Vehicle safety evaluation methods, devices, storage media and computer equipment

Abstract

This application provides a method, apparatus, storage medium, and computer device for evaluating driving safety. The method includes: acquiring a predicted trajectory of a main vehicle, the predicted trajectory including multiple prediction times and predicted driving behavior data of the main vehicle at each prediction time; for each prediction time, generating obstacle avoidance trajectories corresponding to that prediction time based on a preset obstacle avoidance driving mode and the predicted driving behavior data of the main vehicle at that prediction time; and obtaining a safety score for evaluating driving safety based on the obstacle avoidance trajectories corresponding to each prediction time. This application can consider the driving trajectory of the main vehicle under different scenario information and search for possible obstacle avoidance driving behaviors of the autonomous driving algorithm under certain constraints. By combining possible obstacle avoidance driving behaviors for safety evaluation, this application can reduce the probability of false alarms and thus improve the accuracy of the evaluation results.

Description

Technical Field

[0001] This application relates to the field of autonomous driving, and in particular to a method, apparatus, storage medium and computer equipment for evaluating vehicle safety. Background Technology

[0002] In the process of autonomous driving, multiple dimensions can be used to evaluate the driving behavior of autonomous vehicles in order to assess the planning performance of autonomous driving algorithms. Among these, safety, as a highly concerned factor, is one of the most important evaluation dimensions. However, the inventors have found that existing methods for evaluating driving safety suffer from low accuracy. Summary of the Invention

[0003] The purpose of this application is to address at least one of the aforementioned technical defects, particularly the technical defect of low evaluation accuracy in the prior art.

[0004] In a first aspect, embodiments of this application provide a method for evaluating vehicle safety, the method comprising:

[0005] Obtain the predicted trajectory of the main vehicle, which includes multiple prediction times and predicted driving behavior data of the main vehicle at each prediction time.

[0006] For each predicted time, based on the preset obstacle avoidance driving mode and the predicted driving behavior data of the master vehicle at that predicted time, each obstacle avoidance driving trajectory corresponding to that predicted time is generated;

[0007] Based on each obstacle avoidance trajectory corresponding to each predicted time, a safety score is obtained to evaluate driving safety.

[0008] In one embodiment, the step of obtaining a safety score for evaluating driving safety based on each obstacle avoidance driving trajectory corresponding to each of the predicted times includes:

[0009] Based on a preset probability threshold, target obstacle prediction trajectories with a probability greater than the probability threshold are selected from multiple obstacle prediction trajectories.

[0010] For each predicted time, the target trajectory spacing corresponding to the predicted time is determined based on each predicted trajectory of the target obstacle and each obstacle avoidance driving trajectory corresponding to the predicted time. The target trajectory spacing is the minimum trajectory spacing with the largest value. The minimum trajectory spacing refers to the minimum distance between the main vehicle and the obstacle when the main vehicle is traveling along the corresponding obstacle avoidance driving trajectory.

[0011] The safety score is obtained based on the target trajectory spacing corresponding to each of the predicted times.

[0012] In one embodiment, the step of obtaining the safety score based on the target trajectory spacing corresponding to each of the predicted times includes:

[0013] Based on the target trajectory spacing corresponding to each prediction time, an initial score is determined for each prediction time, and the value range of each initial score is [0,1].

[0014] Determine the target probability, and calculate the safety score based on the target probability and the initial score corresponding to each prediction time.

[0015] In one embodiment, the step of determining the target probability includes:

[0016] Acquire multiple hazardous simulation scenarios, multiple safe simulation scenarios, and multiple initial probabilities;

[0017] For each initial probability, a first safety score and a second safety score corresponding to each dangerous simulation scenario are calculated based on the initial probability, and an F1 score corresponding to the initial probability is calculated based on each first safety score and each second safety score.

[0018] The initial probability corresponding to the F1 score with the largest value is taken as the target probability.

[0019] In one embodiment, the step of calculating the safety score based on the target probability and the initial score corresponding to each prediction time includes:

[0020] The security score is calculated based on the following expression:

[0021]

[0022]

[0023]

[0024] In the formula, Let p be the security score, and p be the target probability. The initial score corresponding to the first predicted time point. The initial score corresponding to the second prediction time. The initial score corresponding to the third prediction time. The initial score corresponding to the last predicted time point. To predict the total number of time points, Let be the sum of probabilities.

