Safety evaluation method and device for automatic driving vehicle

By calculating the accident rate and mileage ratio of each driving stage of autonomous driving vehicles, and dividing and weighting the safety indicators of each stage, the problem of complexity of safety evaluation of autonomous driving vehicles is solved, and the accuracy of safety of autonomous driving vehicles is achieved Assessment and risk optimization.

CN119004052APending Publication Date: 2024-11-22YINGCHE XINGCHUANG INTELLIGENT TECH (SHANGHAI) CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410872567.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-01
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The prior art cannot effectively evaluate the safety of autonomous driving vehicles, especially in complex autonomous driving scenarios, and the traditional safety evaluation method of manual driving vehicles cannot be applied to autonomous driving vehicles.

Method used

By determining the accident rate and mileage ratio of autonomous driving vehicles in different driving stages, we calculate the comprehensive safety indicators of each driving stage, including the accident rate of autonomous driving stages and fallback stages, divide different hazard scenarios and weight calculations, and finally based on comprehensive Safety indicators evaluate the safety of a vehicle.

Benefits of technology

The accurate evaluation of the safety of autonomous driving vehicles is achieved, and the safety risks in each driving stage can be more effectively identified and optimized, and the overall safety of autonomous driving vehicles can be improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004052A_ABST
    Figure CN119004052A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of automatic driving, and provides a safety evaluation method and device for an automatic driving vehicle, and the method comprises the steps: determining the accident rate of the automatic driving vehicle in the driving mileage of a stage corresponding to each driving stage, and the driving stage comprises an automatic driving stage and a fallback stage; counting the mileage proportion of the stage mileage corresponding to each driving stage in the total driving mileage; and based on the accident rate of each driving stage and the mileage ratio of each driving stage, determining comprehensive safety indexes of all driving stages, and evaluating the safety of the vehicle according to the comprehensive safety indexes. According to the driving characteristics of the automatic driving vehicle, the driving process is divided into different driving stages, the accident rate of each driving stage in the driving mileage of the corresponding stage is calculated, and the comprehensive safety indexes of all the driving stages are obtained according to the accident rate of each driving stage and the mileage ratio of each driving stage. The safety of the automatic driving vehicle is evaluated according to the comprehensive safety index.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a method and device for evaluating the safety of an autonomous driving vehicle. Background Art

[0002] With the rapid development of smart cars, the application of autonomous driving technology and smart cockpit technology is becoming more and more extensive. In terms of autonomous driving, due to the complexity and long-tail problems of the scene, L4 level autonomous driving (L4 and above are completely driven by the autonomous driving system, which bears all driving responsibilities) is difficult to achieve at full speed on open roads in the short term. At present, the highest level of autonomous driving can reach L3 among the mass-produced models. The characteristic of L3 level is that the autonomous driving system and the driver share the responsibility, that is, L3 vehicles are required to be equipped with a safety officer (or fallback ready user, Fallback Ready User).

[0003] Regardless of the level of autonomous driving, the safety of autonomous driving vehicles is a primary concern. Due to the complexity and variety of autonomous driving scenarios, safety evaluation is more complex than that of traditional manually driven vehicles. Smart cockpit technology has greatly enriched the original human-computer interaction methods and content, and can indirectly support the improvement of driving safety. It can be seen that the safety evaluation of traditional manually driven vehicles cannot be applied to autonomous driving vehicles. Therefore, how to conduct safety evaluation of autonomous driving vehicles is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The present invention provides a method and device for evaluating the safety of an autonomous driving vehicle, which are used to solve the above-mentioned technical problems existing in the prior art.

[0005] The present invention provides a method for evaluating the safety of an autonomous driving vehicle, comprising the following steps.

[0006] Determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, wherein the driving stage includes: an autonomous driving stage and a fallback stage.

[0007] Count the proportion of the mileage corresponding to each driving stage in the total mileage.

[0008] Based on the accident rate in each driving stage and the mileage ratio in each driving stage, a comprehensive safety index for all driving stages is determined, and the vehicle safety is evaluated based on the comprehensive safety index.

[0009] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, for an autonomous driving stage, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is determined, including the following steps.

[0010] Within the stage mileage corresponding to the autonomous driving stage, the scenario accident rate under each dangerous scenario corresponding to the autonomous driving is calculated respectively.

[0011] The accident rate of each scenario is summed up to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage.

[0012] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, for an autonomous driving stage, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is determined, including the following steps.

[0013] Within the stage mileage corresponding to the autonomous driving stage, the scenario accident rate under each dangerous scenario corresponding to the autonomous driving is calculated respectively.

[0014] Based on the accident rate in each scenario and the combined proportion of failures in reminding the driver to enter the driving state, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is determined.

[0015] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, the dangerous scenarios corresponding to the autonomous driving function include the following scenarios.

[0016] Non-sudden potentially dangerous driving scenarios correspond to the accident rate of the first scenario.

[0017] The fault scenario in which there is a danger in the system failure but the diagnosis is not covered and the fallback cannot be triggered corresponds to the accident rate of the second scenario.

[0018] The dangerous scenario of completely passive driving where there is no time to respond corresponds to the third scenario accident rate.

[0019] Unrecognizable or sudden potentially dangerous driving scenarios correspond to the fourth scenario accident rate.

[0020] Among them, within the stage mileage corresponding to the automatic driving stage, the scenario accident rate under each dangerous scenario corresponding to the automatic driving is calculated respectively, including the following steps.

[0021] For the first scenario accident rate, the first occurrence frequency of each dangerous driving behavior in the non-sudden potentially dangerous driving scenario and the first behavior accident rate of the dangerous driving behavior within the stage mileage corresponding to the automatic driving stage are counted, and the first occurrence frequency is used as the weight to sum the weighted first behavior accident rates corresponding to all dangerous driving behaviors to obtain the first scenario accident rate.

[0022] The second scenario accident rate is determined based on the diagnostic coverage of the autonomous driving vehicle.

[0023] The third scenario accident rate is determined based on the statistical probability of the occurrence of a completely passive driving dangerous scenario in which there is no time to respond.

[0024] For the fourth scenario accident rate, the second occurrence frequency of each dangerous driving behavior in the unrecognizable or sudden potentially dangerous driving scenario and the second behavior accident rate of the dangerous driving behavior within the stage mileage corresponding to the automatic driving stage are statistically calculated. The second occurrence frequency is used as a weight, and the weighted sum of the second behavior accident rates corresponding to all dangerous driving behaviors is obtained to obtain the fourth scenario accident rate.

