Warn the driver of the vehicle of a dangerous situation

Abstract

The present invention relates to a method and a driver assistance system for warning a driver of a vehicle of a dangerous situation. The method includes: performing machine learning on a driver profile using driving data of the driver, wherein the driver profile includes driving maneuver dynamics for various driving maneuvers respectively; determining a threshold for a driving maneuver according to the amount of driving data regarding the driving data, wherein the threshold illustrates a deviation from the driving maneuver dynamics of the driver profile; predicting a driving maneuver to be completed based on navigation data and / or vehicle telemetry data; determining the expected dynamics of the driving maneuver to be completed; determining a deviation between the expected dynamics of the driving maneuver to be completed and the driving maneuver dynamics of the driver profile; comparing the deviation of the expected dynamics of the driving maneuver to be completed with the threshold for the driving maneuver; and when the deviation of the expected dynamics of the driving maneuver to be completed exceeds the threshold, outputting a warning prompt to the driver.

Description

Field of the Invention

[0001] The present invention relates to a method and a driver assistance system for warning a driver of a vehicle of a dangerous situation. Background Art

[0002] A driver assistance system (Advanced Driver Assistance System (ADAS) in English, abbreviated as ADAS) assists the driver of a vehicle in certain driving situations. Known driver assistance systems assist the driver in, for example, steering, braking, parking, and / or setting the driving speed. Also known are driver assistance systems that have a so-called Vehicle Monitoring System (VMS), which is also referred to as an In-Vehicle Monitoring System (IVMS) or a Driver Monitoring System (DMS). Such systems typically have sensors for detecting the interior space of the vehicle and / or the environment around the vehicle, and evaluate the data detected by the sensors in order to assist the driver of the vehicle. Sensors used for this purpose are, for example, cameras, radar sensors, and / or lidar sensors. Such systems can in particular be set up to analyze the driver's attention and / or driving behavior and to warn the driver when a lack of attention and / or serious driving behavior is recognized. Such systems can also be set up to analyze the driving situation and to warn the driver when a dangerous driving situation is recognized. Summary of the Invention

[0003] The object on which the present invention is based is to provide a method and a driver assistance system for warning a driver of a vehicle of a dangerous situation, which method and driver assistance system are improved in view of taking into account driver-individual characteristics.

[0004] According to the present invention, this object is solved by a method for warning a driver of a vehicle of a dangerous situation, which method has the following method steps:

[0005] - performing machine learning on a personalized driver profile using driving data regarding the driving of the driver, wherein the driver profile includes driving maneuver dynamics for various driving maneuvers;

[0006] - determining a threshold value for the driving maneuver based on the amount of driving data of the driving data regarding the driving maneuver, wherein the threshold value indicates a deviation from the driving maneuver dynamics of the driver profile;

[0007] - Predict a driving maneuver to be completed based on navigation data and / or vehicle telemetry data;

[0008] - Determine the expected dynamics of the driving maneuver to be completed;

[0009] - Obtain the deviation between the expected dynamics of the driving maneuver to be completed and the driving maneuver dynamics of the driver profile;

[0010] - Compare the deviation of the expected dynamics of the driving maneuver to be completed with a threshold value of the driving maneuver; and

[0011] - When the deviation of the expected dynamics of the driving maneuver to be completed exceeds the threshold value of the driving maneuver, output a warning prompt to the driver.

[0012] Here, a driving maneuver is understood as a specific movement or movement change of a vehicle, which is caused by the driver of the vehicle. Examples of driving maneuvers are cornering of a vehicle, which is characterized by the characteristics of the corner. Such characteristics are, for example, the length, curvature, and / or inclination of the corner. Another example of a driving maneuver is braking of a vehicle in front of a specific traffic infrastructure element, such as braking in front of a traffic light, intersection, tunnel, or bridge.

[0013] The driving maneuver dynamics are characterized by the values of the motion parameters with which the driving maneuver is implemented. Such motion parameters are, for example, the acceleration and speed of the vehicle, especially depending on time and / or relative to position, for example, in the case of cornering relative to the apex of a corner, or in the case of braking of a vehicle in front of a specific traffic infrastructure element with respect to a defined distance from the traffic infrastructure element.

[0014] Navigation data is, for example, the vehicle position of the vehicle and the map data of a digital map, and optionally the route followed by the vehicle. The vehicle position of the vehicle is determined, for example, in the case of using a navigation satellite system, such as in the case of using GPS, GLONASS, Beidou, or Galileo. The map data of the digital map is provided, for example, by a storage unit arranged in the vehicle, or the map data of the digital map is retrieved from a data cloud (English: Cloud), for example, via a radio connection. The route followed by the vehicle is calculated, for example, by a navigation device. The route is, for example, a route used by the navigation device for active destination guidance, and its starting position and / or destination position are obtained, for example, from the following input, and the user of the navigation device inputs this input into the navigation device through an operation unit. Alternatively, for example, in the case of using stored data on the driver's previous driving, the route is obtained, for example, by the navigation device itself.

