Multi-scene switching driving simulator control method, device and storage medium

Abstract

The application provides a multi-scene switching driving simulator control method, device and storage medium, relates to the technical field of driving training, and comprises the following steps: collecting historical training data of a plurality of training scenes of a target trainee within a historical time; analyzing a plurality of training difficulties of the target trainee for the plurality of training scenes according to the historical training data; optimizing training parameters of the target trainee for the plurality of training scenes according to the plurality of training difficulties to obtain a plurality of optimal scene training parameters; and configuring the training parameters of the plurality of training scenes in a driving simulator by using the plurality of optimal scene training parameters, and training and controlling the target trainee. The application solves the technical problem that the existing driving simulator training process lacks pertinence and cannot adjust the training process according to the specific performance of the trainee in different scenes, improves the individualization level and pertinence of driving training, and further improves the efficiency and overall quality of driving training.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of driving training, in particular to a driving simulator control method and device for multiple scene switching and a storage medium. BACKGROUND

[0002] As a kind of efficient and safe training tool, driving simulator can simulate various real driving scenes, such as urban roads, highways and severe weather, and provide a safe and controllable training environment for trainees, and has been widely used in driving schools and professional training.

[0003] The existing driving simulator control method usually adopts preset training scenes and fixed training parameters, and the difficulty and setting of the training scene are unified, which cannot be dynamically adjusted according to the different skill levels and learning progress of trainees. This patterned training mode ignores the individual differences of trainees in the training process and lacks personalized consideration. For example, for trainees with poor foundation, fixed training parameters may be too difficult, affecting the training effect; while for trainees with good foundation, fixed parameters may be too simple, which cannot effectively improve the skills. In addition, this training process lacks sufficient analysis and feedback on the specific performance of trainees in different scenes, and it is difficult to optimize the training parameters according to the actual situation of trainees. Therefore, trainees cannot obtain personalized guidance and targeted training in the training process, resulting in repetition and inadaptability of training content, reducing learning motivation and effect. SUMMARY

[0004] The present application provides a driving simulator control method and device for multiple scene switching and a storage medium, which solves the technical problem that the existing technology ignores the individual differences of trainees, cannot adjust the training process according to the specific performance of trainees in different scenes, and leads to lack of pertinence in driving training, affecting the training efficiency and training effect, achieves the technical effect of improving the individualization level and pertinence of driving training, and further improves the efficiency and overall quality of driving training.

[0005] In view of the above problems, on the one hand, the present application provides a driving simulator control method for multiple scene switching, which comprises: collecting historical training data of a target trainee within a historical time, wherein the historical training data includes training data of multiple training scenes; analyzing multiple training difficulties of the target trainee for the multiple training scenes according to the historical training data; optimizing training parameters of the target trainee for the multiple training scenes according to the multiple training difficulties, to obtain multiple optimal scene training parameters; configuring training parameters of the multiple training scenes in a driving simulator using the multiple optimal scene training parameters, and controlling training of the target trainee.

[0006] In a second aspect, the present application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the driving simulator control method with multiple scene switching when executing the computer program.

[0007] In a third aspect, the present application also provides a computer readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the driving simulator control method with multiple scene switching.

[0008] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0009] The historical training data of the target trainee in multiple training scenes within a historical time are collected, which can reflect the past training of the trainee in different scenes, providing a basis for subsequent analysis of the training difficulty and ability of the trainee. Based on the historical training data, the performance of the trainee in each training scene is analyzed, and the adaptability and difficulty of the trainee to different training scenes are evaluated. This analysis process helps to identify the strengths and weaknesses of the trainee, understand the challenges and difficulties encountered by the trainee in each scene, and further lay the foundation for personalized adjustment of the training scheme. According to the analysis result, the parameters of the training scene are optimized to ensure that the difficulty of each training scene matches the ability of the trainee, and multiple optimal scene training parameters are generated. By adjusting the training difficulty and setting, the training is more targeted and personalized, avoiding mismatch of difficulty in the training process. The training scene of the simulator is configured using the optimized training parameters, so that the trainee can train at an appropriate difficulty. Through this personalized training control method, the training content and difficulty of the simulator can be dynamically adjusted to maximize the training effect.

