An Adaptive Scenario-based Acceleration and Deceleration Control and Switching Method

Abstract

The present invention is applicable to the field of automotive technology and provides a method for adaptive scenario acceleration and deceleration control and switching, including: obtaining vehicle driving environment information and vehicle own state information; establishing an offline scenario recognition model to identify the current driving scenario of the vehicle and output a driving mode; determining a scenario recognition time window and collecting driving scenario data to identify the driving scenario; establishing a driver intention recognition module. This method obtains vehicle driving environment information and vehicle state information through an intelligent perception module, and then uses the perceived vehicle and environment information as the input feature vector for scenario recognition based on a support vector machine to achieve accurate recognition of the current driving scenario of the vehicle. At the same time, a precise and conventional acceleration and deceleration intention recognition module is designed according to scenario requirements. For the precise acceleration and deceleration intention recognition module, the driver can "fine-tune" the longitudinal acceleration and deceleration of the vehicle to meet the precise manipulation requirements of the driver in some scenarios.

Description

Technical Field

[0001] The present invention belongs to the technical field of automobiles, and particularly relates to a method for accelerating, decelerating, controlling, and switching in an adaptive scenario. Background Art

[0002] The automotive manufacturing industry is accelerating its development towards intelligence, and the automotive industry is undergoing a huge transformation. This industrial transformation triggered by innovative technologies is reshaping the value system of the century-old automotive industry. With the continuous evolution of automotive chassis technology towards intelligence, intelligent chassis systems are endowed with higher-level functional requirements. Specifically, the core point is that intelligent chassis must have the ability of intelligent perception, that is, it can accurately identify and distinguish diverse driving scenarios, just as a human driver judges road conditions, traffic environment, etc. based on their own experience and perception. At the same time, users' demands for driving experience are also gradually increasing. Drivers are no longer satisfied with a single and unchanging driving mode, but expect the vehicle to adaptively adjust its driving mode according to the changes in its scenario, so that drivers can more comfortably operate the vehicle in different driving scenarios.

[0003] Currently, there are still limitations in the adaptive switching of acceleration and deceleration modes based on different driving scenarios: The current adaptive chassis mainly focuses on the design of corresponding adaptive mode switching in the suspension system, and there is less research on the adaptive design and switching method of the acceleration and deceleration driving mode. The current vehicle acceleration and deceleration driving mode cannot be adjusted and optimized according to specific driving needs to assist drivers in achieving more accurate and efficient driving operations in some driving scenarios that require precise operation (such as parking scenarios, urban congestion scenarios, low-speed meeting scenarios, low-curve radius scenarios, bad weather, etc.), and improving the overall driving experience and vehicle performance. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for accelerating, decelerating, controlling, and switching in an adaptive scenario, aiming to solve the technical problems existing in the prior art determined in the background art.

[0005] The present invention is implemented as follows. A method for accelerating, decelerating, controlling, and switching in an adaptive scenario, the method includes:

[0006] S1. Obtain vehicle driving environment information and vehicle own state information through in-vehicle sensors;

[0007] S2. Use the data set collected based on a driving simulator to establish an offline scenario recognition model through a support vector machine algorithm. This model can identify the current driving scenario of the vehicle and output a matching driving mode according to the driving scenario;

[0008] S3. Determine the scenario recognition time window of the offline scenario recognition model , and collect the driving scenario data described in S1 within the time window for driving scenario recognition;

[0009] S4. According to the recognized acceleration and deceleration modes, establish a driver intention recognition module and perform driving mode switching through the Sigmoid function.

[0010] As a further aspect of the present invention, the S2 specifically includes:

[0011] Step 1, driving scenario annotation. Classify and annotate the conventional driving scenarios and precise driving scenarios based on the data set collected by the driving simulator. The precise driving scenarios include parking scenarios, urban congestion scenarios, low-speed oncoming vehicle scenarios, low curve radius scenarios, and bad weather scenarios. Further, other scenarios in the data set are annotated as conventional driving scenarios.

[0012] Step 2, driving scenario data preprocessing. Use the data in the precise driving scenarios and conventional driving scenarios in Step 1 as input features, and then normalize these features. Use min-max normalization to unify the dimension of each feature to the interval [0, 1].

[0013] ;

[0014] Among them, represents the maximum value in a single feature, represents the minimum value in a single feature, represents the current feature value, represents the standardized feature value after normalizing the current feature value.

