Driver behavior model generation method and device

Abstract

Embodiments of the present application disclose a driver behavior model generation method and device, the method comprising: obtaining driving behavior information, the driving behavior information comprising road information, vehicle condition information and driver behavior information; determining a target prompt word sequence based on a target prompt word strategy, the road information and the vehicle condition information; and generating a driver behavior model based on the target prompt word sequence and the driver behavior information. The prompt word strategy generates diverse keywords, so that the driver behavior model constructed based on the keywords is diverse and suitable for various vehicle driving conditions.

Description

Technical Field

[0001] This application relates to the field of driver models, and more particularly to a method and apparatus for generating driver behavior models. Background Technology

[0002] With the development of automotive technology, cars are becoming increasingly intelligent, and most cars now have autonomous driving capabilities. Once activated, the vehicle can adopt corresponding driver behavior model generation strategies based on user needs and different scenarios. The driver model is a crucial component in achieving autonomous driving functionality. It simulates the behavior and decision-making process of a real driver. Based on a series of algorithms and parameters, the driver model perceives and analyzes the current vehicle state and outputs corresponding control commands to control the vehicle to achieve the driving effect desired by the user.

[0003] Traditional driver models are constrained by their internal mathematical models and can often only be constructed based on fixed keywords. As a result, the types of drivers that can be simulated by the generated driver models are limited, and they cannot be applied to diverse vehicle driving conditions, thus having certain limitations. Summary of the Invention

[0004] This application provides a method and apparatus for generating a driver behavior model. By generating diverse keywords through a prompt word strategy, the driver behavior model constructed based on the keywords becomes diverse and applicable to various vehicle driving conditions.

[0005] In a first aspect, embodiments of this application provide a method for generating a driver behavior model, including:

[0006] Acquire driving behavior information, including road information, vehicle condition information, and driver behavior information; determine the target prompt word sequence based on the target prompt word strategy, road information, and vehicle condition information; and generate a driver behavior model based on the target prompt word sequence and driver behavior information.

[0007] It can be seen that by determining a target cue word sequence including multiple keywords through the target cue word strategy, and then obtaining a driver behavior model based on this sequence, the generated driver behavior model can be diverse and applicable to various vehicle driving conditions.

[0008] In conjunction with the first aspect, in one possible implementation, the target prompt word sequence is determined based on the target prompt word strategy, road information, and vehicle condition information, including: determining the number of target prompt words in the target prompt word sequence based on road information; determining multiple alternative prompt words based on the target prompt word strategy, road information, and vehicle condition information; and determining the target prompt word sequence based on the number of prompt words and the multiple alternative prompt words.

[0009] Multiple candidate prompts are determined based on the target prompt strategy, road information, and vehicle condition information. Then, a target prompt sequence is obtained based on these multiple candidate prompts. This ensures the diversity of the target prompt sequence, thereby enabling the driver behavior model generated based on the target prompt sequence to be diverse and applicable to various vehicle driving conditions.

[0010] In conjunction with the first aspect, in one possible implementation, the road information includes the complexity of the road, and determining the number of prompt words in the target prompt word sequence based on the road information includes: determining the number of prompt words in the target prompt word sequence based on the complexity of the road.

[0011] By determining the number of target prompt words in the target prompt word sequence based on the complexity of the road, it can be ensured that the generated target prompt word sequence can accurately and completely describe the vehicle's driving status. Then, a driver behavior model is generated based on the target prompt word sequence and driver behavior information, which helps to improve the accuracy of the driver behavior model.

[0012] In conjunction with the first aspect, in one possible implementation, the target prompt word sequence is determined based on the number of target prompt words and multiple alternative prompt words, including: determining the frequency of different prompt words among the multiple alternative prompt words; sorting the multiple alternative prompt words in descending order of frequency to obtain the arrangement order of the alternative prompt words; and determining the target prompt word sequence based on the arrangement order of the alternative prompt words and the number of target prompt words.

[0013] By sorting multiple candidate prompts according to their frequency of occurrence and obtaining the final target prompt sequence, the candidate prompts with higher frequency of occurrence can better reflect the vehicle's driving status, thus enabling the target prompt sequence to more accurately describe the vehicle's driving status. Based on this target prompt sequence and driver behavior information, a driver behavior model is generated, which helps to improve the accuracy of the driver behavior model.

[0014] In conjunction with the first aspect, in one possible implementation, a driver behavior model is generated based on the target prompt word sequence and driver behavior information, including: inputting the target prompt word sequence and driver behavior information into a large language model to generate the driver behavior model.

[0015] By generating driver behavior models based on large language models, the generated driver behavior models are not constrained by internal mathematical models and can be applied in various complex scenarios.

[0016] In conjunction with the first aspect, in one possible implementation, the method further includes: determining the simulated behavior of a first driver based on a first verification dataset and a driver behavior model; determining that the simulated behavior of the first driver differs from the current real behavior of the driver, and adjusting the driver behavior model to obtain an adjusted driver behavior model.