[0025] In one embodiment, the step of determining the target trajectory spacing corresponding to the prediction time based on each predicted target obstacle trajectory and each obstacle avoidance driving trajectory corresponding to the prediction time includes:

[0026] Obtain an obstacle grid map, which is generated based on the predicted trajectories of each of the target obstacles;

[0027] Based on the obstacle grid map and the obstacle avoidance driving trajectories corresponding to the predicted time, the target trajectory spacing corresponding to the predicted time is determined.

[0028] In one embodiment, the step of generating obstacle avoidance trajectories corresponding to each predicted time based on a preset obstacle avoidance driving mode and the predicted driving behavior data of the main vehicle at that predicted time includes:

[0029] For each predicted time, based on the obstacle avoidance driving mode and the driving behavior prediction data of the master vehicle at that predicted time, the obstacle avoidance behavior data of the master vehicle at multiple obstacle avoidance times are predicted respectively. Based on the obstacle avoidance behavior data of the master vehicle at multiple obstacle avoidance times and the driving behavior prediction data no later than the predicted time, each obstacle avoidance driving trajectory corresponding to the predicted time is generated, wherein each obstacle avoidance time is later than the predicted time.

[0030] Secondly, embodiments of this application provide a vehicle safety evaluation device, the device comprising:

[0031] The main vehicle predicted trajectory acquisition module is used to acquire the main vehicle predicted trajectory, which includes multiple prediction times and driving behavior prediction data of the main vehicle at each prediction time.

[0032] The obstacle avoidance driving trajectory acquisition module is used to generate each obstacle avoidance driving trajectory corresponding to each predicted time based on the preset obstacle avoidance driving mode and the driving behavior prediction data of the master vehicle at that predicted time.

[0033] The evaluation module is used to obtain a safety score for evaluating driving safety based on each obstacle avoidance driving trajectory corresponding to each of the predicted times.

[0034] Thirdly, embodiments of this application provide a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the driving safety evaluation method described in any of the above embodiments.

[0035] Fourthly, embodiments of this application provide a computer device, including: one or more processors, and a memory;

[0036] The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the driving safety evaluation method described in any of the above embodiments.

[0037] In the driving safety evaluation method, apparatus, storage medium, and computer equipment of this application, the predicted trajectory of the main vehicle can be obtained. For each prediction time involved in the predicted trajectory of the main vehicle, this application can generate various obstacle avoidance trajectories corresponding to that prediction time based on a preset obstacle avoidance driving mode and the driving behavior prediction data of the main vehicle at that prediction time, so as to reasonably predict the obstacle avoidance driving trajectories that the main vehicle may take. Having obtained the various obstacle avoidance driving trajectories corresponding to each prediction time, this application can perform a safety evaluation based on the various obstacle avoidance driving trajectories corresponding to each prediction time to obtain a safety score for evaluating driving safety. In this way, the driving trajectory of the main vehicle under different scenario information can be considered, and possible obstacle avoidance driving behaviors of the autonomous driving algorithm under certain constraints can be searched. By combining possible obstacle avoidance driving behaviors for safety evaluation, this application can reduce the probability of false alarms and thus improve the accuracy of the evaluation results. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a flowchart illustrating a driving safety evaluation method in one embodiment;

[0040] Figure 2 This is a schematic diagram of the trajectory in one embodiment;

[0041] Figure 3 This is a flowchart illustrating the steps for obtaining a safety score to evaluate driving safety based on each obstacle avoidance driving trajectory corresponding to each of the predicted times, as shown in one embodiment.

[0042] Figure 4 This is a flowchart illustrating the steps for determining the target probability in one embodiment;

[0043] Figure 5 This is a schematic diagram of the vehicle safety evaluation device in one embodiment;

[0044] Figure 6 This is a schematic diagram of the structure of a computer device in one embodiment. Detailed Implementation

[0045] 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, and 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.

[0046] As mentioned in the background, existing vehicle safety evaluation methods cannot accurately assess the driving behavior of autonomous vehicles, resulting in low accuracy. Therefore, based on existing safety evaluation results, it is difficult to quickly and accurately determine whether there are areas for improvement in autonomous driving algorithms and corresponding improvement directions. The inventors' research revealed that the reason for this low accuracy lies in the fact that traditional safety evaluation methods use the predicted trajectory of the main vehicle output by the path planning module as the basis for judgment, determining whether a collision will occur between the main vehicle and obstacles in the future, and using indicators such as the collision judgment result and the collision time as evaluation dimensions to assess driving behavior. However, in actual driving, the main vehicle may adjust its driving based on actual road conditions and may not necessarily follow the predicted trajectory output by the path planning algorithm. Therefore, considering only the predicted trajectory output by the path planning algorithm can easily lead to false alarms, thus affecting the accuracy of the safety evaluation results.