[0015] Vehicle telemetry data is understood to be sensor data detected by sensors arranged in or on the vehicle. Such sensors are, for example: cameras, radar sensors and / or lidar sensors for respectively detecting the interior space of the vehicle and / or the environment around the vehicle; and / or motion sensors for respectively detecting, for example, the motion state of the vehicle such as acceleration and / or speed.

[0016] According to the method of the present invention, it is possible to assist the driver of a vehicle in terms of driving maneuvers taking into account the driver's individual characteristics. For this purpose, the method provides for: creating a personalized driver profile of the driver; evaluating the driving maneuvers to be performed in view of possible dangerous situations using the driver profile; and warning the driver of dangerous situations as required.

[0017] Using the driving data of the driver's previous driving, machine learning is performed on the personalized driver profile of the driver. That is, the driving data of the driver's previous driving is collected and used to create the driver profile. These driving data particularly characterize the dynamics of the driving maneuvers that have been performed in the previous driving. Using these driving data, the driver-specific driving maneuver dynamics are learned for various driving maneuvers and stored in the driver profile. Thus, the driver profile provides, for various driving maneuvers, the driving maneuver dynamics that characterize the typical execution of the driving maneuver for this driver.

[0018] In order to evaluate the driving maneuvers to be performed using the personalized driver profile of the driver, a threshold is determined for the driving maneuver, which threshold indicates the deviation from the driving maneuver dynamics for the driving maneuver stored in the driver profile. This threshold is determined based on the amount of driving data for the driving maneuver, and the driving maneuver dynamics in the driver profile have been learned based on this amount of driving data. For example, a larger threshold is specified in the case of a small amount of driving data than in the case of a large amount of driving data. This advantageously takes into account that the statistical credibility of the driving maneuver dynamics stored in the driver profile increases with the amount of driving data on which it is based.

[0019] In the method according to the invention, the driving maneuvers to be performed are predicted based on navigation data and / or vehicle telemetry data. For example, when the vehicle position approaches a bend recorded on a digital map, the bend driving to be performed is predicted based on the vehicle position and the digital map. If the navigation data includes the route traveled by the vehicle, the bend driving to be performed can also be predicted based on this route. Alternatively or additionally, the bend driving to be performed can be predicted using vehicle telemetry data, for example by evaluating camera images of the vehicle's camera and / or by evaluating sensor signals of the vehicle's radar sensor and / or lidar sensor. The corresponding applies to the prediction of other driving maneuvers.

[0020] For the driving maneuvers to be performed, the expected dynamics are determined. For this purpose, vehicle telemetry data and navigation data are used, for example. For example, the expected speed, the expected lateral acceleration and / or the expected longitudinal acceleration are determined for the bend driving to be performed.

[0021] Then, the deviation of the expected dynamics of the driving maneuver to be performed from the driving maneuver dynamics of the driver profile for this driving maneuver is determined and compared with the threshold value of this driving maneuver. If the deviation of the expected dynamics of the driving maneuver to be performed exceeds the threshold value of this driving maneuver, a warning message is output to the driver of the vehicle.

[0022] For example, visual, acoustic and / or haptic warning signals are output as warning messages. For example, a visual warning message can be output on the display unit of the vehicle. The display unit is, for example, a liquid crystal display (LCD), an OLED (organic light emitting diode) display or a head-up display. For example, in the case where the driving maneuver is bend driving, the visualization of the bend can be highlighted in color as a visual message. Alternatively or additionally, an acoustic warning message can be output using the speaker unit of the vehicle, and / or a haptic signal can be output using the haptic warning unit, for example via the vehicle's steering wheel. A multi-level warning strategy can also be provided. For example, a visual warning message can first be output, and if the driver does not respond or if the threshold value is exceeded continuously, an acoustic and / or haptic warning message can then be output.

[0023] In summary, the method according to the present invention enables an evaluation of the driving maneuver to be performed, which is matched to the individual driving behavior and driving ability of the driver, in view of possible dangerous situations, by determining the expected dynamics of the driving maneuver to be performed and comparing it with the driving maneuver dynamics stored in the driver profile for this driving maneuver. Thereby, in particular, it is possible to avoid outputting unnecessary warning messages when the expected dynamics of the driving maneuver to be performed deviate only slightly from the driving maneuver dynamics stored in the driver profile and it can therefore be assumed that the driver can perform this driving maneuver without danger. On the other hand, when the expected dynamics of the driving maneuver to be performed deviate significantly from the driving maneuver dynamics stored in the driver profile, the driver can be warned. Thereby, in particular, the unreliability or lack of experience of the driver can be taken into account in the case of a specific driving maneuver, or the output of the warning message can be advantageously matched to the experience and driving ability of the driver.