[0010] In summary, the present application controls the driving simulator through the above steps, so that the driving simulator can automatically adjust the difficulty and parameters of the training scene according to the historical performance and ability difference of the trainee, and realize highly personalized training control. Through in-depth analysis of the training data, the training parameters are optimized, so that the trainee can obtain challenges and guidance matching their own ability in the training process, thereby improving the training efficiency and overall quality of the training.

[0011] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented in accordance with the content of the description, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 The flowchart of the driving simulator control method with multiple scene switching provided by the embodiments of the present application is shown.

[0013] Figure 2 A process diagram for analyzing the training difficulty of a target trainee on multiple training scenes according to historical training data is provided in the multi-scene switching driving simulator control method of the embodiments of the present application.

[0014] Figure 3 A process diagram for obtaining multiple optimal scene training parameters is provided in the multi-scene switching driving simulator control method of the embodiments of the present application.

[0015] Figure 4 A structural diagram of a computer device is provided in the embodiments of the present application.

[0016] Legend: bus 300, receiver 301, processor 302, transmitter 303, memory 304, bus interface 305. DETAILED DESCRIPTION

[0017] The embodiments of the present application provide a multi-scene switching driving simulator control method, device and storage medium, collect historical training data of trainees, analyze training difficulty in different scenes, and then optimize training parameters and apply them to the training control of the driving simulator. The technical problem of lack of pertinence in driving training, affecting the training efficiency and training effect due to neglecting the individual differences of trainees and being unable to adjust the training process according to the specific performance of trainees in different scenes is solved. The technical effects of improving the individualization level and pertinence of driving training, and then improving the efficiency and overall quality of driving training are achieved.

[0018] Embodiment one, as shown in the embodiments of the present application, a multi-scene switching driving simulator control method is provided, which comprises: Figure 1

[0019] Step S1: Collecting historical training data of a target trainee in a historical time, wherein the historical training data includes training data of multiple training scenes.

[0020] Specifically, the target trainee refers to a trainee who is receiving driving training. The historical time is a specific time range set when extracting the trainee training data, which can usually be the entire training time experienced by the trainee from the beginning of training to the present. The historical training data is all training records of the trainee in the driving simulator in the historical time, including the specific situation of training, the performance of the trainee and the training content in multiple training scenes. These training scenes can be city driving, night driving, driving in bad weather and various simulation scenarios or tasks.

[0021] ​The driving simulator itself has a data collection function. During the training of the target trainee, the driving simulator automatically records various data related to the trainee's operation. These data cover the trainee's operation in different training scenarios, such as the trainee's reaction time to traffic lights, the operation force of the brake and accelerator in the urban street scenario, the trainee's steering operation and speed control in the mountain road scenario, etc. The data collection system of the interactive driving simulator obtains the historical training data of the target trainee in the past period of time and classifies and organizes these historical data according to specific training scenarios. These historical training data provide a comprehensive data basis for subsequent analysis of the trainee's training in different scenarios, which helps to deeply understand the trainee's driving habits and performance in different scenarios.

[0022] Step S2: According to the historical training data, analyze the training difficulty of the target trainee for the plurality of training scenarios.

[0023] Specifically, the collected historical training data is analyzed, and each operation data of the trainee in each training scenario is counted and evaluated, such as calculating the operation error rate and average reaction time of the trainee in a specific scenario, so as to determine the training difficulty of each training scenario for the trainee. Through the analysis of historical data, the performance of the trainee in different scenarios can be understood, so that the weak links of the trainee in different training scenarios can be accurately found, and a clear direction for subsequent optimization of training parameters can be provided.

[0024] Step S3: According to the plurality of training difficulties, optimize the training parameters of the target trainee for the plurality of training scenarios to obtain a plurality of optimal scenario training parameters.

[0025] Specifically, the optimal scenario training parameter refers to optimizing the training parameter of each training scenario according to the training difficulty of the training scenario, and determining the driving simulator setting parameter most suitable for the training difficulty of the trainee in the current scenario.

[0026] According to the training difficulty of each training scenario for the trainee obtained in the previous step, using a machine learning algorithm, the training parameter of each training scenario is optimized one by one according to the historical scenario training parameter and the training difficulty, and a plurality of optimal scenario training parameters corresponding to a plurality of training scenarios are determined. Through training parameter optimization, targeted difficulty adjustment can be realized, so that the training process is more suitable for the actual ability of the trainee, and the training of the trainee in different scenarios is both challenging and can improve the driving skills of the trainee, thereby improving the effectiveness and efficiency of the training.

