The invention relates to a method for detecting and rating an accident involving a vehicle .

DE 10 2005 021 115 Al discloses a method for triggering a vehicle emergency call . Said triggering comprise s the steps of :

- a trigger signal being delivered to an emergency control unit by an on-vehicle emergency sensor ,

- an emergency signal being transmitted by the emergency control unit by way of a short-range communication ,

- the emergency signal being received by a mobile phone by way of the short-range communication , and

- the vehicle emergency call being transmitted to an external control room by the mobile phone by way of mobile radio communication , and

- the mobile phone , which is embodied as a portable mobile phone designed to receive the short-range communication, being taken into the vehicle by a vehicle occupant , in particular the vehicle driver , before the vehicle emergency call is triggered . Such a method i s unreliable because setup of the short-range connection is often forgotten if the mobile phone is not ready for short-range connections a s standard .

It i s an obj ect of the invention to propose a method for detecting and rating an accident involving a vehicle that can be executed on a mobile device independently of the vehicle and has a high level of reliability a s regards a decision about the occurrence or nonoccurrence of an accident .

This obj ect is achieved by characterizing features of Claim 1 . The subclaims specify advantageous and expedient development s . The method according to the invention for detecting and rating an accident involving a vehicle , the method steps of which method are executed on a mobile device equipped with at least two sensors that i s carried by the vehicle , provides for the following :

- an accident monitoring system is operated on the mobile device in such a way o that the mobile device continually captures sensor data of at least two of its sensors and temporarily stores said data in a memory in the manner of a shift regi ster and o after the sensor data of at least one of the sensors have exceeded a threshold value stipulated for the respective sensor,

■ a time stamp is set in the memory,

■ the available sensor data remain stored and

■ an accident verification routine is launched on the mobile device ,

- wherein the accident verification routine o both evaluates the stored sensor data in such a way that a tiered s ignificance level ( severity level ) is ascertained o and evaluates the sensor data captured from the time stamp onwards in a time interval of predetermined length in such a way that a tiered credibility level ( confidence level ) is repeatedly ascertained, o that the significance level ( severity level ) and/or the currently ascertained credibility level ( confidence level ) are repeatedly used in the time interval to decide

■ either that an accident report is sent or

■ that no accident report is sent and the accident verification routine is executed again or ■ that the accident verification routine is terminated .

The tiered system operated us ing such a method, which system involves a presumption , triggered by the exceeding of a thre shold value stipulated for a sensor , of an accident having occurred being later verified in a sophisticated manner , is capable of deciding about the occurrence or nonoccurrence of a presumed accident with a high level of reliability . This allows incorrect accident reports and incorrect accident warnings to be reliably avoided . Such a method can therefore be used to keep down the number of incorrect accident reports and incorrect accident warnings , and to effectively counteract other threats of overload for control centers , despite an increa sing prevalence of automatic accident reporting systems .

The method also provide s for the accident report to be transmitted from the same mobile device as that on which the method is also executed . This increa ses reliability further , since performance takes place entirely on one mobile device , which means that the method can be hampered neither by a defect of a second device nor by interference on a connection to the second device .

Furthermore , there is provision for the mobile device to be used to capture and evaluate movement data and/or pos ition data a s sensor data in the time interval , this being performed in particular by means of a motion sensor of the mobile device and/or an acceleration sensor of the mobile device and/or a navigation sensor of the mobile device ( e . g . GPS , inter alia ) . On the bas is of movement data and/or position data , it is pos s ible to sense movements of the user of the mobile device and/or movement s of the vehicle , in the time interval , that allow conclusions to be drawn as to what is actually happening . There is also provision for the credibility level ( confidence level ) to be as certained by also using captured event data in addition to the captured sensor data . The evaluation of different data allows what is actually occurring to be detected more extensively and hence al so more reliably .