[0024] In one embodiment of the method according to the present invention, when the amount of driving data of the driving data of a driving maneuver is below a minimum amount of data, additional driving data on the driving of other drivers is also used to perform machine learning on the driver profile for this driving maneuver.

[0025] The above-described embodiment of the method according to the present invention takes into account the following situation: for a driving maneuver, the driving data of the driving of the driver of the vehicle is not sufficient to learn the statistically significant driving maneuver dynamics of this driving maneuver based on these driving data. In view of this situation, this embodiment provides that, in addition to or instead of the driving data on the driving of the driver of the vehicle, additional driving data on the driving of other drivers is used to learn the driving maneuver dynamics of this driving maneuver. Thereby, even in the case of a lack or insufficiency of driving data in the driving of the driver of the vehicle for a driving maneuver, the driving maneuver dynamics of this driving maneuver can be learned.

[0026] In another embodiment of the method according to the present invention, an average driving maneuver dynamics is determined based on the additional driving data on the driving of other drivers, the average driving maneuver dynamics is compared with the driving maneuver dynamics of the driver profile, and the threshold value is also determined based on the deviation of the driving maneuver dynamics from the average driving maneuver dynamics.

[0027] The above-described embodiments of the method according to the invention take into account that a significant deviation between the learned driving maneuver dynamics of a motorized driving maneuver, which has been learned based on driving data regarding the driver's driving, and the average driving maneuver dynamics typical for other drivers may indicate that the learned driving maneuver dynamics may not match the driving maneuver. Thus, this embodiment provides that the threshold is also determined based on the deviation between the driving maneuver dynamics and the average driving maneuver dynamics. For example, as the deviation between the driving maneuver dynamics and the average driving maneuver dynamics increases, the threshold is reduced so that a warning message is output earlier when the deviation is large than when it is small.

[0028] In the above two embodiments of the method according to the invention, the additional driving data is provided, for example, by a data cloud service, that is to say, by a data cloud service that collects driving data regarding the driving of different drivers. This enables the permanent expansion and updating of the additional driving data to be advantageously achieved, as well as access to the updated additional driving data as required, for example via a radio connection to the data cloud service.

[0029] In another embodiment of the method according to the invention, for a driving maneuver, an average value and a variance of the distribution of the values of the motion parameters characterizing the driving maneuver dynamics are determined based on the driving data, and a deviation from the average value depending on the variance is determined as the threshold for the driving maneuver. If the driving maneuver dynamics are characterized by a plurality of motion parameters, the average value and the variance can be determined separately for the distribution of the values of these motion parameters, and the threshold is determined as the deviation from the average value depending on the variance. For example, when the corresponding deviation between the expected dynamics of the driving maneuver to be performed and the driving maneuver dynamics of the driving maneuver exceeds one of these thresholds, a warning message is output.

[0030] Additionally, it can be provided that, for a driving maneuver, the effectiveness of the driver profile for the driving maneuver is evaluated based on the amount of driving data of the driving maneuver, and the threshold for the driving maneuver is determined as the deviation from the average value that also depends on this effectiveness.

[0031] That is, according to the above-described embodiments of the method of the present invention, a statistical evaluation of driving data regarding a driving maneuver is proposed, wherein a threshold value is determined based on the distribution of values of motion parameters characterizing the dynamics of the driving maneuver determined from these driving data. It is also possible to evaluate the effectiveness of a driver profile for the driving maneuver based on the amount of driving data of the driving maneuver, and use this effectiveness to determine the threshold value. Thereby, the threshold value of the driving maneuver can be advantageously matched to the statistical reliability of the driving data regarding the driving maneuver. For example, it can be stipulated that as the variance of the distribution decreases and the effectiveness of the driver profile for the driving maneuver increases, the threshold value is reduced, because the learned driving maneuver dynamics become more reliable as the variance decreases and the effectiveness increases.

[0032] In another embodiment of the method according to the present invention, the threshold value of the driving maneuver is also determined based on the weather conditions at the location of the driving maneuver and / or based on the driver's attention at the time point of the driving maneuver.

[0033] The above-described embodiments of the method according to the present invention take into account that the danger of a driving maneuver depends on the weather conditions at the location of the driving maneuver and the driver's attention at the time point of the driving maneuver. For example, it can be stipulated that when the weather conditions are adverse and / or the driver's attention is limited, the threshold value of the driving maneuver is reduced relative to favorable weather conditions or full driver attention, so that a warning message is output earlier.

[0034] In another embodiment of the method according to the present invention, the driving maneuver dynamics include the longitudinal acceleration, lateral acceleration, and / or speed of the vehicle.

