DE102022104129A1 - Method for determining a health effect, method for training a prognostic function, monitoring device, vehicle and training device - Google Patents

Abstract

The invention relates to a method for determining a health effect on at least one occupant (2) of the motor vehicle (1) in the event of a collision with a motor vehicle (1). A monitoring device evaluates movement data (5) recorded at least in a predetermined period of time after a collision event of the motor vehicle (1) has been detected according to a predetermined prognosis function (19) for determining at least one effect parameter (7) relating to the effects on the at least one motor vehicle occupant (2 ) of the motor vehicle (1) due to the movement that took place in the predetermined period of time, there was at least one health effect. It is provided that a prognosis function (19) is used, which is trained according to a predetermined machine learning method on a training data set (18), the training data set (18) including simulated movement data (15) and simulated impact parameters (7).

Description

The invention relates to a method for determining at least one health effect on at least one occupant of the vehicle in the event of a vehicle collision, a method for training a prognosis function, a monitoring device, a vehicle comprising a monitoring device and a training device.

Collisions in road traffic can not only damage a vehicle, but also injure the vehicle occupants. An injury that can also occur in relatively light collisions is the so-called whiplash. Whiplash is also known as a cervical spine distortion, where the cervical spine stands for. Whiplash occurs during certain movements that usually occur when a vehicle crashes during an accident. The force released during the impact is transmitted to the occupants, which can injure soft and connective tissue in the neck area. A previous movement of the head is characteristic for the occurrence of a whiplash injury. As a result of the impact, the cervical spine is first hyperextended and then severely bent. It is not only the severity of the impact that influences the occurrence of whiplash, but also the seating position of the occupants in the car and their head position during the accident. A subsequent diagnosis of whiplash is difficult, however, since imaging examination methods usually do not allow evidence of changes and injuries to the cervical vertebrae, as are present in whiplash.

Methods and devices are known from the prior art, which enable health data of vehicle occupants to be recorded.

The WO 2019/204949 A1 describes a system for recording personal accident data in a vehicle. The system is set up to record personal accident data in a vehicle. The system includes at least one camera for capturing image data of at least one vehicle occupant. A control unit controls the camera in such a way that storage of image data is terminated at an end time, the end time being a predefined time interval after the receipt of an accident occurrence signal.

The WO 2020/136658 A1 discloses systems, devices, and methods for assisting vehicles after a crash. The system is designed to provide first responders with useful information about the medical condition and injuries of the vehicle occupants after a vehicle accident. The system includes a sensor in the vehicle cabin that includes at least one or more image sensors, depth sensors, and micro-vibration sensors to collect sensory data of the vehicle cabin, including pre-crash, during-crash, and post-crash data. The system also includes at least one processor configured to analyze the sensory data.

A method for preventing whiplash injuries is also known from the prior art.

The DE 10 2008 000 861 B4 discloses a method and a control unit for controlling passenger protection means for a motor vehicle. The method provides for a first signal to be generated as a function of a head position being monitored by means of a contactless sensor system arranged in a headrest. The head position is monitored by influencing an electric field through the head position in the vicinity of the headrest. A second signal is generated by an accident sensor system. The passenger protection arrangement is actuated as a function of the first and second signals. Provision is made for the first signal to be evaluated in order to detect whether a rear-end collision has occurred.

It is an object of the invention to enable whiplash injury to be detected in a vehicle occupant.

The object is solved by the subject matter of the independent patent claims. Advantageous developments of the invention are disclosed by the features of the dependent patent claims, the following description and the figures.

A first aspect of the invention relates to a method for determining at least one health effect on at least one occupant of the motor vehicle in a collision of a motor vehicle. In other words, a movement of the motor vehicle is recorded in movement data by the monitoring device. The movement data can describe, for example, speed, acceleration and/or jerk values of the movement of the motor vehicle. The movement data are monitored by the monitoring device to ensure that a predetermined movement criterion is met. The predetermined movement criterion can describe movements which indicate a collision event of the motor vehicle. If the movement criterion is met by the movement data, the collision event of the motor vehicle is determined in a further step. In other words, the movement data are evaluated by the monitoring device to detect the collision event of the motor vehicle. If the predetermined movement criterion is met by the movement data, the presence of the collision event of the motor vehicle is determined by the monitoring device. It can be provided, for example, that the collision event is detected when predetermined acceleration values are detected by the monitoring device in the movement data and/or a predetermined movement pattern is detected by the monitoring device.