[0047] To address the aforementioned problems, this application provides a driving safety evaluation method, apparatus, storage medium, and computer equipment. It rationally predicts the possible obstacle avoidance trajectories of the main vehicle based on the predicted trajectory of the main vehicle and preset obstacle avoidance driving modes, and obtains a safety score for evaluating driving safety based on each obstacle avoidance trajectory. This allows consideration of the main vehicle's driving trajectory under different scenario information and searches for possible obstacle avoidance driving behaviors of the autonomous driving algorithm under certain constraints. By combining possible obstacle avoidance driving behaviors for safety evaluation, this application can reduce the probability of false alarms and thus improve the accuracy of the evaluation results.

[0048] In one embodiment, this application provides a method for evaluating vehicle safety. The following embodiments illustrate this method applied to a computer device. It is understood that the computer device refers to a device with data processing capabilities, and may be, but is not limited to, a laptop computer, a desktop computer, a single server, or a server cluster consisting of multiple servers. Figure 1 As shown, the vehicle safety evaluation method of this application may include the following steps:

[0049] S102: Obtain the predicted trajectory of the main vehicle, which includes multiple prediction times and driving behavior prediction data of the main vehicle at each prediction time.

[0050] The predicted trajectory of the master vehicle refers to the data output by the path planning algorithm, used to predict the driving behavior and / or trajectory that the master vehicle may take in the future. The predicted trajectory can include multiple prediction times and the predicted driving behavior data of the master vehicle at each prediction time. The predicted driving behavior data can be data that can be used to characterize the vehicle's driving state and / or driving behavior. For example, the predicted driving behavior data of the master vehicle at each prediction time can include the predicted position data, predicted speed data, and predicted acceleration data of the master vehicle at that prediction time. In this case, the predicted trajectory of the master vehicle is {(t1, p1, v1, a1), (t2, p2, v2, a2), ... (ti, pi, vi, ai)}, where i is a positive integer, ti is the i-th prediction time, pi is the predicted position data of the master vehicle at ti, vi is the predicted speed data of the master vehicle at ti, and ai is the predicted acceleration data of the master vehicle at ti.

[0051] In one embodiment, at the same driving time or simulation time, the path planning algorithm can output multiple predicted trajectories of the master vehicle and the future probability corresponding to each predicted trajectories. The computer device can obtain the predicted trajectories of the master vehicle with the highest future probability and execute subsequent steps accordingly.

[0052] S104: For each predicted time, generate each obstacle avoidance trajectory corresponding to that predicted time based on the preset obstacle avoidance driving mode and the driving behavior prediction data of the master vehicle at that predicted time.

[0053] The obstacle avoidance driving mode refers to the driving mode that the main vehicle may adopt when avoiding obstacles, such as accelerating to bypass, decelerating and / or changing direction, etc., which can be determined according to the actual situation, and this application does not impose specific restrictions on it. For ease of description, some embodiments in this document are illustrated using deceleration obstacle avoidance as an example. In one embodiment, in order to make the obstacle avoidance driving trajectory consistent with the actual situation and improve the rationality of the obstacle avoidance driving trajectory, the preset obstacle avoidance driving mode can be matched with vehicle dynamics constraints.

[0054] Specifically, for each predicted time point, the computer equipment can consider the vehicle's obstacle avoidance behavior starting from that predicted time point and obtain at least one obstacle avoidance trajectory corresponding to that predicted time point. For example, in Figure 2 In the trajectory diagram shown, the predicted trajectory of the main vehicle can be represented by the dots. If the main vehicle starts decelerating and avoiding obstacles from second 0, the obstacle avoidance trajectory of the main vehicle can be represented by the triangles.

[0055] To obtain the obstacle avoidance trajectory corresponding to each predicted time, for each predicted time, the computer device can obtain at least one obstacle avoidance trajectory corresponding to that predicted time based on the obstacle avoidance driving mode and the vehicle's driving behavior prediction data at that predicted time. For example, when performing deceleration obstacle avoidance, the computer device can obtain each obstacle avoidance trajectory corresponding to that predicted time based on preset deceleration data, predicted position data corresponding to that predicted time, and predicted speed data corresponding to that predicted time.