Furthermore , there is provision for the accident verification routine to be used to also verify whether or not the measured velocity of the mobile device decrea ses , wherein if the velocity measured by the mobile device decrea ses to a value that in particular is below approximately 10 km/h and preferably is 0 km/h , the significance level ( severity level ) is set to a high value , which is indicative of an accident having occurred . In this way, e . g . the vehicle coasting to a stop following a collision and the vehicle being at a standstill as a result of an accident are reliably detected .

There is also provision for the evaluation to al so involve a comparison of the sensor data and/or the event data with comparison data being performed, wherein the comparison data used are statistical data that in particular have been ascertained on the basi s of analyzed and verified accidents and/or wherein the comparison data used are geodata derived from a map that are taken as a bas is for rating a pos ition or direction of movement of the mobile device . Such a comparison of the captured sensor data and/or the captured event data allows further insight s indicative of or di sputing an accident having occurred to be collected, which increase the reliability of a deci sion about the occurrence or nonoccurrence of an accident further .

Furthermore , there is also provision for the accident verification routine to be used to use comparison data to al so verify whether the measured velocity of the mobile device increases to a value that is plausible for the movement of an undamaged vehicle, such an increase resulting in the credibility level (confidence level) being set to a low reference value, which is indicative of no accident having occurred. Such a comparison allows the reliability of a decision about the occurrence or nonoccurrence of an accident to be increased further .

There is also provision for the sensor data captured to be 3D movement data and/or acceleration data, which preferably have a unique association with the movement in space. Such sensor data can be taken as a basis for detecting e.g. untypical transverse accelerations and overturns.

Furthermore, there is provision for an accident warning to be transmitted after the accident report is sent, wherein the accident warning comprises at least one of the pieces of information cited below: a timestamp, geocoordinates of an accident location, a timestamp for the accident, a velocity measured by the navigation system. This allows other road users to be informed reliably and promptly.

To generate the information mentioned in the accident warning, there is provision for the timestamp and the sensor data to be used. This allows the accident report to be provided on the basis of available data, which means that no additional computing time is required.

Furthermore, there is provision for ascertainment of the significance level (severity level) to be started after a time delay following the launch of the accident verification routine and for the ascertainment of the significance level (severity level) to result in a trend in the velocity since the timestamp was set being taken into consideration to the effect that an increased significance level ( severity level ) is stipulated in the event of a deceleration to a velocity below a limit value or in the event of a standstill and a decreased significance level ( severity level ) i s stipulated in the event of the absence of a deceleration . Such a comparison allows the reliability of a deci sion about the occurrence or nonoccurrence of an accident to be increased further .

There is also provision for the credibility level ( confidence level ) to be as certained from the sensor data and the event data by also drawing conclus ions about a behavior relating to a driver or pas senger handling the mobile device and by evaluating the detected behavior in such a way that the credibility level ( confidence level ) is increased by a behavior that is typical of an accident and that the credibility level ( conf idence level ) is reduced by a behavior that is untypical of an accident . Such a comparison allows the reliability of a deci sion about the occurrence or nonoccurrence of an accident to be increased further .

Furthermore , there is provision for an accident report to be sent whenever the significance level ( severity level ) has reached it s highest reference value and for an accident report to also be sent whenever both the significance level ( severity level ) and the credibility level ( confidence level ) have reached predefined reference values . This ensures that the accident verification routine does not cause any unneces sary delay for accident report .

There is also provision for no accident report to be sent and for the accident verification routine to be executed again if the signif icance level ( severity level ) and the credibility level ( confidence level ) are at reference values that are neither provided for sending an accident report nor provided for terminating the accident verification routine. This reliably prevents an incorrect accident report from being sent.

Finally, there is provision for the accident verification routine to be terminated either if the time interval has expired or if the significance level (severity level) and/or the credibility level (confidence level) are so low that an accident can be ruled out. This reliably prevents an evaluation of sensor data or event data that is too far away in time from the presumption of an accident from leading to incorrect rating of the presumption of an accident.

Within the context of the invention, sensor data are understood to mean data that are captured by sensors of the mobile device, in particular during operation of the accident monitoring system as an application. Such sensor data comprise in particular velocity, preferably including the direction of the velocity, and/or G force, preferably including the direction of the G force, based on individual times and in particular a period of time before the accident and a period of time after the accident .