The monitoring device evaluates the movement data recorded at least in a predetermined period of time from the determination of the collision event of the motor vehicle according to a predetermined prognosis function for determining at least one effect parameter relating to the health effect on the at least one motor vehicle occupant of the motor vehicle. In other words, it is provided that at least the movement data that is recorded within the predetermined period of time from the collision event is evaluated by the monitoring device according to the predetermined prognosis function. The monitoring device uses the prognosis function to determine the at least one effect parameter. The at least one effect parameter describes the at least one health effect of the collision event on the at least one motor vehicle occupant of the motor vehicle, which is caused by the movement that took place in the predetermined period of time. The prognosis function can be provided, for example, to recognize a predetermined movement pattern of the movement of the motor vehicle described in the movement data, which is assigned a predetermined effect on a motor vehicle occupant. The effect parameter can describe, for example, whether the predetermined effect on the at least one motor vehicle occupant is determined according to the prognosis function and/or the probability of the predetermined effect occurring and/or the extent to which the effect on the at least one motor vehicle occupant is determined.

In a subsequent step, at least if a predetermined effect criterion of the effect parameter is met, the effect parameter is output by the monitoring device. In other words, it is provided that the impact parameter is output by the monitoring device if the impact parameter satisfies the predetermined impact criterion. The impact criterion can, for example, describe a required minimum degree of severity or a required minimum probability of the impact occurring. Provision can be made, for example, for the effect parameter to be output only if the effect parameter exceeds a predetermined threshold value.

It is provided that the prognosis function is used, which is trained according to a predetermined machine learning method on a training data set. The training data record includes simulated movement data, the simulated movement data describing a simulated movement of a simulated motor vehicle from a simulated collision event. The training data set includes simulated effect parameters, wherein the simulated effect parameters describe at least one health effect on a simulated motor vehicle occupant of the simulated motor vehicle that occurred due to the movement that took place in the respective simulated time periods.

The training data record includes the movement data of the simulated motor vehicle simulated for the respective simulated time periods. In other words, the prognosis function is trained on the training data set. The training data set includes time periods that include simulated movement data within the time period. Another of the input variables includes simulated effect parameters, which are assigned to the simulated movement data of the respective time periods and describe a simulated health effect on a simulated motor vehicle occupant of the simulated motor vehicle in the respective time periods. In other words, a respective simulated impact parameter is assigned to the respective time periods. The simulated effect parameter describes the health effect that has occurred on a simulated motor vehicle occupant of the simulated motor vehicle on the basis of the simulated movement data. It can be provided, for example, that the training data record was generated using simulations of motor vehicle accidents. The simulations can, for example, run over the respective periods of time and provide the simulated movement data which describe the movement of the simulated motor vehicle during the accident. The health effects on the at least one motor vehicle occupant of the simulated motor vehicle can be simulated in the simulations. The simulated effect parameter can describe, for example, an occurrence of a health effect and/or an extent of a health effect on the simulated motor vehicle occupant during the simulated accident.

The advantage of the invention is that recorded movement data of a motor vehicle are used to identify a predetermined health effect on the at least one motor vehicle occupant.

The invention also includes developments that result in further advantages.

A further development of the invention provides that the effect parameter describes a probability of an exit and/or an extent of a whiplash injury. In other words, it is provided that the effect parameter describes the probability with which whiplash has occurred in the at least one motor vehicle occupant as a result of the movement of the motor vehicle that took place during the period of time. This results in the advantage that it is possible to determine whiplash using the recorded movement data.

A development of the invention provides that the monitoring device evaluates at least the movement data recorded in a predetermined period of time from the determination of the collision event of the motor vehicle according to a predetermined severity determination method in order to determine the severity of the collision event. In other words, it is provided that the monitoring device evaluates the movement data recorded in the predetermined period of time in order to determine the severity of the collision event of the motor vehicle. The degree of severity can, for example, relate to energy converted during the collision, which can be determined from an acceleration detected during the collision and a mass of the vehicle stored in the monitoring device. The aggravated investigation procedure can only relate to the motor vehicle, for example, regardless of the effect on the at least one motor vehicle occupant. The development results in the advantage that a variable for evaluating the collision event is determined.

