Vehicle chassis adjustment method and related apparatus

Abstract

The application discloses a vehicle chassis adjusting method and related equipment, and relates to the technical field of vehicle chassis adjusting. The method comprises the following steps: after the vehicle is started, if it is detected that a vehicle chassis adjusting function is started, collecting multi-source heterogeneous information, wherein the multi-source heterogeneous information comprises multiple items of information such as in-vehicle member related information, information about the environment in which the vehicle is located, vehicle navigation information and vehicle state information; identifying the multi-source heterogeneous information through a preset multi-modal perception model, and outputting corresponding in-cabin scene perception results and out-cabin scene perception results, wherein the in-cabin scene perception results comprise in-vehicle member identity recognition results; and dynamically outputting corresponding vehicle chassis adjusting decisions by fusing the in-cabin scene perception results and the out-cabin scene perception results. The application solves the problem that a single chassis control strategy is difficult to meet control requirements in different scenes.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a vehicle chassis adjustment method and related equipment. Background Technology

[0002] With the development of the automotive industry, vehicle comfort, handling, and safety have increasingly become the focus of consumers' attention. As a key component determining vehicle performance, the chassis system is becoming increasingly adjustable, with features such as adjustable damping shock absorbers, air suspension, variable steering ratio systems, and customizable brake pedal feel.

[0003] In current technology, drivers typically select preset chassis control strategies manually via the central control screen or physical buttons (such as selecting chassis control strategies in comfort, sport, or economy driving modes). Driver preferences are often closely related to specific driving scenarios. For example, when entering a parking garage with a steep incline, drivers may want to raise the suspension to prevent scrapes; in congested areas, they may prefer a more comfortable suspension and a more responsive automatic parking function. However, a single chassis control strategy is difficult to meet the control needs of different scenarios. Summary of the Invention

[0004] The main purpose of this application is to provide a vehicle chassis adjustment method and related equipment, which aims to solve the technical problem that a single chassis control strategy is difficult to meet the control requirements in different scenarios in the current technology.

[0005] To achieve the above objectives, this application proposes a vehicle chassis adjustment method, the vehicle chassis adjustment method comprising:

[0006] After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, multi-source heterogeneous information is collected. The multi-source heterogeneous information includes multiple items such as information related to the occupants of the vehicle, information about the environment in which the vehicle is located, vehicle navigation information, and vehicle status information.

[0007] By using a preset multimodal perception model, the multi-source heterogeneous information is identified, and corresponding in-cabin scene perception results and out-of-cabin scene perception results are output. The in-cabin scene perception results include the identification results of in-cabin occupants.

[0008] By integrating the in-cabin scene perception results and the out-of-cabin scene perception results, the corresponding vehicle chassis adjustment decision is dynamically output.

[0009] In one embodiment, the step of collecting multi-source heterogeneous information after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, includes any one of the following:

[0010] After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and the pre-stored chassis function habit information is detected, it is determined whether the first preset trigger condition is met. If it is met, multi-source heterogeneous information is collected. The first preset trigger condition includes the vehicle driving meeting the requirements and / or the driver identification result meeting the requirements.

[0011] After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and no pre-stored chassis function habit information is detected, the second preset trigger condition is determined by the wheel patrol trigger method. If the condition is met, multi-source heterogeneous information is collected.

[0012] In one embodiment, the second preset triggering condition includes scene recognition meeting requirements, wherein the scene recognition meeting requirements include the scene being a parking garage scene or the scene being a potential congestion scene.

[0013] In one embodiment, the step of identifying the multi-source heterogeneous information by using a preset multimodal perception model and outputting the corresponding in-vehicle scene perception result includes:

[0014] By using a pre-set multimodal perception model, the facial image features of the occupants and the scene state features of the cabin are determined from the multi-source heterogeneous information. The facial image features and the scene state features are spatiotemporally aligned to bind the identity information corresponding to the facial image features of the occupants with the attribute information corresponding to the scene state features.

[0015] Based on the identity and attribute information of the bound occupants, a dynamic profile of each occupant in the cabin is determined, and the dynamic profile of the occupant is used as the corresponding occupant identification result and output.

[0016] In one embodiment, the step of determining the facial image features of the vehicle occupants and the scene state features within the vehicle cabin from the multi-source heterogeneous information using a preset multimodal perception model, and then performing spatiotemporal alignment of the facial image features and the scene state features, includes:

[0017] By using the scene understanding expert module in the preset multimodal perception model, the overall features of each occupant in the vehicle are extracted from the frames of the video stream corresponding to multi-source heterogeneous information. These features serve as the structured scene state features corresponding to the occupant's cabin. The structured scene state features include one or more of the following features: the occupant's age, posture, behavior, body type, and clothing.

[0018] By using the identity expert module in the preset multimodal perception model, facial image features of occupants in the vehicle can be identified from frames of video streams corresponding to multi-source heterogeneous information.

[0019] Spatial location matching is performed on facial image features and corresponding scene state features in the same frame to achieve spatiotemporal alignment of the facial image features and the scene state features.

[0020] In one embodiment, the scene state feature corresponds to a human bounding box;

[0021] The step of matching the spatial location of facial image features and corresponding scene state features in the same frame includes:

[0022] Determine the face detection box corresponding to the facial image features in the same frame, and determine the human body detection box corresponding to the scene state features;

[0023] If the face detection box is located within the corresponding human body detection box, and if it is located within the corresponding human body detection box, and the overlap between the face detection box and the human body detection box is greater than a preset confidence threshold, then it is determined that the facial image features and the corresponding scene state features in the same frame are spatially matched.

[0024] In one embodiment, the step of identifying the multi-source heterogeneous information by using a preset multimodal perception model and outputting the corresponding perception result of the external scene of the vehicle cabin includes:

[0025] By using a preset multimodal perception model, information about the vehicle's environment, vehicle navigation information, and vehicle status information are determined from the multi-source heterogeneous information.

[0026] Based on the information of the vehicle's environment, vehicle navigation information, and vehicle status information, a set of semantic tags for the corresponding external scene is determined, and the set of semantic tags for the external scene is used as the perception result for the corresponding external scene. The information of the vehicle's environment includes one or more of the following: weather type, road type, road features, driving area, and vehicle events. The vehicle navigation information includes one or more of the following: vehicle latitude and longitude coordinates, heading, speed, congestion level on the navigation path, and distance information on the navigation path. The vehicle status information includes one or more of the following: vehicle dynamic parameters, chassis operation, and chassis status information.

[0027] In one embodiment, the step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to output the corresponding vehicle chassis adjustment decision includes:

[0028] Based on the personalized chassis preference submodule in the preset multimodal perception model, the personalized chassis parameter preferences of the in-cabin scene perception results and the out-of-cabin scene perception results are predicted to obtain the personalized chassis preference prediction results. The personalized chassis preference submodule is a pre-trained objective function.

[0029] The personalized chassis preference prediction results are output as the corresponding vehicle chassis adjustment decisions.

[0030] In one embodiment, the step of predicting the personalized chassis parameter preferences of vehicle occupants based on the in-cabin scene perception results and the out-of-cabin scene perception results using the personalized chassis preference submodule in the preset multimodal perception model, to obtain the target chassis preference prediction result, includes:

[0031] If multiple occupants are detected in the cabin, the personalized chassis parameter preferences for each occupant are predicted based on the personalized chassis preference submodule in the preset multimodal perception model, using the cabin scene perception results and the cabin exterior scene perception results.

[0032] If there is a conflict between the personalized chassis preference prediction results of different members, a hierarchical priority arbitration mechanism will be used to arbitrate the personalized chassis preference prediction results of different members, and the arbitration result will be used as the target chassis preference prediction result.

[0033] In one embodiment, the preset hierarchical structure includes a security layer and other layers. The step of arbitrating the personalized chassis preference prediction results of different members through a priority arbitration mechanism of the preset hierarchical structure, and using the arbitration result as the target chassis preference prediction result, includes:

[0034] If there is a conflict between the personalized chassis preference prediction results of different members, the first personalized chassis preference prediction result of the member that does not meet the safety layer shall be removed, and the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result shall be used as the target chassis preference prediction result.

[0035] In one embodiment, the other layers include one or more of a special member care layer, a weighted fusion layer, and a corresponding learning feedback layer among members. The step of using the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result as the target chassis preference prediction result includes:

[0036] Determine the target stratum of the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result;

[0037] Based on the target layer in which it is located, the corresponding arbitration weight is determined, wherein the arbitration weight is dynamically changing;

[0038] Based on other personalized chassis preference prediction results and corresponding arbitration weights, the target chassis preference prediction result is determined.