[0017] By checking the consistency between the simulated behavior generated by the driver behavior model and real driving behavior using the first validation dataset, the accuracy of the generated driver behavior model can be ensured.

[0018] In conjunction with the first aspect, in one possible implementation, the method further includes: determining the simulated behavior of the second driver based on the second verification dataset and the driver behavior model; determining that the simulated behavior of the second driver contains dangerous behavior, and adjusting the driver behavior model to obtain the adjusted driver behavior model.

[0019] The safety of the simulated behavior generated by the driver behavior model is evaluated using a second validation dataset. Based on the evaluation results, it is determined whether the driver behavior model needs to be adjusted, which helps to ensure the reliability of the generated driver behavior model.

[0020] In conjunction with the first aspect, in one possible implementation, the driver behavior model is adjusted to obtain an adjusted driver behavior model, including: adjusting the target prompt word strategy to a modified prompt word strategy; determining a modified prompt word sequence based on the modified prompt word strategy, road information, and vehicle condition information; generating a modified driver behavior model based on the modified prompt word sequence and driver behavior information, and the modified driver behavior model is the adjusted driver behavior model.

[0021] When adjustments to the driver behavior model are needed, the cue word strategy can be adjusted to make the generated driver behavior model more accurately simulate drivers with the corresponding driving styles, thereby making the generated driver behavior model more accurate.

[0022] In conjunction with the first aspect, in one possible implementation, the driver behavior model is adjusted to obtain an adjusted driver behavior model, including: adjusting the driver behavior model by changing the parameters or structure of the driver behavior model to obtain the adjusted driver behavior model.

[0023] Adjusting the driver behavior model by directly changing its parameters or structure can save time and improve the efficiency of the adjustment process.

[0024] In conjunction with the first aspect, in one possible implementation, the driver behavior model is adjusted to obtain an adjusted driver behavior model, including: acquiring corrected driving behavior information, wherein the amount of corrected driving behavior information is greater than the amount of driving behavior information, and the corrected driving behavior information includes corrected road information, corrected vehicle condition information, and corrected driver behavior information; determining a corrected prompt word sequence based on the target prompt word strategy, the corrected road information, and the corrected vehicle condition information; and generating a corrected driver behavior model based on the corrected prompt word sequence and the driver behavior information, wherein the corrected driver behavior model is the adjusted driver behavior model.

[0025] By increasing the amount of training data, the diversity of training data can be increased to a certain extent, thereby enabling the generated driver behavior model to more accurately simulate drivers with different driving styles.

[0026] Secondly, embodiments of this application provide a driver behavior model generation device, the device comprising:

[0027] The acquisition module is used to acquire driving behavior information, which includes road information, vehicle condition information, and driver behavior information.

[0028] The processing module is used to determine the target prompt word sequence based on the target prompt word strategy, road information, and vehicle condition information; and to generate a driver behavior model based on the target prompt word sequence and driver behavior information.

[0029] Thirdly, embodiments of this application provide a driver behavior model generation device, the device comprising:

[0030] The memory, the processor, and executable program code stored in the memory and executable on the processor, the executable program code being configured to implement some or all of the steps as described in any of the methods in the first aspect.

[0031] Fourthly, embodiments of this application provide a computer-readable storage medium storing a driver behavior model generation program, which, when executed by a processor, implements some or all of the steps described in any of the methods in the first aspect.

[0032] Fifthly, embodiments of this application provide a computer program product, wherein the computer program product includes a computer program operable to cause a computer to perform some or all of the steps described in any method of the first aspect of this application. The computer program product may be a software installation package.

[0033] By implementing the embodiments of this application, driving behavior information, including road information, vehicle condition information, and driver behavior information, is first acquired; then, a target prompt word sequence is determined based on the target prompt word strategy, road information, and vehicle condition information; finally, a driver behavior model is generated based on the target prompt word sequence and driver behavior information. The prompt word strategy generates diverse keywords, enabling the driver behavior model constructed based on these keywords to be unrestricted by vehicle driving conditions and possess greater diversity. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application or the background art, the accompanying drawings used in the embodiments of this application or the background art will be described below.

[0035] Figure 1 This is a flowchart of a driver behavior model generation method provided in an embodiment of this application;

[0036] Figure 2 This is a flowchart of a prompt word strategy provided in an embodiment of this application;

[0037] Figure 3 This is an application flowchart of a driver behavior model provided in an embodiment of this application;

[0038] Figure 4 This is a schematic diagram of the structure of a control device provided in an embodiment of this application;

[0039] Figure 5 This is a schematic diagram of another control device provided in an embodiment of this application. Detailed Implementation

[0040] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0041] The terms "first," "second," and "third," etc., used in the specification, claims, and drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0042] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0043] This application provides a method for generating a driver behavior model, which will be described in detail below with reference to the accompanying drawings.