A further development of the invention provides that the monitoring device determines the impact parameter relating to the health impact on the at least one motor vehicle occupant of the motor vehicle only if the determined severity of the collision event satisfies a predetermined severity condition. In other words, it is provided that when a collision event is detected by the monitoring device, the severity of the collision event is determined and the effect parameter is determined only if the predetermined severity condition is met by the severity of the collision event. The development results in the advantage that the effect parameter is only determined if the health effect on the at least one motor vehicle occupant is assumed based on the severity of the collision event. Provision can be made, for example, for the effect parameter not to be determined if the collision event occurred below a predetermined speed of the motor vehicle and/or the energy converted in the collision event fell below a minimum.

A further development of the invention provides that the monitoring device checks the impact parameter for plausibility with the severity of the collision event in a predetermined plausibility process and the impact parameter is only output by the monitoring device if the impact parameter is assessed as plausible in the predetermined plausibility process. In other words, it is provided that the monitoring device checks whether the impact parameter determined is plausible in view of the severity of the collision event. The impact parameter is only output by the monitoring device if the impact parameter is assessed as plausible using the plausibility check method. The advantage results from the further development that any incorrectly recognized effect parameters are recognized using a plausibility check method. Provision can be made, for example, for the impact parameter to have a predetermined relationship to the degree of severity in order to be assessed as plausible.

A further development of the invention provides that the monitoring device sends the effect parameter relating to the health effects on the at least one motor vehicle occupant of the motor vehicle via an antenna device to a central computer external to the vehicle. In other words, it is provided that the impact parameter is transmitted by the monitoring device to the central computer external to the vehicle. The development results in the advantage that, for example, rescue workers are informed about predetermined effects on the at least one motor vehicle occupant when a collision event is detected can become. In addition, this can make it possible to inform insurance companies about the occurrence of an effect. Provision can also be made for the effect parameter to be transmitted together with the movement data recorded during the period in order to enable evaluation by the vehicle-external central computer with greater evaluation accuracy.

A development of the invention provides that the monitoring device detects position data of the at least one motor vehicle occupant, which describes a body posture and/or a body position of the at least one motor vehicle occupant in the motor vehicle. Provision is made for the at least one effect parameter to be determined in the prognosis function as a function of the position data. In other words, it is provided that the impact parameter is determined according to the prognosis function as a function of the movement data and the position data. The position data describe the body posture and/or the body position of the at least one motor vehicle occupant in the motor vehicle. It can be provided, for example, that the posture of a head of the at least one motor vehicle occupant and/or a seating position of the motor vehicle occupant inside the motor vehicle is recorded. Provision can also be made for detecting a head position of the at least one motor vehicle occupant in relation to the motor vehicle seat. The position data can be recorded once, for example, when the collision event is determined. The position data can also be recorded multiple times in the predetermined period of time. As a result, for example, a probability of a whiplash injury occurring or other effects which depend on the posture and/or the position of the body can be determined more precisely.

A further aspect of the invention relates to a method for training a prognosis function for determining at least one effect parameter relating to a health effect on at least one motor vehicle occupant of a motor vehicle using a training device. It is provided that the prognosis function is trained by the training device according to a predetermined machine learning method on a training data set. The training data set can include simulated movement data of a simulated motor vehicle and simulated effect parameters for respective time periods. The simulated effect parameters are assigned to the simulated movement data of the respective time periods and indicate a simulated health effect on the simulated motor vehicle occupant of the simulated motor vehicle. The simulated effect parameters can be traced back to the movements of the motor vehicle simulated in the respective time periods. In other words, provision is made for the training device to train and/or modify the prognosis function according to the predetermined machine learning method. The predetermined machine learning method may include known methods for supervised learning and/or unsupervised learning. The training data record is used by the training device as a basis for the training and/or the modification of the prognosis function. The training device can use the simulated movement data and the simulated impact parameters of the training data set as training data for the predetermined machine learning method in order to train the forecasting function. The training data record includes respective periods of time with respective simulated movement data which describe a simulated movement of a simulated motor vehicle. It may be the movements that the motor vehicle performs in a simulation during a simulated collision event. The respective simulated impact parameters are assigned to the periods of time which have the respective simulated movement data. In other words, the periods of time that comprise the simulated movement data are linked to the respective impact parameters. The respective effect parameters can describe, for example, which health effect was determined for a motor vehicle occupant of the simulated motor vehicle in a collision event on the basis of the simulated movement data. Provision can be made, for example, for the movement data to be generated using a finite element simulation method which simulates a collision event of the motor vehicle. The health effects can be determined, for example, using a finite element simulation method, which outputs effects on the simulated motor vehicle occupants in the simulated motor vehicle. The advantage resulting from the invention is that the prognosis function is based on simulated training data, as a result of which a high level of accuracy is achieved without having to provide the resources required for the simulation in the motor vehicle.