Within the context of the invention, event data are understood to mean data that relate to the use of applications of the mobile device and are related to the use of the applications by a user of the mobile device. Such event data can e.g. log one or more of the events cited below, such as "address book was searched", "telephone number was dialled", "telephone call was conducted", "digital image was taken", "video was recorded", "message was written", "emergency number was dialed", "computer game was played", etc.

Within the context of the invention, the credibility level (confidence level) is a tiered reference value for a likelihood of the "presumed accident" being an actual accident. The credibility level (confidence level) is ascertained on the basis of sensor data captured before, during and after the presumed accident and in particular additionally also on the basis of event data captured before, during and after the presumed accident. The credibility level (confidence level) can also be referred to as "crash likelihood". A high reference value is indicative of a high likelihood of an accident having occurred. A low reference value is indicative of a low likelihood of an accident having occurred.

Within the context of the invention, the significance level (severity level) is a tiered reference value for a likelihood of the "presumed accident" being able to be rated as an accident on the basis of the sensor data. A high reference value is indicative of a high likelihood of an accident having occurred. A low reference value is indicative of a low likelihood of an accident having occurred.

Within the context of the invention, comparison data are understood to mean empirically ascertained data, these being in particular stored empirical data and/or pattern data and/or measurement data that typically characterize modes of behavior of a user of a mobile device, who is in particular the driver of the vehicle or the passenger of the vehicle, or of a vehicle in which the mobile device is located.

Within the context of the invention, carriage of the mobile device by the vehicle is understood to mean at least one of the types of carriage described below:

- carriage of the mobile device in such a way that the mobile device is carried in the vehicle, in particular an automobile, in particular in a passenger compartment or a trunk of the vehicle and is in particular kept or stowed there ;

- carriage of the mobile device in such a way that the mobile device is carried on a vehicle , in particular a two-wheeled vehicle , in particular in a storage compartment or on a bracket ;

- carriage of the mobile device in such a way that the mobile device is carried by a driver of the vehicle ;

- carriage of the mobile device in such a way that the mobile device is carried by a pas senger of the vehicle .

Carriage of a mobile device of a vehicle is understood within the context of the invention to mean either the ca se in which the mobile device is located in or on the vehicle independently of a driver or pas senger or the case in which the mobile device is held by the driver or pas senger or the case in which the mobile device i s located in a pocket of the clothing of the driver or pas senger .

Further details of the invention are described below on the basi s of three exemplary embodiments . Of these ,

- the first exemplary embodiment describes a first sequence of the method , which is typical of the case in which a presumption of an accident is incorrect ;

- the second exemplary embodiment describe s a second sequence of the method, which is typical of the case in which a presumption of an accident i s subsequently rated a s correct ;

- the third exemplary embodiment i s a third sequence of the method, which i s typical of the case in which a presumption of an accident is immediately rated as correct . In accordance with a first exemplary embodiment, a first sequence of the method is described, which is typical of the case in which a presumption of an accident is incorrect. For the method according to the invention for detecting and rating an accident involving a vehicle, which is in the form of an automobile, for example, there is a mobile device located in an interior, e.g. in a passenger compartment, of the vehicle. The mobile device is e.g. a mobile phone of the driver that is in a briefcase of the driver, which has been put behind the driver' s seat. The mobile device comprises e.g. a first sensor, which is formed by an acceleration sensor of the mobile phone, and a second sensor, which is formed by a microphone of the mobile phone, and a third sensor, which is formed by a velocity sensor of the mobile phone. An accident monitoring system is operated as an application on the mobile device. The application continually captures sensor data of the first sensor and of the second sensor and of the third sensor and temporarily stores said data in a memory of the mobile device in the manner of a shift register. The acceleration sensor senses accelerations and the microphone picks up in particular low-frequency sounds, such as occur in the event of a collision between a vehicle and another vehicle or in the event of a collision with a stationary obstacle, e.g. a tree, and the velocity sensor senses the current velocity.

