DE102006036363A1 - Method for steering motor vehicle, involves calculating avoidance path for avoiding before object whereby controller output signals are weighted with one weighting factor which is determined as function of vehicle velocity - Google Patents

Abstract

Steering method involves calculating the avoidance path for avoiding before the object. A controller output signal is determined in accordance with a deviation (e) between an actual position of the motor vehicle and a set point position, predefined on the basis of the avoidance path, in two linear controller modules (501 1-501 N). Controller output signals are weighted with one weighting factor which is determined as a function of vehicle velocity (v). A steering angle of steerable wheels of the motor vehicle is determined by arbitration of the weighted controller output signals. A steering system of the motor vehicle is influenced to measure the determined steering angle. INDEPENDENT CLIAMS are also included for the following: (1) Computer program product; and (2) Steering device.

Description

technical area

The The invention relates to a method for predicting a movement trajectory one in the traffic moving object. Furthermore, the invention relates to a device to predicate a movement trajectory of an object moving in traffic, that is to carry out of the method is suitable.

background the invention

One The goal in the development of motor vehicles is driver assistance systems for accident prevention. These systems monitor the environment of the vehicle, decide if a collision with an object can occur and engage in the steering system or the braking system of the vehicle to avoid the accident by dodging or decelerating. requirement for the decision as to whether an accident can happen or not is the prediction the future Position of the object in a period of a few seconds.

Known prediction methods are based on an extrapolation of the detected path of the object. An example of such a procedure is the Euler extrapolation, in which the position x → (t + T) of the object at a time t + T from the position and the speed at the time t is predicted:

Also known is the polynomial extrapolation. For example, in the third order x → (t + T) = c → three (t + T) three + c → 2 (t + T) 2 + c → 1 (t + T) + c → 0 being used to determine the coefficients c → i the positions recorded in the preceding time steps x → (t - iT) be used. However, in many cases the trajectory of the object can not be predicted with sufficient reliability by such methods. Thus, for example, stopping maneuvers or lane changes of the object can not be correctly predicted, especially for large prediction periods. In the case of a lane change, the methods generally lead rather to the prediction of a cornering and, in the case of a stopping maneuver, the result of the extrapolation is a future reversing of the object.

presentation the invention

It is therefore an object of the present invention, the future position one in the traffic moving object, with a higher reliability to predict. In particular, a reliable prediction also for larger prediction periods are possible.

According to the invention this Task by a method having the features of the claim 1 and by a device having the features of the claim 39 solved.

• – Ermitteln wenigstens einer Bewegungsgröße des Objekts,
• – Ermitteln eines Fahrmanövers, das von dem Objekt ausgeführt wird, anhand einer Bewertung der ermittelten Bewegungsgröße,
• – Auswählen eines Modells für die Bewegung des Objekts in Abhängigkeit von dem ermittelten Fahrmanöver und
• – Berechnen der Bewegungstrajektorie des Objekts anhand des ausgewählten Modells.

Accordingly, it is provided that a method of the aforementioned type is carried out with the following steps:

• Determining at least one movement quantity of the object,
• Determining a driving maneuver executed by the object based on an evaluation of the determined movement amount,
• Selecting a model for the movement of the object as a function of the determined driving maneuver and
• - Compute the movement trajectory of the object based on the selected model.

• – Wenigstens einen Sensor, aus dessen Signalen eine Bewegungsgröße des Objekts ermittelbar ist,
• – eine Fahrmanövererkennungseinrichtung, mit der anhand einer Bewertung der ermittelten Bewegungsgröße ein Fahrmanöver des Objekts ermittelbar ist und
• – eine Prädiktionseinrichtung, mit der anhand eines Modells die Bewegungstrajektorie des Objekts berechenbar ist, wobei das Modell in Abhängigkeit von dem ermittelten Fahrmanöver aus einer Menge von Modellen auswählbar ist.