[0015] Step 3, establish a driving scenario recognition model based on the support vector machine through the driving scenario data obtained in Step 1 and Step 2. Since the RBF kernel performs optimally in non-linearly separable scenarios, the RBF kernel is used as the kernel function of this support vector machine model. At the same time, 70% of the data in the data set is used as the training set, and the remaining 30% of the data is used as the test set to verify the driving scenario recognition effect. The driving scenario recognition model is trained through the training set, and the recognition accuracy is verified through the test set.

[0016] As a further aspect of the present invention, the S3 specifically includes:

[0017] Set the scenario recognition time window based on the recognition requirements of the offline scenario recognition model ;

[0018] Within the recognized time window , continuously collect the vehicle driving environment data and the vehicle's own state data through in-vehicle sensors;

[0019] The vehicle driving environment data and the vehicle's own state data include: the vehicle speeds of surrounding vehicles, the distances from the host vehicle, the distances between the vehicle and obstacles, lane line information, weather conditions, road surface conditions, vehicle load, vehicle speed, acceleration and deceleration data;

[0020] Perform real-time analysis on the collected vehicle driving environment data and the vehicle's own state data, and use the driving scenario recognition model based on support vector machine obtained through S2 to perform driving scenario recognition processing on the vehicle driving environment information and the vehicle's own state information obtained within the recognition time window and output the driving mode corresponding to the current driving environment according to the recognition result, including the normal driving mode and the precise driving mode; if it is recognized that the current driving scenario matches the driving conditions of the precise driving mode, feedback the recognition result and switch the driving mode to the precise driving mode.

[0021] As a further solution of the present invention, the S4 specifically includes:

[0022] Combined with the driving mode corresponding to the current environment, establish a precise acceleration and deceleration intention recognition module and a normal acceleration and deceleration intention recognition module respectively, and the precise acceleration and deceleration intention recognition module and the normal acceleration and deceleration intention recognition module respectively match different driver intentions;

[0023] Both the normal acceleration and deceleration intention recognition module and the precise acceleration and deceleration intention recognition module are the mapping relationships between the opening of the acceleration / deceleration pedal and the desired motion intensity of the vehicle.

[0024] As a further solution of the present invention, the specific expression of the normal acceleration and deceleration intention recognition module is:

[0025] ;

[0026] wherein, is the opening of the normal acceleration / braking pedal, is the acceleration output by the normal acceleration and deceleration intention recognition, is the acceleration gain coefficient;

[0027] The specific expression of the precise acceleration and deceleration intention recognition module is:

[0028] ;

[0029] wherein, is the opening of the precise acceleration / braking pedal, is the acceleration output by the normal acceleration and deceleration intention recognition, and under the same change in the pedal opening, the change in acceleration and deceleration output by the precise acceleration and deceleration intention recognition module is less than the change in acceleration and deceleration output by the normal acceleration and deceleration intention recognition module, is the acceleration gain coefficient, and G is the driving mode fineness coefficient, which represents the scaling ratio compared with the conventional mode.

[0030] As a further solution of the present invention, controlling the mixing ratio between the conventional driving mode and the precise driving mode generates a driving mode switching weight coefficient through the Sigmoid function to achieve, and , controlling the mixing ratio between the conventional driving mode and the precise driving mode:

[0031] ;

[0032] Wherein, is the driving mode switching weight coefficient, is the transition center point, is the transition duration, is the slope parameter, used to control the transition speed between the conventional driving mode and the precise driving mode;

[0033] During the mode transition process, the mixed motion intensity output is:

[0034] ;

[0035] Wherein, is the mixed motion intensity, is the current acceleration / brake pedal opening, and , is the driving mode transition timestamp.

[0036] As a further solution of the present invention, in the conventional driving mode, the driver operation amount is obtained as the input of the conventional acceleration / deceleration intention recognition module, and the output is the expected longitudinal acceleration / deceleration in the conventional driving mode;

[0037] In the precise driving mode, the driver operation amount is obtained as the input of the precise acceleration / deceleration intention recognition module, and the output is the expected longitudinal acceleration / deceleration in the precise driving mode.

[0038] During the longitudinal acceleration / deceleration mode switching process, the driver operation amount is obtained as the input of the precise acceleration / deceleration intention recognition module, and the output is the mixed expected longitudinal acceleration / deceleration during the driving mode switching process.