[0039] In one embodiment, after the step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to dynamically output the corresponding vehicle chassis adjustment decision, the following steps are included:

[0040] If an adjustment to the vehicle chassis adjustment decision is detected by the driver among the members, the preset multimodal perception model is updated based on the adjusted vehicle chassis adjustment decision, so as to make the next decision based on the updated preset multimodal perception model.

[0041] In one embodiment, the adjustment includes adjustments to the vehicle chassis adjustment decision and / or adaptive adjustments to the arbitration weights.

[0042] In one embodiment, after the step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to dynamically output the corresponding vehicle chassis adjustment decision, the following steps are included:

[0043] Determine the smoothing parameters corresponding to the vehicle chassis adjustment decisions;

[0044] Based on the smoothing parameters and the vehicle chassis adjustment decision, the vehicle chassis is adjusted.

[0045] Furthermore, to achieve the above objectives, this application also proposes a vehicle chassis adjustment device, which includes:

[0046] The data acquisition module is used to collect multi-source heterogeneous information after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated. The multi-source heterogeneous information includes multiple items such as information related to the occupants of the vehicle, information about the environment in which the vehicle is located, vehicle navigation information, and vehicle status information.

[0047] The identification module is used to identify the multi-source heterogeneous information through a preset multimodal perception model and output the corresponding in-cabin scene perception results and out-of-cabin scene perception results, wherein the in-cabin scene perception results include the identification results of the in-cabin occupants.

[0048] The fusion module is used to fuse the in-cabin scene perception results and the out-of-cabin scene perception results, and dynamically output the corresponding vehicle chassis adjustment decisions.

[0049] In addition, to achieve the above objectives, this application also proposes a vehicle chassis adjustment device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the vehicle chassis adjustment steps as described above.

[0050] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the vehicle chassis adjustment steps as described above.

[0051] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the vehicle chassis adjustment steps described above.

[0052] One or more technical solutions proposed in this application have at least the following technical effects:

[0053] Compared to related technologies where a single chassis control strategy is insufficient to meet control requirements in different scenarios, this application addresses this issue by collecting multi-source heterogeneous information after vehicle startup if the vehicle chassis adjustment function is detected as activated. This multi-source heterogeneous information includes multiple items such as information related to vehicle occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information. A preset multimodal perception model is used to identify this multi-source heterogeneous information, outputting corresponding in-cabin scene perception results and out-of-cabin scene perception results. The in-cabin scene perception results include occupant identification results. The in-cabin scene perception results and out-of-cabin scene perception results are then fused to dynamically output corresponding vehicle chassis adjustment decisions. It is understood that in this application, after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, the vehicle cabin scene perception result and the external scene perception result are determined based on multi-source heterogeneous information including information about the occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information. Then, based on the vehicle cabin scene perception result and the external scene perception result, the chassis adjustment decision is dynamically determined instead of using a single chassis control strategy in all scenarios. Therefore, the personalized chassis control needs of users in different scenarios are met. Attached Figure Description

[0054] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0055] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a first flowchart illustrating the vehicle chassis adjustment method of this application.

[0057] Figure 2 This is a flowchart illustrating the entire process of the vehicle chassis adjustment method described in this application.

[0058] Figure 3 This is a second process diagram provided for Embodiment 1 of the vehicle chassis adjustment method of this application;

[0059] Figure 4 This is a flowchart illustrating Embodiment 2 of the vehicle chassis adjustment method of this application.

[0060] Figure 5 This is a schematic diagram of the module structure of the vehicle chassis adjustment device according to an embodiment of this application;

[0061] Figure 6 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the vehicle chassis adjustment method in this application embodiment.

[0062] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0063] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0064] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0065] The main solution of this application embodiment is: after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, multi-source heterogeneous information is collected, wherein the multi-source heterogeneous information includes multiple items such as information related to the occupants of the vehicle, information about the environment in which the vehicle is located, vehicle navigation information, and vehicle status information.

[0066] By using a preset multimodal perception model, the multi-source heterogeneous information is identified, and corresponding in-cabin scene perception results and out-of-cabin scene perception results are output. The in-cabin scene perception results include the identification results of in-cabin occupants.

[0067] By integrating the in-cabin scene perception results and the out-of-cabin scene perception results, the corresponding vehicle chassis adjustment decision is dynamically output.

[0068] Compared to related technologies where a single chassis control strategy is insufficient to meet control requirements in different scenarios, this application addresses this issue by collecting multi-source heterogeneous information after vehicle startup if the vehicle chassis adjustment function is detected as activated. This multi-source heterogeneous information includes multiple items such as information related to vehicle occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information. A preset multimodal perception model is used to identify this multi-source heterogeneous information, outputting corresponding in-cabin scene perception results and out-of-cabin scene perception results. The in-cabin scene perception results include occupant identification results. The in-cabin scene perception results and out-of-cabin scene perception results are then fused to dynamically output corresponding vehicle chassis adjustment decisions. It is understood that in this application, after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, the vehicle cabin scene perception result and the external scene perception result are determined based on multi-source heterogeneous information including information about the occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information. Then, based on the vehicle cabin scene perception result and the external scene perception result, the chassis adjustment decision is dynamically determined instead of using a single chassis control strategy in all scenarios. Therefore, the personalized chassis control needs of users in different scenarios are met.

[0069] It should be noted that the executing entity in this embodiment can be a computing service device with vehicle chassis adjustment, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or vehicle chassis adjustment device capable of performing the above functions. The following description uses a vehicle chassis adjustment device as an example to illustrate this embodiment and the subsequent embodiments.

[0070] Based on this, the present application provides a vehicle chassis adjustment method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the vehicle chassis adjustment method of this application.

[0071] In this embodiment, the vehicle chassis adjustment method includes steps S10 to S30:

[0072] Step S10: After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, multi-source heterogeneous information is collected. The multi-source heterogeneous information includes multiple items such as information related to the occupants of the vehicle, information about the environment in which the vehicle is located, vehicle navigation information, and vehicle status information.

[0073] In this embodiment, it should be noted that the executing entity is a vehicle chassis adjustment device, which belongs to a vehicle chassis adjustment system, and the vehicle chassis adjustment system can specifically be an autonomous driving system.

[0074] In this embodiment, a specific application scenario may be:

[0075] In current technology, drivers typically select preset chassis control strategies manually via the central control screen or physical buttons (such as selecting chassis control strategies in comfort, sport, or economy driving modes). Driver preferences are often closely related to specific driving scenarios. For example, when entering a parking garage with a steep incline, drivers may want to raise the suspension to prevent scrapes; in congested areas, they may prefer a more comfortable suspension and a more responsive automatic parking function. However, a single chassis control strategy is difficult to meet the control needs of different scenarios.

[0076] In this embodiment, if the vehicle chassis adjustment function is not detected to be activated after the vehicle is started, the subsequent steps will not be executed.

[0077] In this embodiment, after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, multi-source heterogeneous information is collected. Specifically, the multi-source heterogeneous information may include video data or point cloud data collected by sensors. In addition, the multi-source heterogeneous information may also include text data.

[0078] In this embodiment, the multi-source heterogeneous information includes structured data or unstructured data.

[0079] In this embodiment, the multi-source heterogeneous information includes multiple items such as information related to vehicle occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information.

[0080] For example, the multi-source heterogeneous information may include information related to vehicle occupants, information about the vehicle's environment, and vehicle navigation information; or the multi-source heterogeneous information may include information related to vehicle occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information, etc., without any specific limitation.

[0081] In this embodiment, information related to the occupants includes the number of occupants, their seating posture, and their ages. Information about the vehicle's environment includes the weather, the type of road (e.g., urban roads, rural roads, highways), the road characteristics (e.g., asphalt, cement, dirt roads), the road surface condition, the scenario (e.g., heavy traffic, traffic accidents, construction zones, approaching emergency vehicles), and the driving area.

[0082] It should be noted that all actions involving the acquisition of signals, information, or data in this application are carried out in compliance with the relevant data protection laws and policies of the country where the application is located, and with the authorization granted by the owner of the relevant device.

[0083] In this embodiment, the vehicle navigation information includes the real-time latitude and longitude coordinates (POI information, Point of Interest) of the vehicle, heading, speed, and prior information such as congestion level and distance on the navigation path.

[0084] In this embodiment, the vehicle status information includes vehicle speed, door status, steering wheel angle, acceleration and deceleration, and other vehicle dynamic parameters, as well as corresponding chassis operation and status signals.

[0085] In this embodiment, after collecting multi-source heterogeneous information, the multi-source heterogeneous information can be structured to accelerate the efficiency of subsequent processing.