In the first exemplary case, the sensor data of the acceleration sensor of the mobile device now exceed a stipulated threshold value for the acceleration, because the driver' s briefcase, together with the mobile device accommodated therein, is hurled to the right, as seen in the direction of travel, behind the passenger' s seat on a quickly taken left-hand curve and overturns in the process. The threshold values are stored in the application . According to a variant embodiment , there i s provi sion for the se threshold values to be able to be matched to the type of vehicle , e . g . four-wheeled vehicle or two-wheeled vehicle , automatically or by means of a selection menu . As soon a s at least one of the sensors exceeds the respective as sociated threshold value with one of its sensor value s - as is true of the acceleration sensor in the first exemplary case - the application initiates the steps cited below :

- the sensor data currently stored in the memory of the mobile device remain stored and are not era sed or pushed out of the shift register ;

- an accident verification routine is launched, which comprises the steps cited below : o the accident verification routine evaluates the stored sensor data in such a way that a tiered significance level ( severity level ) is ascertained, wherein in the first exemplary ca se the maximum acceleration sensed by the first sensor doe s not lead to the highest significance level 6 of 6 in the evaluation , but rather only an average signif icance level 3 of 6 i s reached, and wherein in the f irst exemplary case the low-frequency sounds detected by the second sensor lead to an irrelevant s ignificance level 0 of 6 in the evaluation ; o the accident verification routine evaluates the sensor data captured f rom the timestamp onwards in a time interval of predetermined length in such a way that a tiered credibility level ( confidence level ) is repeatedly ascertained on the basis of said data ; the credibility level in the first exemplary ca se is therefore ascertained on the basis of the sensor data that the f irst and the second sensor provide after the time stamp is set , and o the accident verification routine repeatedly uses the significance level ( severity level ) and/or the currently ascertained credibility level ( confidence level ) in the time interval to decide , with regard to the automatic sending of an accident report ,

■ either that an accident report is sent or

■ that no accident report is sent and the accident verification routine is executed again or

■ that the accident verification routine is terminated .

In the first exemplary case , the first sensor ha s triggered the launch of the accident verification routine , the triggering sensor value not being at significance level 6 of 6 , but rather being only 3 of 6 , with the result that an accident report i s not sent immediately, and the credibility level , owing to the further comparatively low sensor values - briefcase again still lying down - , also moving to an increasingly lower credibility level with every f resh calculation , since the briefca se with the mobile device located therein again moves uniformly with the vehicle , which , in the first exemplary case , is traveling on a straight section at a constant velocity of 70 km/h . Ultimately, the fallen credibility level and the only average significance level before the time interval has elapsed lead to termination of the accident verification routine . This means that no accident report is sent and of course no accident warning is output either .

In accordance with a second exemplary embodiment , a second sequence of the method is des cribed, which is typical of the case in which a presumption of an accident is correct . For the method according to the invention for detecting and rating an accident involving a vehicle, which is in the form of an automobile, for example, there is a mobile device located in a passenger compartment of the vehicle. The mobile device is e.g. a mobile phone of the driver that is in a briefcase of the driver, which has been put behind the driver' s seat. The mobile device comprises e.g. a first sensor, which is formed by an acceleration sensor of the mobile phone, and a second sensor, which is formed by a microphone of the mobile phone, and a third sensor, which is formed by a velocity sensor of the mobile phone. An accident monitoring system is operated as an application on the mobile device. The application continually captures sensor data of the first sensor and of the second sensor and of the third sensor and temporarily stores said data in a memory of the mobile device in the manner of a shift register. The acceleration sensor senses in particular accelerations and the microphone picks up in particular low-frequency sounds, such as occur in the event of a collision between a vehicle and another vehicle or in the event of a collision with a stationary obstacle, e.g. a tree.