A development of the invention provides that the prognosis function is a trained neural network. In other words, the prognosis function is designed as an artificial neural network, which is trained in the training data set.

A development of the invention provides that the prognosis function is a statistical data model. In other words, it is intended hen that the prognosis function is designed as a statistical data modeling of the training data.

A development of the invention provides that the predetermined machine learning method includes a regression method. In other words, the training method includes a validation of the prognosis function, with the impact parameters output by the prognosis function, which are determined for the simulated movement data, being compared with the respective simulated impact parameters, and the prognosis function being adapted to reduce a deviation between the simulated impact parameters and the impact parameters determined .

A development of the invention provides that the training data record is generated using a finite element method. In other words, it is provided to use the finite element method to generate the simulated movement data and/or the simulated effect parameters, which generates, for example, the collision event of the simulated motor vehicle and the simulated effects on the motor vehicle occupants during the collision.

The invention also includes a monitoring device for carrying out a method for determining at least one health effect on at least one occupant of the motor vehicle in the event of a collision of a motor vehicle.

The invention also includes a motor vehicle which has a monitoring device. The motor vehicle according to the invention is preferably designed as a motor vehicle, in particular as a passenger car or truck, or as a passenger bus or motorcycle.

The invention also relates to a training device which is set up for training a prognosis function for determining at least one effect parameter relating to a health effect on the at least one motor vehicle occupant of the motor vehicle.

The monitoring device and the training device can each have a data processing device or a processor device that is set up to carry out an embodiment of the method according to the invention. For this purpose, the processor device can have 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). Furthermore, the processor device can have program code which is set up to carry out the embodiment of the method according to the invention when executed by the processor device. The program code can be stored in a data memory of the processor device.

The invention also includes developments of the method according to the invention for training a prognosis function, the monitoring device according to the invention, the motor vehicle according to the invention and the training device according to the invention, which have features as they already exist in connection with the developments of the method according to the invention for determining a collision of a motor vehicle at least one motor vehicle occupant of the motor vehicle took place at least one health effect has been described. For this reason, the corresponding developments of the method according to the invention for training a prognosis function, the monitoring device according to the invention, the motor vehicle according to the invention and the training device according to the invention are not described again here.

The invention also includes the combinations of features of the described embodiments. The invention also includes implementations that each have a combination of the features of several of the described embodiments, provided that the embodiments were not described as mutually exclusive.

• 1 eine schematische Darstellung eines Kraftfahrzeugs;
• 2 eine schematische Darstellung eines Ablaufs eines Verfahrens zur Ermittlung einer bei einer Kollision eines Kraftfahrzeugs erfolgten gesundheitlichen Auswirkung;
• 3 eine schematische Darstellung eines Zusammenwirkens einer Überwachungsvorrichtung und einem fahrzeugexternen Zentralrechner; und
• 4 eine schematische Darstellung eines Ablaufs eines Verfahrens zum Trainieren einer Prognosefunktion.

Exemplary embodiments of the invention are described below. For this shows:

• 1 a schematic representation of a motor vehicle;
• 2 a schematic representation of a sequence of a method for determining an effect on health that has occurred in the event of a collision of a motor vehicle;
• 3 a schematic representation of an interaction of a monitoring device and a vehicle-external central computer; and
• 4 a schematic representation of a sequence of a method for training a prognosis function.

The exemplary embodiments explained below are preferred embodiments of the invention. In the exemplary embodiments, the described components of the embodiments each represent individual, independent gig represent features of the invention to be considered from one another, which also further develop the invention independently of one another. Therefore, the disclosure is also intended to encompass combinations of the features of the embodiments other than those illustrated. Furthermore, the described embodiments can also be supplemented by further features of the invention that have already been described.

In the figures, the same reference symbols designate elements with the same function.