[0035] The above-described embodiments of the method according to the present invention take into account that the longitudinal acceleration, lateral acceleration, and / or speed of the vehicle generally have the most significant influence on the danger of the driving maneuver, and thus are the motion parameters that provide the most important criteria for the output of a warning message.

[0036] In another embodiment of the method according to the present invention, the driving maneuver location of the driving maneuver and the frequency with which the driver has performed the driving maneuver at the driving maneuver location are stored in the driver profile for the driving maneuver. It can also be stipulated that the threshold value of the driving maneuver is determined based on the frequency with which the driver has performed the driving maneuver at the driving maneuver location.

[0037] The above-described embodiment of the method according to the invention takes into account that the frequency with which a driver has carried out a driving maneuver at a specific driving maneuver position is a measure of the driver's familiarity with carrying out the driving maneuver at that driving maneuver position. If, for example, the driver has carried out the driving maneuver very frequently at that driving maneuver position, it can be assumed that the driver is familiar with the driving maneuver position and proficient in the driving maneuver at that driving maneuver position. Thus, it is advantageous to also characterize the driving maneuver in the driver profile by the driving maneuver position and the frequency with which the driver has carried out the driving maneuver at that driving maneuver position; and to also determine the threshold value of the driving maneuver according to the frequency with which the driver has carried out the driving maneuver at that driving maneuver position.

[0038] According to the invention, the task is also solved by a driver assistance system for warning the driver of a vehicle of a dangerous situation, the driver assistance system having the following functional units:

[0039] - A learning unit configured to perform machine learning on a personalized driver profile using driving data of the driver's driving, wherein the driver profile includes driving maneuver dynamics for various driving maneuvers respectively;

[0040] - A threshold determination unit configured to determine a threshold value of the driving maneuver according to the amount of driving data of the driving maneuver, wherein the threshold value indicates a deviation from the driving maneuver dynamics of the driver profile;

[0041] - A prediction unit configured to predict a driving maneuver to be completed based on navigation data and / or vehicle telemetry data;

[0042] - A dynamics determination unit configured to determine the expected dynamics of the driving maneuver to be completed;

[0043] - A deviation calculation unit configured to calculate a deviation between the expected dynamics of the driving maneuver to be completed and the driving maneuver dynamics of the driver profile;

[0044] - A comparison unit configured to compare the deviation of the expected dynamics of the driving maneuver to be completed with the threshold value of the driving maneuver; and

[0045] - An output unit configured to output a warning prompt to the driver when the deviation of the expected dynamics of the driving maneuver to be completed exceeds the threshold value of the driving maneuver.

[0046] Such a driver assistance system can execute the method according to the invention. Thus, the advantages of such a driver assistance system are derived from the advantages of the method according to the invention mentioned above. Description of the Drawings

[0047] The embodiments of the present invention will be further described below with reference to the accompanying drawings. Herein:

[0048] Figure 1 A block diagram of an embodiment of a driver assistance system according to the present invention is shown;

[0049] Figure 2 A flowchart of an embodiment of a method according to the present invention is shown;

[0050] Figure 3 An illustration of a driving maneuver of a vehicle is shown. Detailed Embodiments

[0051] Figure 1 (FIG 1) shows a block diagram of a driver assistance system 100 according to an embodiment of the present invention for a vehicle. The driver assistance system 100 is configured to warn the driver of the vehicle of dangerous situations and has the following functional units depicted in Figure 1 : a learning unit 101, a threshold determination unit 102, a prediction unit 103, a dynamics determination unit 104, a deviation calculation unit 105, a comparison unit 106, and an output unit 107. In addition to the mentioned functional units, the driver assistance system 100 can have other functional units not shown in Figure 1 , such as a storage unit, a communication unit for data exchange with other devices, such as a data cloud, and / or a receiving unit for receiving traffic information.

[0052] The learning unit 101 is configured to perform machine learning on a personalized driver profile using driving data regarding the driving of the driver, where the driver profile includes driving maneuver dynamics for various driving maneuvers.

[0053] The driving maneuver dynamics of a driving maneuver are characterized by the values of one or more motion parameters with which the driving maneuver is performed. Such motion parameters are, for example, the longitudinal acceleration, lateral acceleration, and / or speed of the vehicle, especially depending on time and / or relative to position, for example, in the case of cornering driving relative to the apex of a curve, or in the case of braking of the vehicle with respect to a defined distance from a specific traffic infrastructure element in front of the traffic infrastructure element.

[0054] For example, driving data regarding the driving of the driver are collected in the storage unit of the driver assistance system 100.

[0055] The learning unit 101 can also be configured such that when the amount of driving data of a motor vehicle driving is lower than the minimum amount of data, additional driving data regarding the driving of other drivers is also used to perform machine learning on the driver profile of the motor vehicle driving. Such additional driving data is collected and provided, for example, by a data cloud service, and is called by the driver assistance system 100 from the data cloud service, for example, via a radio connection.

