An autonomous driving scene recognition system and method based on cloud training model

Abstract

The solution belongs to the field of autonomous driving technology, and specifically relates to an autonomous driving scene recognition system based on a cloud training model: information acquisition module: real-time acquisition of obstacles on the current driving section, generating an obstacle index signal and transmitting it to the cloud; the information acquisition module is also used to obtain the current driving speed of the vehicle and the throttle pressing signal of the vehicle. It also relates to an autonomous driving scene recognition method based on a cloud training model, including the following steps: S1: Information acquisition: real-time acquisition of obstacles on the current driving section through on-board sensors such as cameras and radars, and generation of an obstacle index signal; S9: Dynamic response: within the preset threshold time, if the throttle pressing signal is maintained, the warning signal is canceled to allow the vehicle to continue to accelerate; otherwise, the vehicle speed is adjusted to the recommended parameters. It can solve the problem of command conflicts between vehicle computer planning and driver actual operation.

Description

Technical Field

[0001] The present invention belongs to the field of autonomous driving technology, and specifically relates to an autonomous driving scene recognition system and method based on a cloud training model. Background Art

[0002] An autonomous vehicle is an unmanned intelligent vehicle that is realized through an automatic control system. In recent years, the intelligent technology of automobiles has developed rapidly. Assisted driving technology and some autonomous driving technology have entered the industrialization stage; conditional autonomous driving and highly automated driving technology have entered the test and verification stage. Driving scene recognition technology is a key basic technology for intelligent driving assistance systems and unmanned vehicles to perceive the environment. Accurate recognition of driving scenes is conducive to the safe and stable driving of autonomous vehicles.

[0003] The invention patent with application number 201880008862.5 discloses a computer-implemented method for an autonomous driving vehicle, the method comprising: receiving a captured image that perceives the environment of the autonomous driving vehicle from an image capture device of the autonomous driving vehicle, wherein the captured image identifies obstacles in motion near the autonomous driving vehicle; generating a feasible area around the moving obstacle based on the projection of the moving obstacle; if the autonomous driving vehicle is within the feasible area, determining an upper speed limit of the autonomous driving vehicle; and generating a trajectory with a trajectory speed less than the upper speed limit to autonomously control the autonomous driving vehicle according to the trajectory, so that if the autonomous driving vehicle is within the feasible area, the autonomous driving vehicle will slow down; the method also includes: determining the position of the projection of the moving obstacle based on the current position of the moving obstacle and the autonomous driving vehicle, wherein the position of the projection is determined as a perpendicular intersection between the moving obstacle and the moving direction of the autonomous driving vehicle.

[0004] The above technology identifies obstacles in real scenes and plans real-time driving plans, that is, it allows the intelligent model to operate the vehicle in a fully managed manner. However, when the user's intention conflicts with the current driving plan, the driver basically takes over the current driving mode directly. The car computer cannot judge the driver's operation, that is, it cannot determine whether the driver's operation is a false touch, which will lead to a poor driving experience for the driver. This is because the driving plan planned by the car computer after identifying the driving scene is different from the driver's current driving plan, resulting in a contradiction. Now there is a need for a technology that can judge the driver's driving operation during driving to prevent false touches from affecting the driving experience. Summary of the invention

[0005] The purpose of this solution is to provide an autonomous driving scene recognition system and method based on a cloud training model, which can solve the problem of command conflicts between vehicle computer planning and driver actual operation.