[0039] Through the above process, the mapping relationship between the driver operation amount and the expected longitudinal acceleration / deceleration in the two driving modes and during the driving mode switching process can be obtained, and the overall expression is as follows:

[0040] ;

[0041] Wherein, is the expected longitudinal acceleration / deceleration, is the opening of the accelerator / brake pedal.

[0042] The beneficial effects of the present invention are as follows:

[0043] This method establishes an offline scenario recognition model based on the support vector machine algorithm using driving simulator data. This model can identify the current driving scenario of the vehicle. Further, the vehicle driving environment information and vehicle state information are obtained through the intelligent perception module, and then the perceived vehicle and environment information is used as the input feature vector of the offline scenario recognition model based on the support vector machine to achieve accurate recognition of the current driving scenario of the vehicle. At the same time, a precise and conventional acceleration / deceleration intention recognition module is designed according to the scenario requirements. For the precise acceleration / deceleration intention recognition module, the driver can "fine-tune" the longitudinal acceleration and deceleration of the vehicle to meet the precise manipulation requirements of the driver in some scenarios. Further, when the driving mode switch is recognized, this method controls the weight mixing through the Sigmoid function to achieve a smooth and seamless driving mode switch. Description of the Drawings

[0044] Figure 1 is the functional architecture diagram of a method for designing and switching acceleration / deceleration modes for an adaptive scenario provided by an embodiment of the present invention;

[0045] Figure 2 is the flow chart of a method for designing and switching acceleration / deceleration modes for an adaptive scenario provided by an embodiment of the present invention;

[0046] Figure 3 is the mapping relationship between the opening of the accelerator / deceleration pedal and the desired motion intensity of the vehicle in a method for designing and switching acceleration / deceleration modes for an adaptive scenario provided by an embodiment of the present invention;

[0047] Figure 4 is the Sigmoid function as the transition function for switching between the precise driving mode and the standard driving mode. Detailed Embodiments

[0048] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] As Figure 1 and Figure 2 shown, a method for adaptive scenario-based acceleration / deceleration control and switching is provided, and the method includes:

[0050] S1. Comprehensively obtain vehicle driving environment information and vehicle own state information through in-vehicle sensors. Among them, the driving environment information covers multi-dimensional parameters such as the driving speed of surrounding vehicles, the dynamic distance parameters between surrounding vehicles and the vehicle itself, the distance information between the vehicle and various obstacles, as well as road markings, weather conditions, and road surface conditions; the vehicle state information includes parameters such as the dynamic load information carried by the vehicle, the real-time driving speed of the vehicle, and the acceleration and deceleration selection mode information, etc.

[0051] ;

[0052] S2. Use the dataset collected based on the driving simulator to establish an offline scene recognition model through the support vector machine algorithm. This model can identify the current driving scene of the vehicle and output a matching driving mode according to the driving scene;

[0053] Specifically, the vehicle driving environment information (such as parameters of the vehicle speed of surrounding vehicles, the distance from the vehicle itself, the distance between the vehicle and obstacles, lane line information, weather conditions, and road surface conditions, etc.) and the vehicle own state information (such as vehicle load, vehicle speed, acceleration and deceleration mode, etc.) in the driving simulator dataset are used as the input feature vectors of the support vector machine (SVM) algorithm. After the classification and recognition processing of the SVM algorithm, the corresponding driving scene is output, and the corresponding driving mode is also output, including the normal driving mode and the precise driving mode.

[0054] In the example of the present invention, the training and test datasets are extracted from the prior driving simulator driving dataset, and the input feature vectors are the vehicle speed of surrounding vehicles, the distance from the vehicle itself, the distance between the vehicle and obstacles, lane line information, weather conditions, road surface conditions, vehicle load, vehicle speed, acceleration and deceleration; the driving scene recognition model based on SVM can be realized through the following steps:

[0055] Step 1, driving scene annotation, classify and annotate the normal driving scene and the precise driving scene based on the dataset collected by the driving simulator. Among them, the precise driving scenes include parking scenes, urban congestion scenes, low-speed vehicle meeting scenes, low curve radius scenes, and bad weather scenes. Further, other scenes in the dataset are marked as normal driving scenes.

[0056] Step 2, driving scene data preprocessing, use the data in the precise driving scene and the normal driving scene in Step 1 as input features, and then normalize these features. Use min-max normalization to unify the dimension of each feature to the interval [0,1].

