Driving strategy dynamic adjustment method and system based on driving style recognition

Abstract

The invention discloses a driving strategy dynamic adjustment method and system based on driving style recognition. The method comprises the following steps: establishing a driving style classification model; monitoring the behavior data of the driver in real time, completing the confirmation of the driving style of the driver by using a data-driven adaptive algorithm, and dynamically updating the driving style classification model; on the basis of the driving style of the driver, specific behaviors of the driver in different situations are considered, and a scene-driven self-adaptive control strategy is applied to output a current driving strategy of the vehicle; and optimizing the current driving strategy of the vehicle through the interactive feedback of the driver in the driving process. The method can assist the automatic driving system in providing a driving experience mode more conforming to the personal preference of the driver, and improves the comfort and satisfaction of the driver in combination with an intelligent interaction and strategy closed-loop mechanism, and meets the individuation and dynamic adaptability of the automatic driving system.

Description

Technical Field

[0001] The present invention belongs to the technical field of dynamic adjustment of automatic driving driving strategies, and specifically relates to a method and system for dynamic adjustment of driving strategies based on driving style recognition. Background Art

[0002] Autonomous vehicles can ensure driving safety and personalize the driving experience by monitoring and analyzing the driver's driving style.

[0003] At present, autonomous driving systems have made some progress in adapting to driver styles, mainly including driving style classification, driver behavior prediction, driving behavior analysis, etc. However, there are still some pain points, especially when the style is very different from the driver's style, which can easily cause driver dissatisfaction. The main pain points include insufficient adaptability to driving style and inconsistent driving experience.

[0004] Existing systems are not sufficiently adaptable to driver styles, and it is difficult to fully meet the personalized needs of different drivers. For example, some drivers prefer a more aggressive driving style, but the system is often set to a more conservative driving strategy, resulting in a poor driving experience. The system may not perform consistently under different road conditions, weather, and traffic conditions, and there may be a large difference between the system's processing capabilities and the driver's expectations, resulting in a decrease in the driver's trust in the system. Summary of the invention

[0005] In order to solve the above-mentioned problems existing in the prior art, the present invention provides a method and system for dynamically adjusting driving strategies based on driving style recognition, constructs a driving style classification model, combines a data-driven adaptive algorithm to identify the behavior patterns of different drivers, and uses incremental learning and driver behavior dynamic data to update the driving style model. A scenario-driven adaptive control strategy is adopted to assist the automatic driving system in providing a driving experience mode that is more in line with the driver's personal preferences, and combines intelligent interaction with a strategy closed-loop mechanism to improve the driver's comfort and satisfaction, thereby meeting the personalization and dynamic adaptability of the automatic driving system.

[0006] The objective of the present invention is achieved through the following technical solutions:

[0007] In a first aspect, the present invention provides a method for dynamically adjusting a driving strategy based on driving style recognition, comprising the following steps:

[0008] S1. Establish a driving style classification model;

[0009] S2. Data-driven adaptive algorithm: real-time monitoring of driver behavior data, application of data-driven adaptive algorithm to complete the driver's driving style confirmation, and dynamic update of the driving style classification model;

[0010] S3. Application scenario-driven adaptive control strategy: Based on the driver's driving style, the application scenario-driven adaptive control strategy outputs the vehicle's current driving strategy by considering the driver's specific behavior in different situations;

[0011] S4. Intelligent interaction and strategy optimization: Optimize the vehicle’s current driving strategy through the driver’s interactive feedback during the driving process.

[0012] Furthermore, the driver styles are divided into three categories: conservative, moderate, and aggressive.

[0013] Furthermore, the step S2 comprises the following steps:

[0014] S21. Data collection: real-time collection of vehicle driving behavior data and driver status data;

[0015] S22. Data preprocessing: cleaning, denoising and normalizing the collected data;

[0016] S23. Feature extraction: extract features from preprocessed data;

[0017] S24. Training the random forest model: inputting the feature data extracted in step S23 into the random forest model for training;

[0018] S25. Behavior pattern recognition: The trained random forest model recognizes the behavior patterns of different drivers;

[0019] S26. Incremental learning model update: The feature data newly extracted in step S23 is input into the incremental learning model for updating;

[0020] S27. Real-time behavior pattern adjustment: According to the update result of the incremental learning model, the behavior pattern recognition result is adjusted in real time, and the driving style classification model is updated.