[0006] An autonomous driving scene recognition system based on a cloud training model, comprising:

[0007] An information collection module is used to collect the historical passing information of vehicles on the road section, the historical passing information includes the vehicle's driving speed and vehicle suspension parameters, and the vibration signal of the vehicle passing the obstacle, and to sort the vehicles according to the vibration magnitude, to build a training model, and to train the training model with the historical passing information of vehicles on the road section as a training set; to set a vibration threshold, and to use the maximum vehicle speed and vehicle suspension parameters under the vibration threshold as the recommended parameters for passing the obstacle on the road section; and to collect statistics on the vehicle suspension parameters with the smallest vibration under each speed state;

[0008] Information acquisition module: Obtain the obstacle conditions of the current driving section in real time, generate an obstacle index signal and transmit it to the cloud; the information acquisition module is also used to obtain the current driving speed of the vehicle and the accelerator pressing signal of the vehicle; the obstacle conditions include potholes and speed bumps;

[0009] Navigation module: plans the driving route according to the driver's destination. When there is a pothole on the vehicle's driving route, the navigation module is also used to remind the driver to change lanes to avoid it. If it cannot be avoided, it will send a signal that the obstacle is inevitable.

[0010] Data processing module: after receiving the obstacle unavoidable signal, the historical passing information of vehicles on the road section is searched in the information collection module according to the obstacle index signal, and the recommended parameters are obtained and preset in the driving plan of the vehicle computer, which are used to adjust the suspension parameters of the vehicle according to the recommended parameters when passing through the obstacle; before passing the obstacle, if the current driving speed of the vehicle is greater than the speed in the recommended parameters, and the accelerator pressing signal shows that the vehicle is in an accelerating state, a warning reminder signal is issued; within the threshold time, if the accelerator pressing signal is in a maintained state, the reminder signal is canceled, and the vehicle maintains accelerated driving. If the accelerator pressing signal weakens, the vehicle speed is reduced to the recommended parameters; if the current driving speed is less than the speed in the recommended parameters, the vehicle suspension parameters with the smallest vibration at the current vehicle speed are matched, and the suspension parameters of the driving vehicle are adjusted accordingly.

[0011] Basic principles and beneficial effects of this solution: Basic principles: Use on-board sensors to monitor the vehicle's driving status and road obstacles in real time, including vehicle speed, accelerator pedal status, and obstacle information. Collect and analyze the vehicle's historical driving data on a specific road section, including speed, suspension parameters, and vibration signals when passing obstacles. Using historical data as a training set, the cloud-based training model learns the optimal suspension parameters and vehicle speed under different driving conditions to optimize the performance of passing obstacles. Based on the output of the cloud-based training model, the system recommends the optimal suspension parameters and vehicle speed, and adjusts the vehicle settings in real time during driving to adapt to obstacles.

[0012] When potential danger is detected (such as the vehicle approaching an obstacle at too fast a speed) or when the user's operation is different from the plan, the system provides a warning and intervenes to adjust the vehicle speed when necessary.

[0013] Through real-time monitoring and intelligent warnings, this solution can prevent collisions caused by drivers' untimely reactions or misjudgments.

[0014] Automatically adjusting suspension parameters can improve the comfort and handling of the vehicle, especially when facing complex road conditions. The cloud model provides intelligent decision support based on a large amount of historical data, helping drivers to pass obstacles in a safer way. This solution can continuously learn and adapt based on new driving data and user feedback to improve the accuracy of future scene recognition. When facing complex road conditions, this solution can automatically adjust vehicle settings to reduce the driver's operating burden, allowing them to focus more on road conditions and driving. By collecting and analyzing the vehicle's suspension parameters and vibration signals, this solution helps predict vehicle maintenance needs, thereby reducing potential failures and repair costs. This solution provides an effective scene recognition and response strategy for autonomous driving technology, and promotes the development and application of autonomous driving technology. This solution uses the throttle pressing signal, the fact acquisition of the current vehicle speed, and the recommended parameters (recommended vehicle speed and recommended suspension parameters) calculated from historical passing data to alert the driver, and with the driver's permission, reduce the vehicle speed to the recommended parameters and automatically adjust the suspension parameters. If the driver disagrees with adjusting the vehicle speed, the suspension parameters will be adjusted to the suspension parameters with the least vibration in the historical passing data at the current vehicle speed, which can allow the driver to pass the obstacle section safely with the greatest degree of comfort.