[0086] In this embodiment, the step of collecting multi-source heterogeneous information after the vehicle is started if the vehicle chassis adjustment function is detected to be activated includes any one of the following:

[0087] Step S11: After the vehicle is started, if the vehicle chassis adjustment function is detected to be turned on and the pre-stored chassis function habit information is detected, it is determined whether the first preset trigger condition is met. If it is met, multi-source heterogeneous information is collected. The first preset trigger condition includes the vehicle driving meeting the requirements and / or the driver identification result meeting the requirements.

[0088] In this embodiment, as Figure 2 As shown, one of the branches is:

[0089] After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and the pre-stored chassis function habit information is detected (it should be noted that there are multiple chassis function habit information, which are adaptive chassis function habit information formed in different scenarios), then it is further determined whether the first preset trigger condition is met. If it is met, multi-source heterogeneous information is collected. If it is not met, multi-source heterogeneous information is not collected. For example, if the vehicle is stationary, multi-source heterogeneous information is not collected.

[0090] In this embodiment, if the first preset triggering condition is met, multi-source heterogeneous information is collected.

[0091] The first preset trigger condition includes the vehicle driving conditions meeting the requirements, such as a vehicle speed ≥ 7 kph, and / or the first preset trigger condition includes the driver identification result meeting the requirements, wherein the driver identification result meeting the requirements includes the driver being a member corresponding to the adaptive chassis function habit information, and the driver being in the driver's seat from the time the door is opened to the time it is closed.

[0092] It is understood that in this embodiment, when the vehicle is not stationary (e.g., after starting the vehicle), it is necessary to identify the specific driver's identity, and the driver must be a member corresponding to the adaptive chassis function habit information before multi-source heterogeneous information is collected. This avoids the current technology where the vehicle only adjusts the chassis strategy based on the memory function of the account or key (mainly driving habit memory), which makes it impossible to distinguish different drivers under the same account, thus affecting the driver's driving experience. For example, in the current technology, when family members share an account, it is impossible to provide each driver with their own exclusive driving habit memory, but directly use the driving memory corresponding to the same account, making it difficult for different members to make corresponding personalized vehicle chassis adjustments. That is, if there are drivers a and b under the same account, then if the adaptive chassis function habit information corresponds to driver a, and the current driver's seat is driver a (determined by biometrics such as facial recognition), then in this embodiment, multi-source heterogeneous information is collected.

[0093] In other words, in this embodiment, a driver identification locking mechanism based on biometrics is adopted (the condition is triggered only if the driver identification result meets the requirements), which solves the problem of personalization conflict under shared accounts.

[0094] In this embodiment, under specific conditions of vehicle operation (such as vehicle speed exceeding a threshold), the driver's identity is locked by using facial recognition technology decoupled from the account, and then multi-source heterogeneous information is collected, thus solving the problem of personalized conflicts under shared accounts.

[0095] Step S12: After the vehicle is started, if the vehicle chassis adjustment function is detected to be turned on and no pre-stored chassis function habit information is detected, the second preset trigger condition is determined by the wheel patrol trigger method. If it is met, multi-source heterogeneous information is collected.

[0096] In this embodiment, another branch is also provided, such as Figure 2 As shown, specifically, after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and no pre-stored chassis function habit information is detected, that is, if the driver has turned on the vehicle adaptive chassis function but has not saved the adaptive chassis function habit, the first vehicle chassis adaptive adjustment is triggered. Specifically, the second preset trigger condition is determined by the wheel-circling trigger method. If it is met, multi-source heterogeneous information is collected.

[0097] Specifically, the polling triggering method can be as follows: by calling the cameras inside and outside the cabin within a fixed time period, it is determined whether the second preset triggering condition is detected. If the second preset triggering condition is not detected, the camera calls are triggered in a loop until the second preset triggering condition is detected or the 10-minute period expires and the camera is called again.

[0098] The second preset triggering condition includes scene recognition meeting the requirements. The scene recognition meeting the requirements includes the scene being a parking garage scene or the scene being a potential congestion scene. It should be noted that the scene recognition meeting the requirements can also be other situations, which are not specifically limited.

[0099] For example, when the vehicle's current GPS coordinates are detected to have entered the electronic fence of a known underground parking POI (e.g., an area with a radius of 80 meters), or when the navigation destination is an underground parking garage, a preliminary trigger signal for a "potential underground parking garage scenario" is generated. At this time, it is determined that the second preset trigger condition is met. Alternatively, when the navigation system announces that the congestion level of the road segment less than 1km ahead is severe (e.g., Level 2 or 3), a preliminary trigger signal for a "potential congestion scenario" is generated. At this time, it is determined that the second preset trigger condition is met.

[0100] Step S20: The multi-source heterogeneous information is identified by a preset multimodal perception model, and the corresponding in-cabin scene perception results and out-of-cabin scene perception results are output. The in-cabin scene perception results include the in-cabin occupant identification results.

[0101] In this embodiment, a preset multimodal perception model is used to perceive and identify the multi-source heterogeneous information, and output the corresponding in-cabin scene perception results and out-of-cabin scene perception results. The in-cabin scene perception results include the identification results of the in-cabin occupants.

[0102] Specifically, in this embodiment, sensors inside the vehicle, such as a wide-angle infrared camera (covering the entire cabin) and a narrow-angle camera for each seat, are used to determine the perception results of the corresponding cabin scene.

[0103] In this embodiment, a wide-angle camera deployed on the exterior of the vehicle (such as the windshield, bumper, etc.) captures the environmental video stream around the vehicle in real time, thereby obtaining the scene perception results outside the vehicle cabin.

[0104] In this embodiment, the in-cabin scene perception result includes the in-cabin occupant identification result. Specifically, the in-cabin image video stream can be acquired, and the identity ID of the in-cabin occupant can be obtained using a face recognition algorithm (such as FaceNet or ArcFace), thereby determining the in-cabin occupant identification result.

[0105] In this embodiment, the in-cabin scene perception results also include other scene state information such as the location, gender, age, and intention of the in-cabin occupants.

[0106] In this embodiment, the cabin scene perception results also include seat-related data, which includes whether there are occupants in the seats, and can help determine the occupants' body shape and sitting posture based on the fine changes in seat pressure distribution.

[0107] In this embodiment, the in-cabin scene perception results may also include vehicle data, etc.

[0108] Specifically, refer to Figure 3 The step of identifying the multi-source heterogeneous information through a preset multimodal perception model and outputting the corresponding in-vehicle cabin scene perception result includes:

[0109] Step S21: By using a preset multimodal perception model, determine the facial image features of the occupants and the scene state features of the cabin from the multi-source heterogeneous information, and perform spatiotemporal alignment on the facial image features and the scene state features to bind the identity information corresponding to the facial image features of the occupants and the attribute information corresponding to the scene state features.

[0110] In this embodiment, by using a preset multimodal perception model, the facial image features of the vehicle occupants and the scene state features within the vehicle cabin are determined from the multi-source heterogeneous information. The spatiotemporal alignment of the facial image features and the scene state features includes:

[0111] Method 1: By using a preset multimodal perception model, the facial image features of the occupants and the scene state features of the cabin are determined from the multi-source heterogeneous information, with frames at the same timestamp as the unit, and the facial image features and the scene state features are spatiotemporally aligned.

[0112] Method 2: The step of determining the facial image features of the vehicle occupants and the scene state features within the vehicle cabin from the multi-source heterogeneous information using a preset multimodal perception model, and then performing spatiotemporal alignment of the facial image features and the scene state features, includes:

[0113] Step A1: Using the scene understanding expert module in the preset multimodal perception model, extract the overall features of each occupant in the vehicle from the frames of the video stream corresponding to the multi-source heterogeneous information, as the structured scene state features corresponding to the occupant's cabin. The structured scene state features include one or more of the occupant's age, posture, behavior state, body type, and clothing.

[0114] Step A2: Using the identity expert module in the preset multimodal perception model, facial image features of the occupants in the vehicle are identified from the frames of the video stream corresponding to multi-source heterogeneous information.

[0115] Step A3: Spatial location matching is performed on facial image features and corresponding scene state features in the same frame to achieve spatiotemporal alignment of the facial image features and the scene state features.

[0116] In this embodiment, it should be noted that the preset multimodal perception model adopts a parallel perception strategy, that is, it runs two types of AI modules simultaneously to perform spatiotemporal alignment of the facial image features and the scene state features. One type of AI module is an "identity expert" module dedicated to face recognition, which is responsible for quickly identifying which registered user (e.g., Mr. Zhang, ID: 001) is in the camera image. The other type of AI module is a "scene understanding expert" module (e.g., in-cabin VLM), which is responsible for analyzing the overall characteristics of each person in the image, such as determining their age group (adult, infant), their sitting position, and their current state (driving, sleeping).

