Chassis control method and device and electronic equipment

Abstract

The embodiment of the invention relates to the technical field of vehicle control, and discloses a chassis control method and device and electronic equipment, and the method comprises the steps: firstly, inputting vehicle driving data into a control model, and obtaining an initial chassis control decision output by the control model, the vehicle driving data comprising environment data and chassis state data of a vehicle; then, inputting the vehicle driving data and the initial chassis control decision into a large language model, and obtaining a target chassis control decision output by the large language model; and finally, the chassis of the vehicle is controlled according to the target chassis control decision. By means of the technical scheme, the accuracy of chassis control decision can be improved, precise control over a vehicle chassis system is achieved, and higher safety and stability can be provided in a complex and changeable driving environment.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of vehicle control technology, and specifically to a chassis control method, device and electronic equipment. Background Art

[0002] At present, the intelligentization of electric vehicle chassis is one of the important directions of current electric vehicle technology development. Intelligent chassis not only improves the performance and safety of vehicles, but also provides a solid foundation for autonomous driving and future travel. The intelligentization of electric vehicle chassis mainly relies on chassis control algorithms to output target chassis control decisions. However, how to ensure the accuracy of chassis control methods in various environmental conditions is crucial.

[0003] In the prior art, the chassis control method of electric vehicles mainly improves the performance of electric vehicles in complex driving scenarios by using a general reinforcement learning agent method to model the collaborative control between chassis domain systems.

[0004] However, the inference results of the chassis control method in the prior art have poor accuracy, which affects the performance of the vehicle and the driving experience of the user. Summary of the invention

[0005] In view of the above problems, an embodiment of the present invention provides a chassis control method, device and electronic device to solve the problem that the inference results of the chassis control method in the prior art have poor accuracy, affecting the performance of the vehicle and the driving experience of the user.

[0006] According to one aspect of an embodiment of the present invention, a chassis control method is provided, the method comprising:

[0007] Inputting vehicle driving data into a control model to obtain an initial chassis control decision output by the control model, wherein the vehicle driving data includes environmental data and chassis state data of the vehicle;

[0008] Inputting the vehicle driving data and the initial chassis control decision into a large language model to obtain a target chassis control decision output by the large language model;

[0009] The chassis of the vehicle is controlled according to the target chassis control decision.

[0010] In an optional manner, inputting the vehicle driving data and the initial chassis control decision into a large language model to obtain a target chassis control decision output by the large language model includes:

[0011] According to the environmental data, determining whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model;

[0012] If sample vehicle driving data corresponding to the current environment exists in the training set of the control model, the initial chassis control decision is output as the target chassis control decision.

[0013] In an optional manner, the method further includes:

[0014] If there is no sample vehicle driving data corresponding to the current environment in the training set of the control model, the target chassis control decision is output based on a knowledge vector library storing vehicle dynamics limitation information, the initial chassis control decision and the vehicle driving data.

[0015] In an optional manner, the outputting the target chassis control decision based on the knowledge vector library, the initial chassis control decision and the vehicle driving data includes:

[0016] If it is determined according to the knowledge vector library and the vehicle driving data that the initial chassis control decision exceeds the vehicle dynamics limit, the initial chassis control decision is adjusted according to the knowledge vector library to output the target chassis control decision that satisfies the vehicle dynamics limit;

[0017] If it is determined according to the knowledge vector library and the vehicle driving data that the initial chassis control decision does not exceed the vehicle dynamics limit, the initial chassis control decision is output as the target chassis control decision.

[0018] In an optional manner, before inputting the vehicle driving data into the control model and obtaining the initial chassis control decision output by the control model, the method includes:

[0019] The control model is trained according to the training set of the control model, wherein the training set includes a plurality of sample vehicle driving data and a vehicle state corresponding to each sample driving data.

[0020] In an optional manner, judging, based on the environmental data, whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model includes:

[0021] Extracting features from sample environment data of each sample vehicle driving data in the training set to obtain sample environment features;

[0022] Extracting features from the environmental data to obtain environmental features;

[0023] The environmental feature is matched with each sample environmental feature to determine whether there is a sample environmental feature matching the environmental feature, so as to determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model.

[0024] In an optional manner, matching the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature matching the environmental feature includes:

[0025] Calculating a characteristic distance between the environmental characteristic and each sample environmental characteristic;

[0026] It is determined whether the minimum value of the characteristic distance is less than or equal to a preset distance, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0027] In an optional manner, matching the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature matching the environmental feature includes:

[0028] According to each preset environmental interval, the environmental feature and each sample environmental feature, determining a first environmental interval corresponding to the environmental feature and a second environmental interval corresponding to each sample environmental feature;

[0029] It is determined whether there is a second environmental interval that is the same as the first environmental interval, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0030] According to another aspect of an embodiment of the present invention, there is provided a chassis control device, comprising:

[0031] An acquisition module, used for inputting vehicle driving data into a control model to acquire an initial chassis control decision output by the control model, wherein the vehicle driving data includes environmental data and chassis status data of the vehicle;

[0032] An output module, used for inputting the vehicle driving data and the initial chassis control decision into a large language model, and obtaining a target chassis control decision output by the large language model;

[0033] A control module is used to control the chassis of the vehicle according to the target chassis control decision.

[0034] In an optional manner, the output module is further used for:

[0035] According to the environmental data, determining whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model;

[0036] If sample vehicle driving data corresponding to the current environment exists in the training set of the control model, the initial chassis control decision is output as the target chassis control decision.

[0037] In an optional manner, the output module is also used to output the target chassis control decision based on a knowledge vector library, the initial chassis control decision and the vehicle driving data if there is no sample vehicle driving data corresponding to the current environment in the training set of the control model, and the knowledge vector library stores vehicle dynamics limitation information.