[0056] The threshold determination unit 102 is configured to determine a threshold for the motor vehicle driving based on the amount of driving data of the motor vehicle driving, where the threshold indicates a deviation from the driving dynamics of the driver profile of the motor vehicle driving.

[0057] The threshold determination unit 102 is, for example, configured to, for a motor vehicle driving, determine the mean value and variance of the distribution of the values of the motion parameters characterizing the driving dynamics based on the driving data of the driver's driving, and determine the deviation from the mean value depending on the variance as the threshold for the motor vehicle driving. If the driving dynamics of the motor vehicle driving are characterized by multiple motion parameters, the threshold determination unit 102 can also be configured to, for the distribution of the values of these motion parameters, respectively determine the mean value and variance, and respectively determine the threshold as the deviation from the mean value depending on the variance.

[0058] Additionally, the threshold determination unit 102 can be configured to, for a motor vehicle driving, evaluate the effectiveness of the driver profile for the motor vehicle driving based on the amount of driving data of the motor vehicle driving, and determine the threshold for the motor vehicle driving as the deviation from the corresponding mean value that also depends on the effectiveness.

[0059] The threshold determination unit 102 can also be configured to determine the threshold for the motor vehicle driving also based on the weather conditions at the location of the motor vehicle driving, such as the traffic conditions of traffic density and / or traffic flow, and / or based on the driver's attention at the time point of the motor vehicle driving.

[0060] The threshold determination unit 102 can also be configured to determine the threshold for the motor vehicle driving also based on the frequency at which the driver has performed the motor vehicle driving at the motor vehicle driving location. The precondition is that the driving location of the motor vehicle driving and the frequency at which the driver has performed the motor vehicle driving at the motor vehicle driving location are also stored in the driver profile having the driving dynamics of the motor vehicle driving.

[0061] The prediction unit 103 is configured to predict the motor vehicle driving to be completed based on the navigation data and / or vehicle telemetry data.

[0062] The prediction unit 103 uses, for example, the vehicle position of the vehicle and the map data of the digital map, and, if necessary, the route followed by the vehicle as navigation data. The vehicle position of the vehicle is determined, for example, in the case of using a navigation satellite system, for example, in the case of using GPS, GLONASS, Beidou or Galileo. The map data of the digital map is provided, for example, by a storage unit arranged in the vehicle and which is, for example, a functional unit of the driver assistance system 100, or the map data of the digital map is retrieved from a data cloud, for example, via a radio connection. The route followed by the vehicle is calculated, for example, by the navigation device. The route is, for example, a route that is used by the navigation device for active destination guidance and whose starting position and / or destination position are derived, for example, from an input that the user of the navigation device inputs into the navigation device via an operating unit. Alternatively, the route is determined, for example, by the navigation device itself, for example, in the case of using the previously stored driving data of the driver.

[0063] The prediction unit 103 uses sensor data as vehicle telemetry data, which is detected by a sensor arranged in or on the vehicle or a plurality of such sensors. Such sensors are, for example: cameras, radar sensors and / or lidar sensors for respectively detecting the interior space of the vehicle and / or the surrounding environment of the vehicle; and / or motion sensors for respectively detecting the motion state of the vehicle, such as acceleration and / or speed.

[0064] The dynamics determination unit 104 is provided for determining the expected dynamics of the driving maneuver to be performed. For this purpose, the dynamics determination unit 104 uses, for example, vehicle telemetry data and navigation data. For example, the dynamics determination unit 104 determines the expected speed, expected lateral acceleration and / or expected longitudinal acceleration at a position in the curve (for example, at the apex of the curve) or at a plurality of positions in the curve according to the current speed of the vehicle, the current distance of the vehicle from the curve and / or the curve characteristics of the curve, such as the length and curvature of the curve, for the curve driving to be performed.

[0065] The deviation calculation unit 105 is provided for calculating the deviation between the expected dynamics of the driving maneuver to be performed and the driving maneuver dynamics of the driver profile.

[0066] The comparison unit 106 is provided for comparing the deviation of the expected dynamics of the driving maneuver to be performed with a threshold value of the driving maneuver.

[0067] The output unit 107 is provided for outputting a warning prompt to the driver when the deviation of the expected dynamics of the driving maneuver to be performed exceeds the threshold value of the driving maneuver.