[0015] In special cases, such as when there is a pothole obstacle in front of the traffic light, the duration of the green light is not enough to support the vehicle to pass at a low speed; in the past, historical vehicles have passed at a low speed to maintain the stability of the vehicle body, but the current driver observes the road conditions and finds that there are sufficient conditions for safe passing, that is, there are no pedestrians crossing the road and no other vehicles in sight. If the current speed continues, it will not be able to pass before the green light, so it is necessary to accelerate. The accelerated driving state may exceed the threshold speed, and a warning signal will be issued at this time. If the driver still accelerates after receiving the signal, the acceleration signal will be executed to pass the intersection. This process can save the driver's time. This solution can accurately identify the driver's intentions and make secondary judgments based on the user's driving operations, reducing the problem of poor driving experience caused by misoperation and improving safety performance.

[0016] Furthermore, if the obstacle is located at a bend in the road, within the threshold distance before entering the bend, the route planning of the navigation module is used to check whether there is a change of lane to avoid the obstacle. If it cannot be avoided, the data processing module sends a first deceleration signal to remind the driver to slow down to the recommended speed. If the driver slows down, the vehicle suspension parameters are adjusted to the recommended parameters. If the driver does not slow down, after entering the bend, the data processing module sends a second deceleration signal to remind the driver to slow down to the recommended speed. If the driver slows down, the vehicle suspension parameters are adjusted to the recommended parameters. If the driver does not slow down, the suspension parameters are adjusted to the vehicle suspension parameters with the least vibration at the current vehicle speed.

[0017] Beneficial effects: For driving safety, when driving on a normal road, if you encounter a turning road, you should slow down to pass. If the obstacle in this solution appears at the turning, the vehicle will be reminded to slow down before entering the turn, which is convenient for turning at a low speed. At the same time, combined with the recommended parameters, the speed reduction range is controllable, which improves the driver's driving experience and reduces the speed to the right level at one time. Driving is smoother.

[0018] Furthermore, the information acquisition module includes at least a camera and a radar.

[0019] Beneficial effects: Provides multi-sensory input, enhancing the system's ability to identify obstacles in different weather and lighting conditions. The addition of radar particularly enhances performance in adverse weather conditions such as rain and fog, and improves the robustness of the system.

[0020] Furthermore, it also includes a warning module, which is used to transmit the warning reminder signal transmitted by the data processing module to the driver, including sound prompts and text prompts; the information acquisition module at least includes a camera and a radar.

[0021] Beneficial effects: Multiple prompting methods adapt to the preferences of different drivers and improve the efficiency of conveying warning information. Text prompts provide drivers with more intuitive information, which helps them quickly understand the current driving situation.

[0022] Furthermore, the information collection module is also used to classify vehicles that have historically passed the road section and establish a corresponding database of historical passing information according to vehicle brands; the data processing module performs targeted retrieval of corresponding models in the corresponding brands of traveling vehicles based on the obstacle index signal.

[0023] Beneficial effects: Classified data collection enables the training model to be optimized more accurately for different vehicle models. The establishment of databases for different vehicle brands improves the system's adaptability to the characteristics of specific vehicle models.

[0024] Furthermore, the data processing module further includes a real-time analysis submodule for real-time analysis of the vehicle's dynamic response and road conditions, and a prediction algorithm for predicting potential driving risks based on current and historical data, and combining this information to optimize driving planning and suspension parameter adjustment.

[0025] Beneficial effects: Real-time analysis and prediction algorithms enable the system to more accurately assess current and potential driving risks.

[0026] Optimized driving planning and suspension parameter adjustment improve the vehicle's handling and ride comfort.

[0027] Furthermore, the system also includes a user feedback interface that allows the driver to input subjective evaluations of the system warning and adjustment effects, as well as manual input of possible obstacle types and locations, which are fed back to the cloud as auxiliary information for continuous learning and improvement of the training model, as well as real-time updating of the road obstacle database.

[0028] Beneficial effects: The user feedback mechanism provides the system with an opportunity for continuous improvement, enabling the system to better adapt to actual driving needs. Manually input obstacle information helps improve the accuracy of the road obstacle database and enhance the system's predictive capabilities.