[0044] Please see Figure 1 , Figure 1 This is a flowchart of a driver behavior model generation method provided in an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0045] S101, the control device acquires driving behavior information.

[0046] The control device is the vehicle's domain controller or a module within the domain controller. The domain controller acquires driving behavior information in real time.

[0047] The driving behavior information includes road information, vehicle condition information, and driver behavior information. Specifically, road information includes information on road topology, traffic signs, lane lines, etc.; vehicle condition information includes vehicle-related parameters such as vehicle speed, acceleration, steering angle, and brake pressure; and driver behavior information includes driver behavior parameters such as steering wheel rotation, accelerator pedal pressure, and brake pedal pressure.

[0048] Among them, driving behavior information can be relevant data obtained through sensor devices in actual driving scenarios, such as vehicle sensors and cameras; driving behavior information can also be relevant data obtained through simulation in a simulation environment. This method obtains a larger amount of relevant data and is easier to label and debug.

[0049] The vehicle's location information can be obtained through a Global Positioning System (GPS) receiver; the vehicle's acceleration and angular velocity can be obtained through an Inertial Measurement Unit (IMU); vehicle speed, steering angle, and brake pressure can be obtained through vehicle speed sensors, steering sensors, and brake pressure sensors; driver behavior parameters such as steering wheel rotation, accelerator pedal pressure, and brake pedal pressure can be obtained through dedicated sensors, such as steering wheel angle sensors, pedal pressure sensors, or vehicle network communication interfaces; and visual information about the vehicle's surroundings, such as road conditions, traffic signs, vehicles, and pedestrians, can be obtained through onboard cameras. Then, key information from the images is extracted using computer vision technology.

[0050] S102, the control device determines the target prompt word sequence based on the target prompt word strategy, road information and vehicle condition information.

[0051] Please refer to Figure 2 , Figure 2 This is a flowchart of a prompt word strategy provided in an embodiment of this application, such as... Figure 2 As shown, the design of a target prompting word strategy includes the following steps:

[0052] S1021, Define Driving Style: First, it is necessary to define different driving styles or specific types of driver behavior, such as conservative, aggressive, and steady. These driving styles can be categorized based on the driver's habits, preferences, and behavioral characteristics. They can also be defined based on the driver's skill level and experience, such as novice driver, advanced driver, or experienced driver.

[0053] S1022, Cue Word Selection: Select appropriate cue words for each driving style. Cue words are key information input to the large language model, used to guide the model in generating specific types of driver behavior. Cue words should accurately describe the required driving behavior, such as maintaining a safe distance, smooth acceleration, and appropriate lane changes.

[0054] S1023, Cue Sequence Design: Determine the order and number of cues. The cue sequence can be designed based on the complexity of the driving task and the level of detail required for the actions. Simpler tasks may only require one or a few cues, while complex tasks may require a longer cue sequence.

[0055] S1024, Prompt Word Strategy Optimization: Through experimentation and verification, the prompt word strategy is continuously adjusted and optimized to make the generated driver behavior more accurate and realistic. Simulation experiments or tests can be conducted in real driving scenarios to evaluate whether the model-generated behavior is consistent with the expected driving style. By carefully designing and optimizing the prompt word strategy, combined with environmental information, the large language model can be guided to generate control signals that conform to the specified driving style.

[0056] The target cue word sequence includes one or more cue words described in natural language. The cue words can include the driving style and driving behavior of the driving prototype, such as aggressive, conservative, maintaining a safe distance from the vehicle in front, changing lanes and overtaking, etc.

[0057] In one possible implementation, the control device determines the target prompt word sequence based on the target prompt word strategy, road information, and vehicle condition information, including: the control device determining the number of target prompt words in the target prompt word sequence based on the road information; the control device determining multiple alternative prompt words based on the target prompt word strategy, road information, and vehicle condition information; and the control device determining the target prompt word sequence based on the number of prompt words and the multiple alternative prompt words.

[0058] Among them, several alternative prompts describe the driver's style and common driving operations in natural language. For example, when it is determined from road and vehicle information that the vehicle always maintains a certain distance from the vehicle in front and will continuously accelerate when the road ahead is smooth, the alternative prompts can be: maintain a safe distance, accelerate smoothly, and be steady. When it is determined from road and vehicle information that the vehicle likes to change lanes and overtake when permitted, the alternative prompts can be: change lanes reasonably, overtake decisively, and be aggressive.

[0059] As can be seen, in this example, multiple alternative prompt words are determined based on the target prompt word strategy, road information, and vehicle condition information. Then, the target prompt word sequence is obtained based on the multiple alternative prompt words. This ensures the diversity of the target prompt word sequence, thereby making the driver behavior model generated based on the target prompt word sequence diverse and applicable to various vehicle driving conditions.