[0021] Furthermore, in step S21, the vehicle driving behavior data includes vehicle speed, acceleration, braking force, and steering angle.

[0022] Furthermore, in step S21, the driver status data includes driver expression, sight direction, heart rate, and operation frequency.

[0023] Furthermore, in step S24, the training process of the random forest model includes the following steps:

[0024] 1) Data sampling: Multiple subsets are randomly extracted from the original training data set, and each subset is used to train a decision tree;

[0025] 2) Feature selection: When constructing each decision tree, for each node, a portion of features are randomly selected for splitting to increase the diversity of the model;

[0026] 3) Decision tree training: Use each subset to train a decision tree;

[0027] 4) Integrated prediction: For classification tasks, random forest determines the final classification result by majority voting on the prediction results of all decision trees; for regression tasks, the average of the prediction values ​​of all decision trees is taken.

[0028] Furthermore, in step S3, the adaptive strategy is combined with the specific scenario and the driving style, and different driving strategy adjustment schemes are preset for different driving styles in different driving scenarios.

[0029] Furthermore, the step S4 comprises:

[0030] Design feedback interface;

[0031] Obtain user feedback: used to provide feedback on the current driving strategy through the feedback interface;

[0032] Feedback analysis and processing: perform data processing and NLP speech analysis on feedback information to extract key information;

[0033] Optimize the current driving strategy: Optimize the scenario-driven adaptive control strategy based on the analyzed and processed key information.

[0034] Furthermore, in step S4, the feedback interface adopts a human-computer interaction interface designed on a vehicle-mounted display screen.

[0035] In a second aspect, a driving strategy dynamic adjustment system based on driving style recognition is also provided, which is used to implement the method of the present invention, including:

[0036] Basic layer: used to build a driving style classification model;

[0037] Dynamic layer: used to monitor the driver's behavior data in real time, apply data-driven adaptive algorithms to dynamically adjust the driving style model, complete the driver's driving style confirmation and model update;

[0038] Fusion layer: It is used to apply scenario-driven adaptive control strategies based on the dynamic layer, consider the specific behaviors of drivers in different situations, and provide different driving strategies;

[0039] Interaction layer: Optimize adaptive control strategies through real-time user feedback.

[0040] The present invention has the following advantages:

[0041] 1. The present invention can dynamically adjust driving strategies according to different driving styles (such as conservative, moderate, and aggressive) and real-time behavioral data. This personalized adjustment not only improves the driver's comfort and satisfaction, but also significantly improves driving safety and reduces the accident rate. At the same time, the system's adaptability and consistency are enhanced, and efficient performance can be maintained regardless of the driving environment. In addition, energy efficiency is optimized, fuel or electricity consumption is reduced, and green driving is achieved. Accurate feedback and optimization suggestions also help drivers improve their driving behavior and improve their overall driving level, thereby enhancing their trust and reliance on the autonomous driving system;

[0042] 2. The present invention can maintain efficient adaptability in different driving environments and provide a consistent driving experience through scene classification and optimization;

[0043] 3. The present invention can improve driving safety and driver's trust. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings used in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the contents of the embodiments of the present invention and these drawings without paying any creative work.

[0045] Figure 1 This is a flow chart of a method for dynamically adjusting a driving strategy based on driving style recognition according to Embodiment 1 of the present invention;

[0046] Figure 2 This is a flow chart of the data-driven adaptive algorithm described in Example 1 of the present invention;

[0047] Figure 3 This is a flowchart of intelligent interaction and strategy optimization described in Example 1 of the present invention;

[0048] Figure 4 This is a block diagram of a driving strategy dynamic adjustment system based on driving style recognition described in Example 2 of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0050] Example 1

[0051] This embodiment is a method for dynamically adjusting a driving strategy based on driving style recognition, which is characterized by comprising the following steps:

[0052] S1. Establish a driving style classification model:

[0053] S2. Data-driven adaptive algorithm:

[0054] The driver's behavior data is monitored in real time, a data-driven adaptive algorithm is applied to complete the confirmation of the driver's driving style, and the driving style classification model is dynamically updated.