[0025] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, the dangerous scenarios corresponding to the autonomous driving function include the following scenarios.

[0026] Non-sudden potentially dangerous driving scenarios correspond to the accident rate of the first scenario.

[0027] The fault scenario in which there is a danger in the system failure but the diagnosis is not covered and the fallback cannot be triggered corresponds to the accident rate of the second scenario.

[0028] The dangerous scenario of completely passive driving where there is no time to respond corresponds to the third scenario accident rate.

[0029] Unrecognizable or sudden potentially dangerous driving scenarios correspond to the fourth scenario accident rate.

[0030] Among them, based on the accident rate of each scenario and the comprehensive proportion of failure to remind the driver to enter the driving state, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is determined, including: summing the accident rate of the first scenario and the accident rate of the second scenario, and weightedly multiplying them with the comprehensive proportion of failure to remind the driver to enter the driving state, and then summing them with the accident rate of the third scenario and the accident rate of the fourth scenario to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage.

[0031] Alternatively, the accident rate of the first scenario and the accident rate of the second scenario are summed, and the product is weighted by the comprehensive proportion of failures in reminding the driver to enter the driving state, and then summed with the accident rate of the fourth scenario to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage.

[0032] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, for a fallback phase, the accident rate of the autonomous driving vehicle within the phase mileage corresponding to the fallback phase is determined, including the following steps.

[0033] Within the stage mileage corresponding to the fallback stage, the accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered are calculated respectively.

[0034] Based on the fallback trigger frequency and the accident rates in each of the multiple time periods, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined.

[0035] According to a safety evaluation method for an autonomous driving vehicle provided by the present invention, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined based on the fallback trigger frequency and the accident rates of each of the multiple time periods, including: taking the maximum value of the time period accident rates corresponding to the multiple time periods, and then multiplying it by the fallback trigger frequency to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage.

[0036] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, for a fallback phase, the accident rate of the autonomous driving vehicle within the phase mileage corresponding to the fallback phase is determined, including the following steps.

[0037] Within the stage mileage corresponding to the fallback stage, the accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered are calculated respectively.

[0038] The proportion of untaken services corresponding to the multiple time periods is counted.

[0039] Based on the fallback trigger frequency, the accident rate in the time period corresponding to the multiple time periods and the proportion of non-takeover, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined.

[0040] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined based on the fallback trigger frequency, the period accident rate corresponding to multiple time periods, and the non-takeover ratio, including the following steps.

[0041] The weighted value of the time period accident rate corresponding to each time period is calculated using the untaken-over ratio as the weight.

[0042] The maximum value of the weighted values ​​of the accident rates in the time periods corresponding to the multiple time periods is taken, and then the maximum value is multiplied by the fallback trigger frequency to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage.

[0043] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, within the stage mileage corresponding to the fallback stage, the time period accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered are calculated respectively, including the following steps.

[0044] For each of the time periods, the third occurrence frequency of each dangerous driving behavior in the time period and the third behavior accident rate of the dangerous driving behavior within the stage mileage corresponding to the fallback stage are counted.

[0045] The third occurrence frequency is used as a weight to perform a weighted summation of the third behavior accident rates corresponding to all dangerous driving behaviors to obtain the period accident rate of the corresponding time period.

[0046] According to a method for evaluating the safety of an autonomous driving vehicle provided by the present invention, the driving stage also includes: a first manual driving stage within a preset time period after the driver takes over and a second manual driving stage after the first manual driving stage.

[0047] According to a safety evaluation method for an autonomous driving vehicle provided by the present invention, for a first manual driving stage, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the first manual driving stage is calculated, including the following steps.

[0048] Statistics are collected on the proportion of accidental takeover and accident rate of driver’s mistaken takeover. Among them, the driver’s mistaken takeover occurs when the driver takes both hands off the steering wheel after exiting the automatic driving mode.

[0049] Statistics were collected on the takeover accident rate, the frequency of distraction events, and the failure rate of distraction assistance reminders after the driver took over and was distracted.

[0050] Taking the proportion of false takeover as the weight of the false takeover accident rate, taking the frequency of distraction events and the proportion of reminder failure as the weight of the takeover accident rate, and taking the weighted sum of the false takeover accident rate and the takeover accident rate, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the first manual driving stage is obtained.

[0051] According to a safety evaluation method for an autonomous driving vehicle provided by the present invention, a comprehensive safety index of all driving stages is determined based on the accident rate of each driving stage and the mileage proportion of each driving stage, including: taking the mileage proportion of each driving stage as a weight, and weighted summing the accident rate of each driving stage and the corresponding weight to obtain the comprehensive safety index of all driving stages.

[0052] The present invention also provides an autonomous driving vehicle safety evaluation device, comprising the following modules.

[0053] The accident rate determination module is used to determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, and the driving stage includes: an autonomous driving stage and a fallback stage.

[0054] The mileage proportion statistics module is used to count the proportion of the stage mileage corresponding to each driving stage in the total mileage.

[0055] The safety index determination module is used to determine the comprehensive safety index of all driving stages based on the accident rate of each driving stage and the mileage proportion of each driving stage, and evaluate the vehicle safety with the comprehensive safety index.

[0056] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, an autonomous driving vehicle safety evaluation method as described in any one of the above is implemented.

[0057] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the autonomous driving vehicle safety evaluation method as described in any one of the above-mentioned methods is implemented.

[0058] The method and device for evaluating the safety of an autonomous driving vehicle provided by the present invention determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, and the driving stage includes: the autonomous driving stage and the fallback stage; count the mileage proportion of the stage mileage corresponding to each driving stage in the total mileage; based on the accident rate of each driving stage and the mileage proportion of each driving stage, determine the comprehensive safety index of all driving stages, and evaluate the safety of the vehicle with the comprehensive safety index. The present invention divides the driving process into different driving stages according to the driving characteristics of the autonomous driving vehicle, calculates the accident rate of each driving stage within the corresponding stage mileage, and then obtains the comprehensive safety index of all driving stages according to the accident rate of each driving stage and the mileage proportion of each driving stage, evaluates the safety of the autonomous driving vehicle according to the comprehensive safety index, and realizes the safety evaluation of the autonomous driving vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0060] Figure 1It is a flow chart of the autonomous driving vehicle safety evaluation method provided by the present invention.

[0061] Figure 2 It is a schematic diagram of the structure of the autonomous driving vehicle safety evaluation device provided by the present invention.