[0029] A method for autonomous driving scene recognition based on a cloud training model includes the following steps:

[0030] S1: Information acquisition: Obtain the obstacle situation of the current driving section in real time through on-board sensors such as cameras and radars, and generate obstacle index signals;

[0031] S2: Dynamic data collection: collects the vehicle's current speed, accelerator pedal pressure signal, and suspension parameters;

[0032] S3: Historical data integration: Use the information collection module to collect and integrate the historical passing information of vehicles on this road section, including driving speed, suspension parameters and vibration signals when passing obstacles;

[0033] S4: Data sorting and model training: Sort the historical data according to the vibration magnitude when the vehicle passes through obstacles, and use this data as a training set to build and train the machine learning model in the cloud;

[0034] S5: Parameter recommendation: Set the vibration threshold, and extract the maximum vehicle speed and vehicle suspension parameters under the vibration threshold from historical data as recommended parameters;

[0035] S6: Data processing: When the system detects an obstacle, the data processing module searches for the corresponding historical passing information in the information collection module according to the obstacle index signal and obtains the recommended parameters;

[0036] S7: Driving planning: preset the recommended parameters in the driving planning of the vehicle computer to adjust the suspension parameters when the vehicle passes through obstacles;

[0037] S8: Warning generation: before approaching an obstacle, if the vehicle's current speed exceeds the recommended speed and the throttle is in the accelerating state, the warning module sends a warning signal;

[0038] S9: Dynamic response: If the accelerator pressure signal is maintained within the preset threshold time, the warning signal is canceled and the vehicle is allowed to continue accelerating; otherwise, the vehicle speed is adjusted to the recommended parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a system schematic diagram of an embodiment of the present invention;

[0040] Figure 2 A method step diagram of an embodiment of the present invention.

[0041] The following is a further detailed description through specific implementation methods: DETAILED DESCRIPTION

[0042] The embodiment is basically as shown in the attached Figure 1 As shown:

[0043] An automatic driving scene recognition system based on a cloud training model includes: an information acquisition module, an information collection module, an information processing module and a warning module.

[0044] An information collection module; used to collect historical passing information of vehicles on the road section, the historical passing information includes the vehicle's driving speed and vehicle suspension parameters, and the vibration signal of the vehicle passing the obstacle, and sort the information in order according to the vibration size of the vehicle, build a training model, and train the training model with the historical passing information of the vehicles on the road section as a training set; set a vibration threshold, and use the maximum vehicle speed and vehicle suspension parameters under the vibration threshold as recommended parameters for passing the obstacle on the road section; and count the vehicle suspension parameters with the smallest vibration under each speed state; the information collection module at least includes a vehicle speed sensor, a vibration sensor and a vehicle suspension parameter sensor, which respectively measure the vehicle speed, suspension parameters, vibration parameters and other signals of the vehicle; the information collection module also processes the historical passing data of the above-mentioned vehicles.

[0045] The information acquisition module acquires the obstacle conditions of the current driving section in real time, generates an obstacle index signal and transmits it to the cloud; the information acquisition module is also used to acquire the current driving speed of the vehicle and the pressing signal of the vehicle's accelerator; obstacles include potholes and speed bumps; the information acquisition module immediately starts to monitor the road conditions in real time through the on-board camera and radar, and collects the current driving speed of the vehicle and the pressing signal of the accelerator pedal. When the system detects an obstacle ahead in real time, it generates an obstacle index signal and sends it to the cloud.

[0046] In the cloud, the data collection model has been trained using the machine learning model trained with the vehicle data of the road section. The training model in the cloud is trained using a common convolutional neural network model. These historical data include the vehicle's driving speed, suspension parameters, and vibration signals when passing obstacles. The system will sort these data according to the vibration size. By setting the vibration threshold, the system can determine the maximum safe vehicle speed and suspension parameters, and store these as recommended parameters in the cloud.

