Method and system for risk assessment of an overtaking maneuver using a multi-model network

Abstract

The present invention relates to a method for a risk assessment of an overtaking process, in which a multi-model network (21) is formed by at least one sub-model for environmental calculation (23, 25) and at least one sub-model for trajectory calculation (22, 24) and a decision-making model for the risk assessment, in which the multi-model network (21) is trained with training data from a plurality of overtaking processes with actual vehicle trajectories of participating vehicles, wherein the training data are measured by vehicle sensors oriented towards a vehicle environment and stored on a storage unit, in which during a continuously executed drive, measurement data are transmitted to the multi-model network (21) by vehicle sensors, in which a numerical value of the risk assessment is provided by the decision-making model (26),and wherein a warning signal (27) is output when a predetermined threshold value is exceeded. Furthermore, a system is presented with which the method can be carried out.

Description

[0001] The present invention relates to a method for assessing the risk of an overtaking maneuver using a multi-model network. Furthermore, a system is presented with which the method can be implemented.

[0002] Overtaking other road users, also known as "passing," involves maneuvers such as changing lanes or driving around the other road user, usually to position oneself in front of a slower vehicle. Since a collision with an oncoming vehicle at speeds of around 100 km / h has a 90% fatality rate, overtaking maneuvers must be performed with extreme caution.

[0003] The document DE 10 2019 209 432 A1 discloses an overtaking assistance system of a motor vehicle, wherein the trajectory of an overtaking motor vehicle and an oncoming motor vehicle is predicted from sensor signals of a sensor system and a warning message is issued to the drivers if a collision between the motor vehicles is imminent due to the overtaking maneuver.

[0004] From the publication DE 10 2021 213 308 A1, an overtaking assistance system of a motor vehicle is also known, wherein the motor vehicles driving ahead of the overtaking motor vehicle are detected by an environment sensor system of the overtaking motor vehicle and a warning is issued if it is not possible to merge in front of the motor vehicle driving immediately in front of the overtaking motor vehicle.

[0005] The document DE 10 2018 120 942 A1 describes an overtaking assistance system of a motor vehicle, in which an oncoming motor vehicle is detected via an environmental sensor system and, based on this, actively intervenes in the driving operation of the motor vehicle, in particular in the steering and braking device.

[0006] The document DE102019202589A1 discusses a driver information system of a motor vehicle, in which oncoming traffic warning objects are displayed on a display element.

[0007] Against this background, it is an object of the present invention to present a method for assessing the risk of an overtaking maneuver, wherein a warning is to be issued to a vehicle driver if the overtaking maneuver exceeds a certain risk threshold. Furthermore, a system is to be presented with which the method can be carried out.

[0008] To solve the aforementioned problem, a method for risk assessment of an overtaking maneuver is proposed, in which a multi-model network is formed by at least one sub-model for environment calculation and at least one sub-model for trajectory calculation as well as a decision model for the risk assessment. The multi-model network is trained with training data from a large number of overtaking maneuvers with actual vehicle trajectories of participating vehicles, whereby the training data is measured by vehicle sensors aligned to the vehicle's environment and stored on a storage unit. During a journey, measurement data is continuously transmitted to the multi-model network by vehicle sensors. The decision model provides a numerical value of the risk assessment and, if a predetermined threshold is exceeded, a warning signal is issued.

[0009] Such a modular approach, i.e. the division into sub-models, which each feed their separately determined results into the decision model, is advantageous because it provides for or enables the addition of further models in a simple way.

[0010] The specified threshold can, for example, be set to the total uncertainty of all influencing factors when calculating the numerical value of the risk assessment. If no hazard exists, a numerical value greater than zero can still be calculated due to individual measurement uncertainties of the vehicle sensors. Only a numerical value above a specified threshold corresponding to the total uncertainty indicates an actual hazard.