[0117] In this embodiment, the facial image features of the occupants in the vehicle are identified from the frames of the video stream corresponding to multi-source heterogeneous information by using the identity expert module in the preset multimodal perception model. The identity expert module is a pre-trained recognition network.

[0118] In this embodiment, the scene understanding expert module in the preset multimodal perception model is used to extract the overall features of each occupant in the vehicle from the frames of the video stream corresponding to the multi-source heterogeneous information. These features are used as the structured scene state features corresponding to the occupant's cabin. The structured scene state features include one or more of the occupant's age, posture, behavior, body type, and clothing.

[0119] In this embodiment, specifically, when the preset multimodal perception model is a VLM (Visual Language Model) model, the input data received is multi-source heterogeneous information, wherein the multi-source heterogeneous information includes real-time video streams. and vehicle data Please provide a detailed explanation.

[0120] That is, the VLM model takes input data (real-time video stream) as input. and vehicle data The VLM model is processed (where the functionality of the VLM model can be abstracted into a function). The corresponding output is a structured scene state vector, i.e., scene state features. .

[0121]

[0122] The scene state vector or scene state features Includes all passengers in the vehicle Detailed description:

[0123]

[0124] Each passenger A state can be represented as a set of key-value pairs:

[0125]

[0126] in, A unique tracking identifier for a passenger; Indicates location, accurate to the seat, such as RearLeft (left rear seat), RearRight (right rear seat), RearMiddle (middle rear seat).

[0127] It indicates age (Age Class), which is further divided into Infant, Child, Adult, and Senior.

[0128] Indicates posture, such as Sitting Upright, Standing On Seat, Lying Down, Bending Over, etc.

[0129] It indicates the behavior state, dynamically describing what the passenger is doing, such as Sleeping, Playing With Toy, Using Electronics, Reaching For Door;

[0130] It indicates clothing and identifies the type of clothing worn by passengers, such as SummerWear, WinterCoat, and Standard. It can be used to help determine the comfort level inside the vehicle or whether there are any safety hazards (such as clothing obstructing the seat belt).

[0131] In this embodiment, spatial location matching is also performed on facial image features and corresponding scene state features in the same frame to achieve spatiotemporal alignment of the facial image features and the scene state features. Specific alignment methods include:

[0132] Method 1: If there is spatial overlap between the two, it is determined that the facial image features and the corresponding scene state features in the same frame are matched in spatial location.

[0133] Method 2: The scene state features correspond to human bounding boxes;

[0134] The step of matching the spatial location of facial image features and corresponding scene state features in the same frame includes:

[0135] Step A31: Determine the face detection box corresponding to the facial image features in the same frame, and determine the human body detection box corresponding to the scene state features;

[0136] Step A32: Determine whether the face detection box is located within the corresponding human body detection box. If it is located within the corresponding human body detection box, and the overlap between the face detection box and the human body detection box is greater than a preset confidence threshold, then it is determined that the facial image features and the corresponding scene state features in the same frame are spatially matched.

[0137] In this embodiment, the key to matching lies in "spatiotemporal alignment." Temporally, the vehicle chassis adjustment device ensures that both identity and attribute information originate from the same image frame, guaranteeing data consistency. Spatially, the vehicle chassis adjustment device performs position matching, using an algorithm to determine whether a "face detection box" in the image is located within a certain "human detection box." Once it is confirmed that the face box and the human box highly overlap in space (overlap greater than a preset confidence threshold), and that this location corresponds to a specific seat within the vehicle, the identity and attribute information can be firmly and physically bound together.

[0138] Specifically, the process of spatiotemporally aligning the facial image features and the scene state features to bind the identity information corresponding to the facial image features of the vehicle occupants with the attribute information corresponding to the scene state features can be as follows (wherein, the vehicle chassis adjustment device acquires the identity information and attribute information through parallel sensing):

[0139] That is, for frames with the same timestamp t (i.e., the same frame), the vehicle chassis adjustment device can obtain the identity object. (Identity information) and output attribute object (Attribute information), where the identity object is output through the face recognition module (or identity expert module) in the VLM model. The visual attribute module (or scene understanding expert module) in the VLM model outputs attribute objects. :

[0140]

[0141]

[0142] in, Including user ID, For face detection bounding boxes, Includes {age, location, status, ...}, This is the human body detection frame.

[0143] Then, it is determined whether the identity and attribute information corresponding to the same frame can be spatio-temporal bound. That is, spatio-temporal binding is binding the identity and attribute information of the occupants in the vehicle. It should be noted that only when the two objects (identity information and attribute information) are highly consistent in both time and space are they considered to describe the same person. The binding condition can be expressed as:

[0144]

[0145] in, For binding, That is, consistent in time. It is a preset confidence threshold (e.g., 0.9). Its calculation formula is:

[0146]

[0147] Among them, once Once established, the identity information and corresponding attribute information of the occupants in the vehicle are successfully bound together, generating a merged occupant profile. This profile unifies the identity and attribute information for subsequent decision-making.

[0148]

[0149] This file is Clearly linking specific occupants to their precise status within the vehicle provides accurate input for personalized adjustments to the intelligent chassis.

[0150] Step S22: Based on the identity and attribute information of the bound occupants, determine the dynamic profile of each occupant in the cabin, and output the dynamic profile of the occupants as the corresponding occupant identity recognition result.

[0151] In this embodiment, the vehicle chassis adjustment device generates a complete and unified dynamic profile for each occupant in the vehicle. For example, a profile might be: (User: Mr. Zhang, ID: 001, Location: Driver's seat, Age: Adult, Status: Attentive Driving). This comprehensive information, including identity, location, and status, is directly submitted to other subsequent processing modules, thus providing the most reliable basis for the adaptive adjustment of the intelligent chassis.

[0152] The step of identifying the multi-source heterogeneous information through a preset multimodal perception model and outputting the corresponding perception result of the external scene of the vehicle cabin includes:

[0153] Step B1: By using a preset multimodal perception model, determine the information of the vehicle's environment, vehicle navigation information, and vehicle status information from the multi-source heterogeneous information.

[0154] Step B2: Based on the information of the vehicle's environment, vehicle navigation information, and vehicle status information, determine the corresponding set of semantic tags for the external scene of the vehicle cabin, and use the set of semantic tags for the external scene of the vehicle cabin as the perception result of the corresponding external scene of the vehicle cabin; wherein, the information of the vehicle's environment includes one or more of the following: weather type, road type, road features, driving area, and vehicle events; the vehicle navigation information includes one or more of the following: vehicle latitude and longitude coordinates, heading, speed, congestion level on the navigation path, and distance information on the navigation path; the vehicle status information includes one or more of the following: vehicle dynamic parameters, chassis operation, and chassis status information.

[0155] In this embodiment, by using a preset multimodal perception model, information about the vehicle's environment, vehicle navigation information, and vehicle status information are determined from the multi-source heterogeneous information. Based on the information about the vehicle's environment, vehicle navigation information, and vehicle status information, a corresponding set of semantic tags for the external scene is determined. This set of semantic tags for the external scene is used as the corresponding perception result for the external scene. Specifically, in this embodiment, the external visual perception module in the preset multimodal perception model acquires the environmental video stream around the vehicle, and then performs in-depth analysis on the external visual information to output a set of scene semantic tags S with a preset granularity.

[0156] In this embodiment, specifically, a wide-angle camera deployed on the exterior of the vehicle (such as the windshield, bumper, etc.) is used to capture real-time video streams of the environment surrounding the vehicle.

[0157] Among them, a pre-set multimodal perception model, namely the VLM model, is used to perform in-depth analysis on multi-source heterogeneous information, namely external visual information (environmental video stream), and output a set of scene semantic tags S with corresponding granularity.

[0158] The semantic tag set for the exterior scene of the vehicle includes information about the vehicle's environment, S:

[0159]

[0160] in, It can handle sunny days, rainy days, snowfall, dense fog...; Road types include: urban roads, rural roads, highways, etc. Road characteristics include: asphalt, cement, dirt roads, etc. Road surface conditions include: dry, wet, icy / snowy, pothole severity (H / M / L), speed bumps, etc. It can be used for situations with high traffic volume (traffic jams), traffic accidents, construction areas, and when emergency vehicles are approaching. The driving area can be the entrance / exit of an underground parking garage, in front of a tunnel, the toll station area, the steep slope descent area, etc.

[0161] The semantic tag set for the external scene of the vehicle cabin also includes vehicle navigation information, which includes the vehicle's real-time latitude and longitude coordinates (POI information), heading, speed, and prior information such as congestion level and distance on the navigation path.