[0062] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] The autonomous driving vehicle safety evaluation method according to the embodiment of the present invention is as follows: Figure 1 As shown, the following steps S110 to S130 are included.

[0065] Step S110: Determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, wherein the driving stage includes: the autonomous driving stage and the fallback stage. For autonomous driving vehicles, especially autonomous driving vehicles with L4 level and above, the driving stage includes the autonomous driving stage and the fallback stage; and for driving stages with autonomous driving functions of L3 level and below, the autonomous driving journey can be completely decomposed into: the autonomous driving stage (corresponding to DDT in SAE J3016), the fallback stage (corresponding to DDT fallback in SAE J3016, that is, when the system finds a scenario that cannot be solved, it reminds the driver to take over, until the driver completes the takeover or stops the vehicle) and the manual driving stage, a total of three stages. Since the factors or scenarios that affect the accident rate are different in different driving stages, in order to accurately evaluate the safety of the vehicle, in this step, for different driving stages, the accident rate within the stage mileage corresponding to the driving stage is calculated respectively, that is, the comprehensive safety index (or called the comprehensive accident rate) is decomposed into the accident rate corresponding to different driving stages.

[0066] Step S120: Count the proportion of the stage mileage corresponding to each driving stage in the total mileage. In the total mileage of the vehicle, the mileage of different driving stages is different. For vehicles at level L4 and above, automatic driving is the main method. When the automatic driving system finds a scenario that cannot be solved, it will trigger fallback until the vehicle pulls over. Obviously, the stage mileage corresponding to the fallback stage is much smaller than the stage mileage corresponding to the automatic driving stage, and the accident rate of accidents occurring in the fallback stage has little impact on the comprehensive safety index. That is, due to the different stage mileages of different driving stages, the corresponding accident rate has different impacts on the comprehensive safety index. Therefore, in this step, the mileage proportion of the stage mileage corresponding to each driving stage in the total mileage is counted, and the mileage proportion is also used as a consideration factor for the comprehensive safety index, so that the final comprehensive safety index is more in line with the actual situation and the safety evaluation of the vehicle is more accurate.

[0067] Step S130: Based on the accident rate of each driving stage and the mileage ratio of each driving stage, determine the comprehensive safety index of all driving stages, and evaluate the vehicle safety with the comprehensive safety index. The comprehensive safety index is the comprehensive accident rate. The lower the comprehensive safety index, the higher the safety of the autonomous driving vehicle.

[0068] The comprehensive safety index is determined by the accident rate of each driving stage and the mileage proportion of each driving stage. The comprehensive safety assessment result can be obtained by upward aggregation calculation of the accident rate of a single driving stage; or it can be based on a pre-proposed comprehensive safety index and downward decomposition calculation to propose index requirements for the accident rate of a single driving stage, so as to improve certain driving functions corresponding to the single driving stage and reduce the accident rate of a single driving stage.

[0069] The autonomous driving vehicle safety evaluation method of this embodiment divides the driving process into different driving stages according to the driving characteristics of the autonomous driving vehicle, calculates the accident rate of each driving stage within the corresponding mileage, and then obtains the comprehensive safety index of all driving stages according to the accident rate of each driving stage and the mileage proportion of each driving stage. The safety of the autonomous driving vehicle is evaluated according to the comprehensive safety index, thereby realizing the safety evaluation of the autonomous driving vehicle.

[0070] In some embodiments, in step S110, for the autonomous driving stage, determining the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage includes the following steps.

[0071] Step 1: Calculate the scenario accident rate of each dangerous scenario corresponding to the autonomous driving stage within the stage mileage corresponding to the autonomous driving stage.

[0072] Step 2: Sum the accident rates of each scenario to obtain the accident rate of the autonomous driving vehicle within the mileage corresponding to the autonomous driving stage.

[0073] For the autonomous driving stage, different dangerous scenarios lead to different accident rates. Therefore, the scenario accident rates under different dangerous scenarios are calculated separately, and then the different scenario accident rates are statistically summed to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage, so that the final accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is more accurate.

[0074] In the normal autonomous driving stage, accident risks come from several aspects: potentially dangerous driving behaviors in the autonomous driving function, dangerous system failures that are not covered by the diagnosis and cannot trigger fallback, and completely passive driving risks that are too late to respond (such as: the car is driving normally, other vehicles suddenly turn and hit the car, and the car has no time to dodge). Among the above three scenarios, for the first type of scenario, i.e., potentially dangerous driving behaviors in the autonomous driving function, such scenarios can be divided into the following two types: non-sudden potentially dangerous driving scenarios of the autonomous driving function, and unrecognizable or sudden potentially dangerous driving scenarios of the autonomous driving function. Therefore, in some embodiments, the various dangerous scenarios corresponding to the autonomous driving function include: Non-sudden potential dangerous driving scenario, corresponding to the first scenario accident rate S 11 .

[0075] The system failure is dangerous but the diagnosis is not covered, so the fallback cannot be triggered. The corresponding second scenario accident rate S 12 .

[0076] The completely passive driving dangerous scenario where there is no time to respond corresponds to the third scenario accident rate S 13 .

[0077] Unrecognizable or sudden potential dangerous driving scenarios, corresponding to the fourth scenario accident rate S 14 .

[0078] Based on the above four dangerous driving scenarios in the autonomous driving stage, the scenario accident rates of each dangerous scenario corresponding to the autonomous driving are calculated within the corresponding stage mileage of the autonomous driving stage, including: For the first scenario, the accident rate S 11 , the first occurrence frequency of each dangerous driving behavior in non-sudden potential dangerous driving scenarios and the first behavior accident rate of the dangerous driving behavior in the stage mileage corresponding to the automatic driving stage are counted, and the first behavior accident rate corresponding to all dangerous driving behaviors is weighted and summed, with the first occurrence frequency as the weight, to obtain the first scenario accident rate S 11 .

[0079] Specifically, for non-sudden potentially dangerous driving scenarios, they can be mainly divided into the following dangerous driving behavior categories.

[0080] 1) The vehicle’s maximum braking force is insufficient.

[0081] 2) Severe swaying (the vehicle sways continuously from side to side).

[0082] 3) The vehicle drives over the line.

[0083] 4) Speeding.

[0084] 5)…….

[0085] The first occurrence frequencies of the above dangerous driving behaviors are f 111 、f 112 、f 113 、f 114 , ..., the accident rates of the first behavior of the above dangerous driving behaviors in the corresponding stage mileage of the automatic driving stage are S 111 , S 112 , S 113 , S 114 , ..., thus obtaining the first scenario accident rate S 11 .