[0056] In one example, if the preset deceleration data is a deceleration range, the computer device can sample the deceleration range to obtain multiple decelerations, and calculate the obstacle avoidance trajectory corresponding to each deceleration in the manner described above, thereby obtaining multiple obstacle avoidance trajectories corresponding to each prediction time, which can further improve the accuracy of safety evaluation.

[0057] In another example, if the preset deceleration data is a deceleration range, the computer device can determine the maximum deceleration in the deceleration range and calculate the obstacle avoidance trajectory corresponding to the maximum deceleration in the aforementioned manner, thereby obtaining an obstacle avoidance trajectory corresponding to each prediction time, which can reduce the amount of calculation and improve the evaluation efficiency.

[0058] In one embodiment, S104 may include the following steps: for each predicted time, based on the obstacle avoidance driving mode and the driving behavior prediction data of the master vehicle at the predicted time, predict the obstacle avoidance behavior data of the master vehicle at multiple obstacle avoidance times, and generate each obstacle avoidance driving trajectory corresponding to the predicted time based on the obstacle avoidance behavior data of the master vehicle at multiple obstacle avoidance times and driving behavior prediction data no later than the predicted time, wherein each obstacle avoidance time is later than the predicted time.

[0059] Specifically, for each predicted time point, based on the preset obstacle avoidance driving mode and the predicted driving behavior data of the main vehicle at that predicted time point, the obstacle avoidance behavior data of the main vehicle at multiple obstacle avoidance times are predicted, assuming obstacle avoidance begins at that predicted time point. The obstacle avoidance behavior data refers to data that can be used to characterize the vehicle's driving state and / or driving behavior during obstacle avoidance, such as the predicted position data and predicted speed data of the main vehicle at the corresponding obstacle avoidance time point.

[0060] Computer equipment can stitch together the predicted trajectory of the main vehicle and obstacle avoidance behavior data corresponding to multiple obstacle avoidance moments to obtain the obstacle avoidance driving trajectory corresponding to each predicted moment. For example, the predicted trajectory of the main vehicle may include 10 predicted moments t1~t10, and the driving prediction data d1~d10 of the main vehicle at each predicted moment. When obtaining the obstacle avoidance driving trajectory corresponding to t3, the computer equipment can predict the obstacle avoidance behavior data d4'~d13' of the main vehicle at the 10 obstacle avoidance moments t4~t13 based on the obstacle avoidance driving mode and the driving prediction data d3 corresponding to t3, and use t1~t13, d1~d3, and d4'~d13' as the obstacle avoidance driving trajectory corresponding to t3. In this way, the obstacle avoidance driving trajectory can be more reasonable, and safety evaluation can be carried out accordingly, thereby further improving the accuracy of the evaluation results.

[0061] S106: Based on each obstacle avoidance trajectory corresponding to each of the predicted times, a safety score is obtained to evaluate driving safety.

[0062] In this step, after obtaining the obstacle avoidance driving trajectories corresponding to each predicted time, the computer equipment can obtain a safety score based on each obstacle avoidance driving trajectory. This safety score can be used to evaluate the level of driving safety.

[0063] In this application, a computer device can acquire the predicted trajectory of the main vehicle. For each predicted moment involved in the predicted trajectory, this application can generate various obstacle avoidance trajectories corresponding to that predicted moment based on a preset obstacle avoidance driving mode and the predicted driving behavior data of the main vehicle at that predicted moment, so as to reasonably predict the obstacle avoidance driving trajectories that the main vehicle may take. Having obtained the various obstacle avoidance trajectories corresponding to each predicted moment, this application can perform a safety evaluation based on these trajectories to obtain a safety score for evaluating driving safety. Thus, the driving trajectory of the main vehicle under different scenario information can be considered, and possible obstacle avoidance driving behaviors of the autonomous driving algorithm under certain constraints can be searched. By combining possible obstacle avoidance driving behaviors for safety evaluation, this application can reduce the probability of false alarms and thus improve the accuracy of the evaluation results.

[0064] In one embodiment, such as Figure 3 As shown, S106 may include the following steps:

[0065] S202: Based on a preset probability threshold, select the target obstacle prediction trajectory from multiple obstacle prediction trajectories whose probability is greater than the probability threshold.

[0066] Specifically, during driving, for each obstacle around the vehicle, the autonomous driving algorithm can predict the obstacle's trajectory over a future period, obtaining at least one predicted obstacle trajectory. Each predicted obstacle trajectory corresponds to a probability, reflecting the likelihood that the obstacle will move along that predicted trajectory. For each obstacle's predicted trajectory, the computer device can filter them according to a preset probability threshold, and select the predicted obstacle trajectories with a probability greater than the threshold as the target obstacle predicted trajectory.

