Method for analyzing a new vehicle function and server setup

Abstract

The invention relates to a server device (12) and a method for analyzing a new vehicle function. The method comprises the following steps: determining swarm data (14) of a vehicle fleet, wherein the swarm data (14) is provided to a server device (12) and includes at least interaction data with existing vehicle functions; analyzing the swarm data (14) by the server device (12) to detect interaction behavior with the vehicle functions; determining interaction behavior groups from the detected interaction behaviors by the server device (12); generating a digital twin for each interaction behavior group based on the respective detected interaction behaviors; and simulating interactions of the generated digital twins with the new vehicle function and analyzing the interactions.

Description

[0001] The invention relates to a method for analyzing a new vehicle function. Furthermore, the invention relates to a server device configured to carry out the method.

[0002] New vehicle functions are typically validated for usability through user studies before being released in production vehicles. These studies assess how intuitive and easy to understand the functions are to operate. Users test the new functions under real-world conditions and provide feedback on their usability. Based on these studies, the functions are optimized to ensure they meet user requirements and expectations. However, this process requires a significant investment of time and resources, as numerous user studies must be conducted and evaluated.

[0003] US Patent 2023 / 0367688 A1 discloses systems and methods for implementing and using a digital mobility framework, specifically a digital twin, encompassing both a digital and a physical space. The digital mobility framework is implemented on a cloud-based system or platform. Digital twins can represent not only vehicle units but also human or traffic units as data / models of these diverse physical objects / processes. Additionally, the digital mobility framework can utilize data from different timeframes, such as real-time and historical data.

[0004] From WO 2020 / 140895 A1, intelligent vehicle diagnostics using driving behavior modeling and monitoring are known. A data processing system for a vehicle comprises a variety of sensors and a vehicle control unit. The vehicle control unit can sample the output from each of the sensors and compile a data set that can be transmitted to a cloud data center. The cloud data center can apply statistical machine learning algorithms to the data set and training data to generate a model of the user's expected driving behavior. The vehicle can receive the generated model from the cloud computing system via the network, and the vehicle control unit can use the model to monitor the user's driving behavior.In response to the detection of driving behavior that deviates from the expected driving behavior, the vehicle control unit can diagnose the cause of the abnormal behavior and take one or more preventive measures based on the diagnosis.

[0005] German patent DE 10 2018 212 560 A1 discloses a computer-based system for testing a server-based vehicle function. A functional model of the vehicle function is simulated by a first simulator on a server, at least a partial vehicle model is simulated by a second simulator, and the vehicle function is tested while a data connection between the first and second simulators is controlled in a planned manner.

[0006] The purpose of the invention is to simplify the analysis or testing of a new vehicle function.

[0007] This problem is solved by the independent patent claims. Advantageous embodiments of the invention are disclosed in the dependent patent claims, the following description, and the figures.

[0008] The invention is based on the idea that artificial intelligence can simplify and improve the testing and validation of new vehicle functions. AI-supported systems can analyze large amounts of usage data and recognize patterns in user behavior. These insights can be used to create digital twins of users that replicate the behavior and preferences of real users. The digital twins can then be used in simulations to test and validate the new vehicle functions. This reduces the need for extensive human trials and accelerates the validation process.The digital twins can be trained using swarm data, which can encompass the operating behavior of users of all vehicles in a fleet. This operating behavior can be categorized into different clusters to represent different user types. This allows for validating analyses for these different operating types and yielding valuable insights that contribute to optimizing and improving usability and safety.

[0009] One aspect of the invention relates to a method for analyzing a new vehicle function, comprising as steps in determining swarm data of a vehicle fleet, wherein the swarm data is provided to a server facility and includes at least interaction data with existing vehicle functions, analyzing the swarm data by the server facility to detect interaction behavior with the vehicle functions, determining interaction behavior groups from the detected interaction behavior by the server facility, generating a digital twin for each interaction behavior group based on the respective detected interaction behavior, and simulating interactions of the generated digital twins with the new vehicle function and analyzing the interactions.

[0010] In other words, it is possible to collect swarm data from a vehicle fleet, which can then be transferred to a server. This swarm data can include data from sensors and / or functions of the individual vehicles in the fleet, from which at least interaction data with existing vehicle functions is available. Interaction data refers, for example, to operating habits, operating times, and / or operating levels through which a user interacts when using the available vehicle functions. This means that duration, frequency, and other settings, such as preferred use of predefined profiles or individual settings, can be determined.In addition to interaction data, swarm data can also be used to determine other vehicle data, such as traffic situation and / or environmental conditions that may be present during the use of the available vehicle functions.