[0047] Navigation module: plans the driving route according to the driver's destination. When there is a pothole on the vehicle's driving route, the navigation module is also used to remind the driver to change lanes to avoid it. If it cannot be avoided, it will send a signal that the obstacle is inevitable.

[0048] Data processing module: after receiving the obstacle unavoidable signal, the historical passing information of vehicles on the road section is searched in the information collection module according to the obstacle index signal, and the recommended parameters are obtained and preset in the driving plan of the vehicle computer, which are used to adjust the suspension parameters of the vehicle according to the recommended parameters when passing through the obstacle; before passing the obstacle, if the current driving speed of the vehicle is greater than the speed in the recommended parameters, and the accelerator pressing signal shows that the vehicle is in an accelerating state, a warning reminder signal is issued; within the threshold time, if the accelerator pressing signal is in a maintained state, the reminder signal is canceled, and the vehicle maintains accelerated driving. If the accelerator pressing signal weakens, the vehicle speed is reduced to the recommended parameters; if the current driving speed is less than the speed in the recommended parameters, the vehicle suspension parameters with the smallest vibration at the current vehicle speed are matched, and the suspension parameters of the driving vehicle are adjusted accordingly.

[0049] This solution uses the accelerator pressing signal, the current vehicle speed, and the recommended parameters (recommended vehicle speed and recommended suspension parameters) calculated from the historical passing data to provide guidance to the driver. If the driver allows, the vehicle speed will be reduced to the recommended parameters and the suspension parameters will be automatically adjusted. If the driver disagrees with adjusting the speed, the suspension parameters will be adjusted to the suspension parameters with the least vibration in the historical passing data at the current vehicle speed, so that the driver can pass the obstacle section with the greatest degree of comfort and safety.

[0050] When the vehicle approaches an obstacle, the data processing module will quickly retrieve the recommended parameters for the corresponding road section from the cloud and preset them into the vehicle system. If the vehicle's current speed exceeds the recommended speed and the accelerator pedal shows that the driver is accelerating, the warning module will issue a sound and text prompt to inform the driver to slow down. That is, the warning module is an audio or display screen that can display the information processed by the data processing module in the form of sound or text. If the driver does not respond within the preset threshold time, the system will automatically intervene and adjust the vehicle speed and suspension parameters to ensure safe passage through the obstacle.

[0051] The system also includes a real-time analysis submodule that analyzes the vehicle's dynamic response and road conditions, as well as a prediction algorithm that predicts potential driving risks based on current and historical data. This information is used to optimize driving planning and suspension parameter adjustment, further improving the vehicle's passability and comfort. The real-time analysis submodule uses an existing road condition prediction model to analyze road conditions. The specific technical means will not be described in detail in this solution. Through this setting, the user's driving experience can be improved.

[0052] In order to achieve continuous learning and improvement, the system is also equipped with a user feedback interface. Drivers can use this interface to input their evaluation of the system warning and adjustment effects, as well as manual input information such as the type and location of obstacles. These user feedbacks are uploaded to the cloud to update the training model and road obstacle database, allowing the system to continuously adapt to new driving conditions and obstacle types, thereby providing a more accurate and personalized assisted driving experience in the future.

[0053] It also involves an autonomous driving scene recognition method based on a cloud training model: Figure 2 As shown:

[0054] S1: Information acquisition: Obtain the obstacle situation of the current driving section in real time through on-board sensors such as cameras and radars, and generate obstacle index signals;

[0055] S2: Dynamic data collection: collects the vehicle's current speed, accelerator pedal pressure signal, and suspension parameters;

[0056] S3: Historical data integration: Use the information collection module to collect and integrate the historical passing information of vehicles on this road section, including driving speed, suspension parameters and vibration signals when passing obstacles;

[0057] S4: Data sorting and model training: Sort the historical data according to the vibration magnitude when the vehicle passes through obstacles, and use this data as a training set to build and train the machine learning model in the cloud;