[0086] S 11 = S 111 *f 111 +S 112 *f 112 +S 113 *f 113 +S 114 *f 114 +…… (1).

[0087] The first occurrence frequency and the first behavior accident rate can be estimated based on data statistical analysis, that is, obtained through data statistics. For long-tail problems, if there is a difference in order of magnitude, it can be ignored. For example: Here it is assumed that: S 111 *f 111 , S 112 *f 112 >>S 113 *f 113 , S 114 *f 114 , ..., then formula (1) becomes the following formula (2).

[0088] S 11 ≈ S 111 *f 111 +S 112 *f 112 (2).

[0089] From the above formula (1) or (2), we can know that in order to control the accident rate S of the first scenario 11 For each dangerous driving behavior in this scenario, safety can be improved by reducing the first occurrence frequency and the first behavior accident rate. And focusing on optimizing the category with the largest product value is more conducive to significantly reducing the first scenario accident rate S 11 .

[0090] The second scenario accident rate S 12 Based on the diagnosis coverage of the autonomous vehicle, the diagnosis coverage can be determined by referring to ISO-26262, etc. Assuming that the diagnosis coverage of the vehicle is d, the larger the diagnosis coverage d, the more diagnosis cases can be covered and the higher the safety. Therefore, S 12 Related to d, for example: S 12 ≈1-d. Therefore, in order to control the accident rate S in the second scenario 12 , it is necessary to increase diagnostic coverage as much as possible.

[0091] The third scenario accident rate S 13 Based on the statistical probability of the occurrence of a completely passive driving dangerous scenario where there is no time to respond.

[0092] For the fourth scenario accident rate S 14 , the second occurrence frequency of each dangerous driving behavior in the unrecognizable or sudden potential dangerous driving scene and the second behavior accident rate of the dangerous driving behavior in the stage mileage corresponding to the automatic driving stage are counted, and the second occurrence frequency is used as the weight to sum the second behavior accident rates corresponding to all dangerous driving behaviors to obtain the fourth scene accident rate S 14 .

[0093] Specifically, unrecognizable or sudden potentially dangerous driving scenarios can be mainly divided into the following dangerous driving behavior categories.

[0094] 1) Turn the steering wheel sharply.

[0095] 2) Driving in blind spots.

[0096] 3) Heavy brakes.

[0097] 4) Changing lanes seriously crossing the line.

[0098] 5) Unrecognized vehicles / objects.

[0099] 6)…….

[0100] As above, let the second occurrence frequency of the above dangerous driving behavior be f 141 、f 142 、f 143 、f144 、f 145 , ..., the second behavior accident rates of the above dangerous driving behaviors in the corresponding stage mileage of the automatic driving stage are S 141 , S 142 , S 143 , S 144 , S 145 , ..., and thus the accident rate S of the fourth scenario is obtained 14 .

[0101] S 14 = S 141 *f 141 +S 142 *f 142 +S 143 *f 143 +S 144 *f 144 + S 145 *f 145 +…… (3).

[0102] The above second occurrence frequency and second behavior accident rate can be estimated based on data statistical analysis, that is, obtained through data statistics. For long-tail problems, if there is a difference in order of magnitude, it can be ignored. For example: Here it is assumed that: S 141 *f 141 , S 142 *f 142 , S 143 *f 143 >>S 144 *f 144 , S 145 *f 145 , ..., then formula (3) becomes the following formula (4).

[0103] S 14 ≈ S 141 *f 141 +S 142 *f 142 +S 143 *f 143 (4).

[0104] From the above formula (3) or (4), we can know that in order to control the accident rate S of the fourth scenario 14 For each dangerous driving behavior in this scenario, safety can be improved by reducing the second occurrence frequency and the second behavior accident rate. And focusing on optimizing the category with the largest product value is more conducive to significantly reducing the fourth scenario accident rate S 14 .

[0105] After calculating the accident rates in different dangerous scenarios during the autonomous driving stage, the accident rate S1 of the autonomous driving vehicle within the corresponding stage mileage during the autonomous driving stage is expressed by the following formula (5).

[0106] S1= S 11 + S 12 + S 13 + S 14 (5).

[0107] Considering the current road scenario, S 13 The possibility is very small, and because the cause of the danger is not at all due to the vehicle itself, S can be omitted here. 13 , that is, we get formula (6).

[0108] S1≈S 11 + S 12 +S 14 (6).

[0109] In some embodiments, in step S110, for the fallback stage, determining the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage includes the following steps.

[0110] Step 1: In the mileage corresponding to the fallback phase, calculate the accident rates for the preset non-overlapping and continuous time periods after the fallback is triggered. Since the situation that the autonomous driving system is difficult to deal with has occurred, the accident risks will also be significantly different in different time intervals after the fallback is triggered. For example: Assuming that the maximum speed allowed by autonomous driving is 100km / h and the recognition distance of the vehicle's forward sensor is 200m, then in the frame before the fallback is triggered, the forward distance of 200m recognized by the autonomous driving system corresponds to the subsequent driving distance of up to 7.2s. If the condition for triggering the fallback is the failure of the forward recognition ability, then the accident risk ratio within 7.2s and the accident risk ratio above 7.2s will be significantly different. Therefore, after the fallback is triggered, preset multiple non-overlapping and continuous time periods, and calculate the corresponding time period accident rate for each time period, where the number of time periods and the length of each time period can be set according to actual conditions.

[0111] In this embodiment, four time periods are set as follows.

[0112] Time period 21: 0~3s, that is, the first 3 seconds after the fallback is triggered.

[0113] Time period 22: 4~7s, that is, the 4th to 7th seconds after the fallback is triggered.

[0114] Time period 23: 8~15s, that is, the 8th to 15th second after the fallback is triggered.

[0115] Time period 24: 16 seconds or more, that is, from 16 seconds after the fallback is triggered until the vehicle stops.

[0116] Assume that the accident rate corresponding to the above four time periods is S 21 , S 22 , S 23 , S 24 (Usually S 21 ≤S 22 ≤S 23 ≤S 24 ).

[0117] Step 2: Based on the fallback trigger frequency and the accident rates in each of the multiple time periods, determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage.

[0118] Specifically, since the safety risk in the fallback stage is positively correlated with the fallback trigger frequency k, the maximum value of the accident rates corresponding to the multiple time periods is taken, and then multiplied by the fallback trigger frequency k to obtain the accident rate S2 of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage, as shown in the following formula (7).

[0119] S2= max(S 21 , S 22 , S 23 , S 24 )*k(7).