[0068] For example, the output unit 107 is configured to output visual, acoustic and / or tactile warning signals as warning prompts. The visual warning prompt is output by the output unit 107, for example, on a display unit. The display unit is, for example, an LCD display, an OLED display, or a head-up display of the vehicle. Alternatively or additionally, the output unit 107 is configured to output an acoustic warning prompt for the driving maneuver using a speaker unit; and / or to output a tactile signal using a tactile warning unit, for example, via the steering wheel of the vehicle. The output unit 107 can also be configured for a multi-stage warning strategy. For example, a visual warning prompt can be output first, and if the driver does not react or if the threshold value is continuously exceeded, an acoustic and / or tactile warning prompt can then be output.

[0069] The driver assistance system 100 is, for example, a vehicle monitoring system of a vehicle or is part of a vehicle monitoring system of a vehicle. The learning unit 101, the threshold value determination unit 102, the prediction unit 103, the dynamics determination unit 104, the deviation ascertainment unit 105, and the comparison unit 106 each include, for example, a computer program that is executed on a computing unit of the driver assistance system 100.

[0070] Figure 2 FIG2 shows a flow chart 200 of a method for warning a driver of a vehicle of a dangerous situation according to one embodiment of the present invention, with method steps 201 to 207. The method uses the reference Figure 1 The described driver assistance system 100 is implemented within the context of the present invention.

[0071] Then also refer to Figure 3 Method steps 201 to 207 are described below.

[0072] In a first method step 201 , the learning unit 101 of the driver assistance system 100 performs machine learning of a personalized driver profile using driving data about the driver's driving, wherein the driver profile comprises the driving maneuver dynamics for various driving maneuvers.

[0073] Figure 3 FIG3 shows, by way of example, a diagram of a driving maneuver 300 of a vehicle 301. The driving maneuver 300 shown is a curve driving through a curve 302 of a road 303. The driving maneuver 300 is characterized in the driver profile by the curve characteristics of the curve 302, for example, by the length of the curve 302, the average curvature of the curve 302, the maximum curvature of the curve 302 and / or the inclination of the curve 302. In addition, the driving maneuver 300 can be characterized in the driver profile by other data, for example, by the driving maneuver position of the curve 302 and / or the frequency with which the driver has driven through the curve 302.

[0074] In Figure 3 the driving dynamics of the driving maneuver 300 shown, for example, include the speed 304, longitudinal acceleration 305, and / or lateral acceleration 306 of the vehicle 301 at a specific position in the curve 302, such as at the apex of the curve 302 or at multiple positions in the curve 302. In Figure 3 it, the speed 304, longitudinal acceleration 305, and lateral acceleration 306 are each shown by an arrow indicating the direction of the corresponding motion parameter, where it is assumed that the vehicle 301 brakes such that the longitudinal acceleration 305 is in the opposite direction to the speed 304. Correspondingly, the driving dynamics of the driving maneuver 300 are learned using driving data of the driver's previous driving, in which the driver has performed the driving maneuver 300 and the speed 304, longitudinal acceleration 305, and / or lateral acceleration 306 of the vehicle 301 have been detected during the performance of the driving maneuver 300. The driving dynamics of other driving maneuvers are learned correspondingly, such as cornering through a curve with different curve characteristics or braking of the vehicle 301 in front of specific traffic infrastructure elements such as traffic lights, intersections, tunnels, or bridges.

[0075] If the amount of driving data of the driving maneuver is below the minimum amount of data, additional driving data regarding the driving of other drivers can also be used for machine learning of the driver profile of this driving maneuver, such as additional driving data collected and provided by a data cloud service and called by the driver assistance system 100 from the data cloud service.

[0076] In a second method step 202, the threshold determination unit 102 of the driver assistance system 100 determines a threshold for this driving maneuver based on the amount of driving data of the driving maneuver, where the threshold indicates the deviation from the driving dynamics of the driver profile.

[0077] For example, for a driving maneuver, the mean and variance of the distribution of the values of the motion parameters characterizing the driving dynamics are determined based on the driving data of the driver's driving, and a deviation from the mean depending on the variance is determined as the threshold for this driving maneuver. If the driving dynamics are characterized by multiple motion parameters, the mean and variance are determined, for example, separately for the distributions of the values of these motion parameters, and the threshold is determined as the deviation from the mean depending on the variance. That is, for example, for the one in Figure 3In the driving maneuver 300 shown in FIG. 1 , a mean value and a variance of the distribution of the values ​​of the speed 304, the longitudinal acceleration 305 and / or the lateral acceleration 306 of the vehicle 301 at a specific position in a curve 302 or at multiple positions in the curve 302 are determined based on driving data about the driver's driving, and a threshold value is determined as a deviation from the mean value depending on the variance.

[0078] Additionally, the validity of the driver profile for a driving maneuver may be evaluated for the driving maneuver based on the driving data volume of the driving data of the driving maneuver, and a threshold value for the driving maneuver may be determined as a deviation from a corresponding mean value that also depends on the validity.