[0058] S5: Parameter recommendation: Set the vibration threshold, and extract the maximum vehicle speed and vehicle suspension parameters under the vibration threshold from historical data as recommended parameters;

[0059] S6: Data processing: When the system detects an obstacle, the data processing module searches for the corresponding historical passing information in the information collection module according to the obstacle index signal and obtains the recommended parameters;

[0060] S7: Driving planning: preset the recommended parameters in the driving planning of the vehicle computer to adjust the suspension parameters when the vehicle passes through obstacles;

[0061] S8: Warning generation: before approaching an obstacle, if the vehicle's current speed exceeds the recommended speed and the throttle is in the accelerating state, the warning module sends a warning signal;

[0062] S9: Dynamic response: If the accelerator pressure signal is maintained within the preset threshold time, the warning signal is canceled and the vehicle is allowed to continue accelerating; otherwise, the vehicle speed is adjusted to the recommended parameters.

[0063] Embodiment 2

[0064] The difference between the second embodiment and the first embodiment is that: if the obstacle is located at the turning point of the road, within the threshold distance before entering the curve, according to the route planning of the navigation module, whether there is a changeable lane to avoid the obstacle, if it cannot be avoided, the data processing module sends a first deceleration signal to remind the driver to slow down to the recommended speed, if the driver slows down, the vehicle suspension parameters are adjusted to the recommended parameters; if the speed is not reduced, after entering the curve, the data processing module sends a second deceleration signal to remind the driver to slow down to the recommended speed, if the driver slows down, the vehicle suspension parameters are adjusted to the recommended parameters, if the speed is not reduced, the suspension parameters are adjusted to the vehicle suspension parameters with the minimum vibration at the current vehicle speed. For driving safety, in normal road driving, if you encounter a turning road, you should slow down to pass. If the obstacle in this solution appears at the turning point, the vehicle will be reminded to slow down before entering the curve, which is convenient for low-speed cornering. At the same time, combined with the recommended parameters, the deceleration range is controllable, which improves the driver's driving experience and reduces the vehicle speed in one go. Driving is smoother.

[0065] The above is only an embodiment of the present invention, and the common knowledge such as the known specific structure and characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. An autonomous driving scene recognition system based on a cloud training model, characterized in that: include: Information collection module; Used to collect historical passing information of vehicles on the driving section, the historical passing information includes the driving speed and suspension parameters of the vehicle, and the vibration signal of the vehicle passing through the obstacle, and sort the order according to the vibration magnitude of the vehicle, build a training model, and train the training model with the historical passing information of the vehicle on the road section as the training set; Set the vibration threshold, and use the maximum vehicle speed and vehicle suspension parameters under the vibration threshold as the recommended parameters for passing obstacles on this road section; The vehicle suspension parameters with the smallest vibration at each vehicle speed are statistically analyzed; Information acquisition module: Obstacle conditions of the current driving section are acquired in real time, and an obstacle index signal is generated and transmitted to the cloud; the information acquisition module is also used to acquire the current driving speed of the vehicle and the accelerator pressing signal of the vehicle; the obstacle conditions include potholes and speed bumps; Navigation module: plans the driving route according to the driver's destination. When there is a pothole on the vehicle's driving route, the navigation module is also used to remind the driver to change lanes to avoid it. If it cannot be avoided, it will send a signal that the obstacle is inevitable. Data processing module: after receiving the obstacle unavoidable signal, the information collection module searches for the historical passing information of the vehicles on the road section according to the obstacle index signal, obtains the recommended parameters and presets them in the driving plan of the vehicle computer, and adjusts the suspension parameters of the vehicle according to the recommended parameters when passing the obstacle; before passing the obstacle, if the current driving speed of the vehicle is greater than the speed in the recommended parameters, and the accelerator pressing signal shows that the vehicle is in the accelerating state, a warning reminder signal is issued; within the threshold time, if the accelerator pressing signal is in the maintained state, the reminder signal is cancelled, and the vehicle maintains the accelerated driving; if the accelerator pressing signal weakens, the vehicle speed is reduced to the recommended parameters; If the current driving speed is lower than the speed in the recommended parameters, the vehicle suspension parameters with the smallest vibration at the current speed are matched, and the suspension parameters of the driving vehicle are adjusted accordingly.