[0162] The semantic tag set for the external scene of the vehicle cabin also includes vehicle status information. Specifically, in this embodiment, vehicle dynamic parameters such as vehicle speed, door status, steering wheel angle, acceleration and deceleration are obtained through the CAN bus, and chassis operation and chassis status signals are collected.

[0163] Step S30: Combine the in-cabin scene perception results and the out-of-cabin scene perception results to dynamically output the corresponding vehicle chassis adjustment decision.

[0164] In this embodiment, after obtaining the in-cabin scene perception results and the out-of-cabin scene perception results, the in-cabin scene perception results and the out-of-cabin scene perception results are fused, and the corresponding vehicle chassis adjustment decision is dynamically output.

[0165] Specifically, the fusion of the in-cabin scene perception results and the out-of-cabin scene perception results includes weighted fusion or other preset fusion methods.

[0166] In this embodiment, the dynamic output of the corresponding vehicle chassis adjustment decision includes: function operation Action ( ) and the corresponding parameter values ​​( ).

[0167] Specifically, vehicle chassis adjustment decisions can include using softer damping or activating the magic carpet suspension.

[0168] Compared to related technologies where a single chassis control strategy is insufficient to meet control requirements in different scenarios, this application addresses this issue by collecting multi-source heterogeneous information after vehicle startup if the vehicle chassis adjustment function is detected as activated. This multi-source heterogeneous information includes multiple items such as information related to vehicle occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information. A preset multimodal perception model is used to identify this multi-source heterogeneous information, outputting corresponding in-cabin scene perception results and out-of-cabin scene perception results. The in-cabin scene perception results include occupant identification results. The in-cabin scene perception results and out-of-cabin scene perception results are then fused to dynamically output corresponding vehicle chassis adjustment decisions. It is understood that in this application, after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, the vehicle cabin scene perception result and the external scene perception result are determined based on multi-source heterogeneous information including information about the occupants, information about the vehicle's environment, vehicle navigation information, and vehicle status information. Then, based on the vehicle cabin scene perception result and the external scene perception result, the chassis adjustment decision is dynamically determined instead of using a single chassis control strategy in all scenarios. Therefore, the personalized chassis control needs of users in different scenarios are met.

[0169] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 4 The step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to output the corresponding vehicle chassis adjustment decision includes:

[0170] Step C1: Based on the personalized chassis preference submodule in the preset multimodal perception model, predict the personalized chassis parameter preferences of the in-cabin scene perception results and the out-of-cabin scene perception results to obtain personalized chassis preference prediction results. The personalized chassis preference submodule is a pre-trained objective function.

[0171] In this embodiment, based on the personalized chassis preference submodule in the preset multimodal perception model, the personalized chassis parameter preferences of the occupants are predicted based on the in-cabin scene perception results and the out-of-cabin scene perception results, thus obtaining the personalized chassis preference prediction result. Specifically, the personalized chassis preference submodule can be a constructed and updated driving habit mapping model:

[0172] In this embodiment, the driving habit mapping model can perform corresponding learning, arbitration, and decision-making based on fused multi-source heterogeneous information. Specifically, the personalized chassis preference sub-model in the driving habit mapping model... It is used to learn and predict the ideal chassis parameters of member k in a specific scenario S (or scenario state features). .

[0173] Specifically, the data quaternion format recorded by the corresponding vehicle chassis adjustment device is as follows: .in, For the member's ID number, These are the control parameters for operation. This represents the corresponding probability.

[0174] Specifically, based on the driving habit mapping model, or the personalized chassis preference sub-model within the driving habit mapping model. Using sequence models based on Transformer or DNN, corresponding personalized habit mapping functions are constructed. (Depend on Parameterization):

[0175]

[0176] in It is a chassis parameter vector, including suspension height ( ), damping coefficient ( ), convenient for getting on and off the vehicle () ), steep slope descent state ( ) wait.

[0177] In this embodiment, the corresponding personalized habit mapping function is constructed. (Depend on When parameterizing, the training objective is to minimize the error between the predicted preference parameters and the member's historical operating parameters. :

[0178] ;

[0179] Step C2: Output the personalized chassis preference prediction results as the corresponding vehicle chassis adjustment decision.

[0180] In this embodiment, after obtaining the personalized chassis preference prediction result, the personalized chassis preference prediction result (such as...) is... The output is the corresponding vehicle chassis adjustment decision.

[0181] In this embodiment, it should be noted that there is a personalized chassis preference prediction result for each member. That is, when there are 4 members in the vehicle, there can be 4 personalized chassis preference prediction results.

[0182] The step of predicting the personalized chassis preference of vehicle occupants based on the in-cabin scene perception results and the out-of-cabin scene perception results, using the personalized chassis preference submodule in the preset multimodal perception model, to obtain the target chassis preference prediction result, includes:

[0183] Step D1: If multiple occupants are detected in the vehicle cabin, based on the personalized chassis preference submodule in the preset multimodal perception model, predict the personalized chassis parameter preferences of each occupant based on the perception results of the in-cabin scene and the perception results of the out-of-cabin scene.

[0184] Step D2: If there is a conflict between the personalized chassis preference prediction results of different members, the personalized chassis preference prediction results of different members are arbitrated through a preset hierarchical priority arbitration mechanism, and the arbitration result is used as the target chassis preference prediction result.

[0185] Specifically, in this embodiment, when a specific scene is identified... There are multiple members, and their individual chassis preference sets. When conflicts arise, this embodiment activates a multi-user preference conflict decision unit (CDU) to make a conflict decision. The CDU aims to output the optimal set of final chassis control parameters in real time through a hierarchical priority arbitration mechanism. To balance security, regulatory compliance, special member care, and personalized user experience.

[0186] Specifically, in this embodiment, the process of arbitrating the prediction results of different members' personalized chassis preferences through a preset hierarchical priority arbitration mechanism can be as follows:

[0187] Enter the current scene context: ,in, It is the semantic information of the external scene (the perception result of the external scene of the vehicle cabin). Semantic information of the cabin scene (cabin scene perception results).

[0188] The set of member identity information and corresponding scene state features is as follows:

[0189] The set of personalized chassis preference prediction results is as follows: .

[0190] After obtaining the set, the corresponding personalized chassis preference prediction results of different members are arbitrated through a pre-set hierarchical priority arbitration mechanism, and the arbitration result is used as the target chassis preference prediction result.

[0191] Specifically, the preset hierarchical structure includes a security layer and other layers. The step of arbitrating the personalized chassis preference prediction results of different members through a priority arbitration mechanism of the preset hierarchical structure, and using the arbitration result as the target chassis preference prediction result, includes:

[0192] Step D1: If there is a conflict between the personalized chassis preference prediction results of different members, the first personalized chassis preference prediction result of the member that does not meet the safety layer shall be removed, and the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result shall be used as the target chassis preference prediction result.

[0193] Specifically, in this embodiment, the preset hierarchical level includes a security layer. It should be noted that the first layer with the highest priority in the preset hierarchical level is either the security layer or the safety and regulation layer.

[0194] The safety layer, or safety and regulations layer, is a hard constraint layer with the highest veto power. Under all circumstances, the final control parameters of the chassis... It is essential to ensure that the vehicle remains within a safe and legally permissible working envelope. For example, if the vehicle's chassis adjustment mechanism estimates road friction... Forward path curvature estimation (PCLE) indicates a potential risk of loss of control or an impending violation of regulatory limits. In such cases, the personalized chassis preference prediction result corresponding to that layer is not executed. Specifically, the decision formula for the safety layer is: ;

[0195] In this embodiment, the other layers include one or more of the following: a special member care layer, a weighted fusion layer, and a corresponding learning feedback layer among members. The step of using the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result as the target chassis preference prediction result includes:

[0196] Step E1: Determine the target stratum of the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result;

[0197] Step E2: Based on the target layer, determine the corresponding arbitration weight, wherein the arbitration weight is dynamically changing;

[0198] Step E3: Based on other personalized chassis preference prediction results and corresponding arbitration weights, determine the target chassis preference prediction result.

[0199] In this embodiment, the other layers include a second layer: the Vulnerable PassengerCare Layer.

[0200] In other words, if the first layer does not trigger the safety rules, the vehicle chassis adjustment system checks for the presence of a Vulnerable Passenger (VP) inside the vehicle, such as... or .

[0201] In this embodiment, the special member care layer provides care for special members ( The preference assigns a relatively higher priority weight factor (arbitration weight) to influence subsequent weighted fusion. This weight factor It is based on member k. (Age) and (State) mapping generation ensures special members ( Comfort preference is weighted with a relatively high gain.