[0060] In one possible implementation, the road information includes the complexity of the road, and the control device determines the number of prompt words in the target prompt word sequence based on the road information, including: the control device determines the number of prompt words in the target prompt word sequence based on the complexity of the road.

[0061] The complexity of a road can reflect, to some extent, the complexity of the driving task and the level of detail required. The complexity of a road can be determined by factors such as the volume of traffic, the number of lanes, or the complexity of the operable operations at an intersection. For example, a four-lane road is more complex than a two-lane road, a road with high traffic volume is more complex than a road with low traffic volume, and a crossroads is more complex than a straight-ahead intersection.

[0062] As can be seen in this example, by determining the number of target prompt words in the target prompt word sequence based on the complexity of the road, it can be ensured that the generated target prompt word sequence can accurately and completely describe the vehicle's driving status. Then, by generating a driver behavior model based on the target prompt word sequence and driver behavior information, the accuracy of the driver behavior model can be improved.

[0063] In one possible implementation, the control device determines the target prompt word sequence based on the number of target prompt words and multiple alternative prompt words, including: the control device determining the frequency of different prompt words among the multiple alternative prompt words; the control device sorting the multiple alternative prompt words from most frequent to least frequent to obtain the arrangement order of the alternative prompt words; and the control device determining the target prompt word sequence based on the arrangement order of the alternative prompt words and the number of target prompt words.

[0064] Among the multiple candidate prompts, duplicate prompts may appear. Specifically, when driving on different roads, the same or similar driving operations by the driver will result in the generation of the same candidate prompts. For example, if there are two target prompts, the multiple candidate prompts obtained are smooth acceleration, lane change and overtaking, lane change and overtaking, lane change in advance, smooth acceleration, immediate overtaking, and aggressive type. In this case, after sorting the above multiple candidate prompts according to their frequency of occurrence from most to least, since there are two target prompts, the two candidate prompts with the highest frequency of occurrence are selected as the target prompt sequence, which is lane change and overtaking and smooth acceleration.

[0065] As can be seen in this example, by sorting multiple candidate prompts according to their frequency of occurrence to obtain the final target prompt sequence, the candidate prompts with higher frequencies can better reflect the vehicle's driving status, thus enabling the target prompt sequence to more accurately describe the vehicle's driving status. Based on this target prompt sequence and driver behavior information, a driver behavior model is generated, which helps to improve the accuracy of the driver behavior model.

[0066] In one possible implementation, the control device generates a driver behavior model based on a target prompt word sequence and driver behavior information, including: the control device inputs the target prompt word sequence and driver behavior information into a large language model to generate the driver behavior model.

[0067] Large Language Models (LLMs), also known as large-scale language models, are artificial intelligence models designed to understand and generate human language. Trained on massive amounts of text data, LLMs learn and master general language knowledge and abilities through unsupervised, semi-supervised, or self-supervised methods, enabling them to perform a wide range of tasks, including text summarization, translation, and sentiment analysis.

[0068] As can be seen in this example, by generating a driver behavior model based on a large language model, the generated driver behavior model is not constrained by the internal mathematical model and can be applied in various complex scenarios.

[0069] S103, the control device generates a driver behavior model based on the target prompt word sequence and driver behavior information.

[0070] In one possible implementation, the method further includes: the control device determining the simulated behavior of a first driver based on a first verification dataset and a driver behavior model; the control device determining that the simulated behavior of the first driver differs from the current real behavior of the driver, adjusting the driver behavior model to obtain an adjusted driver behavior model.

[0071] The first validation dataset includes complex road information and vehicle condition information encountered by vehicles during driving, such as intersections with high pedestrian traffic, intersections without traffic lights, and roads in front of schools.

[0072] As can be seen in this example, by checking the consistency between the simulated behavior generated by the driver behavior model and the real driving behavior using the first validation dataset, the accuracy of the generated driver behavior model can be ensured.

[0073] In one possible implementation, the method further includes: the control device determining the simulated behavior of the second driver based on the second verification dataset and the driver behavior model; the control device determining that the simulated behavior of the second driver is dangerous, adjusting the driver behavior model to obtain the adjusted driver behavior model.

[0074] The second validation dataset is used to evaluate the safety of the driver behavior model. Dangerous behaviors include driving behaviors that do not comply with traffic rules or safe driving principles. The metrics for measuring dangerous behaviors can be following distance, lane change timing, and acceleration control. For example, dangerous behaviors can include following too closely, sudden acceleration, crossing intersections when the traffic light is red, or not slowing down when there are pedestrians at the crosswalk ahead. After determining that the second driver simulation behavior contains dangerous behaviors, the driver behavior model is adjusted.

[0075] As can be seen, in this example, evaluating the safety of the simulated behavior generated by the driver behavior model using the second validation dataset and determining whether to adjust the driver behavior model based on the evaluation results helps ensure the reliability of the generated driver behavior model.