[0057] ;

[0058] Among them, represents the maximum value in a single feature, Represents the minimum value in a single feature, Represents the current feature value, Represents the standard feature value after normalizing the current feature value.

[0059] Step 3: Establish a driving scenario recognition model based on support vector machine using the driving scenario data obtained in Step 1 and Step 2. Since the RBF kernel performs optimally in non-linearly separable scenarios, the RBF kernel is used as the kernel function of this support vector machine model. At the same time, 70% of the data in the dataset is used as the training set, and the remaining 30% of the data is used as the test set to verify the driving scenario recognition effect. The driving scenario recognition model is trained using the training set, and the recognition accuracy is verified using the test set.

[0060] Through the above machine learning training, an offline SVM driving scenario recognition model can be obtained, that is, an offline scenario recognition model. At the same time, it is defined that the output of the offline scenario recognition model being -1 represents the conventional acceleration and deceleration mode, and +1 represents the precise acceleration and deceleration mode.

[0061] Among them, the conventional driving mode corresponds to the conventional acceleration and deceleration mode, and the precise driving mode corresponds to the precise acceleration and deceleration mode.

[0062] S3. Determine the scenario recognition time window of the offline scenario recognition model and collect driving scenario data within the time window for driving scenario recognition;

[0063] Based on the recognition requirements of the offline scenario recognition model, set the scenario recognition time window ;

[0064] Within a time window, collect the vehicle speed of surrounding vehicles, the distance from the vehicle, the distance between the vehicle and obstacles, lane line information, weather conditions, road surface conditions, the vehicle load, vehicle speed, acceleration and deceleration data through in-vehicle sensors, and use the collected data as the data input for the current driving scenario recognition.

[0065] Since scenario recognition requires real-time performance, the scenario recognition time window should not be too large. After recognizing that the driving scenario conforms to the precise driving mode, it should be immediately fed back to the driver and transition to the precise driving mode. As a preferred embodiment of the present invention, the scenario recognition time window is initially set to 10s.

[0066] Perform real-time analysis on the collected vehicle driving environment data and vehicle own state data, and use the driving scenario recognition model based on support vector machine obtained in S2 for the recognition time window The vehicle driving environment information and the vehicle's own state information obtained are used for driving scenario recognition processing, and according to the recognition result, the driving mode corresponding to the current driving environment is output, including the normal driving mode and the precise driving mode; if it is recognized that the current driving scenario matches the driving conditions of the precise driving mode, the recognition result is fed back and the driving mode is switched to the precise driving mode.

[0067] S4. According to the recognized acceleration and deceleration mode, establish a driver intention recognition module and perform driving mode switching through the Sigmoid function.

[0068] Combined with the driving mode corresponding to the current environment, a precise acceleration and deceleration intention recognition module and a normal acceleration and deceleration intention recognition module are established respectively, and the precise acceleration and deceleration intention recognition module and the normal acceleration and deceleration intention recognition module match different driver intentions respectively;

[0069] Based on the acceleration and deceleration mode recognized in S2, different driver intention recognition modules are matched;

[0070] As Figure 3 shown, both the normal acceleration and deceleration intention recognition module and the precise acceleration and deceleration intention recognition module are the mapping relationship between the opening of the acceleration / deceleration pedal and the desired motion intensity of the vehicle.

[0071] The specific expression of the normal acceleration and deceleration intention recognition module is:

[0072] ;

[0073] where is the opening of the normal acceleration / braking pedal, is the acceleration output by the normal acceleration and deceleration intention recognition, is the acceleration gain coefficient;

[0074] The precise acceleration and deceleration intention recognition module is also the mapping relationship between the opening of the acceleration / deceleration pedal and the desired motion intensity of the vehicle. The precise acceleration and deceleration intention recognition module requires that the mapping relationship between the opening of the acceleration / deceleration pedal and the desired motion intensity of the vehicle needs to be more refined, so as to meet the needs of the driver to precisely control the vehicle. That is, when the driver changes the same opening of the acceleration / deceleration pedal, the acceleration and deceleration change amount output by the precise acceleration and deceleration intention recognition module is smaller than that output by the normal acceleration and deceleration intention recognition module. Therefore, when the driver has a larger operation amount, the acceleration and deceleration change amount of the vehicle is smaller. That is, in the precise acceleration and deceleration mode, the driver can "fine-tune" the longitudinal acceleration and deceleration of the vehicle. Its specific expression is as follows:

[0075] ;

[0076] where is the opening degree of the precise acceleration / brake pedal, is the acceleration output by the conventional acceleration / deceleration intention recognition, and under the same change in pedal opening, the change in acceleration / deceleration output by the precise acceleration / deceleration intention recognition module is less than that of the conventional acceleration / deceleration intention recognition module, is the acceleration gain coefficient, and G is the driving mode refinement coefficient, which represents the scaling ratio compared to the conventional mode.