Furthermore, a device of the aforementioned type is provided which comprises the following devices:

• At least one sensor from whose signals a movement quantity of the object can be determined,
• - A driving maneuver recognition device, with the basis of an assessment of the determined motion size a driving maneuver of the object can be determined and
• A prediction device with which the movement trajectory of the object can be calculated on the basis of a model, the model being selectable from a set of models as a function of the determined driving maneuver.

Advantageous is in the invention first a driving maneuver which is executed by the object. The prediction the movement trajectory of the object is then based on a model, that in dependence from the determined driving maneuver selected becomes. This allows for different determined driving maneuvers the object is used in each case adapted prediction models. Thus, accuracy and reliability of the prediction significantly increased. The term object is here in particular both a motor vehicle to understand as well as an object in the environment of the motor vehicle, at which may be, for example, another vehicle. Under the term movement size are in particular to understand all the driving dynamics state variables of an object. For example, the object speed, a yaw angle, is one Yaw rate, a slip angle and a steering angle of steerable wheels of the Object of the term.

at an embodiment of the method and the device, it is provided that the driving maneuver off a set of predetermined driving maneuvers is selected.

hereby is it possible, for example, the driving maneuver to choose from a crowd the typical driving maneuvers of motor vehicles in road traffic includes. The prediction can be adapted to these typical maneuvers.

A another embodiment of the method and the device provides that the determined driving maneuver one the longitudinal movement describing the driving maneuver and the transverse movement of the object describing driving maneuvers.

By such a separate consideration of the longitudinal and lateral dynamics of the object becomes a particularly simple and thus fast and resource-saving prediction allows.

A Embodiment of the method and the device provides that the driving movement describing the transverse movement of the object a lot selected which is at least a straight ahead, a cornering and a Lane change includes.

hereby are the most common in traffic and in particular occurring on a highway or a motor road maneuvers considered, so that a model-based prediction for almost everyone on such traffic routes carried out by an object lateral dynamics maneuver be performed can.

It has been shown to be a distinction between a lane change and a cornering of the object based on a characteristic Motion size hit which is formed from at least two motion quantities of the object. As a first in the characteristic movement size incoming Movement size of the object is doing a beneficial change used in the direction of movement of the object characterizing size. This may be, for example, the yaw rate of the object or a rate of change act of the price angle of the object.

Therefore is an embodiment of the method and the device thereby characterized in that a distinction between a cornering and a lane change based on a characteristic amount of movement hit is made up of a first movement that is a change the direction of movement of the object characterized, and at least a second amount of movement of the object is formed.

Advantageous allows a correlation between the first and the second movement amount a Distinction between a cornering and a lane change. The first movement size, at in particular, for example, a yaw rate or a Course angle change can act gives the change the direction of movement of the object.

It In particular, it has been shown that already in the initial phase of the driving maneuver between a change the direction of movement and the speed of the object are correlated which is characteristic of a lane change on the one hand and cornering on the other.

Therefore includes an embodiment of the method and the device that it is at the second Movement size of the object is a speed of the object.

Furthermore It turned out that already in the initial stages of the maneuver between a change the direction of movement and the steering angle of steerable wheels of the Object is a correlation that is typical of a lane change on the one hand and cornering on the other hand.

Therefore includes another embodiment of the method and the device that it is at the second Movement size by one Steering angle on steerable wheels of the object.

A another embodiment of the method and the device is characterized in that it at the second movement size by one Steering angle on steerable wheels of the object.

A Configuration of the method and the device is characterized from that the movement trajectory based on the model in dependence is determined by at least one parameter.

at a development of the method and the device it is intended a position of the object predicated on the movement trajectory is compared with a detected position of the object, and that the parameter in dependence adjusted by the result of the comparison.

hereby can the prediction results for one later Time based on a comparison of the measured position and the prediction for one earlier Time to be adjusted. The prediction for the later date becomes thereby further improved.