[0038] In an optional manner, the output module is specifically used to:

[0039] If it is determined according to the knowledge vector library and the vehicle driving data that the initial chassis control decision exceeds the vehicle dynamics limit, the initial chassis control decision is adjusted according to the knowledge vector library to output the target chassis control decision that satisfies the vehicle dynamics limit;

[0040] If it is determined according to the knowledge vector library and the vehicle driving data that the initial chassis control decision does not exceed the vehicle dynamics limit, the initial chassis control decision is output as the target chassis control decision.

[0041] In an optional manner, the chassis control device further includes a training module. Before inputting the vehicle driving data into the control model and obtaining the initial chassis control decision output by the control model, the training module is used to:

[0042] The control model is trained according to the training set of the control model, wherein the training set includes a plurality of sample vehicle driving data and a vehicle state corresponding to each sample driving data.

[0043] In an optional manner, the chassis control device further includes a processing module, which is used to:

[0044] Extracting features from sample environment data of each sample vehicle driving data in the training set to obtain sample environment features;

[0045] Extracting features from the environmental data to obtain environmental features;

[0046] The environmental feature is matched with each sample environmental feature to determine whether there is a sample environmental feature matching the environmental feature, so as to determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model.

[0047] In an optional manner, the processing module is specifically used to:

[0048] Calculating a characteristic distance between the environmental characteristic and each sample environmental characteristic;

[0049] It is determined whether the minimum value of the characteristic distance is less than or equal to a preset distance, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0050] In an optional manner, the processing module is specifically used to:

[0051] According to each preset environmental interval, the environmental feature and each sample environmental feature, determining a first environmental interval corresponding to the environmental feature and a second environmental interval corresponding to each sample environmental feature;

[0052] It is determined whether there is a second environmental interval that is the same as the first environmental interval, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0053] According to another aspect of an embodiment of the present invention, there is provided an electronic device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0054] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform the operation of the chassis control method as described in any one of the first aspects.

[0055] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores at least one executable instruction, wherein the executable instruction enables the electronic device / chassis control device to perform the operation of the chassis control method as described in any one of the first aspects.

[0056] An embodiment of the present invention provides a chassis control method, device and electronic device. In the method, when controlling the chassis of a vehicle, the accuracy and reliability of the control strategy can be greatly improved by combining the existing control model with the large language model to jointly process decisions, thereby ensuring precise control of the vehicle chassis using the control strategy, improving vehicle performance and the user's driving experience.

[0057] The above description is only an overview of the technical solution of the embodiment of the present invention. In order to more clearly understand the technical means of the embodiment of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the accompanying drawings. In the accompanying drawings:

[0059] Figure 1The schematic diagram shows the process of the chassis control method provided by the prior art. Figure 1 ;

[0060] Figure 2 The process diagram of the chassis control method provided by the present invention is shown in FIG. Figure 1 ;

[0061] Figure 3 The process diagram of the chassis control method provided by the present invention is shown in FIG. Figure 2 ;

[0062] Figure 4 The process diagram of the chassis control method provided by the present invention is shown in FIG. Figure 3 ;

[0063] Figure 5 A schematic diagram of the structure of the chassis control device provided by the present invention is shown;

[0064] Figure 6 A schematic structural diagram of an embodiment of an electronic device provided by the present invention is shown. DETAILED DESCRIPTION

[0065] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0066] General artificial intelligence technology represented by large language models is breaking the original technical paradigm and bringing about a technological revolution in the field of artificial intelligence. The productization of large language models has been successfully applied in many applications, such as customer service, chat assistants, office assistants, cockpit assistants, etc. At the same time, large language models also bring new technical integration to robots, such as general household robots combined with large language models. In the field of intelligent driving, large language models can understand different driving needs through natural language and help realize more flexible autonomous driving solutions.

[0067] In the stage where the chassis of electric vehicles is still in the system-by-wire control and a few functions are intelligent, in order to achieve coordinated control between chassis domain systems, the existing technology usually uses the control method of reinforcement learning agents to make control decisions. Specifically, reinforcement learning learns the optimal control strategy under different driving scenarios and environmental conditions through a large amount of simulation training. Then, in actual applications, the agent processes complex sensor data (such as vehicle speed, body posture, road conditions, etc.) through deep neural networks and outputs corresponding control decisions (such as braking force, steering angle, etc.) to improve vehicle performance and safety.

[0068] Figure 1 The schematic diagram shows the process of the chassis control method provided by the prior art. Figure 1 .like Figure 1 As shown, the chassis control method provided by the prior art includes the following steps:

[0069] Step 101: Input chassis status data, road status data, and cloud data including weather, maps, etc. into a control model.

[0070] Step 102: Obtain the final decision output by the control model.

[0071] Step 103: Apply the final decision in the vehicle.

[0072] However, chassis control methods that rely solely on reinforcement learning are difficult to implement. Due to the limitation of the amount of training data, they cannot cover various environmental factors and vehicle conditions, resulting in low accuracy of the output control decisions and failure to meet higher vehicle safety and redundancy requirements.

[0073] Based on this, the present invention proposes a chassis control method, device and electronic device. Considering that the low accuracy of the existing chassis control method is due to the limitation of the training data, the control model may not be able to output an accurate chassis control solution in some extreme cases (for example, extremely cold weather, icy roads, etc.), and the training data in the large language model is comprehensive and large in scale, which can well make up for the lack of data in the existing chassis control method, and can reliably and accurately infer the control method corresponding to the vehicle driving data under various conditions, thereby improving the vehicle driving performance and the driver's driving experience. Therefore, when controlling the vehicle chassis, the existing control model and the large language model can be combined for joint processing and decision-making, which can greatly improve the accuracy and adaptability of the control strategy, thereby achieving precise control of the vehicle chassis system, and then providing higher safety and stability in a complex and changeable driving environment.

[0074] Figure 2 The process diagram of the chassis control method provided by the present invention is shown in FIG. Figure 1 ,like Figure 2 As shown, the method comprises the following steps:

[0075] Step 201: Input vehicle driving data into a control model to obtain an initial chassis control decision output by the control model.