[0055] S3. Application scenario driven adaptive control strategy:

[0056] Based on the driver's driving style and considering the driver's specific behavior in different situations, the scenario-driven adaptive control strategy is applied to output the vehicle's current driving strategy.

[0057] S4. Intelligent interaction and strategy optimization:

[0058] The vehicle's current driving strategy is optimized through the driver's interactive feedback during the driving process.

[0059] In step S1, the driver's style is divided into three categories: conservative, moderate, and aggressive.

[0060] In step S2, the behavior patterns of different drivers are identified using the random forest method, and the driving style of the driver is confirmed through the driving style classification model; the behavior pattern recognition result is dynamically adjusted using the incremental learning method and the driver's real-time behavior data, and the driving style classification model is updated.

[0061] The step S2 comprises the following steps:

[0062] S21. Data collection: Vehicle driving behavior data and driver status data are collected in real time through vehicle-mounted sensors, cameras, etc.

[0063] The vehicle driving behavior data includes vehicle speed, acceleration, braking force, steering angle, etc.; the driver status data includes driver expression, line of sight direction, heart rate, operation frequency, etc.

[0064] S22. Data preprocessing: Clean, denoise and normalize the collected data to ensure data quality and consistency.

[0065] S23. Feature extraction: Extract features from preprocessed data, such as average vehicle speed, acceleration change rate, braking frequency, heart rate fluctuations, etc. Real-time data such as the driver's emotional fluctuations and fatigue level may affect driving behavior and style.

[0066] S24. Training the random forest model: input the feature data extracted in step S23 into the random forest model for training.

[0067] The Random Forest method is an ensemble learning method that improves prediction performance and stability by combining multiple decision tree models. The core idea is to build multiple decision trees and take the majority voting results for classification or regression to reduce the overfitting risk of a single decision tree. The training process of the random forest model includes the following key steps:

[0068] 1) Data sampling: Multiple subsets are randomly selected from the original training data set (using a sampling method with replacement, i.e. Bootstrap sampling), and each subset is used to train a decision tree.

[0069] 2) Feature selection: When constructing each decision tree, for each node, a portion of features is randomly selected for splitting to increase the diversity of the model.

[0070] 3) Decision tree training: A decision tree is trained using each subset. The decision tree can be adjusted according to the classification or regression task. Usually, the constructed tree is not pruned to fully learn the data characteristics.

[0071] 4) Integrated prediction: For classification tasks, random forest determines the final classification result by majority voting on the prediction results of all decision trees; for regression tasks, the average of the prediction values ​​of all decision trees is taken.

[0072] The random forest model has the following advantages: First, it has a strong ability to resist overfitting. Due to the combination of multiple decision trees, the risk of overfitting of a single model is reduced; second, it has efficient computing performance, and can build multiple decision trees in parallel to improve the model training speed; third, it has good robustness to missing data and outliers. At the same time, the random forest model can also provide feature importance metrics to help identify the features that have the greatest impact on the prediction results. In summary, random forests perform well in processing high-dimensional data and complex models, and are widely used in classification, regression, feature selection and other fields.

[0073] S25. Behavioral pattern recognition: The trained random forest model can recognize the behavioral patterns of different drivers, such as conservative, aggressive, moderate, etc.

[0074] S26. Incremental learning model update: The feature data newly extracted in step S23 is input into the incremental learning model for updating.

[0075] Incremental learning is a machine learning method that can continuously learn and update models. It is suitable for scenarios where data streams are constantly updated and dynamically changing. The core idea is to gradually introduce new data to update the model without having to retrain the entire model from scratch, thereby improving learning efficiency and adaptability. The incremental learning method mainly includes the following key steps:

[0076] 1) Initial model training: Use the initial dataset to train a baseline model, which can be any traditional machine learning model, such as decision tree, linear regression, neural network, etc.

[0077] 2) New data collection: As time goes by, the system continuously receives new data, which may contain new features, categories or trends.

[0078] 3) Model update: Input new data into the model and use incremental learning algorithms to update the model. Common incremental learning algorithms include incremental SVM, online gradient descent, incremental decision tree, etc. These algorithms can adjust based on new data without discarding the old model.

[0079] 4) Validation and evaluation: Regularly validate and evaluate the updated model to ensure that it performs well on new data. You can use methods such as cross-validation and sliding window evaluation to monitor the performance of the model.