[0120] In some embodiments, within the stage mileage corresponding to the fallback stage, the period accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered are calculated respectively, including the following steps.

[0121] Step 1: For each time period, the third occurrence frequency of each dangerous driving behavior in the time period and the third behavior accident rate of the dangerous driving behavior in the stage mileage corresponding to the fallback stage are counted. After the fallback is triggered, the vehicle will enter the safe driving state and eventually stop the vehicle in the safe area. In this process, the potential dangerous driving behaviors of the automatic driving may include:

[0122] 1) Unrecognized vehicles / objects.

[0123] 2) Continuous lane change.

[0124] 3) Driving on the line.

[0125] 4) Parking on the line.

[0126] 5)…….

[0127] Taking time period 21 as an example, within this time period 21, the third occurrence frequencies of the above dangerous driving behaviors are f 211 、f 212 、f 213 、f 214 , ..., the third behavior accident rates corresponding to the above dangerous driving behaviors are set as S 211 , S 212 , S 213 , S 214 , .... The above third occurrence frequency and third behavior accident rate can be estimated based on data statistical analysis, that is, obtained through data statistics.

[0128] Step 2: Taking the third occurrence frequency as the weight, weighted sum of the third behavior accident rates corresponding to all dangerous driving behaviors to obtain the period accident rate S of the corresponding time period. 21 .

[0129] S 21 = S 211 *f 211 +S 212 *f 212 +S 213 *f 213 +S 214 *f 214 +…… (8).

[0130] For long-tail problems, if there is a difference in magnitude, it can be ignored. For example: Here we assume that: S 211 *f 211 , S 212 *f 212 >>S 213 *f 213 , S 214 *f 214 , ..., then formula (8) becomes the following formula (9).

[0131] S 21 ≈ S 211 *f 211 +S 212 *f 212 (9).

[0132] Similarly, the accident rate S corresponding to the other three time periods can be obtained 22 , S 23 and S 24 .

[0133] S 22 ≈ S 221 *f 221 +S 222 *f 222 (10).

[0134] S 23 ≈ S 231 *f 231 +S 232 *f 232 (11).

[0135] S 24 ≈ S 241 *f 241 +S 242 *f 242 (12).

[0136] Substituting formulas (9) to (12) into formula (7) can obtain the accident rate S2 of the autonomous driving vehicle within the corresponding stage mileage in the fallback stage.

[0137] In some embodiments, in step S130, based on the accident rate of each driving stage and the mileage ratio of each driving stage, the comprehensive safety index of all driving stages is determined, including: taking the mileage ratio of each driving stage as the weight, weighted summing the accident rate of each driving stage and the corresponding weight to obtain the comprehensive safety index of all driving stages. In the case where the driving stage is the automatic driving stage and the fallback stage, the accident rate corresponding to the automatic driving stage is S1, the mileage ratio is M1, and the accident rate corresponding to the fallback stage is S2, the mileage ratio is M2, then the comprehensive safety index S is represented by the following formula (13).

[0138] S=S1*M1+S2*M2(13).

[0139] Considering S 13 In the case of , substituting formulas (5) and (7) into (13) can obtain the comprehensive safety index S. Without considering S 13 In the case of , the comprehensive safety index S can be obtained by substituting formulas (6) and (7) into (13).

[0140] The above embodiments do not take into account the situation where the driver takes over and subsequent manual driving, and are applicable to the safety evaluation of autonomous driving vehicles of level L4 and above. The following embodiments mainly consider the situation where the driver takes over and there is a manual driving stage, and are applicable to the safety evaluation of autonomous driving vehicles of level L3 and below.

[0141] For autonomous driving vehicles at level L3 and below, considering factors such as vehicle mis-takeover, driver fatigue or distraction, the safety risk when the driver just takes over the vehicle and enters the manual driving stage is significantly higher than the manual driving stage after normal takeover. Therefore, the manual driving stage is divided into the first manual driving stage within a preset time (such as 15s, which can be set according to actual conditions) after the driver takes over and the second manual driving stage after the first manual driving stage. Therefore, autonomous driving vehicles at level L3 and below include four driving stages: autonomous driving stage, fallback stage, first manual driving stage and second manual driving stage.

[0142] Since manual takeover is possible, combined with the vehicle's smart cockpit function, HMI (Human Machine Interface) reminders can be used during the autonomous driving stage and the fallback stage, allowing the driver to take over the vehicle in both the autonomous driving stage and the fallback stage to avoid accidents. Therefore, it is necessary to consider the impact of the proportion of drivers taking over and successfully resolving dangers on the accident rates S1 and S2 corresponding to the autonomous driving stage and the fallback stage, respectively. The accident rates S1 and S2 obtained for these two driving stages are more in line with the actual situation, and are therefore more accurate and reliable.

[0143] In some embodiments, in step S110, for the autonomous driving stage, determining the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage includes the following steps.

[0144] Step 1: Calculate the accident rate of each dangerous scenario corresponding to the autonomous driving within the stage mileage corresponding to the autonomous driving stage. Specifically, the accident rate S of each scenario above can be seen. 11 ~S 14 The calculation method of .

[0145] Step 2: Based on the accident rate of each scenario and the comprehensive proportion of failure to remind the driver to enter the driving state, determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage. There is a certain proportion of drivers accepting HMI reminders to enter the driving state and resolve the accident risk, which mainly depends on whether the HMI reminder can allow the driver to enter the driving state in time. Let the comprehensive proportion of HMI reminders that allow the driver to enter the driving state in time and resolve the accident risk be H', then the comprehensive proportion of failure to remind the driver to enter the driving state is H1, H1=1-H'.

[0146] For dangerous driving scenarios, after providing HMI reminders, it is necessary to give the driver low-load, timely and effective reminders from the perspective of HMI based on the smart cockpit. Here, H' and H1 are more based on statistical results of data analysis rather than calculations.

[0147] The H' indicator can be approximated from the time it takes the DMS (Driver Fatigue Monitor System) module to recover the driver's state under the HMI reminder, and the simulation of the proportion of the system's accident risk under this response time. Because the faster the HMI reminder can drive the driver to respond accurately, the higher the possibility of timely resolving driving risks in AD mode. After statistical analysis, the driver's response time under different HMI reminder strategies and the driving safety risks that can be resolved can be obtained.