[0076] Among them, the vehicle driving data includes environmental data and chassis status data of the vehicle. The environmental data can be the road condition obtained through lidar detection, such as slippery road surface, ice-covered road surface or road slope and other information; it can also be the current weather conditions obtained through cloud data, such as extremely cold temperature and rain; the chassis status data of the vehicle reflects the working status of the current chassis of the vehicle, and may include suspension status, vehicle center of gravity, tire pressure, steering angle, acceleration and other data. The embodiment of the present invention does not impose specific restrictions on the vehicle driving data.

[0077] It should be understood that by inputting the current environmental data and the chassis status data of the vehicle into the control model, the control model will then analyze the real-time input environmental data and vehicle status data, and use the strategy function to make decisions on the current status. Finally, the immediate and long-term returns of different control decisions are evaluated, and the optimal control action, i.e., the initial chassis control decision, is output.

[0078] In one achievable method, before obtaining the initial chassis control decision output by the control model, the control model needs to be trained, specifically:

[0079] The control model is trained according to the training set of the control model.

[0080] The training set includes multiple sample vehicle driving data and the vehicle state corresponding to each sample vehicle driving data.

[0081] It should be noted that the control model can be based on a reinforcement learning framework, a machine learning framework, or a deep learning network based on a neural network, and the present application embodiment does not make specific restrictions. For ease of understanding, this embodiment takes a control model based on a reinforcement learning framework as an example to describe the specific process of the control model in detail.

[0082] First, a large number of sample vehicle driving data in the training set and the vehicle state data corresponding to each sample vehicle driving data together constitute the state space at the current moment and are input into the control model based on the reinforcement learning framework. The intelligent agent in the reinforcement learning generates the corresponding action space, that is, different control strategies, through the policy function according to the current state space (i.e., environmental data, chassis state, and vehicle state); then, each time an action is executed in the action space, the intelligent agent will receive a feedback from the environment, i.e., a reward; the intelligent agent obtains the cumulative reward for a series of actions based on the immediate reward of each step, and the reward continuously updates the policy function. As the number of training times increases, the intelligent agent gradually converges to the optimal chassis control strategy, and makes corresponding control decisions in real time according to different vehicle driving data and the vehicle state corresponding to the vehicle driving data in actual applications to ensure the stability and safety of the vehicle.

[0083] For example, a vehicle is traveling from point A to point B. When it is about to reach point B, it needs an emergency brake. If the vehicle performs an action at this time and completes braking within a safe distance without a collision, the system will provide a higher positive reward to encourage the agent to perform the correct operation. For example, the value of the positive reward can be set to +100 points, which means that the agent has successfully avoided an accident. In addition, if the agent can ensure smooth deceleration of the vehicle (that is, it will not generate excessive braking force or sudden braking) and maintain the stability of the vehicle during braking, the system can also give additional rewards. For example, reducing the discomfort of passengers during braking can add an additional +20 points; if the vehicle has braked too early before reaching point B, a negative reward will be given to let the agent know that the braking timing is inaccurate. For example, braking too early may lead to excessive deceleration and affect the overall driving efficiency of the vehicle, and the reward is -30 points; if the vehicle fails to complete braking within a reasonable safe distance, or even collides, a serious negative reward will be given. For example, a collision accident will result in a negative reward of -200 points.

[0084] It is understandable that by using the training set, the control model can learn how to take the best actions under different conditions, and ultimately achieve precise and efficient chassis control in practical applications, thereby improving vehicle driving safety and user driving experience.

[0085] Step 202: Input the vehicle driving data and the initial chassis control decision into the large language model to obtain the target chassis control decision output by the large language model.

[0086] It should be noted that before the vehicle driving data and the initial chassis control decision are input into the large language model, the vehicle driving data and the initial chassis control decision need to be converted into a language, that is, a natural language that the large language model can understand and process, usually text data. Therefore, the language conversion can be converted into a text description or symbolic information that can be understood by the large language model through a specific conversion model; when the large model and the above-mentioned conversion model are integrated into a multimodal large model, the vehicle driving data and the initial chassis control decision can also be directly input into the large model, and the embodiments of the present application are not specifically limited here.

[0087] It is understandable that since the large language model can handle more complex nonlinear relationships and environmental factors, by inputting vehicle driving data and initial chassis control decisions, the model can combine vehicle driving data and initial control strategies to generate more accurate target chassis control decisions, thereby significantly improving vehicle performance and providing a safer, more comfortable and efficient driving experience.

[0088] It should be understood that the process of the large model determining the target chassis control decision based on the vehicle driving data and the initial chassis control decision will be Figure 3The detailed description is given in the embodiment shown and will not be repeated here.

[0089] Step 203: Control the chassis of the vehicle according to the target chassis control decision.

[0090] Among them, the target chassis control decision can be executed by the vehicle's electronic control unit or a chassis control module (such as active suspension, electronic steering system, etc.).

[0091] It can be understood that by combining the control model with the large language model, the target chassis control decision can be obtained more accurately, so that the stability and safety of the vehicle can be guaranteed.

[0092] The chassis control method provided by the embodiment of the present invention can greatly improve the accuracy and adaptability of the control strategy by combining the existing control model with the large language model for joint processing and decision-making when controlling the vehicle chassis. First, according to the vehicle driving data, an initial chassis control strategy is output by using the control model, and then the large language model, the initial control strategy and the vehicle driving data are further analyzed to output the final target chassis control decision, and finally the target chassis control decision is used to control the vehicle to ensure the accuracy of the decision. This method can not only realize the precise control of the vehicle chassis system, but also provide higher safety and stability in complex and changeable driving environments.

[0093] Figure 3 The process diagram of the chassis control method provided by the present invention is shown in FIG. Figure 2 ,like Figure 3 As shown, this embodiment describes in detail the process of outputting the target chassis control decision by the large language model, and the method includes the following steps:

[0094] Step 301: Based on the environmental data, determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model.