[0080] 5.) Continuous Learning: The model continuously receives new data and is updated to ensure that it always adapts to changes and trends in the data.

[0081] The main advantages of incremental learning include: first, it can efficiently process large-scale data without retraining the entire model, saving computing resources and time; second, the model can adapt to the dynamically changing data environment and maintain the real-time and accuracy of its prediction performance; third, it is suitable for application scenarios such as online learning, real-time prediction and adaptive systems. Through incremental learning, the system can achieve continuous optimization and improvement, improving the overall level of intelligence.

[0082] S27. Real-time behavior pattern adjustment: According to the update result of the incremental learning model, the behavior pattern recognition result is adjusted in real time, and the driving style classification model is updated. For example, if the driver exhibits more aggressive driving behavior during a period of time, the system can adjust the driving strategy accordingly.

[0083] The step S3 comprises:

[0084] The scenario-driven adaptive control strategy proposes an adaptive strategy that combines specific scenarios with driving styles. Different driving strategy adjustment schemes are preset in different driving scenarios for different driving styles (conservative, moderate, and aggressive), including speed control, lane selection, overtaking strategies, etc. The specific strategies are shown in Table 1:

[0085] Table 1 Scenario-driven adaptive control strategy

[0086]

[0087] The step S4 comprises:

[0088] After the strategy is output, an intelligent interaction and closed-loop mechanism needs to be established. Intelligent interaction obtains feedback through the human-computer interaction interface. During the driving process, the driver processes the feedback information and performs NLP voice analysis to extract key information and optimize the current driving strategy. The process is as follows: Figure 3 As shown, including:

[0089] Design feedback interface: Design a human-computer interaction interface on the vehicle display screen for users to provide information feedback to the vehicle computer; design a friendly and intuitive feedback interface on the vehicle display screen so that the driver can easily provide feedback or collect the driver's voice feedback through the voice assistant.

[0090] Obtain user feedback: used to provide feedback on the current driving strategy through the feedback interface; provide feedback on the current driving strategy through the touch screen or voice assistant (such as "speeding too fast", "brakes too hard", etc.).

[0091] Feedback analysis and processing: Perform data processing and NLP speech analysis on feedback information to extract key information.

[0092] Optimize the current driving strategy: Optimize the scenario-driven adaptive control strategy based on the analyzed and processed key information.

[0093] This embodiment analyzes the driver's behavior pattern in depth, and the autonomous driving system can dynamically adjust the driving strategy according to different driving styles (such as conservative, moderate, and aggressive). This personalized adjustment not only improves the driver's comfort and satisfaction, but also significantly improves driving safety and reduces the accident rate. At the same time, the system adaptability and consistency are enhanced, and efficient performance can be maintained regardless of the driving environment. In addition, energy efficiency is optimized, fuel or electricity consumption is reduced, and green driving is achieved. Accurate feedback and optimization suggestions also help drivers improve their driving behavior and improve their overall driving level, thereby enhancing their trust and reliance on the autonomous driving system.

[0094] Example 2

[0095] like Figure 4As shown, this embodiment is a driving strategy dynamic adjustment system based on driving style recognition, which is used to implement the method described in Example 1, including:

[0096] Basic layer: used to establish a driving style classification model. The driving styles are classified into conservative, moderate, aggressive, etc.

[0097] Dynamic layer: used to monitor the driver's behavior data in real time, apply data-driven adaptive algorithms to dynamically adjust the driving style model, complete the driver's driving style confirmation and model update. For example, real-time data such as the driver's mood swings and fatigue level may affect driving behavior and style.

[0098] Fusion layer: Based on the dynamic layer, a scenario-driven adaptive control strategy is applied to consider the driver’s specific behavior in different scenarios. For example, the driver’s driving style and behavior may be different on highways and city roads.

[0099] Interaction layer: Optimize adaptive control strategies through real-time user feedback.

[0100] The dynamic layer includes:

[0101] The data-driven module is used to collect vehicle driving behavior data and driver status data, and perform feature extraction on the collected data;

[0102] Adaptive algorithm module, used to train random forest model based on extracted feature data, identify driver behavior patterns, and confirm driving style type;

[0103] Real-time adjustment of style models, used to adjust behavior patterns in real time through incremental learning models, and to update driving style classification models.