[0077] The Sigmoid function is used as the transition function for switching between the precise driving mode and the standard driving mode. After the recognition output based on S2 indicates the need to switch the acceleration / deceleration mode, directly switching the driving mode will cause a sudden change in the longitudinal acceleration output by the driver intention recognition module, affecting the driving experience of the driver. Therefore, the Sigmoid function is used to switch the acceleration / deceleration mode, so that the longitudinal acceleration output by the driver intention recognition module changes continuously. By controlling the weight mixing through the Sigmoid function, a smooth and seamless driving mode switch can be achieved, while retaining the flexible control of the transition speed and emergency interruption. The driving mode switch weight coefficient is generated through the Sigmoid function , and , controls the mixing ratio between the conventional driving mode and the precise driving mode:

[0078] ;

[0079] Among them, is the driving mode switch weight coefficient, is the transition center point, is the transition duration, is the slope parameter, which is used to control the transition speed between the conventional driving mode and the precise driving mode and can be optimized according to the driver's needs or on-vehicle tests;

[0080] As a preferred embodiment of the present invention, the transition center point is set to 1 s, the transition duration is set to 2 s, and the slope parameter is initially set to 5,

[0081] ;

[0082] The image of using the Sigmoid function as the transition function for switching between the precise driving mode and the standard driving mode is as shown in Figure 4 .

[0083] During the process of mode transition, the mixed expected longitudinal acceleration / deceleration output is:

[0084] ;

[0085] Among them, To mix the expected longitudinal acceleration and deceleration speed, is the current accelerator / brake pedal opening, and , is the driving mode transition timestamp.

[0086] In the normal driving mode, the driver's operation amount is obtained as the input of the normal acceleration and deceleration intention recognition module, and the output is the expected longitudinal acceleration and deceleration speed in the normal driving mode;

[0087] In the precise driving mode, the driver's operation amount is obtained as the input of the precise acceleration and deceleration intention recognition module, and the output is the expected longitudinal acceleration and deceleration speed in the precise driving mode.

[0088] During the longitudinal acceleration and deceleration mode switching process, the longitudinal acceleration and deceleration speed uses the driver's operation amount as the input for both the normal acceleration and deceleration intention recognition and the precise acceleration and deceleration intention recognition at the same time, and the output is the mixed expected longitudinal acceleration and deceleration speed based on the Sigmoid as the transition function.

[0089] Through the above process, the mapping relationship between the driver's operation amount and the expected longitudinal acceleration and deceleration speed in the two driving modes and during the driving mode switching process can be obtained. The overall expression is as follows:

[0090] ;

[0091] Among them, is the expected longitudinal acceleration and deceleration speed, is the accelerator / brake pedal opening.

[0092] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0093] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0094] The above embodiments merely represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

