of the lane detection module is obtained, wherein the hazardous events of the lane detection module are decomposed into lane detection uncertainties caused by different algorithm modules of the lane detection module;

[0080] The probability of hazardous events caused by the lane line detection uncertainty of the lane line detection module is modeled in a scenario to obtain driver models under different hazardous events;

[0081] The quantization targets corresponding to different algorithm modules of the lane detection module are obtained based on driver models under different hazardous events.

[0082] In one embodiment, the above-described electronic device is applied in a safety quantification assessment apparatus for an autonomous driving system, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0083] The safety performance of the autonomous driving system is evaluated using the quantitative targets described in the above implementation method.

[0084] The electronic devices described in this application exist in various forms, including but not limited to:

[0085] (1) Mobile communication devices: These devices are characterized by their mobile communication capabilities and primarily aim to provide voice and data communication. These terminals include: smartphones (e.g., iPhones), multimedia phones, feature phones, and low-end phones, etc.

[0086] (2) Ultra-mobile personal computer devices: These devices fall under the category of personal computers, possessing computing and processing capabilities, and generally also have mobile internet access features. These terminals include PDAs, MIDs, and UMPCs, such as the iPad.

[0087] (3) Portable entertainment devices: These devices can display and play multimedia content. This category includes audio and video players (such as iPods), handheld game consoles, e-book readers, as well as smart toys and portable car navigation devices.

[0088] (4) Server: A device that provides computing services. The components of a server include a processor, hard disk, memory, system bus, etc. Servers are similar to general computer architectures, but because they need to provide highly reliable services, they have higher requirements in terms of processing power, stability, reliability, security, scalability, and manageability.

[0089] (5) Other electronic devices with data interaction functions.

[0090] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0091] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for determining the safety performance quantification index of a lane detection module, used in an autonomous driving system, comprising: The lane detection uncertainty of multiple algorithm modules of the lane detection module is obtained, wherein the hazardous events of the lane detection module are decomposed into lane detection uncertainties caused by different algorithm modules of the lane detection module; The probability of hazardous events caused by the lane line detection uncertainty of the lane line detection module is modeled in a scenario to obtain driver models under different hazardous events; The quantization targets corresponding to different algorithm modules of the lane detection module are obtained based on driver models under different hazardous events.

2. The method according to claim 1, characterized in that, The lane line detection uncertainties of the multiple algorithm modules of the lane line detection module include: When lane detection uncertainties occur in different algorithm modules of the lane detection module, the lane detection uncertainty that leads to the exposure of harmful behavior is obtained, wherein the harmful behavior is uncontrollable.

3. The method according to claim 1, characterized in that, The lane line detection uncertainty of the different algorithm modules of the lane line detection module includes lane line position detection error and lane center line position detection error.

4. The method according to claim 3, characterized in that, The lane line detection uncertainty refers to the lane line position detection error. The process of modeling the probability of hazardous events caused by the lane line detection uncertainty of the lane line detection module to obtain driver models under different hazardous events includes: When there is a vehicle in the adjacent lane of the vehicle's road, a scene model is performed on the scenario of deviation from the vehicle's road caused by the lane line position detection error to obtain a driver model corresponding to the lane line position detection error. The quantization objectives corresponding to the different algorithm modules of the lane detection module, which are based at least on driver models under different hazardous events, include: Based at least on the driver model corresponding to the lane line position detection error, the maximum offset that can keep the vehicle from deviating from its lane by using lateral acceleration in the opposite direction is obtained, and the quantitative target corresponding to the lane line position detection error of the autonomous driving system is obtained.

5. The method according to claim 4, characterized in that, The quantization target corresponding to the lane line position detection error of the autonomous driving system includes: Simulations were performed based on the driver model and actual scene parameters corresponding to the lane line position detection error to obtain the maximum allowable detection error of the lane line detection model at different detection distances. The maximum allowable detection error at different detection distances is used as the quantization target corresponding to the lane line position detection error of the autonomous driving system.

6. The method according to claim 5, characterized in that, The simulation based on the driver model corresponding to the lane line position detection error and the actual scene parameters includes: The lane lines detected at the far end by the lane line detection module are temporarily stored as the detected lane lines; When the vehicle travels to the lane line at the far end, the detected lane line will be taken as the true lane line. The detected lane line is compared with the true lane line to obtain the lane line position detection error.

7. The method according to claim 6, characterized in that, After obtaining the lane line position detection error, the method further includes: Determine whether the current detection error exceeds the maximum detection error for the current detection distance; If the current detection error exceeds the maximum detection error for the current detection distance, record the scenario causing the excessive error.

8. The method according to claim 4, characterized in that, The lane line detection uncertainty refers to the lane centerline position detection error. Scenario modeling is performed on the probability of hazardous events caused by the lane line detection uncertainty of the lane line detection module to obtain driver models under different hazardous events, including: When there is a vehicle in the adjacent lane of the vehicle, a driver model is obtained by performing scene modeling on the deviation from the vehicle's road caused by the lane centerline position detection error, and the driver model is corresponding to the lane centerline position detection error. The quantization objectives corresponding to the different algorithm modules of the lane detection module, which are based at least on driver models under different hazardous events, include: Based at least on the driver model corresponding to the lane centerline position detection error, the maximum offset that can keep the vehicle from deviating from its lane by using lateral acceleration in the opposite direction is obtained, and the quantitative target corresponding to the lane centerline position detection error of the autonomous driving system is obtained, wherein the lane centerline position detection error is obtained by back-calculation of the lane line position detection error.

9. The method according to claim 8, characterized in that, Different types of lane centerline deviation include: The deviation of the two lane lines in the same direction by a distance 'a' results in a deviation of the lane centerline by a distance 'a'. The two lane lines deviate by the same distance 'a' in different directions, resulting in no deviation of the lane centerline; The two lane lines deviate from each other by distances a and b in the same direction, resulting in a lane centerline deviation of (a+b) / 2; and One lane deviates by a distance b in the first direction, and the other lane deviates by a distance a in the second direction, resulting in a lane centerline deviation of |ba| / 2.

10. A method for safety quantification assessment of autonomous driving systems, characterized in that, include: The safety performance of the autonomous driving system is evaluated using the quantification target in any one of claims 1-9.

11. The method according to claim 10, characterized in that, The evaluation of the safety performance of the autonomous driving system includes: The lane line detection error is determined by the detected lane line position; The lane line detection error is compared with the quantization target corresponding to the lane line position detection error. Based on the comparison results, the safety performance of the autonomous driving system is evaluated.

12. The method according to claim 10, characterized in that, The evaluation of the safety performance of the autonomous driving system includes: The lane centerline detection error is determined by the detected lane centerline position; The lane centerline detection error is compared with the quantized target corresponding to the lane centerline position detection error. Based on the comparison results, the safety performance of the autonomous driving system is evaluated.

13. The method according to claim 11 or 12, characterized in that, Also includes: If the security performance assessment does not meet the preset requirements, the corresponding algorithm will be optimized and the test and assessment will be conducted again.

14. The method according to claim 11 or 12, characterized in that, Also includes: If the safety performance assessment meets the preset requirements, the corresponding algorithm will be deployed to the autonomous driving system.

15. An electronic device comprising: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the method according to any one of claims 1-14.

16. A storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps of the method according to any one of claims 1-14.

17. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method described in any one of claims 1-14.

18. A portable platform comprising the electronic device as claimed in claim 15.

Images