1 shows a schematic representation of a motor vehicle 1 which has a monitoring device 3 . The motor vehicle 1 can be a passenger car or a truck, for example, and can be set up to accommodate at least one motor vehicle occupant 2 . The motor vehicle 1 can have a monitoring device 3 which can be set up to determine an effect on the at least one motor vehicle occupant 2 in the event of a collision event. The monitoring device 3 can have a movement sensor unit 4 which can be set up to record movement data 5 of the motor vehicle 1 . The movement data 5 can relate to speed, acceleration and/or jerk data, for example. The movement sensor unit 4 can be set up to record the movement data 5 of the motor vehicle 1 continuously, for example continuously and/or at predetermined intervals, and to transmit them to a control unit 6 of the monitoring device 3 . The control unit 6 can be set up to check the movement data 5 for the occurrence of a collision event of the motor vehicle 1 . Such a collision event can be detected, for example, by detecting predetermined characteristics in the movement data 5, such as a brief strong acceleration. Provision can be made for the control unit 6 to be set up to store the movement data 5 over a predetermined period of time in the event of a collision event and to evaluate it to determine an impact parameter 7 which describes an impact on the at least one motor vehicle occupant 2 . The effect parameter 7 can describe, for example, a probability of the occurrence of a predetermined health impairment of the motor vehicle occupant 2 based on the movement data 5 recorded in the time window. In particular, it can be a matter of a probability and/or a degree of severity 11 of a whiplash injury occurring in the motor vehicle occupant 2 . The impact parameter 7 can be determined by a prognosis function 19 as a function of the movement data 5 . The prognosis function 19 can be trained on simulated training data sets 18 which describe simulated movement data 15 in relation to simulated impact parameters 17 . The monitoring device 3 can have an interior monitoring unit 8 which can be set up to record position data 9 of the at least one motor vehicle occupant 2 . The position data 9 can describe a sitting posture and/or a sitting position of the occupant in the motor vehicle 1, for example. It can be provided, for example, that the position of the head is detected by means of the interior monitoring unit 8 and that this is made available to the control unit 6 as position data 9 for consideration in the prognosis function 19 . The control unit 6 can be set up to evaluate the position data 9 and the movement data 5 to determine the impact parameter 7 using the prognosis function 19 . The control unit 6 can be set up to determine a degree of severity 11 of the collision event and to use the degree of severity 11 for a plausibility check of the effect parameter 7 determined. Provision can be made for the monitoring device 3 to have a transmission unit 10 which is set up to transmit the determined effect parameter 7 to a central computer 12 external to the vehicle. The transmission can take place depending on the determined severity 11, so that it can be provided that the impact parameter 7 is only transmitted to the central computer 12 if a predetermined severity 11 of the collision event is exceeded. The monitoring device 3 can also be set up to transmit the effect parameter 7 together with the movement data 5 and the position data 9 recorded over the period of time to the vehicle-external central computer 12 in order to carry out a more precise analysis and plausibility check of the determined effects on the at least one motor vehicle occupant 2 by the Central computer 12 to allow.

2 1 shows a schematic representation of a sequence of a method for determining an effect on health that has occurred in the event of a collision of a motor vehicle 1 . In a first step S1 , movement data 5 can be recorded by the movement sensor unit 4 . The movement data 5 can include, for example, an acceleration and/or a pressure acting on the motor vehicle 1 and/or derivations of the values over at least 300 milliseconds. The movement data 5 can be recorded at predetermined intervals and transmitted to the control unit 6 . In a step S2, the control unit 6 can regularly monitor a time window of, for example, one to five milliseconds in order to be able to determine the occurrence of a collision event in the movement data 5. In a step S3, by the tax Unit 6 checks whether the collision event is detected in the time window of the movement data 5. If this is not the case, steps S1 to S3 are repeated. In the event that a collision event is detected in method step S3, provision can be made for method step S4 and/or S5 to be carried out by control unit 6. Method step S4 can include an assessment of a severity 11 of the collision event. The degree of severity 11 can be determined using an analytical model, with movement data 5 being determined over a time window of at least 250 milliseconds, which is provided by the movement sensor unit 4 . In method step S5, the impact parameter 7, which can describe a health impact on the at least one motor vehicle occupant 2, can be determined by means of a prognosis function 19 depending on the movement data 5, which was recorded in a time window of, for example, 300 milliseconds from the determination of the collision event. In order to be able to increase the accuracy of the effect parameter 7, it can be provided that an interior monitoring unit 8 records position data 9 of the at least one motor vehicle occupant 2, for example over a certain period of time from the occurrence of the collision event, and provides it to the control unit 6. The position data 9 can be generated, for example, by means of image processing by an occupant camera and calculation of predetermined features of an occupant's head movement. The position data 9 and the movement data 5 can be used in the prognosis function 19 in order to determine the impact parameter 7 . This can, for example, describe a probability of a whiplash injury occurring in the at least one motor vehicle occupant 2 . In a method step S7 it can be provided that the determined severity 11 of the collision event and the impact parameter 7 are compared with one another in order to enable a plausibility check of the determined impact parameter 7 . In other words, it is provided that the control unit 6 determines a plausibility of the determined effect parameter 7 with the determined degree of severity 11 of the collision event. In the event that the determined impact parameter 7 is evaluated as plausible, provision can be made for the impact parameter 7 to be transmitted to a central computer 12 external to the vehicle in a step S8. The transmission can take place, for example, together with the movement data 5 of the period and the position data 9 . The transmission can also be dependent on the determined severity 11, so that the impact parameter 7 is only transmitted if a predetermined threshold value is exceeded.