2. The automatic driving scene recognition system based on a cloud training model according to claim 1, characterized in that: If the obstacle is located at a bend in the road, within the threshold distance before entering the bend, the route planning of the navigation module is used to check whether there is a changeable lane to avoid the obstacle. If it cannot be avoided, the data processing module sends a first speed reduction signal to remind the driver to slow down to the recommended speed. If the driver slows down, the vehicle suspension parameters are adjusted to the recommended parameters; If the speed is not reduced, after entering the curve, the data processing module sends a second speed reduction signal to remind the driver to reduce the speed to the recommended speed. If the driver reduces the speed, the vehicle suspension parameters are adjusted to the recommended parameters. If the speed is not reduced, the suspension parameters are adjusted to the vehicle suspension parameters with the least vibration at the current vehicle speed.

3. The automatic driving scene recognition system based on a cloud training model according to claim 1, characterized in that: It also includes a warning module, which is used to transmit the warning reminder signal transmitted by the data processing module to the driver, including sound prompts and text prompts; the information acquisition module at least includes a camera and a radar.

4. The automatic driving scene recognition system based on a cloud training model according to claim 1, characterized in that: The information collection module is also used to classify vehicles that have passed the road section in the past and establish a database of corresponding historical passing information according to vehicle brands; the data processing module performs targeted retrieval of corresponding models in corresponding brands of traveling vehicles based on obstacle index signals.

5. An autonomous driving scene recognition system based on a cloud training model according to any one of claims 1 to 4, characterized in that: The data processing module further includes a real-time analysis submodule for real-time analysis of the vehicle's dynamic response and road conditions, and a prediction algorithm for predicting potential driving risks based on current and historical data, and combining this information to optimize driving planning and suspension parameter adjustment.

6. The automatic driving scene recognition system based on a cloud training model according to claim 5, characterized in that: The system also includes a user feedback interface that allows the driver to input subjective evaluations of the system warning and adjustment effects, as well as manual input of possible obstacle types and locations, which are then fed back to the cloud as auxiliary information for continuous learning and improvement of the training model, as well as real-time updating of the road obstacle database.

7. A method for autonomous driving scene recognition based on a cloud training model, characterized in that: The following steps are involved: S1: Information acquisition: Obtain the obstacle situation of the current driving section in real time through the vehicle-mounted sensors, including cameras and radars, and generate an obstacle index signal; S2: Dynamic data collection: collects the vehicle's current speed, accelerator pedal pressure signal, and suspension parameters; S3: Historical data integration: Use the information collection module to collect and integrate the historical passing information of vehicles on this road section, including driving speed, suspension parameters and vibration signals when passing obstacles; S4: Data sorting and model training: Sort the historical data according to the vibration magnitude when the vehicle passes through obstacles, and use this data as a training set to build and train the machine learning model in the cloud; S5: Parameter recommendation: Set the vibration threshold, and extract the maximum vehicle speed and vehicle suspension parameters under the vibration threshold from historical data as recommended parameters; S6: Data processing: When the system detects an obstacle, the data processing module searches for the corresponding historical passing information in the information collection module according to the obstacle index signal and obtains the recommended parameters; S7: Driving planning: preset the recommended parameters in the driving planning of the vehicle computer to adjust the suspension parameters when the vehicle passes through obstacles; S8: Warning generation: before approaching an obstacle, if the vehicle's current speed exceeds the recommended speed and the throttle is in the accelerating state, the warning module sends a warning signal; S9: Dynamic response: If the accelerator pressing signal is maintained within the preset threshold time, the warning signal is cancelled and the vehicle is allowed to continue accelerating; Otherwise, adjust the vehicle speed to the recommended parameters.

Images