[0027] Step S4: Use the plurality of optimal scenario training parameters to configure the training parameters of the plurality of training scenarios in the driving simulator, and control the training of the target trainee.

[0028] Specifically, the optimized training parameters are input into the control system of the driving simulator, and the driving simulator reconstructs the training scene according to the new parameters, ensuring that the difficulty, challenge and ability performance of each scene match. Then the trainee trains in this personalized scene. For example, when the curvature and slope of the mountain road scene are adjusted, the trainee will face a training environment that is more suitable for his own ability when he enters the scene again for training. Through the reconfiguration of the training parameters, personalized training control of the trainee is realized, the training effect of the trainee is improved, and the trainee can master the driving skills faster and better.

[0029] Further, the step S1 of the embodiment of the present application comprises:

[0030] Step S11: Collect the training times of the target trainee in the historical time for training in the plurality of training scenes, and obtain a plurality of historical training times.

[0031] Step S12: Collect the training error times of the target trainee in the historical time for training in the plurality of training scenes, and obtain a plurality of historical training error times.

[0032] Step S13: Integrate the plurality of historical training times and the plurality of historical training error times to obtain the historical training data of the target trainee.

[0033] Specifically, in the process of collecting historical training data, first, the total training times of the target trainee in the historical time for training in each training scene are collected, and a plurality of historical training times are obtained. Each historical training time corresponds to a training scene, which can be labeled in the form of scene label to distinguish the data in different training scenes, facilitating subsequent analysis of each training scene. By collecting the historical training times, the participation of the trainee in each training scene can be accurately counted, and the training frequency of the trainee in different scenes can be understood, providing data support for subsequent analysis of the familiarity of the trainee to different scenes.

[0034] In addition to collecting the number of historical training times, the number of training errors in each training scenario also needs to be collected. The scenario label can also be labeled to facilitate subsequent data comparison and calculation. These training error times are the number of times the trainee makes mistakes that do not meet the driving specifications or fails to correctly respond to tasks in the training scenario during the training process in each training scenario. For example, in a highway training scenario, the trainee incorrectly changes lanes 3 times. These 3 times are the training error times in this scenario. During each training process, the driving simulator will determine the trainee's operation according to the preset judgment rules, extract training data with errors in multiple training scenarios, and determine multiple historical training error times corresponding to multiple training scenarios. By obtaining the number of historical training errors, the trainee can clearly identify the areas where problems are likely to occur in different training scenarios, thereby providing a direct basis for analyzing the training difficulty of each scenario for the trainee.

[0035] The obtained number of historical training times and the number of historical training errors in each training scenario are sorted, and the number of historical training times and the number of historical training errors in the same training scenario are correspondingly integrated to form complete historical training data of the target trainee. The sorting process can use a database management system (such as MySQL) or spreadsheet software (such as Excel) to integrate the data, obtaining a historical training data set that comprehensively reflects the training situation of the trainee in each training scenario, and providing a complete data basis for subsequent analysis of the training difficulty of the trainee in different scenarios.

[0036] Further, the step S2 of the embodiment of the present application comprises:

[0037] Step S21: Obtain multiple sample training data of multiple sample trainees performing the multiple training scenarios.

[0038] Step S22: According to the multiple sample training data and the historical training data, calculate and analyze to obtain multiple training difficulties of the target trainee for the multiple training scenarios.

[0039] Specifically, the sample trainee refers to other trainees in addition to the target trainee. The training data of these trainees is used as a reference sample for comparison and analysis with the data of the target trainee. For example, select some other trainees who are in the same training stage as the target trainee and have similar driving experience as the sample trainees in the trainees using the driving simulator. For these sample trainees, the sample training data corresponding to each sample trainee is collected in a similar manner to the collection of the historical training data of the target trainee, including the number of training times, the number of training errors, and other information of each sample trainee in each training scenario, to generate multiple sample training data. The sample training data provides a reference standard for analyzing the training difficulty of the target trainee, enabling the training situation of the target trainee to be evaluated in the context of a group.

