WO2009049814A1

WO2009049814A1 - Method for producing a collision-prevention system for a motor vehicle

Info

Publication number: WO2009049814A1
Application number: PCT/EP2008/008521
Authority: WO
Inventors: André NEUBOHN; André Zander; Frank Keck; Christian Weiss; Reinhard Kaiser; Toni Palau; Simon Sigl; Andreas Kuhn
Original assignee: Volkswagen Aktiengesellschaft; Audi Ag; Andata Entwicklungstechnologie
Priority date: 2007-10-20
Filing date: 2008-10-09
Publication date: 2009-04-23
Also published as: DE102007050254B4; DE102007050254A1

Abstract

The invention relates to a method for producing a collision-prevention system (1) for a motor vehicle (60) or a motor vehicle (60) with a collision-prevention system (1), wherein the collision-prevention system (1) comprises at least one sensor (11, 12, 13) and a collision-prevention module (10) for generating a preventative measure (W, B) at least as a function of an output signal (S1, S2, Sm) which is supplied by the sensor (11, 12, 13), wherein a traffic situation with at least one road user (31) is simulated, wherein a time (TTC) until a collision, a collision probability (pC) and/or an accident severity (sC) is/are determined by means of the traffic simulation, and wherein the collision-prevention module (10) or a part of the collision-prevention module (10) is generated by training with the determined time (TTC) until a collision, the determined collision probability (pC) and/or the determined accident severity (sC) or by training with a preventative measure (W, B) which is derived from the determined time (TTC) until a collision, the determined collision probability (pC) and/or the determined accident severity (sC).

Description

description

Method for producing a collision protection system for a motor vehicle

The invention relates to a method for producing a collision protection system for a motor vehicle or a motor vehicle having a collision protection system, wherein the collision protection system comprises at least one sensor and a collision protection module for generating a protective measure at least in response to an output signal supplied by the sensor.

The known solutions usually assume that a "driving lane" or the movement of the own vehicle and those of the surrounding objects are extrapolated and checked for possible collisions.This is usually very complicated and can usually only be realized for very limited problems become.

It is therefore an object of the invention to provide an improved collision protection system.

The aforementioned object is achieved by a method for producing a collision protection system for a motor vehicle or a motor vehicle with a collision protection system, wherein the collision protection system comprises at least one sensor and a collision protection module for generating a protective measure at least in response to an output signal supplied by the sensor, wherein a traffic simulation of a traffic situation with at least one road user, wherein by means of the traffic simulation a time to a collision, a collision probability and / or an accident severity are determined, and wherein the collision protection module or a part of the collision protection module by training with the determined time to a collision, by training with the collision probability and / or the determined severity or by training with one of the determined time to a collision, the collision probability and / or he mediated accident severity derived protective measure is generated.

A protective measure in the sense of the invention is in particular a warning signal or its optical, acoustic and / or haptic output and / or the generation of a braking intervention or the output of a braking signal for generating a braking intervention. Different braking thresholds can be used (driver brakes himself (increase of brake pressure), driver leaves accelerator pedal, completely autonomous). A protective measure within the meaning of the Alternatively or additionally, steering can be a steering effect, such as a change in the steering ratio.

In a further advantageous embodiment of the invention, the traffic simulation of a traffic situation with at least two road users takes place. In a further advantageous embodiment of the invention carried traffic simulations for a plurality of different traffic situations. In a further advantageous embodiment of the invention, traffic simulations for more than one hundred, in particular for more than one thousand, different traffic situations.

In a further advantageous embodiment of the invention, the simulation is split into different simulation paths in a simulation increment to determine the time to a collision, the collision probability and / or the accident severity by means of the traffic simulation and the time associated with the simulation increment up to a collision, the Time of the simulation increment associated collision probability and / or the time of the simulation increment associated accident severity determined by evaluating the traffic situation at the end of the simulation paths. In this case, the time to a collision is, in particular, a mean time until a collision, and the severity of the accident is, in particular, an average severity of the accident. In a further advantageous embodiment of the invention, a time to a collision, the collision probability and / or an accident severity is determined for a plurality of times.