[0011] The server setup can be an external computing device, in particular a computer network, such as a cloud platform. The respective swarm data can preferably be transmitted wirelessly to the server setup. Such data transmission can be carried out using secure communication protocols, thus ensuring data protection.

[0012] The server setup allows the swarm data to be analyzed to determine interaction behavior with the existing vehicle functions. A program or algorithm can be provided for this purpose, running on the server to analyze the swarm data. Specifically, machine learning algorithms and / or pre-trained artificial intelligence can be used to analyze the swarm data and identify patterns and behaviors with existing vehicle functions. Techniques such as clustering, classification, and anomaly detection can be applied. The swarm data can also be cleaned and pre-processed before analysis. This can include removing outliers, normalizing the data, and / or aggregating relevant information.

[0013] Once interaction behaviors with the available vehicle functions have been determined, they can be categorized into interaction behavior groups. These groups can include those exhibiting similar or predefined interaction behaviors. In other words, the interaction behavior groups can represent different operator types identified through swarm data analysis. This categorization into interaction behavior groups can be automated, particularly through cluster analysis, which identifies and groups similar or related interaction behaviors. Alternatively, user types can be predefined, and the identified interaction behaviors can be classified into the respective interaction behavior groups, each representing a different operator type.

[0014] For each different interaction behavior group, a corresponding digital twin can be created, replicating the respective interaction behavior with vehicle functions. Thus, the digital twin can model the typical behavior and preferences of the respective interaction behavior groups or operator types. The digital twin can be a trained artificial intelligence, in particular a neural network, that has been trained on the respective interaction behavior of the corresponding interaction behavior group.

[0015] Finally, the use of the new vehicle function to be analyzed can be simulated using the respective digital twins created. This means that the digital twins can interact with the new vehicle function to test various scenarios and analyze the impact of the new function on, for example, interaction behavior or driving behavior. The simulation results can then be used to evaluate and optimize the usability, safety, and efficiency of the new vehicle functions.

[0016] In particular, it may be provided that the respective digital twins generated are continuously updated with new swarm data in order to continuously improve them.

[0017] The advantage of the invention is that new vehicle functions can be analyzed and improved more quickly and with less effort, in particular with fewer test subject studies.

[0018] The invention also includes embodiments that offer additional advantages.

[0019] One embodiment provides that the swarm data further includes data on the driving style of individual users of the vehicle fleet, data on reactions to traffic situations, and / or data on environmental conditions. In other words, in addition to interaction data with the available vehicle functions, the swarm data can also include information on when and under what situations the vehicle functions were used. This can include, in particular, sensor data from the respective vehicle, encompassing driving style (e.g., data on speeds and / or accelerations), traffic situations (e.g., traffic volume or vehicle position), and / or environmental conditions (e.g., day / night or weather conditions).This embodiment offers the advantage that an interaction behavior can be assigned to a specific situation, which improves the classification into different interaction behavior groups.

[0020] Another embodiment involves determining the interaction behavior with vehicle functions from the swarm data using machine learning algorithms. This means that pattern recognition can be performed on the swarm data, capturing interactions and behaviors with the vehicle functions. Machine learning algorithms such as clustering, data classification, and / or anomaly detection can be applied for this purpose. In other words, the swarm data can be used to train an artificial intelligence to determine the interaction behavior with the available vehicle functions. This embodiment offers the advantage of faster swarm data analysis.

[0021] Another embodiment involves determining interaction behavior groups based on cluster analysis of the detected interaction behaviors. Cluster analysis allows for the formation of groups (clusters) of detected interaction behaviors that exhibit similar characteristics. This cluster analysis is a method of unsupervised machine learning. The resulting interaction behavior clusters can then be used, for example, for automated classification, particularly to define different user types that exhibit specific interaction behaviors when using vehicle functions. A program or algorithm for cluster analysis can be run on the server to group the interaction behaviors into similar clusters.This offers the advantage that groups with the same or similar interaction behavior can be identified more quickly.

[0022] Another embodiment provides for predefined operator types with predefined properties, with the interaction behavior groups being created based on these predefined operator types. In other words, the interaction behavior groups created using cluster analysis can be classified into the predefined operator types, where operator types have predefined properties when using vehicle functions. Alternatively, the operator types with their predefined properties can be predefined when creating the interaction behavior groups, and the interaction behaviors with the vehicle functions can be classified according to the respective operator type.This allows for the creation of improved digital twins that can exhibit predefined interaction behavior, enabling better testing of the new vehicle function for different operator types.