[0040] By comparing the data differences of the target trainee and the sample trainees in each training scene, the performance of the target trainee in different training scenes and the training difficulty are determined by using statistical analysis methods. For example, if the performance of the target trainee in the night driving scene is significantly different from the performance of other sample trainees, it indicates that this scene has a high difficulty for the target trainee. By calculating and analyzing the differences between the sample training data and the historical training data of the target trainee, the training difficulty of the target trainee in each training scene can be objectively and accurately determined, which provides an important basis for subsequent training parameter optimization for the target trainee.

[0041] Further, as shown in Figure 2 Step S22 includes:

[0042] Step S221: According to the plurality of sample training data, a plurality of sample training frequency sets and a plurality of sample training error frequency sets of the plurality of training scenes are extracted.

[0043] Step S222: The ratios of the plurality of sample training frequency sets and the plurality of sample training error frequency sets are calculated respectively, a plurality of sample error rate sets are obtained, and the average values are calculated to obtain a plurality of average error rates of the plurality of training scenes.

[0044] Step S223: The ratios of the plurality of historical training error frequencies and the plurality of historical training frequencies are calculated respectively, and a plurality of error rates of the plurality of training scenes are obtained.

[0045] Step S224: The ratios of the plurality of error rates and the plurality of average error rates are calculated respectively, and a plurality of basic training difficulties are obtained.

[0046] Step S225: The average values of the plurality of sample training frequency sets are calculated respectively, and a plurality of average training frequencies are obtained.

[0047] Step S226: The ratios of the plurality of historical training frequencies and the plurality of average training frequencies are calculated respectively, a plurality of frequency correction coefficients are obtained, and the plurality of basic training difficulties are multiplied respectively to complete the correction calculation, and a plurality of training difficulties are obtained.

[0048] Specifically, when analyzing the training difficulty of the target trainee, since the difficulties of different training scenes are not the same, not only the performance of the target trainee itself needs to be considered, but also the performance of other trainees needs to be compared in order to accurately and objectively evaluate the training difficulty of the target trainee.

[0049] First, the sample training times and sample training failure times of each sample trainee under each training scene are extracted from multiple sample training data to generate multiple sample training time sets and multiple sample training failure time sets for multiple training scenes. The sample training time set contains the set of training times of all sample trainees in a specific training scene. The sample training failure time set contains the set of training failure times of all sample trainees in a specific training scene. During the sample data extraction process, the corresponding sample trainee number and training scene annotation can be added to each data to indicate the data source, facilitating subsequent comparison and data calculation.

[0050] For each training scene, each element in the sample training failure time set is divided by the corresponding element in the sample training time set to obtain a sample failure rate set. For example, in the urban road scene, the training times of sample trainee A are 10 times, and the failure times are 2 times, with a failure rate of 20%; the training times of sample trainee B are 15 times, and the failure times are 3 times, with a failure rate of 20%. Similarly, the failure rates of all sample trainees in this scene are calculated to form a sample failure rate set. Then, the average value of the sample failure rate set is calculated to obtain the average failure rate of the scene. The sample training data of other training scenes is processed in the same way to obtain multiple average failure rates for multiple training scenes. The average failure rate represents the average failure level of the sample trainees in the corresponding training scene, providing a reference standard for subsequent evaluation of the training difficulty of the target trainee.

[0051] For the target trainee, multiple failure rates for multiple training scenes are calculated in the same way as the sample trainees. That is, the historical training failure times of the target trainee in each training scene are divided by the historical training times to obtain the failure rate in that scene. For example, the historical training times of the target trainee in the urban road scene are 8 times, and the failure times are 3 times, with a failure rate of 37.5%. The failure rate of the target trainee reflects its actual failure in each training scene.

[0052] The failure rate of the target trainee in each training scene is divided by the corresponding average failure rate to obtain the basic training difficulty. The larger the ratio of the failure rate to the average failure rate, the greater the basic training difficulty, indicating that the training scene is more difficult for the target trainee. When the failure rate of the target trainee is greater than the average failure rate of the multiple sample trainees, the basic training difficulty is greater than 1, otherwise it is less than or equal to 1. For example, in the urban road scene, the failure rate of the target trainee is 37.5%, and the average failure rate is 20%, so the basic training difficulty is 37.5% / 20% = 1.875. This means that the training difficulty of the target trainee in this scene is 87.5% higher than the average level of the sample trainees.