[0095] It should be understood that since the accuracy of the initial chassis control decision output by the control model for a known environment is high, the accuracy of the initial chassis control decision output by the control model cannot be guaranteed for an unknown environment. Therefore, by judging whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model, it is possible to determine whether the initial chassis control decision output by the control model is effective and accurate, and provide a basis for whether the large language model needs to further analyze the initial chassis control decision. In this way, the problem of incorrect chassis control caused by directly applying the initial chassis control strategy output by the control model in an unfamiliar environment can be avoided, thereby improving the accuracy of the vehicle.

[0096] Step 302: If there is sample vehicle driving data corresponding to the current environment in the training set of the control model, the initial chassis control decision is output as the target chassis control decision.

[0097] It is understandable that by judging from the environmental data that there are sample vehicle driving data corresponding to the current environment in the training set of the control model, it means that the control model has learned the effective control strategy in the current environment through the previous training process, so the output initial chassis control decision is effective and has high accuracy, and can be directly applied to the current vehicle chassis control. In this way, the large language model is avoided from making repeated chassis control decisions, computing resources are saved, and delayed response is reduced, further improving the user experience.

[0098] Step 303: If there is no sample vehicle driving data corresponding to the current environment in the training set of the control model, a target chassis control decision is output based on the knowledge vector library, the initial chassis control decision and the vehicle driving data.

[0099] The knowledge vector library stores vehicle dynamics restriction information, which can be introduced through a large language model using the Chain of Thought (COT) framework.

[0100] Furthermore, if there is no corresponding vehicle dynamics restriction information in the knowledge vector library, the relevant vehicle dynamics restriction information may be obtained through an external search interface, which is not specifically limited in this embodiment.

[0101] Vehicle dynamics limitation information refers to various limitations and characteristic information related to the dynamic behavior of the vehicle, which may include basic dynamic parameters (the position of the vehicle's center of gravity, suspension system characteristics, normal operating temperature of vehicle components, etc.), motion limitations (maximum acceleration, maximum deceleration, turning radius, roll angle, etc.), environmental impacts (the maximum speed at which tires can drive on ice without slipping, etc.) and safety margins (the safety range of the vehicle in extreme situations such as sharp turns and emergency braking).

[0102] It is understandable that the large language model makes certain corrections and adjustments to the content of the initial chassis control decision based on the knowledge vector library and vehicle driving data, comprehensively considers various factors, ensures that the control decision is in line with the vehicle's dynamic characteristics, and prevents the vehicle from losing control or having an accident risk in a complex environment; in addition, the large language model can also refer to similar cases contained in the knowledge vector library, draw on existing successful experiences, and further increase the reliability of the target chassis control decision.

[0103] In one achievable method, when the large language model outputs the target chassis control decision, it needs to judge the initial chassis control decision based on the knowledge vector library. The specific process is as follows:

[0104] If it is determined based on the knowledge vector library and the vehicle driving data that the initial chassis control decision exceeds the vehicle dynamics limit, the initial chassis control decision is adjusted based on the knowledge vector library to output a target chassis control decision that meets the vehicle dynamics limit; if it is determined based on the knowledge vector library and the vehicle driving data that the initial chassis control decision does not exceed the vehicle dynamics limit, the initial chassis control decision is output as the target chassis control decision.

[0105] It should be noted that using the knowledge vector library and vehicle driving data to judge the initial chassis control decision can ensure that the final output control decision meets the vehicle's dynamic limitations and effectively avoid potential safety issues caused by decisions that exceed the vehicle's dynamic capabilities.

[0106] For example, suppose a car is driving on a slippery uphill road. The vehicle's driving data includes the current speed, road friction coefficient, slope, weather, etc. The initial chassis control decision generated by the control model recommends accelerating by 5% to keep the speed constant. However, the large language model obtains the vehicle's dynamic limitations based on the slippery road conditions, indicating that under the current friction coefficient and slope, accelerating by 5% may cause the vehicle to lose control or slip, and can only accelerate by 2% at most. In this case, the current initial chassis control decision needs to be corrected to be within the range of the vehicle's dynamic limitations, such as adjusting the acceleration of 5% to 2% to ensure that the tire's grip on the slippery road does not exceed the limit. For another example, if the initial chassis control decision generated by the control model recommends accelerating by 1%, the large language model determines based on the basic knowledge base that the initial chassis control decision does not exceed the vehicle's dynamic limitations, so there is no need to correct the initial decision, and the initial decision of accelerating by 1% is directly output as the target chassis control decision.

[0107] It is understandable that by combining the knowledge vector library to judge the initial chassis control decision, it is ensured that the generated target decision meets the vehicle dynamics constraints. This method not only improves the vehicle driving safety, but also improves the vehicle's ability to adapt to complex environments, providing more accurate decision support for the intelligent driving system.

[0108] The chassis control method provided by the embodiment of the present invention determines whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model based on the environmental data. If so, the initial decision is directly adopted to reduce the calculation overhead and improve the decision-making efficiency. If not, the large language model is further corrected and adjusted to make up for the shortcomings of the control model and output a target control decision that is more suitable for the current environment. In this way, the control model is prevented from making wrong decisions in an environment that has never been seen before, thereby ensuring the accuracy of the vehicle's chassis control strategy in various complex situations, and significantly improving the safety of vehicle driving and the user's driving experience.

[0109] It should be noted that when the large language model determines whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model, feature engineering is needed to make the judgment. The specific implementation process is as follows:

[0110] First, feature extraction is performed on the sample environmental data of each sample vehicle driving data in the training set to obtain the sample environmental features; then, feature extraction is performed on the environmental data to obtain the environmental features; finally, the environmental features are matched with each sample environmental feature to determine whether there are sample environmental features that match the environmental features, so as to determine whether there are sample vehicle driving data corresponding to the current environment in the training set of the control model.