[0148] For example: Assuming that under the four smart cockpit HMI strategies of HMI1, HMI2, HMI3, and HMI4, the DMS detects that the driver needs 1s, 3s, 6s, and 10s to enter the driving state, respectively, and assuming that the simulation results show that this time plus the appropriate reaction delay, 1.5s, 3.5s, 6.5s, and 10.5s can solve 99%, 95%, 90%, and 50% of driving safety risk events, respectively, then H' is 99%, 95%, 90%, and 50%, respectively, and H1 is 1%, 5%, 10%, and 50%, respectively. Therefore, by optimizing the smart cockpit HMI reminder strategy, improving the driver's response time to dangerous scenarios or faults can also improve overall driving safety.

[0149] Specifically, the non-sudden potential dangerous driving scenario corresponds to the first scenario accident rate S 11 Since it is a non-sudden potentially dangerous driving scenario, the driver can be reminded to take over through the HMI, that is, the accident rate of the first scenario is S 11 Will be affected by H1.

[0150] The system failure is dangerous but the diagnosis is not covered, so the fallback cannot be triggered. The corresponding second scenario accident rate S 12 In this scenario, fallback cannot be triggered, but the autonomous driving system can be used to warn of a fault. Therefore, the accident rate of the second scenario is S 12 It will also be affected by H1.

[0151] The completely passive driving dangerous scenario where there is no time to respond corresponds to the third scenario accident rate S 13 , not affected by H1.

[0152] Unrecognizable or sudden potential dangerous driving scenarios, corresponding to the fourth scenario accident rate S 14 Due to the unrecognizable or sudden characteristics, the accident rate of the fourth scenario is S 14 Also not affected by H1.

[0153] Based on the accident rate of each scenario and the comprehensive proportion of failure to remind the driver to enter the driving state, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is determined, including: the first scenario accident rate S 11 and the second scenario accident rate S 12 Sum, and take the comprehensive proportion of failure to remind the driver to enter the driving state H1 as the weight to calculate the weighted product, and then add it to the accident rate S of the third scenario 13 and the fourth scenario accident rate S 14 The sum is calculated to obtain the accident rate S1 of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage.

[0154] S1= (S 11 +S 12 )*H1+S 13 +S 14 (14).

[0155] Or, ignoring the third scenario accident rate S 13 In the case of the first scenario, the accident rate S 11 and the second scenario accident rate S 12 Sum, and take the comprehensive proportion of failure to remind the driver to enter the driving state H1 as the weight to calculate the weighted product, and then add it to the accident rate S of the fourth scenario 14 The sum is calculated to obtain the accident rate S1 of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage.

[0156] S1≈ (S 11 +S 12 )*H1+S 14 (15).

[0157] In this embodiment, for the automatic driving stage, the influence of the proportion of drivers taking over and successfully resolving dangers on the accident rate S1 corresponding to the automatic driving stage is taken into account, so that S1 is more in line with the actual situation and thus more accurate and reliable.

[0158] In some embodiments, in step S110, for the fallback stage, determining the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage includes the following steps.

[0159] Step 1: Within the mileage corresponding to the fallback phase, calculate the accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered. In this embodiment, the setting of multiple time periods can refer to the four time periods set in the above embodiment, and the corresponding time period accident rates are S 21 , S 22 , S 23 and S 24 .

[0160] Step 2: Count the corresponding untaken-over ratios in the multiple time periods. The untaken-over ratios in the above four time periods are H. 21 , H 22 , H 23 and H 24 The proportion of untaken is H 21 , H 22 , H 23 and H 24 It can be obtained through data statistics.

[0161] Step 3: Based on the fallback trigger frequency and the accident rate and non-takeover ratio of each of the multiple time periods, determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage.

[0162] Specifically, the weighted value of the time period accident rate corresponding to each time period is calculated using the untaken-over ratio as a weight.

[0163] The maximum value of the weighted values ​​of the accident rates in the time periods corresponding to the multiple time periods is taken, and then the maximum value is multiplied by the fallback trigger frequency to obtain the accident rate S2 of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage.

[0164] S2= max(S 21 *H 21 , S 22 *H 22 , S 23 *H 23 , S 24 *H 24 )*k (16).

[0165] In this embodiment, S 21 , S 22 , S 23 and S 24 It can be obtained by referring to the above formulas (9) to (12).

[0166] In this embodiment, for the fallback stage, the influence of the proportion of drivers not taking over on the accident rate S2 corresponding to the fallback stage is considered, so that S2 is more in line with the actual situation and thus more accurate and reliable.

[0167] In some embodiments, in step S110, for the first manual driving stage, determining the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the first manual driving stage includes the following steps.

[0168] Step 1: Count the proportion of drivers taking over by mistake H 31 and the accident rate of wrong takeover S31 , where the driver's hands leaving the steering wheel after exiting the automatic driving mode indicates that the driver's mis-takeover event has occurred, that is, the driver's hands are off the steering wheel in the first manual driving stage. The Hands off Detection (HOD) data can show that the driver's hands are off the steering wheel, which is considered to be a mis-takeover event. The HOD data can also be used to calculate the proportion of mis-takeover events H 31 .

[0169] Step 2: Count the accident rate S of drivers taking over after they take over and are distracted (fatigue or distraction) 32 , the frequency of distraction events f 32 And the failure rate of attention-distracted auxiliary reminders H 32 It is possible to determine whether the driver is severely fatigued or distracted 10 seconds before exiting the autonomous driving mode, and to indicate that the driver may still be fatigued or distracted after exiting the autonomous driving mode, that is, a distraction event may occur.

[0170] Step 3: Take the misconnection ratio H 31 is the weight of the accident rate of wrong takeover 31 , with the frequency of distraction events f 32 and reminder failure ratio H 32 is the takeover accident rate S 32 The weight of the accident rate S 31 and takeover accident rate S 32 By weighted summation, we can obtain the accident rate S3 of the autonomous driving vehicle within the mileage corresponding to the first manual driving stage.

[0171] S3= f 31 *S 31 +S 32 *f 32 *H 32 (17).

[0172] From the above formula (17), it can be seen that in order to control the accident rate S3 within the stage mileage corresponding to the first manual driving stage, the mis-takeover ratio H can be reduced. 31 , reduce the frequency of distraction events 32 , improve the driver status assistance of the smart cockpit after exiting the automatic driving mode (i.e. reduce H 32 ) to improve security in three directions.

[0173] For the second manual driving stage, that is, entering the full manual driving stage, the full manual driving stage is the same as the traditional manual driving vehicle. The corresponding accident rate S4 can be obtained by referring to industry data.