[0202] Specifically, an example of defining the weighting factor (arbitration weight) for special members is as follows:

[0203]

[0204] in, It is a preset value much greater than 1, used to amplify the preference influence or arbitration weight of special member VPs in the next layer, for example, .

[0205] In this embodiment, the other layers may also include a third layer: a weighted fusion and neutralization layer.

[0206] In this embodiment, at the weighted fusion layer, the vehicle chassis adjustment device incorporates the personalized preferences of all occupants. Dynamic weighted fusion is performed to calculate the final chassis control actions. Or the target chassis preference prediction result.

[0207] In this embodiment, the dynamic weighted fusion process involves corresponding arbitration weights, which are dynamically changing, i.e., the final weight (arbitration weight) of each member. By weight function Dynamic calculations show that this function takes into account:

[0208] First: Member K's identity (Drivers have a higher baseline weight) );

[0209] Second: Current specific scenario (For example, in traffic jam scenarios, passenger comfort is given more weight.)

[0210] Third: Special Passenger Care Factors (From the second layer).

[0211] Furthermore, based on other personalized chassis preference prediction results and corresponding arbitration weights, a weighted fusion calculation is performed to determine the target chassis preference prediction result, i.e., the final chassis parameters. It is a weighted average of the preferences of all members:

[0212]

[0213] in, It is the learning preference vector or preference prediction result of member k in the current scene S (e.g. ), That is the total number of people in the vehicle. It is a normalization factor. For suspension height, The damping coefficient is... For easy access when getting on and off the vehicle, etc.

[0214] In this embodiment, an example is given to illustrate how to determine the target chassis preference prediction result based on other personalized chassis preference prediction results and corresponding arbitration weights. If driver D's preference decision... Passenger P (infant) Preference .

[0215] like Output .

[0216] but: The results significantly favored the infant's comfort needs.

[0217] In this embodiment, it should be noted that the other layers also include a fourth layer, namely the learning feedback layer (DriverDominance & Learning Layer).

[0218] It should be noted that the learning feedback layer is the learning and feedback layer of the CDU. That is, if automatic decision-making... After execution, by the driver Manual preference decision-making The vehicle chassis adjustment device treats this intervention behavior as a strong supervision signal, and therefore needs to learn and update the corresponding model or module based on this strong supervision signal.

[0219] That is, after the step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to dynamically output the corresponding vehicle chassis adjustment decision, the following steps are included:

[0220] If an adjustment to the vehicle chassis adjustment decision is detected by the driver among the members, the preset multimodal perception model is updated based on the adjusted vehicle chassis adjustment decision, so as to make the next decision based on the updated preset multimodal perception model.

[0221] Specifically, in this embodiment, if an adjustment to the vehicle chassis adjustment decision by the driver is detected, it should be noted that the adjustment includes the adjustment of the vehicle chassis adjustment decision and / or the adaptive adjustment of the arbitration weight. Based on the adjusted vehicle chassis adjustment decision, the personalized chassis preference submodule in the preset multimodal perception model is updated to make the next decision based on the updated preset multimodal perception model.

[0222] The update of the personalized chassis preference submodule can be:

[0223] Vehicle chassis adjustment device use ( As training samples, the driver's own preference model Perform reinforcement learning or gradient descent-based learning to update its personalized chassis parameter preferences in the current scenario S.

[0224] in,

[0225] It should be noted that, in this embodiment, when updating its personalized chassis parameter preferences in the current scenario S, it also involves the adaptive adjustment of the arbitration weight. Specifically, the adaptive adjustment of the arbitration weight can be: the arbitration weight function is adaptive, that is, the vehicle chassis adjustment device further analyzes the current multi-person scenario ( ) and driver correction amount Adjust the weighting function appropriately underlying parameters .

[0226]

[0227] In this embodiment, the vehicle chassis adjustment device does not only learn "what the driver wants" (preference model) It can also learn "how drivers tend to arbitrate in multi-user conflicts" (weighting function). This allows its future automated decision-making in similar multi-person scenarios to be more closely aligned with the actual handling methods of that specific driver.

[0228] In this embodiment, specific embodiments are also provided:

[0229] Example 1: Personalized accessibility adjustment for underground parking garage scenarios;

[0230] Scenario: The vehicle is driving to the "underground parking garage entrance" area ( ).

[0231] Passenger ID: Driver ,passenger (aldult).

[0232] Preference learning outcomes:

[0233] Preferences:

[0234] Preferences:

[0235] The conflict arbitration process through the CDU can be as follows:

[0236] In this embodiment, CDU corresponds to the first layer, namely the security layer: not triggered or not involved.

[0237] As for the second tier, namely the VP care tier: no special passengers ( ).

[0238] For the third layer, the weighted fusion layer: assuming the baseline weight of driver ID_A... , .

[0239] ;

[0240] Therefore, in this embodiment, the output of the target chassis preference prediction result is: the suspension height is close to the "high" setting of ID_A, and the damping coefficient is slightly higher than the "soft" setting of ID_A.

[0241] In this embodiment, a second embodiment is also provided:

[0242] Prioritizing the comfort of special passengers and learning feedback;

[0243] Specific scenario: A pre-set multimodal perception model, such as the VLM model, identifies the road surface condition as "road surface pothole level M," and the cabin identifies a "baby" passenger. .

[0244] Passenger ID: Driver ,passenger .

[0245] The personalized chassis preference prediction result is as follows:

[0246] I Preferences: .

[0247] Preference (Unified Comfort Strategy): .

[0248] The conflict arbitration process through the CDU can be as follows:

[0249] In this embodiment, CDU corresponds to the first layer, namely the security layer: not triggered or not involved.

[0250] VP Care Layer: Existence .

[0251] Corresponding weighted fusion layer: Assumption .

[0252]

[0253] Therefore, in this embodiment, the output of the target chassis preference prediction result is: damping. The soft damping setting is extremely baby-friendly, maximizing comfort.

[0254] In this embodiment, a learning feedback layer is also involved: assuming ID_C considers 0.256 too soft, it is manually adjusted to... .

[0255] The vehicle chassis adjustment device then immediately updates the ID_C preference model. .

[0256] Additionally, it should be noted that the vehicle chassis adjustment system also learns that in the scenario of an "infant" passenger and a "pothole-prone road surface M", the driver's ID_C tends to be slightly higher than the VP-care damping. Vehicle chassis adjustment system adjustment. In similar situations in the future, appropriately reduce The weight of ID_C, or increase the weight of ID_C, so that It is closer to 0.4.

[0257] In this embodiment, based on the personalized chassis preference submodule in a preset multimodal perception model, the personalized chassis parameter preferences of vehicle occupants are predicted from the in-cabin scene perception results and the out-of-cabin scene perception results, resulting in a personalized chassis preference prediction result. The personalized chassis preference submodule is a pre-trained objective function. The personalized chassis preference prediction result is then output as the corresponding vehicle chassis adjustment decision, thereby meeting the user's needs.

[0258] In this embodiment, after the step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to dynamically output the corresponding vehicle chassis adjustment decision, the following steps are included:

[0259] Step F1: Determine the smoothing parameters corresponding to the vehicle chassis adjustment decision;

[0260] Step F2: Based on the smoothing parameters and the vehicle chassis adjustment decision, adjust the vehicle chassis.

[0261] In this embodiment, after determining the vehicle chassis adjustment decision, the saved mapping function F is obtained to acquire the set of prediction parameters. After that, it is also necessary to... Perform a smooth transition to avoid abrupt chassis movements, for example:

[0262] ;

[0263] in, Here, t represents the smoothing coefficient, and t represents time. In other words, in this embodiment, the vehicle chassis is also adjusted by determining the smoothing parameters corresponding to the vehicle chassis adjustment decision, thereby improving the user experience.

[0264] This application also provides a vehicle chassis adjustment device, please refer to... Figure 5 The vehicle chassis adjustment device includes:

[0265] The acquisition module 10 is used to acquire multi-source heterogeneous information after the vehicle is started if the vehicle chassis adjustment function is detected to be activated. The multi-source heterogeneous information includes multiple items such as information related to the occupants of the vehicle, information about the environment in which the vehicle is located, vehicle navigation information, and vehicle status information.

[0266] The identification module 20 is used to identify the multi-source heterogeneous information through a preset multimodal perception model and output the corresponding in-cabin scene perception results and out-of-cabin scene perception results, wherein the in-cabin scene perception results include the identification results of the in-cabin occupants.

[0267] The fusion module 30 is used to fuse the in-cabin scene perception results and the out-of-cabin scene perception results, and dynamically output the corresponding vehicle chassis adjustment decision.