[0111] It should be noted that the feature extraction method can be used to extract features of the sample environmental data in each sample vehicle driving data (for example: temperature, humidity, road friction coefficient, road slope, traffic conditions, etc.), so that the sample environmental data in the training data can be represented in a more structured and clearer manner, which is convenient for subsequent matching judgment; similarly, the same feature extraction method is used to extract features from the environmental data to obtain environmental features, ensuring that the feature extraction method of the current environmental data is consistent with the feature extraction method of the samples in the training set.

[0112] It is understandable that through precise feature extraction and matching, the large language model can automatically determine whether there is sample data matching the current environment in the training set of the control model, and based on this, it can process the chassis control decisions more efficiently, thereby improving the overall performance and safety of autonomous driving.

[0113] In one achievable manner, when determining whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model, the specific implementation manner includes:

[0114] First, the characteristic distance between the environmental feature and each sample environmental feature is calculated; then, a determination is made as to whether the minimum value in the characteristic distance is less than or equal to a preset distance, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0115] It should be noted that the method for calculating the characteristic distance between the environmental feature and each sample environmental feature can be to use methods such as Euclidean distance, Manhattan distance or cosine similarity. The purpose is to determine whether there is similarity between the environmental feature and each sample environmental feature. The embodiment of the present application does not make specific restrictions here. Next, the calculation process of the characteristic distance between the environmental feature and each sample environmental feature using Euclidean distance and cosine similarity is described in detail.

[0116] For example, the smaller the distance between two features calculated using the Euclidean distance, the more similar the two features are. Assuming there are three features: temperature, humidity, and road friction coefficient, the sample environment feature vector is A = [5,3,2], and the environment feature vector is B = [4,2,3], then the calculated Euclidean distance is:

[0117]

[0118] At this time, by comparing with the preset distance, for example, the preset distance is 1.2, and 1.732>1.2, it indicates that there is a certain distance between the current environment characteristics and the sample environment characteristics, so there is no sample data matching the current environment in the training set.

[0119] Continuing with the previous example, the angle between two vectors calculated using cosine similarity, the smaller the angle, the closer the cosine value is to 1, indicating that the two feature vectors are more similar. The specific cosine similarity calculation process is:

[0120]

[0121] Among them, A·B represents the dot product of vector A and vector B; ||A|| and ||B|| respectively represent the modulus (i.e., vector length) of vector A and vector B. At this time, by comparing with the preset distance, for example, the preset distance is 1.1, and 0.964<1.1 indicates that there is a certain degree of similarity between the current environment characteristics and the sample environment characteristics, so there is sample data matching the current environment in the training set.

[0122] It can be understood that by calculating the feature distance between the environmental feature and each sample environmental feature, the similarity between them can be quantified and the feature distance can be judged within a preset range, rather than being limited to the difference in absolute values. Instead, it focuses on the relative changes in the features, thereby providing a more accurate matching result.

[0123] In another achievable manner, when judging the environmental features and each sample environmental feature, the specific implementation manner may also include:

[0124] First, according to each preset environmental interval, environmental feature and each sample environmental feature, determine the first environmental interval corresponding to the environmental feature and the second environmental interval corresponding to each sample environmental feature; then, determine whether there is a second environmental interval that is the same as the first environmental interval, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0125] It should be noted that each preset environmental interval is a pre-division of the value range of each environmental feature. For example, the temperature is discretized according to a certain interval (such as 0-10℃ for the low temperature interval, 10-20℃ for the medium temperature interval, etc.). It no longer relies on specific values, but matches according to the interval, which helps to reduce the impact of feature noise and avoid excessive sensitivity to slight numerical changes.

[0126] For example, assume that the environmental characteristics and preset ranges are: temperature range: 0-10°C, 10-20°C, 20-30°C; humidity range: 0-20%, 20-40%, 40-60%; road friction coefficient range: 0-0.5, 0.5-1.0.

[0127] The first environmental intervals corresponding to the environmental features are: temperature: 15°C (belonging to 10-20°C); humidity: 30% (belonging to 20-40%); friction coefficient: 0.6 (belonging to 0.5-1.0); the second environmental intervals corresponding to the sample environmental features are: temperature: 25°C (belonging to 20-30°C); humidity: 50% (belonging to 40-60%); friction coefficient: 0.8 (belonging to 0.5-1.0). Since the first environmental interval does not match the second environmental interval, it is considered that there is no sample environmental feature that matches the environmental feature.

[0128] Furthermore, if the first environmental intervals corresponding to the environmental features are: temperature: 15°C (belonging to 10-20°C); humidity: 30% (belonging to 20-40%); friction coefficient: 0.6 (belonging to 0.5-1.0); the second environmental intervals corresponding to the sample environmental features are: temperature: 12°C (belonging to 10-20°C); humidity: 35% (belonging to 20-40%) friction coefficient: 0.7 (belonging to 0.5-1.0). Since the first environmental interval matches the second environmental interval, it is considered that there is a sample environmental feature that matches the environmental feature.

[0129] It is understandable that through this interval matching method, the large language model can quickly and effectively determine whether the current environment has corresponding sample data in the training set of the control model. This method avoids misjudgment caused by tiny numerical fluctuations.

[0130] Optionally, when judging whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model, statistical features can also be used, including:

[0131] First, according to each preset statistical threshold, statistical feature, environmental feature and each sample environmental feature, determine the first statistical data corresponding to the environmental feature and the second statistical data corresponding to each sample environmental feature; then, determine whether the difference between the first statistical data and the second statistical data is less than or equal to the preset statistical threshold, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0132] The preset statistical features may include statistics such as mean, variance, and median.

[0133] For example, assuming that the environmental features are the speed and temperature of the car, the statistical features are the mean and variance, and the preset statistical thresholds are that the mean difference does not exceed 10km / h and the variance difference does not exceed 5. If the corresponding first statistical data calculated based on the environmental features is: the mean speed is 60km / h and the variance is 15; the corresponding second statistical data calculated based on the sample environmental features is: the mean speed is 55km / h and the variance is 13; since the mean speed difference between the first statistical data and the second statistical data is 5km / h and the variance difference is 2, both of which are less than the preset statistical thresholds, it indicates that there are sample data in the training set that match the current environment.