[0174] In some embodiments, in step S130, based on the accident rate of each driving stage and the mileage ratio of each driving stage, the comprehensive safety index of all driving stages is determined, including: taking the mileage ratio of each driving stage as the weight, weighted summing the accident rate of each driving stage and the corresponding weight to obtain the comprehensive safety index of all driving stages. In the case where the driving stage is the automatic driving stage and the fallback stage, the accident rate corresponding to the automatic driving stage is S1, and the mileage ratio is M1, the accident rate corresponding to the fallback stage is S2, and the mileage ratio is M2, the accident rate corresponding to the first manual driving stage is S3, and the mileage ratio is M3, and the accident rate corresponding to the second manual driving stage is S4, and the mileage ratio is M4, then the comprehensive safety index S is represented by the following formula (18).

[0175] S=S1*M1+S2*M2+S3*M3+S4*M4 (18).

[0176] Considering S 13 In the case of , substituting formulas (14), (16) and (17) into (18) can obtain the comprehensive safety index S. Without considering S 13 In the case of , the comprehensive safety index S can be obtained by substituting formulas (15), (16) and (17) into (18).

[0177] It should be noted that the above-mentioned accident rates can be converted into the industry standard accident rate per million kilometers, and the safety of autonomous driving vehicles can be evaluated based on the comprehensive accident rate per million kilometers of autonomous driving vehicles.

[0178] The autonomous driving vehicle safety evaluation device provided by the present invention is described below. The autonomous driving vehicle safety evaluation device described below and the autonomous driving vehicle safety evaluation method described above can be referenced to each other.

[0179] The automatic driving vehicle safety evaluation device according to the embodiment of the present invention is as follows Figure 2 As shown, it includes the following modules.

[0180] The accident rate determination module 210 is used to determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, and the driving stage includes: an autonomous driving stage and a fallback stage.

[0181] The mileage proportion statistics module 220 is used to count the mileage proportion of the stage mileage corresponding to each driving stage in the total driving mileage.

[0182] The safety index determination module 230 is used to determine the comprehensive safety index of all driving stages based on the accident rate of each driving stage and the mileage proportion of each driving stage, and evaluate the vehicle safety with the comprehensive safety index.

[0183] The autonomous driving vehicle safety evaluation device of this embodiment divides the driving process into different driving stages according to the driving characteristics of the autonomous driving vehicle, calculates the accident rate of each driving stage within the corresponding mileage, and then obtains the comprehensive safety index of all driving stages according to the accident rate of each driving stage and the mileage proportion of each driving stage. The safety of the autonomous driving vehicle is evaluated according to the comprehensive safety index, thereby realizing the safety evaluation of the autonomous driving vehicle.

[0184] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute the autonomous driving vehicle safety assessment method, which includes the following steps.

[0185] Determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, wherein the driving stage includes: an autonomous driving stage and a fallback stage.

[0186] Count the proportion of the mileage corresponding to each driving stage in the total mileage.

[0187] Based on the accident rate in each driving stage and the mileage ratio in each driving stage, a comprehensive safety index for all driving stages is determined, and the vehicle safety is evaluated based on the comprehensive safety index.

[0188] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0189] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the autonomous driving vehicle safety evaluation method provided by the above-mentioned methods, which includes the following steps.

[0190] Determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, wherein the driving stage includes: an autonomous driving stage and a fallback stage.

[0191] Count the proportion of the mileage corresponding to each driving stage in the total mileage.

[0192] Based on the accident rate in each driving stage and the mileage ratio in each driving stage, a comprehensive safety index for all driving stages is determined, and the vehicle safety is evaluated based on the comprehensive safety index.

[0193] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for evaluating the safety of an autonomous driving vehicle provided by the above-mentioned methods is implemented, and the method includes the following steps.

[0194] Determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, wherein the driving stage includes: an autonomous driving stage and a fallback stage.

[0195] Count the proportion of the mileage corresponding to each driving stage in the total mileage.

[0196] Based on the accident rate in each driving stage and the mileage ratio in each driving stage, a comprehensive safety index for all driving stages is determined, and the vehicle safety is evaluated based on the comprehensive safety index.

[0197] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0198] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0199] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating the safety of an autonomous driving vehicle, characterized in that: include: Determine an accident rate of the autonomous driving vehicle within a stage mileage corresponding to each driving stage, wherein the driving stage includes: an autonomous driving stage and a fallback stage; Count the proportion of the stage mileage corresponding to each driving stage in the total driving mileage; Based on the accident rate in each driving stage and the mileage ratio in each driving stage, a comprehensive safety index for all driving stages is determined, and the vehicle safety is evaluated based on the comprehensive safety index.

2. The method for evaluating the safety of an autonomous driving vehicle according to claim 1, characterized in that: For the autonomous driving stage, determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage, including: Within the stage mileage corresponding to the autonomous driving stage, the scenario accident rate under each dangerous scenario corresponding to the autonomous driving is calculated respectively; The accident rate of each scenario is summed up to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage.

3. The method for evaluating the safety of an autonomous driving vehicle according to claim 1, characterized in that: For the autonomous driving stage, determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage, including: Within the stage mileage corresponding to the autonomous driving stage, the scenario accident rate under each dangerous scenario corresponding to the autonomous driving is calculated respectively; Based on the accident rate in each scenario and the combined proportion of failures in reminding the driver to enter the driving state, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is determined.

4. The method for evaluating the safety of an autonomous driving vehicle according to claim 2 or 3, characterized in that: The dangerous scenarios corresponding to the autonomous driving function include: Non-sudden potentially dangerous driving scenarios, corresponding to the first scenario accident rate; The fault scenario in which the system is dangerous but the diagnosis does not cover it, so the fallback cannot be triggered, corresponds to the accident rate of the second scenario; The completely passive driving dangerous scenario where there is no time to respond corresponds to the third scenario accident rate; Unrecognizable or sudden potentially dangerous driving scenarios correspond to the fourth scenario accident rate; Among them, within the stage mileage corresponding to the autonomous driving stage, the scene accident rate under each dangerous scene corresponding to the autonomous driving is calculated respectively, including: For the first scenario accident rate, the first occurrence frequency of each dangerous driving behavior in the non-sudden potential dangerous driving scenario and the first behavior accident rate of the dangerous driving behavior within the stage mileage corresponding to the automatic driving stage are counted, and the first occurrence frequency is used as the weight to weight the first behavior accident rates corresponding to all dangerous driving behaviors to obtain the first scenario accident rate; The second scenario accident rate is determined based on the diagnostic coverage of the autonomous driving vehicle; The third scenario accident rate is determined based on the statistical probability of the occurrence of a completely passive driving dangerous scenario where there is no time to respond; For the fourth scenario accident rate, the second occurrence frequency of each dangerous driving behavior in the unrecognizable or sudden potentially dangerous driving scenario and the second behavior accident rate of the dangerous driving behavior within the stage mileage corresponding to the automatic driving stage are statistically calculated. The second occurrence frequency is used as a weight, and the weighted sum of the second behavior accident rates corresponding to all dangerous driving behaviors is obtained to obtain the fourth scenario accident rate.

