Method for operating a motor vehicle, computer program product, control unit and motor vehicle

Abstract

A method for operating a motor vehicle (20) comprising a bus system (3) via which data (5) are transmitted that influence at least one driving function of the motor vehicle (20), characterized in that data (5) relevant to the driving function are selected in a decision tree (30) in the form of influencing factors (31-36) and prioritized with respect to the driving function in order to significantly reduce the amount of data (5) to be transmitted via the bus system (3), wherein the motor vehicle (20) is considered as a cyber-physical system in its environment with all data (5) acquired in the motor vehicle (20) and its environment, wherein attributes with the greatest influencing factor (31-36) are extracted and explored, wherein data (5) from the environment of the vehicle (20) are structured as attributes of a layered model (17), wherein the decision tree (30) shows a weighting of certain attributes with respect to a malfunction of a driving function.

Description

[0001] The invention relates to a method for operating a motor vehicle, comprising a bus system through which data is transmitted that influences at least one driving function of the motor vehicle. The invention further relates to a computer program, a control unit, and a motor vehicle.

[0002] International patent application WO 2020 / 079074 A2 discloses a planning method for an autonomous vehicle, in which sensor signals are processed to determine a driving scenario and a tree search algorithm is executed to determine a sequence of maneuvers. Chinese patent application CN 109677406 A discloses a vehicle lane control system with risk monitoring, in which environmental information data, road information data, and vehicle information data are collected and the collected data are classified according to a clustering analysis model to obtain all types of factor influence coefficients. American patent application US 2022 / 028306 A1 discloses a method for data reduction in a storage device for machine learning, in which data used for training in a tree-based adapted iteration session is stored in the storage device.German patent application DE 10 2018 205 248 A1 discloses a functional system for providing environmental information to a driver assistance system of a motor vehicle based on information from multiple environmental sensors. German patent application DE 102 07 993 A1 discloses a traffic data information system for use in vehicles, which records traffic-related data of various categories, correlates it according to predefined criteria, and sorts the combined data according to relevance and makes it available to the driver via means of information transmission. The criteria according to which data can be evaluated, sorted with respect to relevance, and assigned to different data categories or different relevance values ​​are defined.German patent application DE 10 2019 204 691 A1 discloses a method for determining the health status of at least one occupant of a vehicle due to driving conditions, wherein the vehicle has at least one sensor for the sensorial detection of at least one driving dynamics parameter during the vehicle's operation, and wherein the sensor data supplied by the at least one sensor are preprocessed in the vehicle to minimize their data volume for transmission to the external computer. German patent application DE 10 2020 133 262 A1 discloses a method comprising: assigning at least one entry in a queue of a queue management device, wherein the at least one entry comprises a memory segment allocation from a memory segment cache or a page stack; and receiving, at the queue management device, a request to allocate a memory region.The generic German patent application DE 10 2019 105 853 A1 discloses a method for processing vehicle data, comprising the steps of acquiring the vehicle data during vehicle operation; transmitting a subset of the vehicle data to an external node; examining the transmitted vehicle data to determine whether a predefined driving event has occurred at the external node; and requesting additional vehicle data from the vehicle if the predefined driving event is present in the subset of vehicle data, the method being designed to ensure that no irrelevant vehicle data needs to be transmitted between the vehicle and the node and that the volume of data to be transmitted is reduced, whereby vehicle data is filtered and analyzed in a defined manner for this purpose.

[0003] The object of the invention is to increase the efficiency of operating a motor vehicle comprising a bus system through which data is transmitted that influences at least one driving function of the motor vehicle.

[0004] The problem is solved by a method with the features of claim 1. This effectively reduces the amount of data transmitted via the bus system. As a result, the efficiency of vehicle operation, particularly in the operation of a large number of vehicles in a fleet, can be increased. Data from the vehicle's environment is structured as attributes of a layered model. A conventional layered model with, for example, six layers can be used.Such a layered model comprises, for example, a road network and road surface conditions in a first layer; information and attributes from the vehicle's environment in a second layer, such as houses and / or trees at the roadside; temporary events, such as construction sites that persist for an extended period, in a third layer; dynamic objects in a fourth layer; environmental data, such as weather conditions and lighting conditions, in a fifth layer; and communication information, such as internet connectivity and / or mobile network coverage, in a sixth layer. The decision tree illustrates the weighting of specific attributes in relation to a malfunction of a driving function. This means, for example, that the influence of a given attribute on the classification of a fault or passivation of the affected function is greatest depending on its position in the decision tree.The higher up in the decision tree the attribute is, the greater its influence on the classification of the error or the passivation of the function.

[0005] A preferred embodiment of the method is characterized in that the motor vehicle and its environment are considered as a cyber-physical system, including all data acquired within the vehicle and its environment, whereby attributes with the greatest influence are extracted and explored. This leads to a significant reduction of the relevant data space.

[0006] Another preferred embodiment of the method is characterized in that vehicle bus data is acquired during real-world driving. The acquired vehicle bus data is stored either internally or externally within the vehicle. The acquired vehicle bus data can be processed either internally or externally, i.e., in a backend.

[0007] Another preferred embodiment of the method is characterized in that the incoming vehicle bus data includes information such as error throw-offs or passivation by a driver. An error throw-off or passivation by the driver is generated, for example, when an automated parking process is interrupted by the driver.

[0008] Another preferred embodiment of the method is characterized in that a machine learning model is trained on the basis of the acquired vehicle bus data in the decision tree for the classification of the acquired vehicle bus data. This allows, for example in the back end, the identification of which data is relevant for a previous error drop using historical data.