[0134] Figure 4 The process diagram of the chassis control method provided by the present invention is shown in FIG. Figure 3 ,like Figure 4 As shown, the method comprises the following steps:

[0135] Step 401: Input vehicle driving data into the control model.

[0136] Step 402: Input the initial chassis control decision output by the control model and the vehicle driving data into the large language model.

[0137] Step 403: The large model outputs the target chassis control decision based on the knowledge vector library.

[0138] Step 404: Apply the target chassis control decision in the vehicle.

[0139] It should be noted that the specific implementation process is Figure 2 as well as Figure 3 The above has been described in the embodiments of the present invention, and the embodiments of the present application will not be described in detail here.

[0140] Figure 5 The schematic diagram of the structure of the chassis control device provided by the present invention is shown as follows: Figure 5 As shown, the chassis control device 50 includes: an acquisition module 501 , an output module 502 , and a control module 503 .

[0141] An acquisition module 501 is used to input vehicle driving data into a control model to acquire an initial chassis control decision output by the control model, wherein the vehicle driving data includes environmental data and chassis status data of the vehicle;

[0142] An output module 502 is used to input the vehicle driving data and the initial chassis control decision into the large language model to obtain the target chassis control decision output by the large language model;

[0143] The control module 503 is used to control the chassis of the vehicle according to the target chassis control decision.

[0144] In an optional manner, the output module 502 is further configured to:

[0145] Based on the environmental data, determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model;

[0146] If there is sample vehicle driving data corresponding to the current environment in the training set of the control model, the initial chassis control decision is output as the target chassis control decision.

[0147] In an optional manner, the output module 502 is also used to: if there is no sample vehicle driving data corresponding to the current environment in the training set of the control model, then based on the knowledge vector library, the initial chassis control decision and the vehicle driving data, output the target chassis control decision, and the knowledge vector library stores vehicle dynamics limitation information.

[0148] In an optional manner, the output module 502 is specifically configured to:

[0149] If it is determined that the initial chassis control decision exceeds the vehicle dynamics limit according to the knowledge vector library and the vehicle driving data, the initial chassis control decision is adjusted according to the knowledge vector library to output a target chassis control decision that meets the vehicle dynamics limit;

[0150] If it is determined based on the knowledge vector library and the vehicle driving data that the initial chassis control decision does not exceed the vehicle dynamics limit, the initial chassis control decision is output as the target chassis control decision.

[0151] In an optional manner, the chassis control device further includes a training module 504. Before inputting the vehicle driving data into the control model and obtaining the initial chassis control decision output by the control model, the training module 504 is used to:

[0152] The control model is trained according to a training set of the control model, where the training set includes a plurality of sample vehicle driving data and a vehicle state corresponding to each sample driving data.

[0153] In an optional manner, the chassis control device further includes a processing module 505, which is used to:

[0154] Perform feature extraction on the sample environment data of each sample vehicle driving data in the training set to obtain the sample environment features;

[0155] Extract features from environmental data to obtain environmental features;

[0156] The environmental features are matched with each sample environmental feature to determine whether there is a sample environmental feature that matches the environmental feature, so as to determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model.

[0157] In an optional manner, the processing module 505 is specifically configured to:

[0158] Calculate the characteristic distance between the environmental characteristic and each sample environmental characteristic;

[0159] Determine whether the minimum value in the feature distance is less than or equal to the preset distance to determine whether there is a sample environmental feature that matches the environmental feature.

[0160] In an optional manner, the processing module 505 is specifically configured to:

[0161] According to each preset environmental interval, environmental feature and each sample environmental feature, determine a first environmental interval corresponding to the environmental feature and a second environmental interval corresponding to each sample environmental feature;

[0162] It is determined whether there is a second environmental interval that is the same as the first environmental interval, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0163] As can be seen from the above, the chassis control device 50 provided in the embodiment of the present invention can greatly improve the accuracy and adaptability of the control strategy by combining the existing control model with the large language model for joint processing and decision-making when controlling the vehicle chassis. First, according to the vehicle driving data, an initial chassis control strategy is output using the control model, and then the large language model, the initial control strategy and the vehicle driving data are further analyzed to output the final target chassis control decision, and finally the target chassis control decision is used to control the vehicle to ensure the accuracy of the decision. This method can not only achieve precise control of the vehicle chassis system, but also provide higher safety and stability in complex and changeable driving environments.

[0164] Figure 6 The schematic diagram of the structure of an embodiment of the electronic device provided by the present invention is shown, and the specific embodiment of the present invention does not limit the specific implementation of the electronic device.

[0165] like Figure 6 As shown, the electronic device 60 may include: a processor (processor) 601 , a memory 602 , a communication interface (Communications Interface) 603 , and a communication bus 604 .

[0166] The processor 601, the communication interface 603, and the memory 602 communicate with each other via the communication bus 604. The communication interface 603 is used to communicate with other devices such as a client or other server network elements. The processor 602 is used to execute the program 605, which can specifically execute the relevant steps in the above-mentioned chassis control method embodiment.

[0167] Specifically, the program may include program code including computer executable instructions.

[0168] The processor 602 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the electronic device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0169] The memory 602 is used to store the program 605. The memory 602 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0170] The program 605 can be specifically called by the processor 602 to enable the electronic device 60 to perform the following operations:

[0171] Inputting vehicle driving data into the control model to obtain an initial chassis control decision output by the control model, wherein the vehicle driving data includes environmental data and chassis status data of the vehicle;

[0172] Inputting vehicle driving data and initial chassis control decision into the large language model to obtain the target chassis control decision output by the large language model;

[0173] The chassis of the vehicle is controlled according to the target chassis control decision.

[0174] In an optional manner, the vehicle driving data and the initial chassis control decision are input into the large language model to obtain the target chassis control decision output by the large language model, including:

[0175] Based on the environmental data, determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model;

[0176] If there is sample vehicle driving data corresponding to the current environment in the training set of the control model, the initial chassis control decision is output as the target chassis control decision.