[0076] In one possible implementation, the driver behavior model is adjusted to obtain an adjusted driver behavior model, including: the control device adjusts the target prompt word strategy to a modified prompt word strategy; the control device determines a modified prompt word sequence based on the modified prompt word strategy, road information, and vehicle condition information; the control device generates a modified driver behavior model based on the modified prompt word sequence and driver behavior information, and the modified driver behavior model is the adjusted driver behavior model.

[0077] The adjustment methods for the prompt word strategy include data analysis, user feedback, and expert evaluation. Specifically, data analysis involves analyzing driver behavior data to understand commonly used prompt words or phrases in different situations and the relationship between these prompt words and specific behaviors. Based on the data analysis results, the prompt word strategy can be adjusted, such as increasing or decreasing the frequency of certain prompt words. User feedback involves collecting driver feedback on the prompt word strategy through user surveys, questionnaires, or on-site observations. Based on the user feedback, the prompt word strategy can be optimized and adjusted to better suit drivers' actual needs and habits. Expert evaluation involves consulting experts in relevant fields to evaluate and provide suggestions on the prompt word strategy. Experts can offer targeted optimization suggestions based on their experience and knowledge to improve the effectiveness and practicality of the prompt word strategy.

[0078] As can be seen in this example, when it is necessary to adjust the driver behavior model, adjusting the prompt word strategy can make the generated driver behavior model more accurately simulate the driver with the corresponding driving style, thereby making the generated driver behavior model more accurate.

[0079] In one possible implementation, the driver behavior model is adjusted to obtain an adjusted driver behavior model, including: the control device adjusts the driver behavior model by changing the parameters or structure of the driver behavior model to obtain the adjusted driver behavior model.

[0080] The performance of driver behavior models is influenced by factors such as hyperparameters, data quality, and feature selection. Hyperparameters are parameters manually set within the model; these cannot be directly learned from training data but must be manually configured. When adjusting a driver behavior model, more critical parameters, such as the learning rate and number of iterations, should be adjusted first, followed by adjustments to other parameters. Specifically, when adjusting the parameters of a driver behavior model, cross-validation can be used. K-fold cross-validation can be employed to evaluate the performance of the driver behavior model under different parameter combinations, thereby selecting the optimal parameter combination.

[0081] The driver behavior model includes a multi-layered structure, with each layer containing multiple substructures. When the driver behavior model needs adjustment, the substructures under each layer can be adjusted first, such as replacing or deleting existing substructures, or adding substructures.

[0082] As can be seen in this example, by directly changing the parameters or structure of the driver behavior model to adjust the driver behavior model, the time required for adjusting the driver behavior model can be saved, and the adjustment efficiency of the driver behavior model can be improved.

[0083] In one possible implementation, the driver behavior model is adjusted to obtain an adjusted driver behavior model, including: a control device acquiring corrected driving behavior information, wherein the amount of corrected driving behavior information is greater than the amount of driving behavior information, and the corrected driving behavior information includes corrected road information, corrected vehicle condition information, and corrected driver behavior information; the control device determining a corrected prompt word sequence based on a target prompt word strategy, corrected road information, and corrected vehicle condition information; and the control device generating a corrected driver behavior model based on the corrected prompt word sequence and driver behavior information, wherein the corrected driver behavior model is the adjusted driver behavior model.

[0084] As can be seen in this example, by increasing the amount of training data, the diversity of training data can be increased to a certain extent, thereby enabling the generated driver behavior model to more accurately simulate drivers with different driving styles.

[0085] Please see Figure 3 , Figure 3 This is an application flowchart of a driver behavior model provided in an embodiment of this application, such as... Figure 3 As shown, the method includes the following steps:

[0086] S301, The control device acquires environmental data.

[0087] The control device is the vehicle's domain controller or a module within the domain controller. The domain controller acquires driving behavior information in real time, while the control device acquires environmental data collected by sensors in real time. This data may include information such as the vehicle's position, speed, acceleration, steering wheel angle, and brake pressure, as well as environmental information such as road conditions, traffic signals, and obstacle detection.

[0088] Among them, environmental information can be relevant data obtained through sensor devices in actual driving scenarios, such as vehicle sensors and cameras; environmental information can also be relevant data obtained through simulation of the environment. This method obtains a larger amount of relevant data and is easier to label and debug.

[0089] The vehicle's location information can be obtained through a Global Positioning System (GPS) receiver; the vehicle's acceleration and angular velocity can be obtained through an Inertial Measurement Unit (IMU); vehicle speed, steering angle, and brake pressure can be obtained through vehicle speed sensors, steering sensors, and brake pressure sensors; driver behavior parameters such as steering wheel rotation, accelerator pedal pressure, and brake pedal pressure can be obtained through dedicated sensors, such as steering wheel angle sensors, pedal pressure sensors, or vehicle network communication interfaces; and visual information about the vehicle's surroundings, such as road conditions, traffic signs, vehicles, and pedestrians, can be obtained through onboard cameras. Then, key information from the images is extracted using computer vision technology.