[0009] Another preferred embodiment of the method is characterized in that only data identified by the model as relevant influencing factors are transmitted to a back end via the bus system. This effectively increases the efficiency of vehicle operation, particularly with regard to the amount of data to be handled.

[0010] The invention further relates to a computer program product comprising instructions, the execution of which by a computing device follows a previously described data-driven procedure. Data-driven means, in particular, that the causes of malfunctions are identified more easily and effectively.

[0011] The invention further relates to a control unit with such a computer program product. The control unit is, for example, an in-vehicle control unit. However, the term "control unit" also includes a control unit that operates in the back end.

[0012] Further advantages, features and details of the invention will become apparent from the following description, in which various embodiments are described in detail with reference to the drawing.

[0013] They show: Fig. 1. A schematic representation of a control unit with data that is transmitted via a bus system during the operation of a motor vehicle; and Fig. 2 a schematic representation of the control unit Fig. 1 with a motor vehicle and a decision tree to illustrate the claimed method.

[0014] In Fig. 1 is a control unit 1 with a large amount of data, shown in a highly simplified manner. Control unit 1 is, for example, an in-vehicle control unit located in a motor vehicle that is in Fig. Control unit 2, labeled with 20, is located there. However, control unit 1 can also be an external control unit located in a backend, i.e., outside the vehicle.

[0015] The data 5 includes, for example, map data 2, bus system data 3, and weather data 4. The bus system data 3 is provided via a bus system 6, which is connected to vehicle-internal components such as sensors.

[0016] The data (5) can be provided via an Operation Design Domain (10). Operation Design Domain (10) is also abbreviated with the capital letters ODD. ODD 10 includes, for example, information from the environment or surroundings.

[0017] The data are advantageously structured in a layered model 17. Such a layered model 17 comprises, for example, six layers 11 to 16. A first layer 11 comprises, for example, topology or geometry data. A second layer 12 comprises, for example, traffic data. A third layer 13 comprises, for example, load data or transport data. A fourth layer 14 comprises, for example, dynamic object data. A fifth layer 15 comprises, for example, environmental data. A sixth layer 16 comprises, for example, communication data, such as mobile network and traffic light data.

[0018] In Fig. Figure 2 illustrates how the data 5 are trained in a machine learning model 18 with regard to the driving function of the vehicle 20. The search space of factors from the vehicle's environment 20 that affect the behavior of a driving function becomes increasingly complex as the operational domains of the driving functions expand. This makes the influencing factors increasingly difficult for humans to assess. The more environmental factors are present, the greater the test complexity and the number of test cases and simulations to be executed.

[0019] The claimed method provides a selection procedure for a driving function, for which vehicle bus data is acquired during real-world driving. This data also includes information such as error rejections or passivation by the driver.

[0020] First, the machine learning model 18 is trained on the basis of a decision tree 30 to classify these error rejections or passivations. Data from the vehicle's environment 20 serve as input parameters. This data is, as in Fig. 1 illustrated, structured as attributes of the layer model 17.

[0021] The structure of decision tree 30 with influencing factors 31 to 36 after training with machine learning model 18 reveals the weighting of certain attributes in relation to the malfunction of a driving function. The higher up in decision tree 30 an attribute is, the greater its influence on the classification of a fault or a passivation of the function. Attributes with the greatest influencing factor can thus be extracted and explored, leading to a reduction in the test space. Reference sign 1 control unit 2 map data 3 Bus system data 4 weather data 5 data 6 bus system 10 Operation Design Domain 11 first shift 12 second shift 13 third shift 14 fourth shift 15 fifth shift 16 sixth shift 17-layer model 18 Model for machine learning 20 motor vehicles 30 Decision tree 31 influencing factors 32 influencing factors 33 influencing factors 34 influencing factors 35 influencing factors 36 influencing factors

Claims

[1] Method for operating a motor vehicle (20) comprising a bus system (3) through which data (5) are transmitted which influences at least one driving function of the motor vehicle (20), characterized by , that data (5) relevant to the driving function are selected in the form of influencing factors (31-36) in a decision tree (30) and prioritized with respect to the driving function in order to significantly reduce the amount of data (5) to be transmitted via the bus system (3), wherein the motor vehicle (20) in its environment is considered as a cyber-physical system with all data (5) captured in the motor vehicle (20) and its environment, wherein attributes with the greatest influencing factor (31-36) are extracted and explored, wherein data (5) from the environment of the vehicle (20) are structured as attributes of a layer model (17), wherein the decision tree (30) shows a weighting of certain attributes with respect to a malfunction of a driving function. [2] Method according to claim 1, characterized by , that vehicle bus data is collected during real-world driving tests. [3] Method according to claim 2, characterized by that the recorded vehicle bus data includes information such as error messages or passivations by the driver. [4] Method according to claim 2 or 3, characterized by , that a machine learning model (18) is trained on the basis of the incoming vehicle bus data in the decision tree (30) for the classification of the incoming vehicle bus data. [5] Method according to claim 4, characterized by , that only data (5) identified by the model (18) as relevant influencing factors (31-36) are transmitted to a backend via the bus system (3). [6] Computer program product comprising instructions, the execution of which by a computing device is a data-driven method according to one of the preceding claims. [7] Control unit (1) with a computer program product according to claim 6. [8] Motor vehicle with a bus system (3) and a control unit (1) according to claim 7.

Images