[0023] In particular, the predefined operator types may include: an active operator who frequently and intensively interacts with vehicle functions; a passive operator who rarely interacts with vehicle functions; an occasional operator who interacts with vehicle functions only occasionally and in specific situations; an intensive operator who interacts with vehicle functions very frequently and intensively; a safety-oriented operator who primarily uses safety-related functions; a comfort-oriented operator who primarily uses comfort functions; and a tech-savvy operator who frequently changes settings on vehicle functions. By creating these interaction behavior groups, a digital twin can be generated for each operator type.These digital twins can then be used to validate the new vehicle function with regard to operation, for example, to record and analyze the type and duration of interaction and individual behaviors. The specified operator types allow for the mapping of commonly occurring behaviors, thus improving the testing of the new vehicle function.

[0024] Another embodiment involves designing each digital twin as an artificial neural network that is trained on the interaction behaviors of the respective interaction behavior groups. In other words, the interaction behaviors of the digital twins with the new vehicle function can be simulated by training an artificial neural network for each interaction behavior group or operator type using the respective interaction behaviors. Weights within the artificial neural network can be adjusted using these interaction behaviors to generate decisions and predictions for the respective digital twin. For example, a new vehicle function, particularly a new user interface, can be defined that interacts with the digital twin or the trained artificial neural network. This interaction then allows the new vehicle function to be validated.

[0025] Another embodiment provides that the swarm data is determined by the existing vehicle functions and / or sensors of the vehicles in the fleet and transmitted to the server in real time or at predefined intervals. In particular, secure and reliable communication protocols can be used for this purpose. Real-time transmission can mean that the data is shared with the server via a data connection immediately after measurement. Alternatively, the measured data can be temporarily stored in each vehicle and then transmitted to the server at predefined intervals, such as once a day. This provides the server with a large amount of current swarm data, which can be used to continuously adapt and improve the respective digital twins.

[0026] Another embodiment provides for the swarm data to be provided to the server in anonymized form. This means that before the respective data is transmitted from the vehicle to the server, personal information can be removed or obscured from the data within the vehicle itself, thus ensuring data protection.

[0027] Another aspect of the invention relates to a server system configured to analyze provided swarm data from a vehicle fleet, which includes at least interaction data with existing vehicle functions, to recognize interaction behavior with the vehicle functions, to determine interaction behavior groups from the recognized interaction behavior, to generate a digital twin for each interaction behavior group based on the respective recognized interaction behavior, to simulate interactions of the digital twins with a new vehicle function, and to analyze the interactions. This offers the same advantages and variations as the method.

[0028] For use cases or application situations that may arise during the procedure and are not explicitly described here, it may be provided that, according to the procedure, an error message and / or a request for user feedback is issued and / or a default setting and / or a predetermined initial state is set.

[0029] A training platform for training a machine learning model can comprise a computer or a computer network in which a machine learning model, such as an artificial neural network or a decision tree model, can be trained in a known manner using a training algorithm based on training data. Such a training algorithm can, for example, be based on the backpropagation algorithm. The training data can be, for example, sensor data sets of the same type that might later be present in a system where the machine learning model is intended to be used by the user in a product. The training data can be supplemented in a known manner with so-called labels or ground truth information that specify a desired processing result of the model, i.e., the target output data.The training data can include training datasets that describe different training situations as well as the associated ground truth information (labels). By selecting appropriate training datasets, the behavior of the trained model can be influenced or controlled.

[0030] The invention also includes the control device for the server device. The control device can comprise a data processing device or a processor circuit configured to perform an embodiment of the method according to the invention. For this purpose, the processor circuit can comprise at least one microprocessor and / or at least one microcontroller and / or at least one FPGA (Field Programmable Gate Array) and / or at least one DSP (Digital Signal Processor). In particular, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an NPU (Neural Processing Unit) can be used as the microprocessor. Furthermore, the processor circuit can comprise program code configured to perform the embodiment of the method according to the invention when executed by the processor circuit.The program code can be stored in a data memory of the processor device. The processor device can be based, for example, on at least one circuit board and / or on at least one SoC (System on Chip).

[0031] The invention also includes further developments of the server device according to the invention, which have features already described in connection with the further developments of the method according to the invention. For this reason, the corresponding further developments of the server device according to the invention are not described again here.

[0032] The motor vehicles in the fleet are preferably designed as motor vehicles, in particular as passenger cars or trucks, or as passenger buses or motorcycles.