In a further advantageous embodiment of the invention, the collision protection system is implemented in a motor vehicle. It is provided in a further advantageous embodiment of the invention that the determined by the collision protection system time to collision and / or the (expected) accident severity input to a controller for an airbag system or a belt tensioner.

Motor vehicle in the sense of the invention is in particular a land vehicle which can be used individually in road traffic. Motor vehicles according to the invention are not limited in particular to land vehicles with internal combustion engine. Further advantages and details emerge from the following description of exemplary embodiments. Showing:

1 shows an embodiment of a collision protection system for a motor vehicle,

2 shows an embodiment of a method for producing a collision protection system for a motor vehicle,

3 a traffic situation,

4 is a table as a result of a simulation,

5 shows an exemplary embodiment of a limit value for generating a brake signal or a warning signal,

6 shows a further embodiment of a limit value for generating a brake signal or a warning signal,

7 shows an embodiment of a training table,

8 an embodiment for training a collision protection module,

Fig. 9 shows an embodiment of a motor vehicle and

10 shows another embodiment of a collision protection system for a motor vehicle.

Fig. 1 shows a collision protection system 1 for a motor vehicle. The collision protection system 1 comprises at least one sensor 11 for predicting a collision or detecting an environment of the motor vehicle. In the present embodiment, at least three sensors 11, 12, 13 are provided for detecting an environment of the motor vehicle. S ₁ denotes the output signal of the sensor 11, S ₂ the output signal of the sensor 12, and S _m the output signal of the sensor 13. The sensors 11, 12 and / or 13 can be designed, for example, as cameras.

The collision protection system 1 also includes a collision protection module 10 for generating a warning signal W and a brake signal B in response to the output signals S ₁ , S ₂ and S _{m of} the sensors 11, 12 and 13 and the speed v of the motor vehicle and the acceleration a of the motor vehicle. The brake signal B and the warning signal W are exemplary embodiments of a protective measure in the sense of the claims. The acceleration a of the motor vehicle also includes a deceleration (negative acceleration) of the motor vehicle, for example due to braking.

2 shows a method for producing the collision protection system 1 according to FIG. 1 or for producing a motor vehicle with the collision protection system 1, wherein a traffic simulation of a plurality of traffic situations with one or more traffic participants is shown. takes place and wherein by means of the traffic simulations a time to a collision, a collision probability and an accident severity are determined. In the method for producing the collision protection system 1 shown in FIG. 2, initially various traffic situations to be simulated are compiled in a step 21. For this purpose, it is first determined which protective measures should be simulated with which sensor signals for which traffic situations and should be trained later.

3 shows an embodiment of such a traffic situation to be simulated. Here, reference numeral 31 denotes the own motor vehicle and reference numeral 310 its route. Reference numeral 32 denotes another motor vehicle and reference numeral 320 indicates its route. Reference numeral 35 denotes the location of a possible collision between the two motor vehicles 31 and 32.

The step 21 is followed by a step 22, in which the different traffic situations are simulated. For example, the crash simulation program PC-Crash can be used for this purpose. This simulates a multitude of traffic situations and automatically simulates them for different processes using Monte Carlo simulation. Several thousand processes are simulated. For this purpose, the conventional simulation procedure (forward simulation) is modified. In this case, the nominal simulation is stopped in a defined time interval ti, t ₂ , t ₃ , U ■ t ■, as shown in FIG. 4. Subsequent simulations are then carried out for the resulting state by further continuation and additional variation in accordance with the preconditions in order to estimate the collision probability pC and the expected severity sC and, if appropriate, the time TTC up to a collision (incremental simulation).