[0067] S204: For each predicted time, based on each predicted trajectory of the target obstacle and each obstacle avoidance driving trajectory corresponding to the predicted time, determine the target trajectory spacing corresponding to the predicted time, wherein the target trajectory spacing is the minimum trajectory spacing with the largest value, and the minimum trajectory spacing refers to the minimum distance between the master vehicle and the obstacle when the master vehicle is driving along the corresponding obstacle avoidance driving trajectory.

[0068] Specifically, for each predicted time, the computer device can search for all possible obstacle avoidance trajectories corresponding to that predicted time, and select the obstacle avoidance trajectory with the largest minimum distance to surrounding obstacles from all the obstacle avoidance trajectories corresponding to that predicted time. The minimum distance between the selected obstacle avoidance trajectory and the surrounding obstacles is recorded as the target trajectory distance corresponding to that predicted time. For example, if there are two obstacle avoidance trajectories corresponding to the predicted time t0, and the minimum distance between one obstacle avoidance trajectory and the surrounding obstacles is dist_min1, and the minimum distance between the other obstacle avoidance trajectory and the surrounding obstacles is dist_min2, then the target trajectory distance corresponding to the predicted time t0 is max[dist_min1, dist_min2], where max[] is the operation of selecting the maximum value.

[0069] S206: Obtain the safety score based on the target trajectory spacing corresponding to each of the predicted times.

[0070] Since the selected obstacle avoidance trajectory represents the scenario where the vehicle can maintain the maximum safe distance from the obstacle during obstacle avoidance maneuvers, the autonomous driving algorithm is highly likely to follow this trajectory during actual driving or simulation. The computer determines the safety score based on the minimum distance corresponding to the selected obstacle avoidance trajectory, which reduces computational load while ensuring accuracy, thus balancing accuracy and processing efficiency.

[0071] In one embodiment, S206 may include the following steps:

[0072] Based on the target trajectory spacing corresponding to each prediction time, an initial score is determined for each prediction time, and the value range of each initial score is [0,1].

[0073] Determine the target probability, and calculate the safety score based on the target probability and the initial score corresponding to each prediction time.

[0074] Specifically, for each predicted time interval corresponding to the target trajectory spacing, the computer device determines the initial score of 0-1 quantization corresponding to that predicted time interval based on the target trajectory spacing corresponding to that predicted time. It is understood that the initial score can be calculated in any way. For example, for the predicted time ti, the computer device can use the formula 1-min[1, di] to determine the initial score corresponding to ti, where di is the target trajectory spacing corresponding to ti.

[0075] After obtaining the initial scores corresponding to each prediction time, the computer device can calculate the final safety score based on the target probability and each initial score. It is understood that this application can employ any method to calculate the final safety score based on the target probability and each initial score. In one embodiment, the safety score can be calculated using the following expression to obtain a more accurate safety score:

[0076]

[0077]

[0078]

[0079] In the formula, Let p be the security score, and p be the target probability. The initial score corresponding to the first predicted time point. The initial score corresponding to the second prediction time. The initial score corresponding to the third prediction time. The initial score corresponding to the last predicted time point. To predict the total number of time points, Let be the sum of probabilities.

[0080] This embodiment uses a 0-1 quantitative scoring method, which makes the evaluation results more continuous and less likely to produce large differences in safety scores under similar conditions.

[0081] In one embodiment, such as Figure 4 As shown, the step of determining the target probability includes:

[0082] S302: Obtain multiple hazardous simulation scenarios, multiple safe simulation scenarios, and multiple initial probabilities;

[0083] S304: For each of the initial probabilities, calculate the first safety score corresponding to each of the dangerous simulation scenarios and the second safety score corresponding to each of the safe simulation scenarios based on the initial probability, and calculate the F1 score corresponding to the initial probability based on each of the first safety scores and each of the second safety scores;

[0084] S306: The initial probability corresponding to the F1 score with the largest value is taken as the target probability.

[0085] Specifically, to make the target probability more accurate and further improve the accuracy of the evaluation results, this application can adopt a data-driven approach to select the target probability from multiple initial probabilities. This application can pre-collect multiple hazardous simulation scenarios, multiple safe simulation scenarios, and multiple initial probabilities. For each initial probability, a safety score (i.e., a first safety score) and a safety score (i.e., a second safety score) are calculated for each hazardous simulation scenario using that initial probability. The F1 score (i.e., the balanced F-score) corresponding to each initial probability is then determined based on each first safety score and each second safety score. Furthermore, when calculating the F1 score for each initial probability, the computer device can use hazardous simulation scenarios with a first safety score greater than 0.5 as True Positives, hazardous simulation scenarios with a first safety score less than 0.5 as False Negatives, and safe simulation scenarios with a second safety score greater than 0.5 as False Positives, thereby obtaining the F1 score for each initial probability.