[0090] S302, the control device generates a driver behavior model based on environmental data.

[0091] The control device first obtains a sequence of prompt words based on environmental data, and then obtains a driver behavior model based on the sequence of prompt words and a large language model.

[0092] Large Language Models (LLMs), also known as large-scale language models, are artificial intelligence models designed to understand and generate human language. Trained on massive amounts of text data, LLMs learn and master general language knowledge and abilities through unsupervised, semi-supervised, or self-supervised methods, enabling them to perform a wide range of tasks, including text summarization, translation, and sentiment analysis.

[0093] The target cue word sequence includes one or more cue words described in natural language. The cue words can include the driving style and driving behavior of the driving prototype, such as aggressive, conservative, maintaining a safe distance from the vehicle in front, changing lanes and overtaking, etc.

[0094] The number of prompts in the prompt sequence is determined by the complexity of the road. The complexity of the road can reflect the complexity of the driving task and the level of detail of the required actions to a certain extent. The complexity of the road can be determined by the traffic volume, the number of lanes, or the complexity of the executable operations at the intersection. For example, a four-lane road is more complex than a two-lane road, a road with high traffic volume is more complex than a road with low traffic volume, and a crossroads is more complex than a straight-ahead intersection.

[0095] The process of generating a driver behavior model includes preprocessing, feature extraction, and transformation of the input environmental data, converting real-time environmental data into natural language so that it can be input into the driver model for prediction.

[0096] S303, the control device obtains the driver's operation based on environmental information and driver behavior model.

[0097] In this process, after obtaining the driver behavior model, the control device uses the acquired environmental information as the input to the driver behavior model, and then obtains the output of the driver behavior model, which is the driver's operation.

[0098] S304, the control device generates corresponding vehicle control signals based on the prediction results of the driver behavior model.

[0099] Generating the corresponding vehicle control signals involves calculating control parameters such as steering angle, accelerator pedal pressure, and brake pressure to simulate driver behavior.

[0100] S305, the control device transmits the generated vehicle control signals to the vehicle model to achieve simulation of vehicle control.

[0101] Transmitting vehicle control signals to the vehicle model involves converting the control signals into a specific communication protocol or data format to facilitate interaction with vehicle simulation software.

[0102] S306, the control device adjusts the parameters of the driver behavior model based on the feedback information.

[0103] The feedback information includes vehicle status and sensor data. The parameters of the driver behavior model are adjusted based on the feedback information to ensure that the generated driver behavior is consistent with the actual vehicle behavior.

[0104] As can be seen in this example, the generation of diverse keywords through the prompt word strategy enables the driver behavior model built based on the keywords to be more diverse and not limited by the vehicle's driving conditions.

[0105] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of a control device provided in an embodiment of this application, such as... Figure 4 As shown, the control device 400 includes:

[0106] The acquisition module 401 is used to acquire driving behavior information, which includes road information, vehicle condition information and driver behavior information.

[0107] The processing module 402 is used to determine the target prompt word sequence based on the target prompt word strategy, road information and vehicle condition information; and to generate a driver behavior model based on the target prompt word sequence and driver behavior information.

[0108] In one possible implementation, in determining the target prompt word sequence based on the target prompt word strategy, road information, and vehicle condition information, the processing module 402 is specifically used to: determine the number of target prompt words in the target prompt word sequence based on road information; determine multiple alternative prompt words based on the target prompt word strategy, road information, and vehicle condition information; and determine the target prompt word sequence based on the number of prompt words and the multiple alternative prompt words.

[0109] In one possible implementation, the processing module 402 is specifically used to determine the number of prompt words in the target prompt word sequence based on the road information, which includes the complexity of the road.

[0110] In one possible implementation, in determining the target prompt word sequence based on the number of target prompt words and multiple alternative prompt words, the processing module 402 is specifically used to: determine the frequency of different prompt words among the multiple alternative prompt words; sort the multiple alternative prompt words according to their frequency of occurrence from most to least to obtain the arrangement order of the alternative prompt words; and determine the target prompt word sequence based on the arrangement order of the alternative prompt words and the number of target prompt words.

[0111] In one possible implementation, in generating a driver behavior model based on the target prompt word sequence and driver behavior information, the processing module 402 is specifically used to: input the target prompt word sequence and driver behavior information into a large language model to generate a driver behavior model.

[0112] In one possible implementation, the processing module 402 is further configured to: determine the simulated behavior of a first driver based on the first verification dataset and the driver behavior model; compare the simulated behavior of the first driver with the real behavior of the driver; and when it is determined that there is a difference between the simulated behavior of the first driver and the current real behavior of the driver, adjust the driver behavior model to obtain the adjusted driver behavior model.