[0053] The basic training difficulty initially reflects the training difficulty for the target learner relative to the sample learners in each training scenario. Theoretically, as the number of training sessions in each scenario increases, the learner's proficiency in that scenario will increase, and the error rate will decrease. If a learner fails the exam after multiple training sessions in a particular scenario, it indicates that the training scenario is more difficult for the learner. Considering the impact of the number of training sessions on the training results, the basic training difficulty needs to be adjusted based on the number of training sessions to more accurately reflect the training difficulty of each training scenario for the target learner.

[0054] First, the average number of training iterations for each training scenario is calculated to obtain the average number of training iterations. This average number of iterations reflects the average training effort of the trainees in each scenario, providing a reference for subsequent iteration adjustment. Next, the ratio of the target trainee's historical training iterations to the average training iterations in each scenario is calculated to obtain multiple iteration adjustment coefficients. Then, these coefficients are multiplied by the corresponding base training difficulty to obtain multiple adjusted training difficulties. Each training difficulty corresponds to a training scenario. For example, in the urban road scenario, if the target trainee's historical training iterations are 8 and the average is 12, the iteration adjustment coefficient is 8 / 12 = 0.667. The base training difficulty is 1.875, so the adjusted training difficulty is 1.875 × 0.667 = 1.247. Through this adjustment calculation, the training difficulty can be automatically adjusted based on the trainee's training iterations, thus more accurately reflecting the target trainee's training difficulty in each scenario and making subsequent training parameter adjustments more consistent with the target trainee's actual performance.

[0055] Furthermore, such as Figure 3 As shown, step S3 in this embodiment includes:

[0056] Step S31: Configure multiple scene weights according to the multiple training difficulties.

[0057] Step S32: Obtain the total training time of the target student during driving simulation training.

[0058] Step S33: Randomly allocate the total training time to the multiple training scenarios to obtain multiple first scenario training times, which are used as multiple first scenario training parameters.

[0059] Step S34: Based on the multiple training difficulties and multiple first scene training times, predict the training success rate to obtain multiple first scene training success rates.

[0060] Step S35: Calculate the first training fitness of the training parameters of the multiple first scenes based on the multiple scene weights and the multiple first scene training success rates.

[0061] Step S36: Continue the optimization of the scene training parameters of the plurality of training scenarios until convergence, output the plurality of scene training parameters with the maximum training fitness, and obtain the plurality of optimal scene training parameters.

[0062] Specifically, the scene weight refers to the relative importance or priority of the training difficulty in different training scenarios. According to the level of training difficulty, a scene weight is assigned to each training scenario, and the size of the weight reflects the training difficulty or training value of the target student in that scenario. Training scenarios with higher training difficulty will be assigned a larger weight, and subsequent training time for this training scenario will be relatively longer, while training scenarios with lower training difficulty will be assigned a smaller weight to ensure that training time is reasonably allocated according to the difficulty of each scenario, thereby optimizing the training effect of the target student. The specific weight allocation can be determined according to actual conditions and experience, or optimized through data analysis. For example, the training difficulty is taken as input, and a linear function or nonlinear function is used to calculate the weight.

[0063] Total training time refers to the total duration of the target student's training on the driving simulator. The interactive driving simulator determines the total duration of the target student's training, i.e. the total training time, according to the student's training needs and plan.

[0064] The first scene training time refers to the preliminary training time randomly allocated to each training scenario as the basis for subsequent training parameter optimization. The total training time of the target student is randomly allocated to different training scenarios to obtain a plurality of first scene training times corresponding to a plurality of training scenarios, and the plurality of first scene training times are used as a plurality of first scene training parameters. For example, a random number generator can be used to allocate an initial training time to each scenario. Randomly allocating training time provides an initial solution space for subsequent optimization.

[0065] The first scene training success rate refers to the probability of the target student achieving success in each scenario according to the preliminary allocated training time and training difficulty. Based on the training difficulty and the first scene training time, a prediction model is trained based on historical data, and the prediction model is used to estimate the first scene training success rate in each scenario. The prediction model can be trained using statistical models such as linear regression, logistic regression, etc., or machine learning models such as decision trees, support vector machines, etc., to predict the training success rate of the target student in the corresponding training scenario according to the set first scene training time. The prediction of the training success rate can help to evaluate the rationality of the current training parameter setting.