3 shows a schematic representation of an interaction between the monitoring device 3 and a vehicle-external central computer 12. In a first step T1, a collision between the motor vehicle 1, which includes the monitoring device 3, and another motor vehicle 1 can occur. Acceleration of the motor vehicle 1 can occur as a result of the collision. The acceleration can result in a characteristic pattern being recognized by the control unit 6 in the continuously recorded movement data 5, T2. By recognizing the characteristic movement data 5, the collision event can be determined by the control unit 6 and movement data 5, which was recorded over a period of, for example, 300 milliseconds from the determination of the collision event, can be evaluated. The effect parameter 7 , which can describe an effect of the collision on the motor vehicle occupant 2 , can be determined by means of the prognosis function 19 . In a step T3, the effect parameter 7 can be transmitted together with movement and position data 9 to the central computer 12 external to the vehicle. In this case, the central computer 12 external to the vehicle can carry out a new evaluation of the movement data 5 and/or the position data 9 in order to check and/or specify the determined effect parameter 7 . In a step T5 it can be provided that the effect parameter 7 is transmitted by the vehicle-external central computer 12 to a target computer 13 of a rescue service and/or an insurance company.

4 shows a schematic representation of a sequence of a method for training a prognosis function 19. The method can be carried out by a training device 14. Provision can be made for finite element simulations to be carried out in order to generate a training data set 18, it being possible for the training data set 18 to include simulation results from simulated collisions. The training data set 18 can include simulated movement data 15 of a simulated motor vehicle 1 and the impact parameters 7 determined in the simulations on a simulated motor vehicle occupant 2. Simulated movement data 15, simulated position data 16 and the simulated impact parameters 17 can be provided to the prognosis function 19 in order to adapt the prognosis function 19 and/or to exercise. An accuracy of the prognosis function 19 can be achieved by providing output variables, with the determined variable of the impact parameter 7 being compared with the simulated impact parameters 17 by the prognosis function 19 . Through ongoing adjustment, a person can Formance assessment and an adjustment of the forecast function 19 can be achieved.

Accidents happen again and again on German roads. Many of these are rear-end collisions, for example at traffic lights. This often results not only in property damage, but possibly also in injuries.

One of the most common injuries is whiplash, also known as a cervical sprain. In this case, HWS stands for “cervical spine”. This trauma is caused by the impact. The released force is transmitted to the occupants, injuring the soft tissue and connective tissue in the neck area.

In some cases, those affected by a cervical spine distortion are awarded compensation for pain and suffering. Evidence of whiplash injuries is difficult today. An automatic assessment of the severity of the injury and possible whiplash is not possible due to the vehicle sensors and processing.

An accident is recognized today by activating the vehicle's inherent inertial sensors. An emergency call signal (eCall) is not initiated until a serious accident with airbag deployment is detected at the earliest.

Using the recorded accident data requires manual reading from the accident vehicle and is therefore only carried out in a few exceptional cases by accident research institutions such as AARU (Audi Accident Research Unit) or GIDAS (German In-Depth Accident Study). The number of such surveys only covers a small part of the number of accidents that actually occurred due to the effort involved, the lack of accident researchers in the vicinity of possible accident locations and the often very short time window available for an investigation. Data on the course of the accident, which is based on descriptions by eyewitnesses or recalculations based on skid marks, etc., are subject to the uncertainties that are usual in such procedures. An automated transmission of data from the pre-, in- and possibly also the post-crash phase does not yet exist.