[0113] In one possible implementation, the processing module 402 is further configured to: determine the second driver's simulated behavior based on the second verification dataset and the driver behavior model; and after determining that the second driver's simulated behavior contains dangerous behavior, adjust the driver behavior model to obtain an adjusted driver behavior model.

[0114] In one possible implementation, in adjusting the driver behavior model to obtain the adjusted driver behavior model, the processing module 402 is specifically used to: adjust the target prompt word strategy to a modified prompt word strategy; determine a modified prompt word sequence based on the modified prompt word strategy, road information, and vehicle condition information; generate a modified driver behavior model based on the modified prompt word sequence and driver behavior information, and the modified driver behavior model is the adjusted driver behavior model.

[0115] In one possible implementation, in adjusting the driver behavior model to obtain an adjusted driver behavior model, the processing module 402 is specifically used to: adjust the driver behavior model by changing the parameters or structure of the driver behavior model to obtain an adjusted driver behavior model.

[0116] In one possible implementation, in adjusting the driver behavior model to obtain the adjusted driver behavior model, the processing module 402 is specifically used to: acquire corrected driving behavior information, wherein the amount of data of the corrected driving behavior information is greater than the amount of data of the driving behavior information, and the corrected driving behavior information includes corrected road information, corrected vehicle condition information, and corrected driver behavior information; determine a corrected prompt word sequence based on the target prompt word strategy, the corrected road information, and the corrected vehicle condition information; and generate a corrected driver behavior model based on the corrected prompt word sequence and the driver behavior information, wherein the corrected driver behavior model is the adjusted driver behavior model.

[0117] It is worth noting that the specific functional implementation of the control device 400 is described above. Figure 1 The description of the driver behavior model generation method shown includes, for example, the acquisition module 401 for implementing the relevant content of S101, and the processing module 402 for implementing the relevant content of S102-S103. Each unit or module in the control device 400 can be individually or entirely merged into one or more other units or modules, or some of the units or modules can be further divided into multiple functionally smaller units or modules. This achieves the same operation without affecting the technical effect of the embodiments of the present invention. The above-mentioned units or modules are based on logical function division. In practical applications, the function of one unit (or module) is implemented by multiple units (or modules), or the function of multiple units (or modules) is implemented by one unit (or module).

[0118] Based on the description of the above method embodiments and related device embodiments, please refer to... Figure 5 , Figure 5 This is a schematic diagram of another control device provided in an embodiment of this application. Figure 5 The control device 500 shown includes a processor 501, a memory 502, a communication interface 503, and a bus 504. The processor 501, memory 502, and communication interface 503 are interconnected via the bus 504.

[0119] Optionally, the memory 502 can be a ROM, a static storage device, a dynamic storage device, or RAM.

[0120] Memory 502 stores executable program code. When the executable program code stored in memory 502 is executed by processor 501, processor 501 and communication interface 503 are used for execution. Figure 2 The various steps of the video frame processing method in the illustrated embodiment.

[0121] The processor 501 employs a general-purpose CPU, microprocessor, application-specific integrated circuit (ASIC), GPU, or one or more integrated circuits to execute relevant programs to perform the image quality processing method of the method embodiments of this application.

[0122] The processor 501 can also be an integrated circuit chip with signal processing capabilities. In implementation, each step of the video frame processing method of this application can be completed through the integrated logic circuitry in the hardware of the processor 501 or through software instructions. Optionally, the processor 501 can be a general-purpose processor, DSP, ASIC, FPGA, or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The processor can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor is a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. Optional software modules are located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory 502. The processor 501 reads the information in memory 502 and, in conjunction with its hardware, completes the functions required by the units included in the control device 400 of this application embodiment, or executes the video frame processing method of the method embodiment of this application.

[0123] Communication interface 503 uses transceiver-related devices such as, but not limited to, transceivers.

[0124] Bus 504 may include a pathway for transmitting information between various components of control device 500 (e.g., memory 502, processor 501, communication interface 503).

[0125] It should be noted that, although Figure 5 The control device 500 shown only illustrates the memory, processor, and communication interface. However, those skilled in the art should understand that in specific implementations, the control device 500 may also include other devices necessary for normal operation. Furthermore, based on specific needs, those skilled in the art should understand that the control device 500 may also include hardware devices for implementing other additional functions. Moreover, those skilled in the art should understand that the control device 500 may only include the devices necessary for implementing the embodiments of this application, and may not necessarily include... Figure 5 All the devices shown.

[0126] This application provides a computer-readable storage medium storing a computer program for electronic data interchange. The computer program includes execution instructions for performing some or all of the steps of any of the driver behavior model generation methods described in the above-described driver behavior model generation method embodiments. The computer includes an electronic terminal device.

[0127] This application provides a computer program product, which includes a computer program operable to enable a computer to perform some or all of the steps of any driver behavior model generation method described in the above method embodiments. The computer program product may be a software installation package.

[0128] It should be noted that, for the sake of simplicity, each of the aforementioned embodiments of the driver behavior model generation method is described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to this application.