[0086] After obtaining the F1 score corresponding to each initial probability, the computer device can use the initial probability corresponding to the largest F1 score as the target probability and use the target probability to evaluate driving safety.

[0087] In one embodiment, the step of determining the target trajectory spacing at the prediction time based on each predicted target obstacle trajectory and each obstacle avoidance trajectory at the prediction time includes:

[0088] Obtain an obstacle grid map, which is generated based on the predicted trajectories of each of the target obstacles;

[0089] Based on the obstacle grid map and the obstacle avoidance driving trajectories corresponding to the predicted time, the target trajectory spacing corresponding to the predicted time is determined.

[0090] Specifically, computer equipment can generate an obstacle grid map based on the predicted trajectories of each target obstacle. In the obstacle grid map, the grid information corresponding to each grid can be used to reflect the presence of obstacles in that grid at each prediction time. For example, it can reflect that an obstacle is located in that grid at prediction time t1, but is not located in that grid at prediction time t10.

[0091] For each predicted time point, when searching for the target trajectory spacing corresponding to that predicted time point, the computer device can acquire an obstacle grid map and determine the target trajectory spacing corresponding to that predicted time point based on the obstacle grid map and the obstacle avoidance driving trajectories. This avoids redundant calculations in the minimum distance calculation process, thereby reducing computational load and improving processing efficiency.

[0092] Furthermore, for each prediction time, the computer device can also use a depth-first search algorithm to determine the target trajectory spacing corresponding to that prediction time. In other words, the process of determining the target trajectory spacing can be regarded as searching for feasible solutions on a future state tree. When a parent node is traversed and it is clear that the performance of each descendant node of that parent node is worse than the performance of the parent node, the computer device can terminate the subsequent search of that parent node, that is, there is no need to traverse each descendant node under that parent node, so as to improve processing efficiency.

[0093] The driving safety evaluation device provided in the embodiments of this application is described below. The driving safety evaluation device described below can be referred to in correspondence with the driving safety evaluation method described above.

[0094] In one embodiment, this application provides a vehicle safety evaluation device 400. For example... Figure 5 As shown, the device 400 includes a main vehicle predicted trajectory acquisition module 410, an obstacle avoidance driving trajectory acquisition module 420, and an evaluation module 430. Wherein:

[0095] The main vehicle predicted trajectory acquisition module 410 is used to acquire the main vehicle predicted trajectory, which includes multiple prediction times and driving behavior prediction data of the main vehicle at each prediction time.

[0096] The obstacle avoidance driving trajectory acquisition module 420 is used to generate each obstacle avoidance driving trajectory corresponding to each predicted time based on the preset obstacle avoidance driving mode and the driving behavior prediction data of the master vehicle at the predicted time.

[0097] Evaluation module 430 is used to obtain a safety score for evaluating driving safety based on each obstacle avoidance driving trajectory corresponding to each of the predicted times.

[0098] In one embodiment, the evaluation module 430 of this application may include a target obstacle prediction trajectory acquisition unit, a target trajectory spacing acquisition unit, and a safety score calculation unit. The target obstacle prediction trajectory acquisition unit is used to filter target obstacle prediction trajectories with a probability greater than a preset probability threshold from multiple obstacle prediction trajectories. The target trajectory spacing acquisition unit is used to determine the target trajectory spacing corresponding to each prediction time based on each target obstacle prediction trajectory and each obstacle avoidance driving trajectory corresponding to that prediction time. The target trajectory spacing is the minimum trajectory spacing with the largest value, where the minimum trajectory spacing refers to the minimum distance between the main vehicle and the obstacle when traveling along the corresponding obstacle avoidance driving trajectory. The safety score calculation unit is used to obtain the safety score based on the target trajectory spacing corresponding to each prediction time.

[0099] In one embodiment, the score calculation unit of this application includes an initial score determination unit and a calculation unit. The initial score determination unit is used to determine an initial score corresponding to each prediction time based on the target trajectory spacing corresponding to each prediction time, wherein the value range of each initial score is [0,1]. The calculation unit is used to determine the target probability and calculate the safety score based on the target probability and the initial scores corresponding to each prediction time.