[0066] The scene weight and the first scene training success rate are combined to calculate a first training fitness of each first scene training parameter. The first training fitness can reflect the pros and cons of the current training parameter setting. The fitness can be calculated using a weighted average method. For example, the training success rate of each scene is multiplied by its corresponding scene weight, and then summed to obtain the fitness. The calculation of the first training fitness provides an evaluation standard for subsequent parameter optimization.

[0067] An optimization algorithm, such as a genetic algorithm, a particle swarm optimization algorithm, etc., is used to optimize the parameters of the training scenes. In each iteration, the distribution of training time is adjusted according to the fitness until the fitness converges to a stable value. Finally, multiple scene training parameters with the maximum training fitness are output as multiple optimal scene training parameters, providing the most personalized and effective training arrangement for the target trainee.

[0068] The above steps optimize the scene training parameters by setting scene weights and fitness functions, continuously adjust the time distribution of each training scene, determine the optimal scene training parameters of each training scene, and maximize the training effect.

[0069] Further, step S34 includes:

[0070] Step S341: According to the historical training data of the plurality of sample trainees, a plurality of sample training difficulty sets and a plurality of sample scene training time sets of the plurality of training scenes are collected, and the proportion of the number of training successes after training is collected to obtain a plurality of sample scene training success rate sets.

[0071] Step S342: The plurality of sample training difficulty sets, the plurality of sample scene training time sets, and the plurality of sample scene training success rate sets are respectively used as a plurality of supervised training data to train a plurality of training success prediction branches to obtain a training success prediction model.

[0072] Step S343: The plurality of training difficulties and the plurality of first scene training times are respectively input into the corresponding training success prediction branches to predict the output to obtain a plurality of first scene training success rates.

[0073] Specifically, the sample training difficulty set is a set composed of training difficulties of multiple sample trainees in various training scenarios. For example, for the urban road training scenario, the training difficulty data of all sample trainees in this scenario form a sample training difficulty set. The sample scenario training time set is a set composed of training times of multiple sample trainees in various training scenarios. The sample scenario training success rate set is a set composed of proportions of the number of training successes to the total number of training times of multiple sample trainees in various training scenarios, reflecting the training success of sample trainees in different scenarios. The training success prediction branch is a model branch for predicting the training success probability constructed for different training scenarios, and each branch is specially designed to process data prediction in a specific scenario. The training success prediction model is an overall model composed of multiple training success prediction branches, which is used to predict the training success rate according to the input training difficulty and training time.

[0074] The historical training data of sample trainees is classified and sorted according to training scenarios. For each training scenario, the training difficulty data of sample trainees is extracted to form a sample training difficulty set, and the training time data is extracted to form a sample scenario training time set. At the same time, the proportion of the number of training successes to the total number of training times of each sample trainee in the training scenario is calculated, and these proportion data are combined to form a sample scenario training success rate set. For multiple training scenarios, multiple sample training difficulty sets, multiple sample scenario training time sets, and multiple sample scenario training success rate sets can be obtained.

[0075] For each training scenario, the corresponding sample training difficulty set, sample scenario training time set, and sample scenario training success rate set are used as a piece of supervised training data. These data are used to train the training success prediction branch for the training scenario. For example, for the urban road scenario, the sample training difficulty set, the sample scenario training time set, and the sample scenario training success rate set are used to train a branch that is specially designed to predict the urban road training success rate. The training process can use regression models, neural networks, decision trees, and other models for training. Multiple training success prediction branches are combined to form the training success prediction model.