The rescue control center does not have live access to rescue-related accident and/or vehicle data.

The previous movement of the head is typical for a whiplash injury. As a result of the impact, the cervical spine is first overstretched and then severely bent. Not only does the severity of the impact have an impact on the trauma, but also the seating position in the car and the position of the head during the accident.

The diagnosis is difficult, however, since imaging examination methods usually do not show any changes or injuries to the cervical vertebrae. Thus, clear evidence is lacking. The doctor must therefore fully rely on the patient's description and derive the diagnosis and further treatment measures from it.

In principle, the whiplash injury must be medically examined and confirmed in writing. Past court hearings show that insurance companies often argue that a speed change of 10 to 15 kilometers per hour is not enough to trigger whiplash. In such cases, a written medical report is essential.

This limit, which the insurance companies draw here, is called the "harmlessness limit". According to the Federal Court of Justice, a diagnosis should be made despite the harmlessness limit, because compensation for whiplash is not excluded despite the limit in traffic law and is decided on a case-by-case basis.

The prognosis of whiplash. The procedure can enable whiplash injuries to be evaluated in all collision directions, even in accidents with low crash severity. The sensor system can be used for accident detection and the prognosis of a whiplash injury can be sent to a backend. The entire course of the crash up to 300 ms can be used, not just the first ms as is usual with the deployment of the restraint system. The method may provide for using a whiplash prediction function that is parameterized using a finite element human model and finite element crash simulation data. Use of new neck load characteristics from simulations. The prognosis can be validated by an analytical crash severity estimator. Optionally, features from the image processing of an indoor camera can be used to improve the forecast quality. Image processing enables higher accuracy of the whiplash detection especially for all occupant positions (e.g. new seat positions in the frame bed position). The occupant positions may be described in the position data.

The method enables whiplash injury to be recognized and verified in an accident that has occurred for all vehicle occupants, and automatic reporting to the treating doctor and insurance company.

Using classic crash sensors (acceleration sensors, pressure hose, etc.), the monitoring device can detect an accident and assess the mechanical severity of the accident. If an accident occurs with a high degree of certainty and is sufficiently serious, the risk of a whiplash injury can be predicted. If the mechanical crash severity and the probability of a whiplash injury match, the whiplash injury is then sent to the central computer external to the vehicle.

It is proposed to predict whiplash injury using a machine-learning trained prediction function. In this case, the prognosis can classify the collision, in particular with regard to an expected probability of whiplash injury.

The prognosis function can be parameterized by a number of processing parameters, which are determined by the prognosis function being trained by a number of training data sets as part of machine learning. In addition or as an alternative, parameter values and/or curves of the neck load can also be used, e.g. B. occurring accelerations or acceleration curves. It is proposed to use a data-based model as a forecasting function, which is trained using training data sets. This model can be, for example, appropriately trained artificial neural networks or statistical data models. For example, an artificial neural network based on radial basis functions can be used. It is also possible to provide several, in particular non-correlated, decision trees, the results of which are combined. In particular when at least one numerical prognosis result is determined, the prognosis function can also be parameterized by a regression method.

The training data records can each include input data and output data, with the prognosis function being trained, for example, by error feedback, also known as backpropagation of error, in such a way that output data are generated by applying the prognosis function to the input data, which output data are as similar as possible to the output data of the respective training data records.

The training data sets can be generated by FEM crash simulations. Here, derived features from crash pulses from full-vehicle crashes and a finite element human model can be used to determine the neck load.

• • Crash Pulse Features (t0-300ms)
• • Beschleunigungszeitreihen und Ableitungen
• • Druckmessungszeitreihen und Ableitungen
• • OLC
• • dV
• • Nackenbelastung (t0-300ms)
• • Kopf Bewegung in Kamera Bilder: Kopfverlagerungen über Zeit und Ableitungen

Input features of the training set can include:

• • Crash Pulse Features (t0-300ms)
• • Acceleration time series and derivatives
• • Pressure measurement time series and derivatives
• • OLC
• • dV
• • Neck strain (t0-300ms)
• • Head Movement in Camera Images: Head displacements over time and derivatives

• • Erkennung Whiplash (Schleudertrauma)
• • Nacken Belastungskennwerte

Output features of the training set:

• • Whiplash detection (whiplash)
• • Neck Stress Characteristics

Taken together, the examples show how a whiplash detection solution can be provided.