[0100] In one embodiment, the calculation unit includes an initial data acquisition unit, an F1 score determination unit, and a target probability selection unit. The initial data acquisition unit acquires multiple hazardous simulation scenarios, multiple safe simulation scenarios, and multiple initial probabilities. The F1 score determination unit calculates a first safety score for each hazardous simulation scenario and a second safety score for each safe simulation scenario based on the initial probability, and calculates the F1 score corresponding to that initial probability based on each of the first and second safety scores. The target probability selection unit selects the initial probability corresponding to the largest F1 score as the target probability.

[0101] In one embodiment, the calculation unit is further configured to calculate the security score based on the following expression:

[0102]

[0103]

[0104]

[0105] In the formula, Let p be the security score, and p be the target probability. The initial score corresponding to the first predicted time point. The initial score corresponding to the second prediction time. The initial score corresponding to the third prediction time. The initial score corresponding to the last predicted time point. To predict the total number of time points, Let be the sum of probabilities.

[0106] In one embodiment, the target trajectory spacing acquisition unit includes a grid map acquisition unit and a spacing determination unit. The grid map acquisition unit acquires an obstacle grid map, which is generated based on the predicted trajectories of each target obstacle. The spacing determination unit determines the target trajectory spacing corresponding to the predicted time based on the obstacle grid map and the obstacle avoidance trajectories corresponding to the predicted time.

[0107] In one embodiment, the obstacle avoidance trajectory acquisition module 420 is used to predict obstacle avoidance behavior data of the main vehicle at multiple obstacle avoidance times for each predicted time, based on the obstacle avoidance driving mode and the driving behavior prediction data of the main vehicle at the predicted time, and generate each obstacle avoidance trajectory corresponding to the predicted time based on the obstacle avoidance behavior data of the main vehicle at multiple obstacle avoidance times and driving behavior prediction data no later than the predicted time, wherein each obstacle avoidance time is later than the predicted time.

[0108] In one embodiment, this application also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the driving safety evaluation method described in any of the above embodiments.

[0109] In one embodiment, this application also provides a computer device. The computer device stores computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the driving safety evaluation method described in any of the above embodiments.

[0110] Indicatively, Figure 6 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of this application. In one example, the computer device can be a server. (Refer to...) Figure 6The computer device 900 includes a processing component 902, which further includes one or more processors, and memory resources represented by memory 901 for storing instructions, such as application programs, that can be executed by the processing component 902. The application programs stored in memory 901 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 902 is configured to execute instructions to perform the steps of the driving safety evaluation method described in any of the above embodiments.

[0111] The computer device 900 may also include a power supply component 903 configured to perform power management of the computer device 900, a wired or wireless network interface 904 configured to connect the computer device 900 to a network, and an input / output (I / O) interface 905. The computer device 900 may operate on an operating system stored in memory 901, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0112] Those skilled in the art will understand that the internal structure of the computer device shown in this application is merely a block diagram of a portion of the structure related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0113] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this document, "a," "an," "the," "the," and "its" may also include plural forms unless the context clearly indicates otherwise. "Multiple" refers to at least two, such as 2, 3, 5, or 8, etc. "And / or" includes any and all combinations of the related listed items.

[0114] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0115] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for evaluating driving safety, characterized in that, The method includes: Obtain the predicted trajectory of the main vehicle, which includes multiple prediction times and predicted driving behavior data of the main vehicle at each prediction time. For each predicted time, based on the preset obstacle avoidance driving mode and the predicted driving behavior data of the master vehicle at that predicted time, each obstacle avoidance driving trajectory corresponding to that predicted time is generated; Based on each obstacle avoidance trajectory corresponding to each predicted time, a safety score is obtained to evaluate driving safety. The step of obtaining a safety score for evaluating driving safety based on each obstacle avoidance trajectory corresponding to each predicted time includes: Based on a preset probability threshold, target obstacle prediction trajectories with a probability greater than the probability threshold are selected from multiple obstacle prediction trajectories. For each predicted time, the target trajectory spacing corresponding to the predicted time is determined based on each predicted trajectory of the target obstacle and each obstacle avoidance driving trajectory corresponding to the predicted time. The target trajectory spacing is the minimum trajectory spacing with the largest value. The minimum trajectory spacing refers to the minimum distance between the main vehicle and the obstacle when the main vehicle is traveling along the corresponding obstacle avoidance driving trajectory. Based on the target trajectory spacing corresponding to each prediction time, an initial score is determined for each prediction time, and the value range of each initial score is [0,1]. Acquire multiple hazardous simulation scenarios, multiple safe simulation scenarios, and multiple initial probabilities; For each initial probability, a first safety score and a second safety score corresponding to each dangerous simulation scenario are calculated based on the initial probability, and an F1 score corresponding to the initial probability is calculated based on each first safety score and each second safety score. The initial probability corresponding to the F1 score with the largest value is taken as the target probability; The safety score is calculated based on the target probability and the initial score corresponding to each prediction time.