[0076] With a linear regression model as an example, for each training scene, its training data includes a set of sample training difficulties under the corresponding training scene (denoted as X1), a set of sample scene training time (denoted as X2), and a set of sample scene training success rate (denoted as Y). For training difficulty and training time data, the minimum-maximum normalization method is used for normalization processing, so that the numerical range is within a suitable interval, facilitating model training. The basic form of the linear regression model is y = β0 + β1x1 + β2x2 + e, where y is the predicted training success rate (taken from Y), x1 is the training difficulty (taken from X1), x2 is the training time (taken from X2), β0 is the intercept, β1 and β2 are the regression coefficients, and e is the error term. Initialize the regression coefficients β0, β1, and β2 of the linear regression model. Select mean square error (MSE) as the loss function. Select the gradient descent algorithm as the optimization algorithm to iteratively update the regression coefficients to minimize the loss function. In each iteration of the training process, according to the current regression coefficients β0, β1, and β2, the model's predicted value y i for the training set samples is calculated, and then the loss function is calculated to update the regression coefficients according to the gradient descent algorithm. Repeat the iteration process until the stopping condition is reached. The stopping condition can be reaching the maximum number of iterations, or the value of the loss function being less than a certain pre-set threshold. Output the trained linear regression model at this time, then use the trained model to predict the test set data, which is a portion of data previously divided from the training data. For each sample (x 1i , x 2i ) in the test set, calculate the predicted training success rate y i = β0 + β1x 1i + β2x 2i + e. Use the root mean square error as an evaluation indicator to evaluate the prediction results, and the smaller the value of the root mean square error, the better the prediction effect of the model. The model that meets the expected effect is used as the training success prediction branch.

[0077] The training difficulty and the first scene training time of each training scene of the target student are input into the corresponding training success prediction branch of the training success prediction model. For example, the training difficulty and the first scene training time data of the target student in the urban road scene are input into the branch that specifically predicts the urban road training success rate. The model predicts the output of the first scene training success rate in this scene according to the relationship learned before.

[0078] Through the training success prediction model, the first scene training success rate of the target student in each training scene can be quickly and accurately predicted, providing data support for subsequent calculation of the first training adaptability, making the allocation of training time and the adjustment of training content more scientific and reasonable, and thus improving the overall training quality.

[0079] Further, the calculation formula in step S35 is as follows: wherein FCJ is a training fitness, M is the number of multiple training scenes, w i is the scene weight of the i-th training scene, K i is the scene training success rate of the i-th training scene.

[0080] Specifically, the multiple scene training success rates K i obtained in step S34 and the scene weights w i configured for each training scene in step S31, and the total number M of training scenes are substituted into the fitness calculation formula: to calculate the first training fitness of the multiple first scene training parameters. In the whole optimization process, after each iteration, the scene training success rates of the respective training scenes are predicted according to the adjusted scene training parameters, and then the training fitness under the corresponding configuration is calculated by substituting into the fitness calculation formula, so as to gradually find the scene training parameter combination with the highest fitness, thereby realizing the maximization of training efficiency and training quality, and finally improving the training success rate and overall learning effect of the target trainee in multiple training scenes.

[0081] In summary, the multi-scene switching driving simulator control method provided by the embodiments of the present application has the following technical effects:

[0082] The embodiments of the present application collect the historical training data of the trainee, analyze the training difficulty in different scenes, and then optimize the training parameters and apply them to the training control of the driving simulator. This method fully considers the individual differences of the trainee, and the optimized training parameters are more in line with the actual situation of the trainee, so as to facilitate the trainee to train under conditions more suitable for his own learning progress, which helps the trainee to master driving skills faster and reduces the wasted training time under parameters not suitable for himself, thereby improving the efficiency of the whole training process. At the same time, the personalized training parameters and the targeted optimization of the scene can enable the trainee to receive more effective training in each training scene, so as to facilitate the trainee to better adapt to the driving requirements in different scenes, comprehensively improve the driving skills, and thus improve the overall quality of the training.

[0083] Embodiment two, based on the same inventive concept as the multi-scene switching driving simulator control method in the aforementioned embodiment one, the present application further provides a computer device, comprising: at least one processor, a memory communicatively connected with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the method described in the aforementioned embodiment one.

[0084] As Figure 4As shown, the bus architecture is represented with a bus 300, which can include any number of interconnected buses and bridges, the bus 300 connects various circuits including one or more processors represented by processor 302 and memory represented by memory 304. The bus 300 can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and thus, will not be further described herein. A bus interface 305 provides an interface between the bus 300 and the receiver 301 and transmitter 303. The receiver 301 and transmitter 303 can be the same device, i.e., a transceiver, providing a unit for communicating with various other apparatuses over a transmission medium. The processor 302 is responsible for managing the bus 300 and general processing, while the memory 304 can be used for storing data used by the processor 302 in executing operations.

[0085] In the third embodiment, based on the same inventive concept of the driving simulator control method of the first embodiment, the present application further provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement each process of the driving simulator control method of the multi-scene switching embodiment and achieve the same technical effects. To avoid repetition, details are not described herein.