[0177] In an optional manner, the program 605 may also be called by the processor 602 to enable the electronic device 60 to perform the following operations:

[0178] If there is no sample vehicle driving data corresponding to the current environment in the training set of the control model, the target chassis control decision is output based on the knowledge vector library, the initial chassis control decision and the vehicle driving data. The knowledge vector library stores vehicle dynamics limitation information.

[0179] In an optional manner, based on the knowledge vector library and the initial chassis control decision, a target chassis control decision is output, including:

[0180] If it is determined that the initial chassis control decision exceeds the vehicle dynamics limit according to the knowledge vector library and the vehicle driving data, the initial chassis control decision is adjusted according to the knowledge vector library to output a target chassis control decision that meets the vehicle dynamics limit;

[0181] If it is determined based on the knowledge vector library and the vehicle driving data that the initial chassis control decision does not exceed the vehicle dynamics limit, the initial chassis control decision is output as the target chassis control decision.

[0182] In an optional manner, before inputting the vehicle driving data into the control model and obtaining the initial chassis control decision output by the control model, the program 605 can also be called by the processor 602 to enable the electronic device 60 to perform the following operations:

[0183] The control model is trained according to a training set of the control model, where the training set includes a plurality of sample vehicle driving data and a vehicle state corresponding to each sample driving data.

[0184] In an optional manner, judging whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model according to the environmental data includes:

[0185] Perform feature extraction on the sample environment data of each sample vehicle driving data in the training set to obtain the sample environment features;

[0186] Extract features from environmental data to obtain environmental features;

[0187] The environmental features are matched with each sample environmental feature to determine whether there is a sample environmental feature that matches the environmental feature, so as to determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model.

[0188] In an optional manner, matching the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature that matches the environmental feature includes:

[0189] Calculate the characteristic distance between the environmental characteristic and each sample environmental characteristic;

[0190] Determine whether the minimum value in the feature distance is less than or equal to the preset distance to determine whether there is a sample environmental feature that matches the environmental feature.

[0191] In an optional manner, matching the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature that matches the environmental feature includes:

[0192] According to each preset environmental interval, environmental feature and each sample environmental feature, determine a first environmental interval corresponding to the environmental feature and a second environmental interval corresponding to each sample environmental feature;

[0193] It is determined whether there is a second environmental interval that is the same as the first environmental interval, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0194] Therefore, the electronic device 60 provided by the embodiment of the present invention takes into account that the general chassis control model is generally trained based on the commonly used automobile driving data, resulting in that the chassis control model may not be able to output an accurate chassis control solution for some extreme cases during use, while the training data scale in the large language model is large, and the control method corresponding to the vehicle driving data under various conditions can be reliably and accurately inferred, thereby improving the performance of the vehicle form and the driving experience of the driver. Therefore, when controlling the vehicle chassis, the accuracy and adaptability of the control strategy can be greatly improved by combining the existing control model with the large language model for joint processing and decision-making. First, according to the vehicle driving data, an initial chassis control strategy is output using the control model, and then the initial control strategy and the vehicle driving data are further analyzed using the large language model to output the final target chassis control decision, and finally the target chassis control decision is used to control the vehicle to ensure the accuracy and reliability of the decision, thereby improving the vehicle performance and the user's driving experience.

[0195] An embodiment of the present invention provides a computer-readable storage medium, which stores at least one executable instruction. When the executable instruction runs on the electronic device 60 / chassis control device 50, the electronic device 60 / chassis control device 50 executes the chassis control method in any of the above method embodiments.

[0196] The executable instructions may be specifically used to enable the electronic device 60 / chassis control device 50 to perform the following operations:

[0197] Inputting vehicle driving data into the control model to obtain an initial chassis control decision output by the control model, wherein the vehicle driving data includes environmental data and chassis status data of the vehicle;

[0198] Inputting vehicle driving data and initial chassis control decision into the large language model to obtain the target chassis control decision output by the large language model;

[0199] The chassis of the vehicle is controlled according to the target chassis control decision.

[0200] In an optional manner, the vehicle driving data and the initial chassis control decision are input into the large language model to obtain the target chassis control decision output by the large language model, including:

[0201] Based on the environmental data, determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model;

[0202] If there is sample vehicle driving data corresponding to the current environment in the training set of the control model, the initial chassis control decision is output as the target chassis control decision.

[0203] In an optional manner, the executable instruction may also be used to enable the electronic device 60 / chassis control device 50 to perform the following operations:

[0204] If there is no sample vehicle driving data corresponding to the current environment in the training set of the control model, the target chassis control decision is output based on the knowledge vector library, the initial chassis control decision and the vehicle driving data. The knowledge vector library stores vehicle dynamics limitation information.

[0205] In an optional manner, based on the knowledge vector library and the initial chassis control decision, a target chassis control decision is output, including:

[0206] If it is determined that the initial chassis control decision exceeds the vehicle dynamics limit according to the knowledge vector library and the vehicle driving data, the initial chassis control decision is adjusted according to the knowledge vector library to output a target chassis control decision that meets the vehicle dynamics limit;

[0207] If it is determined based on the knowledge vector library and the vehicle driving data that the initial chassis control decision does not exceed the vehicle dynamics limit, the initial chassis control decision is output as the target chassis control decision.

[0208] In an optional manner, before inputting the vehicle driving data into the control model and obtaining the initial chassis control decision output by the control model, the executable instructions may be specifically used to enable the electronic device 60 / chassis control device 50 to perform the following operations:

[0209] The control model is trained according to a training set of the control model, where the training set includes a plurality of sample vehicle driving data and a vehicle state corresponding to each sample driving data.