5. The method for evaluating the safety of an autonomous driving vehicle according to claim 3, characterized in that: The dangerous scenarios corresponding to the autonomous driving function include: Non-sudden potentially dangerous driving scenarios, corresponding to the first scenario accident rate; The fault scenario in which the system is dangerous but the diagnosis does not cover it, so the fallback cannot be triggered, corresponds to the accident rate of the second scenario; The completely passive driving dangerous scenario where there is no time to respond corresponds to the third scenario accident rate; Unrecognizable or sudden potentially dangerous driving scenarios correspond to the fourth scenario accident rate; Among them, based on the accident rate of each scenario and the comprehensive proportion of failure to remind the driver to enter the driving state, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage is determined, including: The accident rate of the first scenario and the accident rate of the second scenario are summed, and after weighted product is calculated using the comprehensive proportion of failures in reminding the driver to enter the driving state as the weight, the sum is then added with the accident rate of the third scenario and the accident rate of the fourth scenario to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage; Alternatively, the accident rate of the first scenario and the accident rate of the second scenario are summed, and the product is weighted by the comprehensive proportion of failures in reminding the driver to enter the driving state, and then summed with the accident rate of the fourth scenario to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the autonomous driving stage.

6. The method for evaluating the safety of an autonomous driving vehicle according to claim 1, characterized in that: For the fallback stage, determine the accident rate of the autonomous driving vehicle within the mileage corresponding to the fallback stage, including: Within the stage mileage corresponding to the fallback stage, calculate the accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered; Based on the fallback trigger frequency and the accident rates in each of the multiple time periods, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined.

7. The method for evaluating the safety of an autonomous driving vehicle according to claim 6, characterized in that: Based on the fallback trigger frequency and the accident rates of the respective time periods of the multiple time periods, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined, including: The maximum value of the accident rates in the time periods corresponding to the multiple time periods is taken, and then the maximum value is multiplied by the fallback trigger frequency to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage.

8. The method for evaluating the safety of an autonomous driving vehicle according to claim 1, characterized in that: For the fallback stage, determine the accident rate of the autonomous driving vehicle within the mileage corresponding to the fallback stage, including: Within the stage mileage corresponding to the fallback stage, calculate the accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered; Counting the proportion of untaken over corresponding to multiple time periods; Based on the fallback trigger frequency, the accident rate in the time period corresponding to the multiple time periods and the proportion of non-takeover, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined.

9. The method for evaluating the safety of an autonomous driving vehicle according to claim 8, characterized in that: Based on the fallback trigger frequency, the accident rate and the non-takeover ratio corresponding to the multiple time periods, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage is determined, including: Taking the untaken-over ratio as the weight, calculate the weighted value of the time period accident rate corresponding to each time period; The maximum value of the weighted values ​​of the accident rates in the time periods corresponding to the multiple time periods is taken, and then the maximum value is multiplied by the fallback trigger frequency to obtain the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the fallback stage.

10. The method for evaluating the safety of an autonomous driving vehicle according to any one of claims 6 to 9, characterized in that: Within the mileage corresponding to the fallback phase, the accident rates corresponding to multiple non-overlapping and continuous time periods preset after the fallback is triggered are calculated, including: For each of the time periods, the third occurrence frequency of each dangerous driving behavior in the time period and the third behavior accident rate of the dangerous driving behavior within the stage mileage corresponding to the fallback stage are counted; The third occurrence frequency is used as a weight to perform a weighted summation of the third behavior accident rates corresponding to all dangerous driving behaviors to obtain the period accident rate of the corresponding time period.

11. The method for evaluating the safety of an autonomous driving vehicle according to claim 1, characterized in that: The driving phase also includes: a first manual driving phase within a preset time period after the driver takes over and a second manual driving phase after the first manual driving phase.

12. The method for evaluating the safety of an autonomous driving vehicle according to claim 11, characterized in that: For the first manual driving stage, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the first manual driving stage is calculated, including: The percentage of driver mis-takeover and the accident rate of mis-takeover are counted. When the driver takes both hands off the steering wheel after exiting the autonomous driving mode, it indicates that the driver has mis-taken over. Statistics are collected on the accident rate of takeover after the driver takes over and is distracted, the frequency of distraction events, and the failure rate of distraction assistance reminders; Taking the proportion of false takeover as the weight of the false takeover accident rate, taking the frequency of distraction events and the proportion of reminder failure as the weight of the takeover accident rate, and taking the weighted sum of the false takeover accident rate and the takeover accident rate, the accident rate of the autonomous driving vehicle within the stage mileage corresponding to the first manual driving stage is obtained.

13. The method for evaluating the safety of an autonomous driving vehicle according to claim 1, characterized in that: Based on the accident rate of each driving stage and the mileage ratio of each driving stage, the comprehensive safety index of all driving stages is determined, including: The mileage proportion of each driving stage is used as a weight, and the accident rate of each driving stage and the corresponding weight are weighted and summed to obtain a comprehensive safety index of all driving stages.

14. An automatic driving vehicle safety evaluation device, characterized in that: include: An accident rate determination module is used to determine the accident rate of the autonomous driving vehicle within the stage mileage corresponding to each driving stage, wherein the driving stage includes: an autonomous driving stage and a fallback stage; The mileage proportion statistics module is used to count the proportion of the stage mileage corresponding to each driving stage in the total mileage; The safety index determination module is used to determine the comprehensive safety index of all driving stages based on the accident rate of each driving stage and the mileage proportion of each driving stage, and evaluate the vehicle safety with the comprehensive safety index.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, it implements the autonomous driving vehicle safety evaluation method as described in any one of claims 1 to 13.

16. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the autonomous driving vehicle safety assessment method as described in any one of claims 1 to 13 is implemented.

Citation Information

Patent Citations

  • Method and device for determining ASIL level information of autonomous vehicle and electronic equipment

    CN114802279A

  • Self-Driving Safety Evaluation Method, Apparatus, and System

    US20200406911A1