In this way, not only the nominal ideal sensor values but also distribution data of the collision probability pC and the accident severity sC over the time t are obtained for a sequence. The result of a simulation run is a table shown in FIG. 4, in the individual times ti, t ₂ , t ₃ , t ₄ ... T _n values for the speed V ₁ , V ₂ , V ₃ , V ₄ . V _{n of} the motor vehicle, the acceleration a ^ a ₂ , a ₃ , a ₄ ... a _{π of} the motor vehicle, the output signal Sn, Si ₂ , S ₁₃ , Si ₄ ... S _{1n of} the sensor 11 of the output signal S ₂₁ , S ₂₂ , S ₂₃ , S ₂₄ ... S _{2n of} the sensor 12, the output signal S _m1 , S _m2 , S _m3 , S _m4 ... S _mn , the time TTC ₁ , TTC ₂ , TTC ₃ , TTC ₄ . .. TTC _n up to a collision, the collision probability PC ₁ , pC ₂ , pC ₃ , pC ₄ ... pC _n and the accident severity SC ₁ , sC ₂ , sC ₃ , sC ₄ ... sC _{n be} assigned.

- A - The step 22 is followed by a step 23, in which a warning signal W and a brake signal B are derived as a function of the collision probability pC and the accident severity sC. In a simple case, the product of collision probability pC and accident severity sC can be compared with a limit value. If, for example, the product of collision probability pC and accident severity sC exceeds a first limit value, the warning signal W is set from a value 0 to a value 1. If the product of collision probability pC and accident severity sC exceeds a second limit value which is greater than the first limit value, then the brake signal is set from a value 0 to a value 1. In this case, the value 1 for the warning signal W means that an optical, acoustic and / or haptic warning is output. If the brake signal B is set to a value 1, a braking intervention takes place in the motor vehicle. Various braking thresholds can also be used (driver brakes himself (increase in brake pressure), driver leaves accelerator pedal, completely autonomous). In this case, in addition to the value 0 for the brake signal B, three further values are provided. Alternatively or additionally, a steering influence, such as a change in the steering ratio to be provided as a brake signal B. With an additional steering influence, such as the change of the steering ratio, as a brake signal B, the brake signal is a vector:

wherein BB corresponds to the described brake signal B without steering control and LB corresponds to the steering influence.

For determining the warning signal W and the brake signal B as a function of pairs of values of the collision probability pC and the severity of severity sC, more complex relationships can also be selected, as illustrated, for example, in FIGS. 5 and 6. If a point defined by the collision probability pC and the accident severity sC is right above the line 42 or 46, the warning signal is set from a value 0 to a value 1. If a point defined by the collision probability pC and the accident severity sC lies to the right above the line 41 or 45, a brake signal is set from a value 0 to a value of 1.

The result of step 23 is a table shown in FIG. 7, which is stored in a further step 24. Step 24 is followed by a query 25 as to whether enough training data is present or whether all traffic situations that were to be simulated have been processed. If further data is to be generated or traffic situations need to be simulated, query 25 is followed by step 21; otherwise, query 25 is followed by a step 26, in which the Collision protection module is trained by means of a system shown in Fig. 8. In this case, reference numeral 52 corresponds to the table shown in FIG. 7 or to all the tables for the simulated traffic situations. Reference numeral 51 denotes a training algorithm for training the collision protection module 10 as a function of the data stored in the tables 52. Reference symbol B 'in FIG. 8 designates the brake signal generated during training of the collision protection module 10. Reference symbol W denotes the warning signal generated during training of the collision protection system 10.

In addition, it can be provided that the collision protection system 10 also outputs the time TTC until a collision. In this case, the time TTC is to be included in the training until a collision occurs. The time TTC output by the collision protection system 10 until the collision can - as shown in FIG. 9 - be used in a motor vehicle 60 for controlling a belt tensioner 65 or an airbag 66. For example, it may be provided that when the time TTC falls below 200 ms until a collision, the triggering threshold of the airbag 66 or the belt tensioner 65 is lowered or the belt tensioner 65 or the airbag 66 is ignited earlier.

The step 26 shown in FIG. 2 may be followed by a step 27, in which the collision protection system 1 with its sensors 11, 12 and 13 as well as its collision protection module 10 is integrated in the motor vehicle 60. For this purpose, the collision protection system 1 is connected to an output device 61 for outputting a warning signal and to the brake 62 of the motor vehicle 60. In addition, the collision protection system 1 is connected to a controller 63 for providing a value of the speed v of the motor vehicle 60 and a value of its acceleration a. In addition, the collision protection system 1 is connected to the airbag 66 and the belt tensioner 65.