[0011] In one embodiment of the method according to the invention, the at least one submodel for trajectory calculation is designed to calculate a trajectory of at least one vehicle from the following list: ego vehicle during continued driving movement, ego vehicle during overtaking maneuver, at least one vehicle driving ahead, oncoming vehicle, turning-in vehicle, vehicle merging from a parking position.

[0012] In a further embodiment of the method according to the invention, the at least one submodel for environment calculation is designed to calculate at least one vehicle environment from the following list: road surface, road type, lane markings, road layout, traffic signs, country-specific regulations, and occupancy grids based on trajectories determined by the at least one submodel for trajectory calculation. Thus, depending on the country traveled through, it is conceivable to add a model with the locally applicable regulations, e.g., for maximum speeds.

[0013] In a further embodiment of the method according to the invention, the measurement data from the vehicle sensors oriented toward the vehicle's surroundings are sent to the multi-model network via a FlexRay data bus system. At least one vehicle sensor is selected from the following list: long-range radar, medium-range radar, nano-radar, vehicle camera, LiDAR. The FlexRay data bus system advantageously provides the necessary transmission speed for the measurement data so that the computing unit can determine the numerical value of the risk assessment in real time.

[0014] In yet another embodiment of the method according to the invention, when initiating or executing an overtaking maneuver despite the presence of a warning signal, at least one measure from the following list is carried out: haptic, acoustic and / or optical driver warning, initiation of a driver assistance system to intervene in the vehicle control.

[0015] Furthermore, a system is claimed which comprises an ego vehicle, at least one vehicle sensor oriented towards a vehicle environment, a computing unit, and a memory unit. The computing unit is configured to execute a multi-model network for calculating a risk assessment of an overtaking maneuver, wherein the multi-model network is formed by at least one sub-model for calculating the environment and at least one sub-model for calculating trajectories, as well as a decision model for the risk assessment. The multi-model network is trained with training data from a plurality of overtaking maneuvers with actual vehicle trajectories of participating vehicles, wherein the training data is measured by vehicle sensors and stored on a memory unit. The at least one vehicle sensor is configured to continuously transmit measurement data to the multi-model network during a journey.The multi-model network is configured to provide a numerical value of the risk assessment using the decision model and to issue a warning signal when a specified threshold is exceeded.

[0016] In one embodiment of the system according to the invention, the at least one submodel for trajectory calculation is designed to calculate a trajectory of at least one vehicle from the following list: ego vehicle during continued driving movement, ego vehicle during overtaking maneuver, at least one vehicle driving ahead, oncoming vehicle, turning-in vehicle, vehicle merging from a parking position.

[0017] In a further embodiment of the system according to the invention, the at least one sub-model for calculating the environment is designed to calculate at least one vehicle environment from the following list: road condition, road type, lane markings, road layout, traffic signs, country-specific regulations, occupancy grid on the basis of trajectories determined by the at least one sub-model for calculating the trajectory.

[0018] In a further embodiment of the system according to the invention, it comprises a FlexRay data bus system, wherein the FlexRay data bus system is configured to transmit the measurement data from the vehicle sensors oriented toward the vehicle's surroundings to the multi-model network. In the system, at least one vehicle sensor is selected from the following list: long-range radar (LRR), medium-range radar (MRR), nano-radar, vehicle camera, LiDAR.

[0019] In yet another embodiment of the system according to the invention, the system is configured to execute at least one measure from the following list when an overtaking maneuver is initiated or executed despite a warning signal being present: haptic, acoustic, and / or visual driver warning, triggering a driver assistance system to intervene in vehicle control. It is conceivable to implement the system according to the invention as part of a driver assistance system.

[0020] Further advantages and embodiments of the invention will become apparent from the description and the accompanying drawings.

[0021] It is understood that the features mentioned above and those to be explained below can be used not only in the combination specified in each case, but also in other combinations or on their own, without departing from the scope of the present invention. Fig. 1 schematically shows various traffic situations with an embodiment of the method according to the invention. Fig. 2 shows a schematic flowchart for a further embodiment of the method according to the invention.