[0268] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0269] After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and the pre-stored chassis function habit information is detected, it is determined whether the first preset trigger condition is met. If it is met, multi-source heterogeneous information is collected. The first preset trigger condition includes the vehicle driving meeting the requirements and / or the driver identification result meeting the requirements.

[0270] After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and no pre-stored chassis function habit information is detected, the second preset trigger condition is determined by the wheel patrol trigger method. If the condition is met, multi-source heterogeneous information is collected.

[0271] In one embodiment, the second preset triggering condition includes scene recognition meeting requirements, wherein the scene recognition meeting requirements include the scene being a parking garage scene or the scene being a potential congestion scene.

[0272] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0273] By using a pre-set multimodal perception model, the facial image features of the occupants and the scene state features of the cabin are determined from the multi-source heterogeneous information. The facial image features and the scene state features are spatiotemporally aligned to bind the identity information corresponding to the facial image features of the occupants with the attribute information corresponding to the scene state features.

[0274] Based on the identity and attribute information of the bound occupants, a dynamic profile of each occupant in the cabin is determined, and the dynamic profile of the occupant is used as the corresponding occupant identification result and output.

[0275] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0276] By using the scene understanding expert module in the preset multimodal perception model, the overall features of each occupant in the vehicle are extracted from the frames of the video stream corresponding to multi-source heterogeneous information. These features serve as the structured scene state features corresponding to the occupant's cabin. The structured scene state features include one or more of the following features: the occupant's age, posture, behavior, body type, and clothing.

[0277] By using the identity expert module in the preset multimodal perception model, facial image features of occupants in the vehicle can be identified from frames of video streams corresponding to multi-source heterogeneous information.

[0278] Spatial location matching is performed on facial image features and corresponding scene state features in the same frame to achieve spatiotemporal alignment of the facial image features and the scene state features.

[0279] In one embodiment, the scene state feature corresponds to a human bounding box;

[0280] The vehicle chassis adjustment device is used to achieve:

[0281] Determine the face detection box corresponding to the facial image features in the same frame, and determine the human body detection box corresponding to the scene state features;

[0282] If the face detection box is located within the corresponding human body detection box, and if it is located within the corresponding human body detection box, and the overlap between the face detection box and the human body detection box is greater than a preset confidence threshold, then it is determined that the facial image features and the corresponding scene state features in the same frame are spatially matched.

[0283] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0284] By using a preset multimodal perception model, information about the vehicle's environment, vehicle navigation information, and vehicle status information are determined from the multi-source heterogeneous information.

[0285] Based on the information of the vehicle's environment, vehicle navigation information, and vehicle status information, a set of semantic tags for the corresponding external scene is determined, and the set of semantic tags for the external scene is used as the perception result for the corresponding external scene. The information of the vehicle's environment includes one or more of the following: weather type, road type, road features, driving area, and vehicle events. The vehicle navigation information includes one or more of the following: vehicle latitude and longitude coordinates, heading, speed, congestion level on the navigation path, and distance information on the navigation path. The vehicle status information includes one or more of the following: vehicle dynamic parameters, chassis operation, and chassis status information.

[0286] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0287] Based on the personalized chassis preference submodule in the preset multimodal perception model, the personalized chassis parameter preferences of the in-cabin scene perception results and the out-of-cabin scene perception results are predicted to obtain the personalized chassis preference prediction results. The personalized chassis preference submodule is a pre-trained objective function.

[0288] The personalized chassis preference prediction results are output as the corresponding vehicle chassis adjustment decisions.

[0289] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0290] If multiple occupants are detected in the cabin, the personalized chassis parameter preferences for each occupant are predicted based on the personalized chassis preference submodule in the preset multimodal perception model, using the cabin scene perception results and the cabin exterior scene perception results.

[0291] If there is a conflict between the personalized chassis preference prediction results of different members, a hierarchical priority arbitration mechanism will be used to arbitrate the personalized chassis preference prediction results of different members, and the arbitration result will be used as the target chassis preference prediction result.

[0292] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0293] If there is a conflict between the personalized chassis preference prediction results of different members, the first personalized chassis preference prediction result of the member that does not meet the safety layer shall be removed, and the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result shall be used as the target chassis preference prediction result.

[0294] In one embodiment, the other layers include one or more of a special member care layer, a weighted fusion layer, and a corresponding learning feedback layer among members. The step of using the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result as the target chassis preference prediction result includes:

[0295] Determine the target stratum of the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result;

[0296] Based on the target layer in which it is located, the corresponding arbitration weight is determined, wherein the arbitration weight is dynamically changing;

[0297] Based on other personalized chassis preference prediction results and corresponding arbitration weights, the target chassis preference prediction result is determined.

[0298] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0299] If an adjustment to the vehicle chassis adjustment decision is detected by the driver among the members, the preset multimodal perception model is updated based on the adjusted vehicle chassis adjustment decision, so as to make the next decision based on the updated preset multimodal perception model.

[0300] In one embodiment, the adjustment includes adjustments to the vehicle chassis adjustment decision and / or adaptive adjustments to the arbitration weights.

[0301] In one embodiment, the vehicle chassis adjustment device is used to achieve:

[0302] Determine the smoothing parameters corresponding to the vehicle chassis adjustment decisions;

[0303] Based on the smoothing parameters and the vehicle chassis adjustment decision, the vehicle chassis is adjusted.

[0304] The vehicle chassis adjustment device provided in this application, employing the vehicle chassis adjustment method described in the above embodiments, can solve the technical problems of vehicle chassis adjustment devices. Compared with the prior art, the beneficial effects of the vehicle chassis adjustment device provided in this application are the same as those of the vehicle chassis adjustment method provided in the above embodiments, and other technical features in the vehicle chassis adjustment device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0305] This application provides a vehicle chassis adjustment device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the vehicle chassis adjustment method in the above embodiment 1.

[0306] The following is for reference. Figure 6 The diagram illustrates a structural schematic suitable for implementing the vehicle chassis adjustment device in the embodiments of this application. The vehicle chassis adjustment device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, tablets, digital radio receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The vehicle chassis adjustment device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of this application.

[0307] like Figure 6 As shown, the vehicle chassis adjustment device may include a processing unit 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the vehicle chassis adjustment device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the vehicle chassis adjustment equipment to communicate wirelessly or wiredly with other devices to exchange data. Although the figures show vehicle chassis adjustment equipment with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0308] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0309] The vehicle chassis adjustment device provided in this application, employing the vehicle chassis adjustment method described in the above embodiments, can solve the technical problem. Compared with the prior art, the beneficial effects of the vehicle chassis adjustment device provided in this application are the same as those of the vehicle chassis adjustment method provided in the above embodiments, and other technical features of this vehicle chassis adjustment device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0310] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0311] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0312] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to perform the vehicle chassis adjustment method in the above embodiments.

[0313] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0314] The aforementioned computer-readable storage medium may be included in the vehicle chassis adjustment equipment; or it may exist independently and not be installed in the vehicle chassis adjustment equipment.

[0315] The aforementioned computer-readable storage medium carries one or more programs. When the one or more programs are executed by the vehicle chassis adjustment device, the vehicle chassis adjustment device: collects access information from different application sources through an access acquisition program and stores the access information in a message queue, wherein the access acquisition program is deployed in the runtime environment of the application; performs service call dependency deduction on the access information in the message queue to obtain the call relationship between the different applications, and generates an application topology architecture based on the call relationship.

[0316] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0317] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0318] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0319] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described vehicle chassis adjustment method, thereby solving the technical problem of vehicle chassis adjustment. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the vehicle chassis adjustment method provided in the above embodiments, and will not be repeated here.

[0320] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the vehicle chassis adjustment method described above.

[0321] The computer program product provided in this application can solve the technical problem of vehicle chassis adjustment. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the vehicle chassis adjustment method provided in the above embodiments, and will not be repeated here.

[0322] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.