[0104] The interaction layer is a human-computer interaction interface.

[0105] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for dynamically adjusting driving strategy based on driving style recognition, characterized in that: The following steps are involved: S1. Establish a driving style classification model; S2. Data-driven adaptive algorithm: real-time monitoring of driver behavior data, application of data-driven adaptive algorithm to complete the driver's driving style confirmation, and dynamic update of the driving style classification model; S3. Application scenario-driven adaptive control strategy: Based on the driver's driving style, the application scenario-driven adaptive control strategy outputs the vehicle's current driving strategy by considering the driver's specific behavior in different situations; S4. Intelligent interaction and strategy optimization: Optimize the vehicle’s current driving strategy through the driver’s interactive feedback during the driving process.

2. The method for dynamic adjustment of driving strategy based on driving style recognition according to claim 1, characterized in that: The driver styles are divided into three categories: conservative, moderate, and aggressive.

3. The method for dynamic adjustment of driving strategy based on driving style recognition according to claim 1, characterized in that: The step S2 comprises the following steps: S21. Data collection: real-time collection of vehicle driving behavior data and driver status data; S22. Data preprocessing: cleaning, denoising and normalizing the collected data; S23. Feature extraction: extract features from preprocessed data; S24. Training the random forest model: inputting the feature data extracted in step S23 into the random forest model for training; S25. Behavior pattern recognition: The trained random forest model recognizes the behavior patterns of different drivers; S26. Incremental learning model update: The feature data newly extracted in step S23 is input into the incremental learning model for updating; S27. Real-time behavior pattern adjustment: According to the update result of the incremental learning model, the behavior pattern recognition result is adjusted in real time, and the driving style classification model is updated.

4. The method for dynamically adjusting driving strategy based on driving style recognition as claimed in claim 3, characterized in that: In step S21, the vehicle driving behavior data includes vehicle speed, acceleration, braking force, and steering angle.

5. The method for dynamic adjustment of driving strategy based on driving style recognition as claimed in claim 3, characterized in that: In step S21, the driver status data includes driver's expression, sight direction, heart rate, and operation frequency.

6. The method for dynamic adjustment of driving strategy based on driving style recognition as claimed in claim 3, characterized in that: In step S24, the training process of the random forest model includes the following steps: 1) Data sampling: Multiple subsets are randomly extracted from the original training data set, and each subset is used to train a decision tree; 2) Feature selection: When constructing each decision tree, for each node, a portion of features are randomly selected for splitting to increase the diversity of the model; 3) Decision tree training: Use each subset to train a decision tree; 4) Integrated prediction: For classification tasks, random forest determines the final classification result by majority voting on the prediction results of all decision trees; for regression tasks, the average of the prediction values ​​of all decision trees is taken.

7. The method for dynamic adjustment of driving strategy based on driving style recognition according to claim 1, characterized in that: In step S3, the adaptive strategy of the specific scene and driving style is combined, and different driving strategy adjustment schemes are preset in different driving scenes for different driving styles.

8. The method for dynamically adjusting driving strategy based on driving style recognition according to claim 1, characterized in that: The step S4 comprises: Design feedback interface; Obtain user feedback: used to provide feedback on the current driving strategy through the feedback interface; Feedback analysis and processing: perform data processing and NLP speech analysis on feedback information to extract key information; Optimize the current driving strategy: Optimize the scenario-driven adaptive control strategy based on the analyzed and processed key information.

9. The method for dynamically adjusting driving strategy based on driving style recognition as claimed in claim 7, characterized in that: In the step S4, the feedback interface adopts a human-computer interaction interface designed on a vehicle-mounted display screen.

10. A driving strategy dynamic adjustment system based on driving style recognition, used to implement the method described in any one of claims 1 to 8, characterized in that: include: Basic layer: used to build a driving style classification model; Dynamic layer: used to monitor the driver's behavior data in real time, apply data-driven adaptive algorithms to dynamically adjust the driving style model, complete the driver's driving style confirmation and model update; Fusion layer: It is used to apply scenario-driven adaptive control strategies based on the dynamic layer, consider the specific behaviors of drivers in different situations, and provide different driving strategies; Interaction layer: Optimize adaptive control strategies through real-time user feedback.

Images