[0033] As a further solution, the invention also includes a computer-readable storage medium comprising program code which, when executed by a computer or a computer network, causes it to execute an embodiment of the method according to the invention. The storage medium can be provided at least partially as a non-volatile data storage medium (e.g., as flash memory and / or as an SSD - solid state drive) and / or at least partially as a volatile data storage medium (e.g., as RAM - random access memory). The storage medium can be located within the computer or computer network. However, the storage medium can also be operated, for example, as a cloud server on the internet.

[0034] The computer or computer network can provide a processor circuit with, for example, at least one microprocessor. The program code can be provided as binary code and / or assembly code and / or source code of a programming language (e.g., C) and / or as a program script (e.g., Python). Alternatively, the computer-readable storage medium can be implemented as a signal containing computer-readable data, e.g., a time-varying voltage signal and / or a radio signal.

[0035] The invention also includes combinations of the features of the described embodiments. The invention therefore also includes realizations that each exhibit a combination of the features of several of the described embodiments, provided that the embodiments have not been described as mutually exclusive.

[0036] The following are exemplary embodiments of the invention described. This is illustrated by: Fig. 1 a system comprising motor vehicles and a server device; Fig. 2 a process diagram for a procedure for analyzing a new vehicle function.

[0037] The exemplary embodiments described below are preferred embodiments of the invention. In these exemplary embodiments, the described components each represent individual features of the invention, which can be considered independently of one another and each further develops the invention independently. Therefore, the disclosure is intended to include combinations of features of the embodiments other than those shown. Furthermore, the described embodiments can also be supplemented by further features of the invention already described.

[0038] In the figures, identical reference symbols denote functionally equivalent elements.

[0039] In Fig. Figure 1 shows motor vehicles 10 and a server facility 12. The motor vehicles 10 can belong to a fleet and be connected to the server facility 12 via a wireless data connection. The server facility 12 can be a central server or a cloud platform to which the motor vehicles 10 can transmit swarm data 14 in real time or at predefined intervals.

[0040] In particular, swarm data can include 14 interaction data, encompassing interactions of motor vehicle users with vehicle functions. This means that the interaction data can contain information about user interactions with vehicle functions, such as usage duration, frequency of use, and type of use.

[0041] Furthermore, swarm data can include vehicle data and / or environmental data, which may encompass the conditions under which the respective vehicle functions were used and the type of usage carried out under the respective environmental conditions. For example, it can be provided that certain vehicle functions are primarily used under predefined conditions or situations.

[0042] The server setup 12 allows the swarm data 14 to be evaluated, particularly for analyzing new vehicle functions. For this purpose, the swarm data can be preprocessed to remove outliers, normalize the data, and aggregate relevant information. Using machine learning algorithms, the swarm data can then be analyzed to identify patterns and behaviors, especially to determine interaction behavior with vehicle functions. This means that, firstly, interaction behavior with the existing vehicle functions can be automatically determined, specifically in which situations and under which conditions the vehicle functions are used and how.

[0043] Once the interaction behaviors have been extracted, they can be categorized into interaction behavior groups. This can be achieved, in particular, by performing cluster analysis so that similar or identical interaction behaviors are assigned to the same group. These interaction behavior groups can then be combined into predefined user types, which can have predefined characteristics. These predefined characteristics can, in particular, define predefined operator types who typically use vehicle functions. For example, an active operator can be defined who interacts frequently and intensively with the vehicle functions, such as frequent use of infotainment systems, navigation, and driver assistance systems. A passive operator can be defined who rarely interacts with the vehicle functions and prefers a more passive approach.An occasional user may only use vehicle functions occasionally and in specific situations, for example, mainly on longer journeys and in unfamiliar areas. An intensive user may interact with vehicle functions very frequently and intensively, particularly by extensively customizing settings. A safety-oriented user may primarily use safety-related functions, such as driver assistance systems and warning systems. A comfort-oriented user may primarily use comfort functions, such as air conditioning, seat heating, and infotainment systems. Finally, a tech-savvy user can be defined, who enjoys using new technologies and functions.

[0044] For each of these operator types or interaction behavior groups, a digital twin can then be created that exhibits the interaction behavior of the respective group. Artificial intelligence, in particular an artificial neural network, can be trained to create a digital twin by replicating the behavior of the respective operator types.