[0086] The above description of disclosed embodiments enables one of ordinary skill in the art to make or use the application. Various modifications to these embodiments will be readily apparent to those of ordinary skill in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended 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 of controlling a driving simulator with multiple scenario switching, characterized in that, The method comprises: collecting historical training data of a target student in a historical time, wherein the historical training data comprises training data of multiple training scenes; analyzing multiple training difficulties of the target student for the multiple training scenes according to the historical training data; optimizing training parameters of the target student for the multiple training scenes according to the multiple training difficulties, to obtain multiple optimal scene training parameters; configuring training parameters of the multiple training scenes in a driving simulator according to the multiple optimal scene training parameters, and controlling training of the target student; optimizing training parameters of the target student for the multiple training scenes according to the multiple training difficulties, to obtain multiple optimal scene training parameters, comprising: configuring multiple scene weights according to the multiple training difficulties; obtaining a total training time of the target student for driving simulation training; randomly allocating the total training time to the multiple training scenes to obtain multiple first scene training times as multiple first scene training parameters; performing training success rate prediction according to the multiple training difficulties and the multiple first scene training times, to obtain multiple first scene training success rates; calculating first training fitness of the multiple first scene training parameters according to the multiple scene weights and the multiple first scene training success rates; continuing optimization of scene training parameters of the multiple training scenes until convergence, and outputting multiple scene training parameters with maximum training fitness to obtain the multiple optimal scene training parameters.

2. The multi-scenario switching driving simulator control method according to claim 1, characterized in that, Collecting historical training data of a target student in a historical time comprises: collecting training times of the target student for the multiple training scenes in the historical time to obtain multiple historical training times; collecting training failure times of the target student for the multiple training scenes in the historical time to obtain multiple historical training failure times; integrating the multiple historical training times and the multiple historical training failure times to obtain the historical training data of the target student.

3. The multi-scenario switching driving simulator control method according to claim 2, wherein, Analyzing multiple training difficulties of the target student for the multiple training scenes according to the historical training data comprises: obtaining multiple sample training data of multiple sample students for the multiple training scenes; calculating and analyzing the multiple training difficulties of the target student for the multiple training scenes according to the multiple sample training data and the historical training data.

4. The multi-scenario switching driving simulator control method according to claim 3, wherein, Calculating and analyzing the multiple training difficulties of the target student for the multiple training scenes according to the multiple sample training data and the historical training data comprises: extracting multiple sample training time sets and multiple sample training failure time sets of the multiple training scenes according to the multiple sample training data; calculating ratios of the multiple sample training time sets and the multiple sample training failure time sets respectively to obtain multiple sample failure rate sets, and calculating average values to obtain multiple average failure rates of the multiple training scenes; calculating ratios of the multiple historical training failure times and the multiple historical training times respectively to obtain multiple failure rates of the multiple training scenes. respectively, to obtain a plurality of basic training difficulties; respectively, to obtain a plurality of average training times; respectively, to obtain a plurality of training difficulties.

5. The multi-scenario switching driving simulator control method of claim 1, wherein, According to the plurality of training difficulties and the plurality of first scene training times, a training success rate prediction is performed to obtain a plurality of first scene training success rates, including: According to the historical training data of a plurality of sample trainees, a plurality of sample training difficulty sets and a plurality of sample scene training time sets of the plurality of training scenes are collected, and the proportion of the number of successful training after training is collected to obtain a plurality of sample scene training success rate sets; respectively, to obtain a plurality of training difficulties. According to the plurality of scene weights and the plurality of first scene training success rates, a first training fitness of the plurality of first scene training parameters is calculated, as follows:

6. The multi-scenario switching driving simulator control method of claim 1, wherein, The processor executes the computer program to implement the steps of the multi-scene switching driving simulator control method of any one of claims 1-6. ； wherein FCJ is a training fitness, M is a number of training scenarios, is a scenario weight of the i-th training scenario, is a scenario training success rate of the i-th training scenario. 7.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-6. The computer program is executed by the processor to implement the steps of the multi-scene switching driving simulator control method of any one of claims 1-6.

8. A computer readable storage medium having stored thereon a computer program, characterized in that, ​

Images