In the second exemplary case, the sensor data of the acceleration sensor of the mobile device now exceed a stipulated threshold value for the acceleration, because the driver' s briefcase, together with the mobile device accommodated therein, is hurled to the right, as seen in the direction of travel, behind the passenger' s seat and overturns in the process. In this instance, this has been triggered by the driver having taken a left-hand curve too quickly on a slippery road and the vehicle having slid laterally from the curve onto a downhill meadow and having come to a standstill 50 m away from the road.

In respect of the threshold values and in respect of the steps triggered as a result of at least one of the threshold values being exceeded, explicit reference i s made to the explanations pertaining to the first exemplary embodiment .

In the second exemplary case , the acceleration sensed by the first sensor does not lead to the highest s ignif icance level 6 of 6 in the evaluation , but rather reaches only an average significance level 4 of 6 , and the low-frequency sounds detected by the second sensor lead only to an irrelevant significance level 0 of 6 in the evaluation . This also applie s to the values of the velocity sensor .

The credibility level i s ascertained in the second exemplary case on the bas is of the sensor data that the first sensor , the second sensor , the third sensor and a fourth sensor , which i s in the form of a GPS sensor , deliver after the time stamp is set .

In the second exemplary case , the first sensor has triggered the launch of the accident verification routine , the triggering sensor value not being at significance level 6 of 6 , with the result that an accident report is not sent immediately, and the credibility level continuing to be comparatively low when the vehicle coasts to a stop and the vehicle is at a standstill , since when it i s coasting to a stop over the meadow the unevennes ses that are present there result in the acceleration sensor capturing acceleration values that are below the thre shold value s stored for accelerations and lead only to a significance level 2 of 6 when the credibility level is calculated . On the basis of the sensor data of the fourth sensor , however , the comparison with a digital road map available on the mobile device can then detect that the mobile device is initially situated 50 m away from a sealed road and i s then being moved in an area that i s within a radius of 10 m around the last location . Since this indicates that the mobile device has been removed f rom the vehicle and is being moved around the vehicle, this ultimately leads to a credibility level 5 of 6 in the evaluation, which credibility level, together with the significance level 3 of 6, leads to an accident report being sent within the time interval and an accident warning being transmitted .

In accordance with a third exemplary embodiment, a third sequence of the method is described, which is typical of the case in which a presumption of an accident is correct and a serious accident has occurred. For the method according to the invention for detecting and rating an accident involving a vehicle, which is in the form of an automobile, for example, there is a mobile device located in a passenger compartment of the vehicle. The mobile device is e.g. a mobile phone of the driver that is in a briefcase of the driver, which has been stowed in a trunk of the vehicle. The mobile device comprises e.g. a first sensor, which is formed by an acceleration sensor of the mobile phone, and a second sensor, which is formed by a microphone of the mobile phone, and a third sensor, which is in the form of a velocity sensor of the mobile phone. An accident monitoring system is operated as an application on the mobile device. The application continually captures sensor data of the first sensor and of the second sensor and of the third sensor and temporarily stores said data in a memory of the mobile device in the manner of a shift register. The acceleration sensor senses accelerations and the microphone picks up in particular low-frequency sounds, such as occur in the event of a collision between a vehicle and another vehicle or in the event of a collision with a stationary obstacle, e.g. a tree.

In the third exemplary case, the sensor data of the acceleration sensor of the mobile device now exceed a stipulated threshold value for the acceleration, because the driver' s briefcase, together with the mobile device accommodated therein, undergoes an extremely high deceleration in the trunk as seen in the direction of travel . This is caused by a frontal impact by the vehicle against a tree .

In respect of the threshold values and in respect of the steps triggered as a result of at least one of the threshold value s having been exceeded, explicit reference is made to the explanations pertaining to the first exemplary embodiment .

In the third exemplary case , the acceleration sensed by the first sensor leads to the highest significance level 6 of 6 in the evaluation and the low-frequency sounds detected by the second sensor lead to a significance level 5 of 6 in the evaluation . The abrupt change of velocity sensed by the third sensor also leads to a signif icance level 6 of 6 in the evaluation .