Claims

1. A method for adjusting a vehicle chassis, characterized in that, The vehicle chassis adjustment method includes: After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, multi-source heterogeneous information is collected. The multi-source heterogeneous information includes multiple items such as information related to the occupants of the vehicle, information about the environment in which the vehicle is located, vehicle navigation information, and vehicle status information. By using a preset multimodal perception model, the multi-source heterogeneous information is identified, and corresponding in-cabin scene perception results and out-of-cabin scene perception results are output. The in-cabin scene perception results include the identification results of in-cabin occupants. By integrating the in-cabin scene perception results and the out-of-cabin scene perception results, the corresponding vehicle chassis adjustment decision is dynamically output. The step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to output the corresponding vehicle chassis adjustment decision includes: If multiple occupants are detected in the cabin, based on the personalized chassis preference submodule in the preset multimodal perception model, the personalized chassis parameter preferences of each occupant are predicted based on the cabin scene perception results and the cabin exterior scene perception results. The personalized chassis preference submodule is a pre-trained objective function. If there is a conflict between the personalized chassis preference prediction results of different members, the personalized chassis preference prediction results of different members are arbitrated through a preset hierarchical priority arbitration mechanism, and the arbitration result is used as the target chassis preference prediction result. The preset hierarchical level includes a weighted fusion layer, which is used to dynamically weight and fuse the personalized chassis preference prediction results of different members based on the arbitration weight. The arbitration weight is determined based on the identity information of the in-vehicle members, the perception results of the scene outside the vehicle cabin, and special passenger care factors. The personalized chassis preference prediction results are output as the corresponding vehicle chassis adjustment decisions.

2. The vehicle chassis adjustment method as described in claim 1, characterized in that, The step of collecting multi-source heterogeneous information after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated, includes any one of the following: After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and the pre-stored chassis function habit information is detected, it is determined whether the first preset trigger condition is met. If it is met, multi-source heterogeneous information is collected. The first preset trigger condition includes the vehicle driving meeting the requirements and / or the driver identification result meeting the requirements. After the vehicle is started, if the vehicle chassis adjustment function is detected to be activated and no pre-stored chassis function habit information is detected, the second preset trigger condition is determined by the wheel patrol trigger method. If the condition is met, multi-source heterogeneous information is collected.

3. The vehicle chassis adjustment method as described in claim 2, characterized in that, The second preset triggering condition includes meeting the scene recognition requirements, which include the scene being a parking garage scene or a potential congestion scene.

4. The vehicle chassis adjustment method as described in claim 2, characterized in that, The step of identifying the multi-source heterogeneous information through a preset multimodal perception model and outputting the corresponding in-vehicle cabin scene perception result includes: By using a pre-set multimodal perception model, the facial image features of the occupants and the scene state features of the cabin are determined from the multi-source heterogeneous information. The facial image features and the scene state features are spatiotemporally aligned to bind the identity information corresponding to the facial image features of the occupants with the attribute information corresponding to the scene state features. Based on the identity and attribute information of the bound occupants, a dynamic profile of each occupant in the cabin is determined, and the dynamic profile of the occupant is used as the corresponding occupant identification result and output.

5. The vehicle chassis adjustment method as described in claim 4, characterized in that, The step of determining the facial image features of vehicle occupants and the scene state features within the vehicle cabin from the multi-source heterogeneous information using a preset multimodal perception model, and then performing spatiotemporal alignment of the facial image features and the scene state features, includes: By using the scene understanding expert module in the preset multimodal perception model, the overall features of each occupant in the vehicle are extracted from the frames of the video stream corresponding to multi-source heterogeneous information. These features serve as the structured scene state features corresponding to the occupant's cabin. The structured scene state features include one or more of the following features: the occupant's age, posture, behavior, body type, and clothing. By using the identity expert module in the preset multimodal perception model, facial image features of occupants in the vehicle can be identified from frames of video streams corresponding to multi-source heterogeneous information. Spatial location matching is performed on facial image features and corresponding scene state features in the same frame to achieve spatiotemporal alignment of the facial image features and the scene state features.

6. The vehicle chassis adjustment method as described in claim 5, characterized in that, The scene state features correspond to human bounding boxes; The step of matching the spatial location of facial image features and corresponding scene state features in the same frame includes: Determine the face detection box corresponding to the facial image features in the same frame, and determine the human body detection box corresponding to the scene state features; If the face detection box is located within the corresponding human body detection box, and if it is located within the corresponding human body detection box, and the overlap between the face detection box and the human body detection box is greater than a preset confidence threshold, then it is determined that the facial image features and the corresponding scene state features in the same frame are spatially matched.

7. The vehicle chassis adjustment method as described in claim 2, characterized in that, The step of identifying the multi-source heterogeneous information through a preset multimodal perception model and outputting the corresponding perception result of the external scene of the vehicle cabin includes: By using a preset multimodal perception model, information about the vehicle's environment, vehicle navigation information, and vehicle status information are determined from the multi-source heterogeneous information. Based on the information of the vehicle's environment, vehicle navigation information, and vehicle status information, a set of semantic tags for the corresponding external scene is determined, and the set of semantic tags for the external scene is used as the perception result for the corresponding external scene. The information of the vehicle's environment includes one or more of the following: weather type, road type, road features, driving area, and vehicle events. The vehicle navigation information includes one or more of the following: vehicle latitude and longitude coordinates, heading, speed, congestion level on the navigation path, and distance information on the navigation path. The vehicle status information includes one or more of the following: vehicle dynamic parameters, chassis operation, and chassis status information.

8. The vehicle chassis adjustment method as described in claim 1, characterized in that, The preset hierarchical structure includes a security layer and other layers. The step of arbitrating the personalized chassis preference prediction results of different members through a priority arbitration mechanism of the preset hierarchical structure, and using the arbitration result as the target chassis preference prediction result, includes: If there is a conflict between the personalized chassis preference prediction results of different members, the first personalized chassis preference prediction result of the member that does not meet the safety layer shall be removed, and the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result shall be used as the target chassis preference prediction result.

9. The vehicle chassis adjustment method as described in claim 8, characterized in that, The other layers include one or more of the following: a special member care layer, a weighted fusion layer, and a corresponding learning feedback layer among members. The step of using the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result as the target chassis preference prediction result includes: Determine the target stratum of the other personalized chassis preference prediction results after removing the first personalized chassis preference prediction result; Based on the target layer in which it is located, the corresponding arbitration weight is determined, wherein the arbitration weight is dynamically changing; Based on other personalized chassis preference prediction results and corresponding arbitration weights, the target chassis preference prediction result is determined.

10. The vehicle chassis adjustment method as described in claim 9, characterized in that, After the step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to dynamically output the corresponding vehicle chassis adjustment decision, the following steps are included: If an adjustment to the vehicle chassis adjustment decision is detected by the driver among the members, the preset multimodal perception model is updated based on the adjusted vehicle chassis adjustment decision, so as to make the next decision based on the updated preset multimodal perception model.

11. The vehicle chassis adjustment method as described in claim 10, characterized in that, The adjustments include adjustments to the vehicle chassis adjustment decisions and / or adaptive adjustments to the arbitration weights.

12. The vehicle chassis adjustment method as described in claim 1, characterized in that, After the step of fusing the in-cabin scene perception results and the out-of-cabin scene perception results to dynamically output the corresponding vehicle chassis adjustment decision, the following steps are included: Determine the smoothing parameters corresponding to the vehicle chassis adjustment decisions; Based on the smoothing parameters and the vehicle chassis adjustment decision, the vehicle chassis is adjusted.

13. A vehicle chassis adjustment device, characterized in that, The vehicle chassis adjustment device includes: The data acquisition module is used to collect multi-source heterogeneous information after the vehicle is started, if the vehicle chassis adjustment function is detected to be activated. The multi-source heterogeneous information includes multiple items such as information related to the occupants of the vehicle, information about the environment in which the vehicle is located, vehicle navigation information, and vehicle status information. The identification module is used to identify the multi-source heterogeneous information through a preset multimodal perception model and output the corresponding in-cabin scene perception results and out-of-cabin scene perception results, wherein the in-cabin scene perception results include the identification results of the in-cabin occupants. The fusion module is used to fuse the in-cabin scene perception results and the out-of-cabin scene perception results, and dynamically output the corresponding vehicle chassis adjustment decisions. The vehicle chassis adjustment device is used to achieve: If multiple occupants are detected in the cabin, the personalized chassis parameter preferences for each occupant are predicted based on the personalized chassis preference submodule in the preset multimodal perception model, using the cabin scene perception results and the cabin exterior scene perception results. If there is a conflict between the personalized chassis preference prediction results of different members, the personalized chassis preference prediction results of different members are arbitrated through a preset hierarchical priority arbitration mechanism, and the arbitration result is used as the target chassis preference prediction result. The preset hierarchical level includes a weighted fusion layer, which is used to dynamically weight and fuse the personalized chassis preference prediction results of different members based on the arbitration weight. The arbitration weight is determined based on the identity information of the in-vehicle members, the perception results of the scene outside the vehicle cabin, and special passenger care factors. The personalized chassis preference prediction results are output as the corresponding vehicle chassis adjustment decisions.

14. A vehicle chassis adjustment device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the vehicle chassis adjustment method as described in any one of claims 1 to 12.

15. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the vehicle chassis adjustment method as described in any one of claims 1 to 12.

Images