[0022] In the Fig. 1 schematically shows various traffic situations using an embodiment of the method according to the invention. In an exemplary traffic situation 11, an ego vehicle 1 is traveling safely behind two other vehicles 2 traveling in front of it. A truck 3 is approaching in the opposite direction. In a first dangerous traffic situation 12, the driver of the ego vehicle 1 attempts to perform an overtaking maneuver or overtaking process. Using the embodiment of the method according to the invention, trajectories for the overtaking maneuver and the other vehicles are calculated or numerically predicted, and a hazard for the overtaking maneuver is determined based on a likewise calculated occupancy grid. The driver is given a visual driver warning that an attempt to overtake poses a hazard - indicated on the ego vehicle 1 by an exclamation mark.In a second dangerous traffic situation 13, the driver of the ego vehicle 1 attempts to perform the overtaking maneuver despite a displayed driver warning. It is conceivable that further measures, such as haptically making it more difficult to move the steering wheel or the accelerator pedal, could further encourage the driver to abort the overtaking maneuver and return to a safe side of the road. It is also conceivable, if the method according to the invention is implemented in a driver assistance system, to gain control of the ego vehicle 1 and return to the safe side of the road. Such a decision by the driver assistance system can, for example, be based on the magnitude of the numerical value of the hazard calculated using the method according to the invention, e.g., if this value exceeds 50% of its maximum possible value. In a third dangerous traffic situation 14, a traffic sign 5 indicates a no-overtaking zone.Here, it is conceivable that even in the case of very slow vehicles 2 and very slow oncoming trucks 3, i.e., when the occupancy matrix for a trajectory of the ego vehicle is "free", a driver assistance system will, due to legal regulations, take the same or similar measures as in traffic situation 13. In the first safe traffic situation 15, however, at the current speed of the ego vehicle 3, there is sufficient time and space for the ego vehicle 1 to return to the safe lane after overtaking another vehicle 2. In the second safe traffic situation 16, in which a faster car 4 is approaching, the overtaking maneuver can be safely completed and no driver warning is displayed. In a third safe traffic situation 17, overtaking is possible without oncoming traffic. No driver warning is displayed.

[0023] In Fig.Figure 2 schematically shows a flowchart 20 for a further embodiment of the method according to the invention. A multi-model network 21 has, as submodels, a trajectory calculation for other vehicles 22, a calculation of an occupancy grid 23, a trajectory calculation for overtaking maneuvers of the ego vehicle 24, and a classification of road conditions 25. The submodels are fed with measured values ​​from the vehicle sensors and forward their results to a decision model 26, which provides a numerical value for the risk assessment. Depending on whether this numerical value exceeds a predetermined threshold, a warning signal is generated as output 27. List of reference symbols 1 ego vehicle 2 Other vehicles to be overtaken 3 Oncoming truck 4 Oncoming car 5 No overtaking traffic sign 11 Traffic situation with driver warning 12 First dangerous overtaking maneuver 13 Second dangerous overtaking maneuver 14 Third dangerous overtaking maneuver 15 First safe overtaking maneuver 16 Second safe overtaking maneuver 17 Third safe overtaking maneuver 20 Risk assessment flowchart 21 Multi-model network for risk assessment / overtaking 22 Trajectory calculation for third-party vehicles 23 Calculation of occupancy grid 24 Trajectory calculation for overtaking maneuvers 25 Classification of road conditions 26 Decision model 27th issue QUOTES CONTAINED IN THE DESCRIPTION

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

[0000] DE 10 2019 209 432 A1

[0003] DE 10 2021 213 308 A1

[0004] DE 10 2018 120 942 A1

[0005] DE 102019202589A1

[0006]