[0079] Furthermore, the threshold value for the driving maneuver can also be determined in addition to weather conditions at the location of the driving maneuver, traffic conditions such as traffic density and / or traffic volume, and / or driver attentiveness of the driver at the time of the driving maneuver. For example, if atypical driving maneuvers of the driver are detected, in particular those that deviate from the driver's profile, such as atypically frequent lane changes or abnormal lateral or longitudinal acceleration of the vehicle, or if driving maneuvers that deviate significantly from the driver's profile occur frequently, then insufficient driver attentiveness of the driver is inferred.

[0080] Furthermore, the threshold value of the driving maneuver may also be determined based on the frequency with which the driver has performed the driving maneuver in the driving maneuver position.

[0081] In a third method step 203 , prediction unit 103 of driver assistance system 100 predicts a driving maneuver to be performed based on navigation data and / or vehicle telemetry data.

[0082] For example, the vehicle position of the vehicle and the map data of a digital map and, if necessary, the route followed by the vehicle are used as navigation data. The vehicle position of the vehicle is determined, for example, using a navigation satellite system, such as GPS, GLONASS, BeiDou or Galileo. For example, the map data of the digital map are provided by a storage unit arranged in the vehicle and, for example, a functional unit of the driver assistance system 100, or the map data of the digital map are called up from a data cloud, for example, via a radio connection. For example, the route followed by the vehicle is calculated by a navigation device. This route is, for example, a route that is used by the navigation device for active destination guidance and whose starting position and / or destination position are derived, for example, from the following input, which is entered into the navigation device by the user of the navigation device via an operating unit. Alternatively, the route is determined, for example, by the navigation device itself, for example, using stored data about the driver's previous driving.

[0083] Using sensor data detected by one sensor arranged in or on a vehicle or a plurality of such sensors as vehicle telemetry data. Such sensors are, for example: cameras, radar sensors and / or lidar sensors for respectively detecting the interior space of the vehicle and / or the environment around the vehicle; and / or motion sensors for respectively detecting, for example, the motion state of the vehicle such as acceleration and / or speed.

[0084] For example, the prediction unit 103 identifies that the vehicle is approaching a bend with specific bend characteristics based on navigation data and / or vehicle telemetry data, and predicts the corresponding bend driving as a driving maneuver.

[0085] In the fourth method step 204, the dynamics determination unit 104 of the driver assistance system 100 determines the expected dynamics of the driving maneuver to be performed.

[0086] For this purpose, the dynamics determination unit 104 uses, for example, vehicle telemetry data and navigation data. For example, for the driving maneuver 300 to be performed as shown in Figure 3 , the expected speed 304, the expected lateral acceleration 305 and / or the expected longitudinal acceleration 306 at a position in the bend 302 (for example, at the bend apex) or at multiple positions in the bend are determined based on the current speed of the vehicle, the current distance of the vehicle from the bend 302, and the bend characteristics of the bend 302, such as the length and curvature of the bend 302.

[0087] In the fifth method step 205, the deviation calculation unit 105 of the driver assistance system 100 calculates the deviation between the expected dynamics of the driving maneuver to be performed and the driving maneuver dynamics of the driver profile.

[0088] For example, in the case of the driving maneuver 300 to be performed as shown in Figure 3 , the deviation between the expected speed 304, the expected longitudinal acceleration 305 and / or the expected lateral acceleration 306 of the vehicle 301 at a characteristic position in the bend 302 or at multiple positions in the bend 302 and the respective corresponding values stored in the driving maneuver dynamics of the driver profile for the driving maneuver 300 is determined.

[0089] In the sixth method step 206, the comparison unit 106 of the driver assistance system 100 compares the deviation of the expected dynamics of the driving maneuver to be performed with the threshold value of this driving maneuver.

[0090] In the seventh method step 207, when the deviation of the expected dynamics of the driving maneuver to be performed exceeds the threshold value of this driving maneuver, the output unit 107 of the driver assistance system 100 outputs a warning prompt to the driver.

[0091] For example, output visual, audible, and / or haptic warning signals as warning cues. For example, output a visual warning cue on a display unit. The display unit is, for example, an LCD display, an OLED display, or a head-up display of a vehicle. For example, in the case of the driving maneuver 300 to be completed shown in Figure 3 , visualize the bend 302, for example, by highlighting it in color as a visual warning cue.

[0092] Output an audible warning cue, for example, using the speaker unit of the vehicle. Output a haptic warning cue, for example, via the steering wheel of the vehicle.

[0093] It is also possible to output warning cues in multiple levels. For example, first output a visual warning cue and then output an audible and / or haptic warning cue if the driver does not respond or if the threshold is exceeded continuously.