[0210] In an optional manner, judging whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model according to the environmental data includes:

[0211] Perform feature extraction on the sample environment data of each sample vehicle driving data in the training set to obtain the sample environment features;

[0212] Extract features from environmental data to obtain environmental features;

[0213] The environmental features are matched with each sample environmental feature to determine whether there is a sample environmental feature that matches the environmental feature, so as to determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model.

[0214] In an optional manner, matching the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature that matches the environmental feature includes:

[0215] Calculate the characteristic distance between the environmental characteristic and each sample environmental characteristic;

[0216] Determine whether the minimum value in the feature distance is less than or equal to the preset distance to determine whether there is a sample environmental feature that matches the environmental feature.

[0217] In an optional manner, matching the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature that matches the environmental feature includes:

[0218] According to each preset environmental interval, environmental feature and each sample environmental feature, determine a first environmental interval corresponding to the environmental feature and a second environmental interval corresponding to each sample environmental feature;

[0219] It is determined whether there is a second environmental interval that is the same as the first environmental interval, so as to determine whether there is a sample environmental feature that matches the environmental feature.

[0220] Therefore, in the embodiment of the present invention, considering that the reason why the existing chassis control method has low accuracy is the incompleteness of the training data, the control model may not be able to output an accurate chassis control solution in some extreme cases (for example, extremely cold weather, icy road surface, etc.), and the training data in the large language model is comprehensive and large in scale, which can well make up for the lack of data in the existing chassis control method, and can reliably and accurately infer the control method corresponding to the vehicle driving data under various circumstances, improve the performance of the vehicle form and the driving experience of the driver. Therefore, when controlling the vehicle chassis, by combining the existing control model with the large language model for joint processing and decision-making, the accuracy and adaptability of the control strategy can be greatly improved. First, according to the vehicle driving data, an initial chassis control strategy is output by using the control model, and then the initial control strategy and the vehicle driving data are further analyzed by using the large language model to output the final target chassis control decision, and finally the target chassis control decision is used to control the vehicle to ensure the accuracy of the decision. This method can not only achieve accurate control of the vehicle chassis system, but also provide higher safety and stability in a complex and changeable driving environment.

[0221] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system or other device. In addition, the embodiments of the present invention are not directed to any particular programming language.

[0222] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. Similarly, in order to simplify the present invention and help understand one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, the various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. Wherein, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself is a separate embodiment of the present invention.

[0223] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive.

[0224] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of the devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

Claims

1. A chassis control method, characterized in that: The method comprises: Inputting vehicle driving data into a control model to obtain an initial chassis control decision output by the control model, wherein the vehicle driving data includes environmental data and chassis state data of the vehicle; Inputting the vehicle driving data and the initial chassis control decision into a large language model to obtain a target chassis control decision output by the large language model; The chassis of the vehicle is controlled according to the target chassis control decision.

2. The method according to claim 1, characterized in that The step of inputting the vehicle driving data and the initial chassis control decision into a large language model to obtain a target chassis control decision output by the large language model includes: According to the environmental data, determining whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model; If sample vehicle driving data corresponding to the current environment exists in the training set of the control model, the initial chassis control decision is output as the target chassis control decision.

3. The method according to claim 1 or 2, characterized in that: The method further comprises: If there is no sample vehicle driving data corresponding to the current environment in the training set of the control model, the target chassis control decision is output based on a knowledge vector library storing vehicle dynamics limitation information, the initial chassis control decision and the vehicle driving data.

4. The method according to claim 1 or 2, characterized in that: The outputting the target chassis control decision based on the knowledge vector library, the initial chassis control decision and the vehicle driving data includes: If it is determined according to the knowledge vector library and the vehicle driving data that the initial chassis control decision exceeds the vehicle dynamics limit, the initial chassis control decision is adjusted according to the knowledge vector library to output the target chassis control decision that satisfies the vehicle dynamics limit; If it is determined according to the knowledge vector library and the vehicle driving data that the initial chassis control decision does not exceed the vehicle dynamics limit, the initial chassis control decision is output as the target chassis control decision.

5. The method according to claim 2, characterized in that: Before inputting the vehicle driving data into the control model and obtaining the initial chassis control decision output by the control model, the method includes: The control model is trained according to the training set of the control model, wherein the training set includes a plurality of sample vehicle driving data and a vehicle state corresponding to each sample driving data.

6. The method according to claim 2, characterized in that The determining, based on the environmental data, whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model comprises: Extracting features from sample environment data of each sample vehicle driving data in the training set to obtain sample environment features; Extracting features from the environmental data to obtain environmental features; The environmental feature is matched with each sample environmental feature to determine whether there is a sample environmental feature matching the environmental feature, so as to determine whether there is sample vehicle driving data corresponding to the current environment in the training set of the control model.

7. The method according to claim 6, characterized in that The matching of the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature matching the environmental feature includes: Calculating a characteristic distance between the environmental characteristic and each sample environmental characteristic; It is determined whether the minimum value of the characteristic distance is less than or equal to a preset distance, so as to determine whether there is a sample environmental feature that matches the environmental feature.

8. The method according to claim 6, characterized in that The matching of the environmental feature with each sample environmental feature to determine whether there is a sample environmental feature matching the environmental feature includes: According to each preset environmental interval, the environmental feature and each sample environmental feature, determining a first environmental interval corresponding to the environmental feature and a second environmental interval corresponding to each sample environmental feature; It is determined whether there is a second environmental interval that is the same as the first environmental interval, so as to determine whether there is a sample environmental feature that matches the environmental feature.

9. A chassis control device, characterized in that: The device comprises: An acquisition module, used for inputting vehicle driving data into a control model to acquire an initial chassis control decision output by the control model, wherein the vehicle driving data includes environmental data and chassis status data of the vehicle; An output module, used for inputting the vehicle driving data and the initial chassis control decision into a large language model, and obtaining a target chassis control decision output by the large language model; A control module is used to control the chassis of the vehicle according to the target chassis control decision.

10. An electronic device, characterized in that: include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the operation of the chassis control method as described in any one of claims 1-8.

Images