In the third exemplary case , the first pas s of the loop results in an accident report being sent immediately, since the significance level is sufficiently high for it not to be nece s sary to as certain the credibility level , and a rapid accident report and a rapid accident warning take priority .

It i s true of all of the exemplary embodiments that the accident monitoring system can evaluate sensor data from further sensors of the mobile device . In this instance , the accident monitoring system is programmed flexibly enough for it to be able to work with different groups of sensors , depending on the sensors that are available f rom the mobile device .

Furthermore , there is also provision for event data to be evaluated in addition , with the result that a low credibility level can be derived for example from the circumstance that a pas senger is playing a computer game in the time interval after the timestamp is set or it is established that the passenger was playing the computer game without interruption before the timestamp was set, during the timestamp and after the timestamp was set.

Referring to an embodiment introduced in Figures 1 to 3, the method according to the invention is furthermore explained.

Fig. 1 shows a vehicle 1 carrying a mobile device 2, the method steps being executed on the mobile device 2. The mobile device 2 is equipped with at least three sensors 1, 2 and 3. The sensor 1 is a velocity sensor, sensor 2 is a microphone and sensor 3 is an acceleration sensor. Also other sensors like navigation sensors are possible.

The accident monitoring system is operated on the mobile device 2 in such a way that the mobile device 2 continually captures sensor data 6, from sensor 3, sensor data 7 from sensor 4 and sensor data 8 from sensor 5. The sensor data 6, 7 and 8 is temporarily stored over a time window 11 in a memory 9 in the manner of a shift register 10. The time windows 11 and accordingly the time dependent storage of the shift register 10 moves along the time axis 12 as a dataset 13 in time. The invention is not limited to three sensors but can use the sensor data and shift register dataset of any kind and amount of sensors provided by the mobile device 2.

Fig. 2 shows the shift register 10 according to Fig. 1 with the data 6, 7 and 8 from the sensors 3, 4 and 5. When a potential accident 20 occurs, the sensor data 6, 7 or 8 of at least one of the sensors have exceeded a threshold value. This is depicted by symbols 21, 22 and 23 for each data stream of sensor data 6, 7 and 8 stipulated for the respective sensor. Threshold values do not have to be exceeded on all sensors or in the same dimension. Also single pea ks or events can be sufficient to initiate the further steps of the method . When threshold value 21 and/or 22 and/or is exceeded , a timestamp 24 i s set in the memory 9 of the shift regi ster 10 , and the available sensor data 6 , 7 and 8 remain stored . A continusous enlargement 39 of the stored data is provided in order to allow the sensor data 6 , 7 and 8 to be continuous ly used .

An accident verification routine 30 as also shown in Fig . 3 is launched on the mobile device 2 , wherein the accident verification routine 30 evaluates the stored sensor data in such a way that a tiered significance level 31 ( severity level ) i s ascertained . The accident verification routine 30 also evaluates the sensor data captured from the timestamp onwards in a time interval of predetermined length in such a way that a tiered credibility level 32 ( confidence level ) is repeatedly ascertained . These being continuously calculated in the routine 30 the accident verification routine 30 use s the s ignificance level 31 ( severity level ) and/or the currently a scertained credibility level 32 ( confidence level ) repeatedly 33 for the significance level 31 and repeatedly 34 for the creditability level 32 in the time interval from the time stamp 24 to the time of verification to decide

■ either that an accident report 35 is sent or

■ that no accident report 36 is sent and the accident verification routine is executed again or

■ that the accident verification routine is terminated 37 .

As mentioned the report 35 can be transmitted by the same mobile device 2 a s that on which the method is also executed . It i s a pos sible embodiment that the verification routine 30 for evaluation also comprises a comparison of the sensor data and/or the event data 40 with comparison data 41 , wherein the comparison 41 data used are statistical data that in particular have been ascertained on the basis of analyzed and verified accidents and/or wherein the compari son data used are geodata derived from a map that are taken as a basi s for rating a position or direction of movement of the mobile device 2 .