[0095] The above is only the preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An acceleration and deceleration control and switching method for an adaptive scenario, characterized in that The method includes: S1. Obtain vehicle driving environment information and vehicle own state information through on-vehicle sensors; S2. Use the dataset collected based on a driving simulator to establish an offline scenario recognition model through a support vector machine algorithm. The offline scenario recognition model is used to identify the current driving scenario of the vehicle and output a matching driving mode according to the driving scenario, where the normal driving mode corresponds to the normal acceleration and deceleration mode, and the precise driving mode corresponds to the precise acceleration and deceleration mode; The S2 includes: driving scenario annotation, classifying and annotating the normal driving scenario and the precise driving scenario based on the dataset collected by the driving simulator. The precise driving scenarios include parking scenarios, urban congestion scenarios, low-speed vehicle meeting scenarios, scenarios with a low curve radius of the road, and bad weather scenarios, and other scenarios in the dataset are annotated as normal driving scenarios; S3. Determine the scene recognition time window of the offline scene recognition model , and collect the driving scene data described in S1 within the time window, perform driving scene recognition, and output the driving mode corresponding to the current driving environment according to the recognition result, including the normal driving mode and the precise driving mode; S4. Establish a driver intention recognition module according to the recognized acceleration and deceleration mode, and perform driving mode switching through the Sigmoid function; The S4 specifically includes: Combined with the driving mode corresponding to the current environment, establish a precise acceleration and deceleration intention recognition module and a normal acceleration and deceleration intention recognition module respectively. The precise acceleration and deceleration intention recognition module and the normal acceleration and deceleration intention recognition module match different driver intentions respectively; Both the normal acceleration and deceleration intention recognition module and the precise acceleration and deceleration intention recognition module are the mapping relationships between the opening of the acceleration / deceleration pedal and the expected motion intensity of the vehicle; The specific expression of the normal acceleration and deceleration intention recognition module is: ； Among them, is the opening of the conventional acceleration / brake pedal, is the acceleration output by the conventional acceleration / deceleration intention recognition, is the acceleration gain coefficient; The specific expression of the precise acceleration and deceleration intention recognition module is: ； Among them, is the opening degree of the precise acceleration / brake pedal, is the acceleration output by the conventional acceleration / deceleration intention recognition, and under the same change in the pedal opening degree, the change in acceleration and deceleration output by the precise acceleration / deceleration intention recognition module is less than that output by the conventional acceleration / deceleration intention recognition module, is the acceleration gain coefficient, and G is the driving mode refinement coefficient, which represents the scaling ratio with the conventional mode; Controlling the mixing ratio between the conventional driving mode and the precise driving mode generates a driving mode switching weight coefficient through the Sigmoid function is achieved by, and , controlling the mixing ratio between the conventional driving mode and the precise driving mode: ； Among them, is the weight coefficient for driving mode switching, is the transition center point, is the transition duration, is the slope parameter, which is used to control the transition speed between the normal driving mode and the precise driving mode; During the process of mode transition, the mixed expected longitudinal acceleration and deceleration output is: ； Among them, is the mixed expected longitudinal acceleration and deceleration speed, is the current accelerator / brake pedal opening, and , is the driving mode transition timestamp.

2. The method according to claim 1, characterized in that, The S2 also includes: Driving scenario data preprocessing, taking the data in the precise driving scenario and the normal driving scenario as input features and performing normalization processing. Use min-max normalization to unify the dimension of each feature to the interval [0, 1]; ； Among them, represents the maximum value in a single feature, represents the minimum value in a single feature, represents the current feature value, represents the standard feature value after normalizing the current feature value; Establish a driving scenario recognition model based on the support vector machine through the obtained driving scenario data. At the same time, take 70% of the data in the dataset as the training set, and the remaining 30% of the data as the test set. Use the training set to train the driving scenario recognition model and use the test set to verify the recognition accuracy.

3. The method according to claim 2, wherein The S3 specifically includes: Set the scene recognition time window based on the recognition requirements of the offline scene recognition model ; Within the identified time window During this period, vehicle-mounted sensors continuously collect data on the vehicle's driving environment and its own status; The vehicle driving environment data and vehicle own state data include: the vehicle speed of surrounding vehicles, the distance from the vehicle itself, the distance between the vehicle and obstacles, lane line information, weather conditions, road surface conditions, vehicle load, vehicle speed, acceleration and deceleration data; Perform real-time analysis on the collected vehicle driving environment data and vehicle own state data, and use the driving scenario recognition model based on support vector machine obtained through S2 to perform driving scenario recognition processing on the vehicle driving environment information and vehicle own state information obtained in the recognition time window and output the driving mode corresponding to the current driving environment according to the recognition result, including the normal driving mode and the precise driving mode; if it is recognized that the current driving scenario matches the driving conditions of the precise driving mode, feedback the recognition result and switch the driving mode to the precise driving mode.

4. The method according to claim 1, wherein In the normal driving mode, obtain the driver operation amount as the input of the normal acceleration and deceleration intention recognition module, and the output is the expected longitudinal acceleration and deceleration in the normal driving mode; In the precise driving mode, obtain the driver operation amount as the input of the precise acceleration and deceleration intention recognition module, and the output is the expected longitudinal acceleration and deceleration in the precise driving mode; During the process of longitudinal acceleration and deceleration mode switching, obtain the driver operation amount as the input of the precise acceleration and deceleration intention recognition module, and the output is the mixed expected longitudinal acceleration and deceleration during the driving mode switching.

Images