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• WO 2019/204949 A1 [0004]
• WO 2020/136658 A1 [0005]
• DE 102008000861 B4 [0007]

Claims (15)

Method for determining at least one health effect on at least one vehicle occupant (2) of the motor vehicle (1) in the event of a collision of a motor vehicle (1), the method comprising at least the following steps carried out by a monitoring device (3): - Continuous detection of Movement data (5) which describe a movement of the motor vehicle (1), - continuous monitoring of the movement data (5) for fulfillment of a predetermined movement criterion which indicates a collision event of the motor vehicle (1), - determination of the predetermined collision event of the motor vehicle ( 1) if the predetermined movement criterion is met by the movement data (5), - evaluation of at least the movement data (5) recorded in a predetermined period of time from the determination of the collision event according to a predetermined prognosis function (19) to determine at least one impact parameter (7), relating to the at least one health effect on the at least one motor vehicle occupant (2) of the motor vehicle (1) as a result of the movement that took place in the predetermined period of time, and - outputting the at least one effect parameter (7) if a predetermined effect criterion is met by the at least one effect parameter (7), characterized in that the prognosis function (19) is used, which has previously been trained on a training data set (18) according to a predetermined machine learning method, the training data set (18) containing simulated movement data (15) and simulated effect parameters (7) comprises, wherein - the simulated movement data (15) describe a simulated movement of a simulated motor vehicle (1) from a simulated collision event and - the simulated effect parameters (17) at least one health effect on a motor vehicle occupant that occurred on the basis of the simulated movement that took place in the predetermined simulated time periods (2) of the motor vehicle (1) describe. procedure after claim 1 , characterized in that the effect parameter (7) describes a probability of occurrence and/or an extent of whiplash. procedure after claim 1 or 2 , characterized in that the monitoring device (3) evaluates at least the movement data (5) recorded in a predetermined period of time from the determination of the collision event of the motor vehicle (1) according to a predetermined severity determination method in order to determine a severity (11) of the collision event. procedure after claim 3 , characterized in that the monitoring device (3) only evaluates the impact parameter (17, 7) relating to the health impact on the at least one motor vehicle occupant (2) of the motor vehicle (1) if the determined degree of severity (11) of the Collision event a predetermined severity condition is met. procedure after claim 3 or 4 , characterized in that the monitoring device (3) checks the impact parameter (17, 7) in a predetermined plausibility check method for plausibility with the severity (11) of the collision event, and the impact parameter (7) is only sent by the monitoring device (3). if the impact parameter (7) is evaluated as plausible in the predetermined plausibility check method. Method according to one of the preceding claims, characterized in that the monitoring device (3) transmits the effect parameter (7) relating to the health effect on the at least one motor vehicle occupant (2) of the motor vehicle (1) via an antenna device of the monitoring device (3). a vehicle-external central computer (12) is sent. Method according to one of the preceding claims, characterized in that the monitoring device (3) detects position data (9) of the at least one motor vehicle occupant (2) which indicates a posture and/or position of the at least one motor vehicle occupant (2) in the motor vehicle (1 ) describe, and the at least one impact parameter (7) is determined according to the prognosis function (19) as a function of the position data (9). Method for training a prognosis function (19) by a training device (14), the prognosis function (19) being trained according to a predetermined machine learning method on a training data set (18), characterized in that the training data record (18) includes movement data (15) of a simulated motor vehicle (1) simulated for respective time periods, and the training data record (18) includes simulated effect parameters (17) which are assigned to the simulated movement data (15) of the respective time periods and respective simulated health parameters Describe the effect on a motor vehicle occupant (2) of the motor vehicle (1) of the respective time periods. Method according to one of the preceding claims, characterized in that the prognosis function (19) is a trained artificial neural network. Method according to one of the preceding claims, characterized in that the prognosis function (19) is a statistical data modelling. Procedure according to one of Claims 8 until 10 , characterized in that the predetermined machine learning method comprises a regression method. Procedure according to one of Claims 8 until 12 , characterized in that the training data record (18) is generated using a finite element method. Monitoring device (3), set up to carry out a method according to one of Claims 1 until 7 . Motor vehicle (1) comprising a monitoring device (3). Claim 13 . Training device (14), set up to carry out a method according to one of Claims 8 until 12 .

Images