[0045] The digital twins can then be used for analyzing and testing new vehicle functions. This means that a simulation environment can be provided using the digital twins, allowing them to interact with the new vehicle function. Such a simulation environment makes it possible to test various scenarios and analyze the impact of new functions, particularly on driving behavior. The simulation results can then be used to evaluate and optimize the usability, safety, and efficiency of the new vehicle functions. The simulation environment can be provided by the server unit 12 or on external computing units (not shown), which serve as a development environment for new vehicle functions.

[0046] In Fig. Figure 2 shows a schematic process diagram for analyzing a new vehicle function.

[0047] In step S10, swarm data from a vehicle fleet can be determined, wherein the swarm data is provided to a server facility 12 and includes at least interaction data with existing vehicle functions. Transmission of the swarm data to the server facility 12 can preferably be carried out via secure connections to prevent unauthorized access and data loss. Furthermore, the swarm data can be anonymized before transmission to the server facility 12 to comply with data protection regulations and to protect the privacy of users of the vehicles 10 in the vehicle fleet.

[0048] In step S12, the swarm data can be evaluated to determine interaction behavior with the vehicle functions. This can be done, for example, using techniques such as clustering, classification, and anomaly detection.

[0049] In step S14, the detected interaction behaviors can be categorized into different, particularly predefined, interaction behavior groups that can correspond to operating types of vehicle functions. The operating types can have predefined properties that correspond to a group of users with similar interaction behavior with the vehicle functions.

[0050] In step S16, a digital twin can be created for each of these interaction behavior groups or operating types, whereby the digital twin can be a digital model of a respective operating type.

[0051] Finally, in step S18, interactions of the respective digital twins generated with a new vehicle function can be simulated to test the new function and evaluate aspects such as user-friendliness, safety, or efficiency. Based on this validation, vehicle functions can then be optimized, particularly for the respective predefined operating types.

[0052] Overall, the examples show how the use of swarm data and artificial intelligence to create digital twins can be used to validate new vehicle functions. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] US 2023 / 0367688 A1

[0003] WO 2020 / 140895 A1

[0004] DE 10 2018 212 560 A1

[0005]

Claims

[1] Method for analyzing a new vehicle function, comprising the following steps: - Determining swarm data (14) of a vehicle fleet, wherein the swarm data (14) is provided to a server facility (12) and includes at least interaction data with existing vehicle functions; - Analyzing the swarm data (14) by the server facility (12) to detect interaction behavior with the vehicle functions; - Determining interaction behavior groups from the detected interaction behaviors by the server setup (12); - Creating a digital twin for each interaction behavior group based on the respective identified interaction behaviors; - Simulating interactions of the generated digital twins with the new vehicle function and analyzing the interactions. [2] Method according to claim 1, wherein the swarm data (14) further comprise data on the driving style of respective users of the vehicle fleet, data on reactions to traffic situations and / or data on environmental conditions. [3] Method according to one of the preceding claims, wherein the interaction behaviors with the vehicle functions are determined from the swarm data (14) using machine learning algorithms. [4] Method according to one of the preceding claims, wherein the interaction behavior groups are determined by means of a cluster analysis of the identified interaction behaviors. [5] Method according to one of the preceding claims, wherein operator types are specified which have predefined properties, wherein the interaction behavior groups are created based on the specified operator types. [6] Method according to claim 5, wherein the specified operator types include: - an active operator who frequently and intensively interacts with vehicle functions; - a quiet operator who rarely interacts with the vehicle's functions; - a casual operator who only interacts with the vehicle's functions occasionally and in specific situations; - an intensive operator who interacts very frequently and intensively with the vehicle functions; - a safety-oriented operator who primarily uses safety-related functions; - a comfort-oriented operator who primarily uses comfort features; and - a technically savvy operator who frequently changes settings on the vehicle functions. [7] Method according to any of the preceding claims, wherein the respective digital twin is designed as an artificial neural network which is trained with the interaction behavior of the respective interaction behavior groups. [8] Method according to one of the preceding claims, wherein the swarm data (14) are determined by the existing vehicle functions and / or sensors of motor vehicles (10) of the vehicle fleet and are transmitted to the server facility (12) in real time or at predetermined time intervals. [9] Method according to any of the preceding claims, wherein the swarm data are provided to the server facility (12) in anonymized form. [10] Server facility (12) designed to analyze provided swarm data (14) of a vehicle fleet, which includes at least interaction data with existing vehicle functions, to detect interaction behavior with the vehicle functions, to determine interaction behavior groups from the detected interaction behaviors, to generate a digital twin for each interaction behavior group based on the respective detected interaction behaviors, to simulate interactions of the digital twins with a new vehicle function and to analyze the interactions.

Images