[0129] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principle and implementation of a driver behavior model generation method and device of this application. The description of the above embodiments is only for the purpose of helping to understand the method and its core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the ideas of a driver behavior model generation method and device of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

[0130] This application is described with reference to flowchart illustrations and / or block diagrams of methods, hardware products, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0131] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. Memory may include: flash drives, read-only memory (ROM), random access memory (RAM), hard disks or optical disks, etc.

[0132] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce a good effect.

[0133] Those skilled in the art will understand that all or part of the steps in the various method embodiments of any of the above-described driver behavior model generation methods can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a disk, or an optical disk, etc.

[0134] It is understood that any product that is controlled or configured to execute the processing method of the flowchart described in an embodiment of the driver behavior model generation method of this application, such as the apparatus and computer program product of the above flowchart, falls within the scope of the related products described in this application.

[0135] Obviously, those skilled in the art can make various modifications and variations to the driver behavior model generation method and apparatus provided in this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. A method for generating a driver behavior model, characterized in that, The method includes: Acquire driving behavior information, which includes road information, vehicle condition information, and driver behavior information; The number of target prompt words in the target prompt word sequence is determined based on the road information; Multiple alternative prompt words are determined based on the target prompt word strategy, the road information, and the vehicle condition information; A target prompt word sequence is determined based on the number of target prompt words and multiple candidate prompt words; The target prompt word sequence and the driver behavior information are input into a large language model to generate a driver behavior model.

2. The method as described in claim 1, characterized in that, The road information includes the complexity of the road, and determining the number of prompt words in the target prompt word sequence based on the road information includes: The number of prompt words in the target prompt word sequence is determined based on the complexity of the road.

3. The method as described in claim 1, characterized in that, Determining the target prompt word sequence based on the number of target prompt words and multiple candidate prompt words includes: Determine the frequency of different prompt words among the multiple candidate prompt words; The multiple candidate prompt words are sorted from most frequent to least frequent to obtain the arrangement order of the candidate prompt words; The target prompt word sequence is determined based on the order of the candidate prompt words and the number of target prompt words.

4. The method as described in claim 1, characterized in that, The method further includes: The simulated behavior of the first driver is determined based on the first verification dataset and the driver behavior model. If it is determined that there is a difference between the simulated behavior of the first driver and the actual behavior of the driver, the driver behavior model is adjusted to obtain an adjusted driver behavior model.

5. The method as described in claim 1, characterized in that, The method further includes: The simulated behavior of the second driver was determined based on the second verification dataset and the driver behavior model. If the second driver simulation behavior is determined to be dangerous, the driver behavior model is adjusted to obtain an adjusted driver behavior model.

6. The method as described in claim 4 or 5, characterized in that, The adjustment of the driver behavior model to obtain the adjusted driver behavior model includes: Adjust the target prompting strategy to a modified prompting strategy; The sequence of correction prompt words is determined based on the correction prompt word strategy, the road information, and the vehicle condition information; A corrected driver behavior model is generated based on the corrected prompt word sequence and the driver behavior information, and the corrected driver behavior model is the adjusted driver behavior model.

7. The method as described in claim 4 or 5, characterized in that, The adjustment of the driver behavior model to obtain the adjusted driver behavior model includes: The driver behavior model is adjusted by changing its parameters or structure to obtain an adjusted driver behavior model.

8. The method as described in claim 4 or 5, characterized in that, The adjustment of the driver behavior model to obtain the adjusted driver behavior model includes: Obtain corrected driving behavior information, wherein the amount of data for corrected driving behavior information is greater than the amount of data for driving behavior information, and the corrected driving behavior information includes corrected road information, corrected vehicle condition information, and corrected driver behavior information; The corrected prompt word sequence is determined based on the target prompt word strategy, the corrected road information, and the corrected vehicle condition information; A corrected driver behavior model is generated based on the corrected prompt word sequence and the driver behavior information, and the corrected driver behavior model is the adjusted driver behavior model.

9. A driver behavior model generation device, characterized in that, The device includes: The acquisition module is used to acquire driving behavior information, which includes road information, vehicle condition information, and driver behavior information. The processing module is configured to: determine the number of target prompt words in the target prompt word sequence based on the road information; determine multiple alternative prompt words based on the target prompt word strategy, the road information, and the vehicle condition information; determine the target prompt word sequence based on the number of target prompt words and the multiple alternative prompt words; and input the target prompt word sequence and the driver behavior information into a large language model to generate a driver behavior model.

10. A driver behavior model generation device, characterized in that, The device includes: The method comprises a memory, a processor, and executable program code stored in the memory and executable on the processor, wherein the processor executes the executable program code to perform the steps of the driver behavior model generation method as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a driver behavior model generation program, the driver behavior model generation program including execution instructions for performing the steps of the driver behavior model generation method as described in any one of claims 1-8.

Images