2. The driving safety evaluation method according to claim 1, characterized in that, The step of calculating the safety score based on the target probability and the initial score corresponding to each prediction time includes: The security score is calculated based on the following expression: In the formula, Let p be the security score, and p be the target probability. The initial score corresponding to the first predicted time point. The initial score corresponding to the second prediction time. The initial score corresponding to the third prediction time. The initial score corresponding to the last predicted time point. To predict the total number of time points, Let be the sum of probabilities.

3. The method for evaluating driving safety according to claim 1 or 2, characterized in that, The step of determining the target trajectory spacing at the predicted time based on each predicted target obstacle trajectory and each obstacle avoidance trajectory at the predicted time includes: Obtain an obstacle grid map, which is generated based on the predicted trajectories of each of the target obstacles; Based on the obstacle grid map and the obstacle avoidance driving trajectories corresponding to the predicted time, the target trajectory spacing corresponding to the predicted time is determined.

4. The method for evaluating driving safety according to claim 1 or 2, characterized in that, The step of generating obstacle avoidance trajectories for each predicted time based on a preset obstacle avoidance driving mode and the predicted driving behavior data of the main vehicle at that predicted time includes: For each predicted time, based on the obstacle avoidance driving mode and the driving behavior prediction data of the master vehicle at the predicted time, the obstacle avoidance behavior data of the master vehicle at multiple obstacle avoidance times are predicted respectively. Based on the obstacle avoidance behavior data of the master vehicle at multiple obstacle avoidance times and the driving behavior prediction data no later than the predicted time, each obstacle avoidance driving trajectory corresponding to the predicted time is generated, wherein each obstacle avoidance time is later than the predicted time.

5. A vehicle safety evaluation device, characterized in that, The device includes: The main vehicle predicted trajectory acquisition module is used to acquire the main vehicle predicted trajectory, which includes multiple prediction times and driving behavior prediction data of the main vehicle at each prediction time. The obstacle avoidance driving trajectory acquisition module is used to generate each obstacle avoidance driving trajectory corresponding to each predicted time based on the preset obstacle avoidance driving mode and the driving behavior prediction data of the master vehicle at that predicted time. An evaluation module is used to obtain a safety score for evaluating driving safety based on each obstacle avoidance trajectory corresponding to each predicted time point; wherein, the evaluation module includes: The target obstacle prediction trajectory acquisition unit is used to filter out target obstacle prediction trajectories with a probability greater than the preset probability threshold from multiple obstacle prediction trajectories according to the preset probability threshold. The target trajectory spacing acquisition unit is used to determine the target trajectory spacing corresponding to each prediction time based on each predicted trajectory of the target obstacle and each obstacle avoidance driving trajectory corresponding to the prediction time. The target trajectory spacing is the minimum trajectory spacing with the largest value. The minimum trajectory spacing refers to the minimum distance between the main vehicle and the obstacle when the main vehicle is driving along the corresponding obstacle avoidance driving trajectory. The initial score determination unit is used to determine the initial score corresponding to each prediction time according to the target trajectory spacing corresponding to each prediction time, and the value range of each initial score is [0,1]. The initial data acquisition unit is used to acquire multiple hazardous simulation scenarios, multiple safe simulation scenarios, and multiple initial probabilities. The F1 score determination unit is used to calculate, for each of the initial probabilities, a first safety score corresponding to each of the dangerous simulation scenarios and a second safety score corresponding to each of the safe simulation scenarios, and to calculate the F1 score corresponding to the initial probability based on each of the first safety scores and each of the second safety scores. The target probability selection unit is used to select the initial probability corresponding to the F1 score with the largest value as the target probability. The calculation unit is used to calculate the safety score based on the target probability and the initial score corresponding to each prediction time.

6. A storage medium, characterized in that, The storage medium stores computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the driving safety evaluation method as described in any one of claims 1 to 4.

7. A computer device, characterized in that, include: One or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the driving safety evaluation method as described in any one of claims 1 to 4.

Images