Claims

[1] A method for a risk assessment of an overtaking maneuver, in which a multi-model network (21) is formed by at least one sub-model for calculating the environment (23, 25) and at least one sub-model for calculating trajectories (22, 24), as well as a decision model for the risk assessment, in which the multi-model network (21) is trained with training data from a plurality of overtaking maneuvers with actual vehicle trajectories of participating vehicles (1, 2, 3, 4), the training data being measured by vehicle sensors aligned with a vehicle's surroundings and stored on a memory unit, in which, during a journey in continuous execution, measurement data is transmitted to the multi-model network (21) by vehicle sensors, in which a numerical value of the risk assessment is provided by the decision model (26), and in which a warning signal (27) is output when a predetermined threshold value is exceeded. [2] Method according to claim 1, wherein the at least one sub-model for trajectory calculation (22, 24) is designed for calculating a trajectory of at least one vehicle (1, 2, 3, 4) from the following list: ego vehicle (1) during continued driving movement, ego vehicle (1) during overtaking maneuver, at least one vehicle (2) driving ahead, oncoming vehicle (3, 4), turning-in vehicle, vehicle merging from a parking position. [3] Method according to claim 2, wherein the at least one sub-model for calculating the environment (23, 25) is designed for calculating at least one vehicle environment from the following list: road condition (25), road type, lane markings, road course, traffic signs (5), country-specific regulations, occupancy grid (23) on the basis of trajectories determined by the at least one sub-model for calculating the trajectories (22, 24). [4] Method according to one of the preceding claims, in which the measurement data of the vehicle sensors aligned to the vehicle environment are sent to the multi-model network (21) via a FlexRay data bus system and in which at least one vehicle sensor is selected from the following list: long-range radar, medium-range radar, nano-radar, vehicle camera, LiDAR. [5] Method according to one of the preceding claims, in which, when initiating or executing an overtaking maneuver despite the presence of a warning signal (27), at least one measure from the following list is carried out: haptic, acoustic and / or optical driver warning, initiation of a driver assistance system for vehicle control intervention. [6] System comprising an ego vehicle (1), at least one vehicle sensor aligned with a vehicle environment, a computing unit and a memory unit, wherein the computing unit is configured to execute a multi-model network (21) for calculating a risk assessment of an overtaking maneuver, wherein the multi-model network (21) is formed by at least one sub-model for calculating the environment (23, 25) and at least one sub-model for calculating trajectories (22, 24) as well as a decision model (26) for the risk assessment, wherein the multi-model network (21) is trained with training data from a plurality of overtaking maneuvers with actual vehicle trajectories of participating vehicles (1, 2, 3, 4), wherein the training data are measured by vehicle sensors and stored on a memory unit, wherein the at least one vehicle sensor is configured toto transmit measurement data to the multi-model network (21) during a continuous journey, wherein the multi-model network (21) is configured to provide a numerical value of the risk assessment by means of the decision model (26) and to output a warning signal (27) if a predetermined threshold value is exceeded. [7] System according to claim 6, wherein the at least one submodel for trajectory calculation (22, 24) is designed to calculate a trajectory of at least one vehicle (1, 2, 3, 4) from the following list: ego vehicle (1) during continued driving movement, ego vehicle (1) during overtaking maneuver, at least one preceding vehicle (2), oncoming vehicle (3, 4), turning-in vehicle, vehicle merging from a parking position. [8] System according to claim 7, wherein the at least one sub-model for calculating the environment (23, 25) is designed to calculate at least one vehicle environment from the following list: road condition (25), road type, lane markings, road course, traffic signs (5), country-specific regulations, occupancy grid (23) on the basis of trajectories determined by the at least one sub-model for calculating the trajectories (22, 24). [9] System according to one of claims 6 to 8, which comprises a FlexRay data bus system which is configured to transmit the measurement data of the vehicle sensors oriented towards the vehicle environment to the multi-model network (21) and in which at least one vehicle sensor is selected from the following list: long-range radar, medium-range radar, nano-radar, vehicle camera, LiDAR. [10] System according to one of claims 6 to 9, wherein the system is configured to carry out at least one measure from the following list when initiating or executing an overtaking maneuver despite the presence of a warning signal (27): haptic, acoustic and / or optical driver warning, initiation of a driver assistance system to intervene in the vehicle control.

Images