Claims

1. A method for warning a driver of a vehicle (301) of a dangerous situation, the method having the following method steps: - Performing machine learning on a personalized driver profile using driving data regarding the driving of the driver, wherein the driver profile includes driving maneuver dynamics for various driving maneuvers (300) respectively; - Determining a threshold value for the driving maneuver (300) according to the amount of driving data of the driving data regarding the driving maneuver (300), wherein the threshold value illustrates a deviation from the driving maneuver dynamics of the driver profile. For the driving maneuver (300), an average driving maneuver dynamics is determined based on additional driving data regarding the driving of other drivers, the average driving maneuver dynamics is compared with the driving maneuver dynamics of the driver profile, and the threshold value is also determined according to the deviation between the driving maneuver dynamics and the average driving maneuver dynamics. The threshold value is reduced as the deviation between the driving maneuver dynamics of the driver profile and the average driving maneuver dynamics increases, so as to output a warning prompt earlier when the deviation is large than when the deviation is small; - Predicting the driving maneuver (300) to be completed based on navigation data and / or vehicle telemetry data; - Determining the expected dynamics of the driving maneuver (300) to be completed; - Obtaining the deviation between the expected dynamics of the driving maneuver (300) to be completed and the driving maneuver dynamics of the driver profile; - Comparing the deviation of the expected dynamics of the driving maneuver (300) to be completed with the threshold value of the driving maneuver (300); And - When the deviation of the expected dynamics of the driving maneuver (300) to be completed exceeds the threshold value of the driving maneuver (300), outputting a warning prompt to the driver.

2. The method according to claim 1, wherein When the amount of driving data of the driving data regarding the driving maneuver (300) is lower than the minimum data amount, additional driving data regarding the driving of other drivers is also used to perform machine learning on the driver profile of the driving maneuver (300).

3. The method according to claim 1 or 2, wherein, The additional driving data is provided by a data cloud service.

4. The method according to claim 1 or 2, wherein For the driving maneuver (300), the average value and variance of the distribution of the values of the motion parameters characterizing the driving maneuver dynamics are obtained based on the driving data, and the deviation from the average value depending on the variance is determined as the threshold value of the driving maneuver (300).

5. The method according to claim 4, wherein For the driving maneuver (300), the effectiveness of the driver profile for the driving maneuver (300) is evaluated based on the amount of driving data of the driving maneuver (300), and the threshold value of the driving maneuver (300) is determined as the deviation from the average value depending on the effectiveness.

6. The method according to claim 1 or 2, wherein The threshold value of the driving maneuver (300) is determined according to the weather conditions at the location of the driving maneuver (300), and / or according to the driver's attention at the time point of the driving maneuver (300).

7. The method according to claim 1 or 2, wherein The driving maneuver dynamics includes the longitudinal acceleration (305), lateral acceleration (306) and / or speed (304) of the vehicle (301).

8. The method according to claim 1 or 2, wherein Store the driving position of the motor vehicle (300) and the frequency with which the driver has carried out the motor vehicle operation (300) at the driving position of the motor vehicle (300) in the driver profile for the motor vehicle (300).

9. The method according to claim 8, wherein, Determine a threshold value for the motor vehicle operation (300) based on the frequency with which the driver has carried out the motor vehicle operation (300) at the driving position of the motor vehicle (300).

10. A driver assistance system (100) for warning a driver of a vehicle (301) of a dangerous situation, the driver assistance system having the following functional units: - A learning unit (101) configured to perform machine learning on a personalized driver profile using driving data regarding the driving of the driver, wherein the driver profile includes driving dynamics for various motor vehicle operations (300) respectively; - A threshold determination unit (102) configured to determine a threshold value for the motor vehicle operation (300) based on the amount of driving data regarding the motor vehicle operation (300), wherein the threshold value indicates a deviation from the driving dynamics of the driver profile. For the motor vehicle operation (300), an average driving dynamics is determined based on additional driving data regarding the driving of other drivers, the average driving dynamics is compared with the driving dynamics of the driver profile, and the threshold value is also determined based on the deviation between the driving dynamics and the average driving dynamics. The threshold value is reduced as the deviation between the driving dynamics of the driver profile and the average driving dynamics increases, so that a warning message is output earlier when the deviation is large than when the deviation is small; - A prediction unit (103) configured to predict a motor vehicle operation (300) to be performed based on navigation data and / or vehicle telemetry data; - A dynamics determination unit (104) configured to determine the expected dynamics of the motor vehicle operation (300) to be performed; - A deviation calculation unit (105) configured to calculate the deviation between the expected dynamics of the motor vehicle operation (300) to be performed and the driving dynamics of the driver profile; - A comparison unit (106) configured to compare the deviation of the expected dynamics of the motor vehicle operation (300) to be performed with the threshold value of the motor vehicle operation (300); and - An output unit (107) configured to output a warning message to the driver when the deviation of the expected dynamics of the motor vehicle operation (300) to be performed exceeds the threshold value of the motor vehicle operation (300).

Images