10 shows an alternative embodiment of a collision protection system 100 with an alternatively configured collision protection module 110. The alternatively configured collision protection module 110 supplies values for the collision probability pC and the accident severity sC instead of the brake signal B and the warning signal W. To train the collision protection module 110, the training shown in FIG. 8 is to be modified accordingly. In this case, the collision protection module 110 is not trained with the data denoted by D ₁ and D ₃ in FIG. 7, but with the data denoted by D ₁ and D ₂ . If it is provided that the collision protection module 110 should also output the time TTC until the collision, this is included in the training. The collision protection system 100 also includes an assessment module 115, which generates a brake signal B or a warning signal W on-board or on-line from the collision probability pC and the accident severity sC. To produce the collision protection system 100 shown in FIG. 10, step 23 is omitted.

The method described with reference to FIG. 2 may initially be carried out for ideal sensors with regard to steps 21, 22, 23, 24 and 26 as well as query 25. Such an ideal sensor has, for example, an opening angle of 180 °, a range of 100 m and is noise-free. Subsequently, the method can be repeated for models of different sensors, so that the sensors can be assessed in terms of their performance. Similarly, action concepts and executions can be evaluated. The method described can also be combined with the method described in WO 2005/037612 A1.

LIST OF REFERENCE NUMBERS

1, 100 Column protection system

10, 110 collision protection module

11, 12, 13 sensor

21, 22, 23, 24,

26, 27 step

25 query

31, 32, 60 motor vehicle

35 Location of a possible collision

41, 42, 45, 46 line

51 training algorithm

52 Table or plurality of tables

61 output device

62 brake

63 control

65 belt tensioners

66 airbag

115 assessment module

310, 320 Route a acceleration

B, B ¹ Brake signal

D data pC collision probability

S Output signal of a sensor sC Accident severity t Time or time grid

TTC time to collision

V speed

W ₁ W warning signal

Claims

claims

1. A method for producing a collision protection system (1) for a motor vehicle (60) or a motor vehicle (60) with a collision protection system (1), wherein the collision protection system (1) at least one sensor (1 1, 12, 13) and a collision protection module ( 10) for generating a protective measure (W, B) at least as a function of an output signal (S ₁ , S ₂ , S _m ) supplied by the sensor (11, 12, 13), wherein a traffic simulation of a traffic situation with at least one road user (31 ), wherein by means of the traffic simulation a time (TTC) up to a collision, a collision probability (pC) and / or an accident severity (sC) are determined, and wherein the collision protection module (10) or a part of the collision protection module (10) by training with the determined time (TTC) up to a collision, the ascertained collision probability (pC) and / or the ascertained accident severity (sC) or by training with one of the Time (TTC) until a collision, the determined collision probability (pC) and / or the determined accident severity (sC) derived protective measure (W, B) is generated.

2. The method according to claim 1, characterized in that the traffic simulation of a traffic situation with at least two road users (31, 32) takes place.

3. The method according to claim 1 or 2, characterized in that carried out traffic simulations for a plurality of different traffic situations.

4. The method according to claim 1, 2 or 3, characterized in that traffic simulations are carried out for more than a thousand different traffic situations.

5. The method according to any one of the preceding claims, characterized in that for determining the time (TTC) up to a collision, the collision probability (pC) and / or the severity (sC) by means of the traffic simulation, the simulation in a simulation increment split into different simulation paths is and the time (ti) of the simulation increment associated time (TTCi) up to a collision, the collision probability associated with the time K ₁ ) of the simulation increment (pCi) and / or the accident severity (SC ₁ ) associated with the time (ti) of the simulation increment is determined by evaluating the traffic situation at the end of the simulation paths. A method according to claim 5, characterized in that for a plurality of times (U, t ₂ , t ₃ , U. - _■ t _n ) a time (TTC ₁ , TTC ₂ , TTC ₃ , TTC ₄ , ..., TTC _n ) up to a collision, a collision probability CpC ₁ , pC ₂ , pC ₃ , pC ₄ , ..., pC _n ) and / or an accident severity (SC ₁ , sC ₂ , sC ₃ , sC ₄ , ... ^' , sC _n ) is determined.