The The invention is particularly suitable for the prediction of the movement trajectories of objects in the environment of a motor vehicle. Therefore, one embodiment of the proceedings and the device characterized in that it is at the object is located in the environment of a motor vehicle Environment object, where the motion size of the environment object means an environment sensor of the motor vehicle is determined.

at The environment object may be, in particular, an environment of the "own" motor vehicle Another motor vehicle act.

A Design of the method and the device is characterized in that that's the longitudinal movement driving maneuvers describing the environment object as a function of a comparison of speeds present at the successive times of the environment object is determined from an amount that at least a uniform Movement and emergency braking.

This allows an identification of relevant the longitudinal movement of the environment object descriptive maneuver even if the state of motion of the environment object - for example due to errors in the capture - only with a lower Accuracy can be determined. This is often the case if the speed or acceleration of the environment object by means of an environment sensor based on a comparison of position data must be determined.

A another embodiment of the method and the device provides that a direction of movement of the environment object at successive times and that this describes the transverse movement of the environment object maneuvers based on a comparison of from the determined directions of movement determined course angles is determined.

In an embodiment of the method and the device, it is additionally provided that the driving maneuver describing the transverse movement of the environment object is based on a comparison of the size (γ Obj, 0 (t) - γ Obj, 0 (t 0 )) · Ν Obj, 0 with at least one threshold value, where γ _{Obj, 0} (t) is the heading angle in a fixed axbox at an observation time, γ _{Obj, 0} (t ₀ ) is the heading angle in the stationary axbox at the beginning of a change in heading angle and ν _{Obj, 0} denotes the speed of the environment object.

This makes it possible to carry out a particularly simple and reliable identification of the driving maneuver carried out by an environment object. Under the observation time, the time or Time step understood in which the course angle γ _{Obj, 0} (t) has been detected and in which the identification of the driving maneuver is made.

A further embodiment of the method and the device includes that a lane change of the environment object is determined when the course angle of the object changes to a predetermined extent and when the size (γ Obj, 0 (t) - γ Obj, 0 (t 0 )) · Ν Obj, 0

values within a given interval.

Especially allows an evaluation of the size mentioned thus an identification of a lane change already in its initial phase.

On the basis of the size, lane changes of cornering of the environment object can already be distinguished in the initial phase of the maneuver. Therefore, an embodiment of the method and apparatus provides that cornering of the environment object is determined when the heading angle of the object changes by a predetermined amount and when the magnitude (γ Obj, 0 (t) - γ Obj, 0 (t 0 )) · Ν Obj, 0

values outside of the predetermined interval.

Further is an embodiment the method and the device characterized in that the Movement trajectory of the environment object based on a predicated Speed of the environment object and a predicted course angle of the environment object is determined.

A Embodiment of the method and the device includes that the motion trajectory of the environment object is calculated recursively, wherein a position of the environment object is in a prediction step based on the predicated Speed and the predicated Course angle from the determined in a previous prediction step Position is determined.

Farther it is in one embodiment the method and the device provided that the predicated Speed of the surrounding object based on the selected the longitudinal movement of the environment object descriptive model.

moreover sees an embodiment of the method and the device that the predicated Course angle of the surrounding object based on the selected the transverse movement of the Environment descriptor model is calculated.

A Further development of the method and the device is characterized from that predicated Course angle of the surrounding object according to a sigmoid function is calculated when driving a maneuver Lane change has been found, the shape of the sigmoid function is determined by at least one parameter which depends on determined by the velocity of the environment object.

Under a sigmoid or a sigmoid function is in the usual Meaning an approximately S-shaped, real, continuously differentiable, monotone and limited function with a turning point Understood. Examples of this are functions of the form of a hyperbolic tangent function f (x) = αtanh (β (x - γ)), a logistic Function f (x) = α / (1 + exp (-β (x - γ)) or one Arctangent function f (x) = αarctan (β (x - γ)) with at least partially dynamically determined parameters α, β, γ. Based on such functions the trajectory can be specified closed. Furthermore, the Parameters that determine the shape of the sigmoid in a simpler and efficiently determinable on the basis of few constraints.

In addition, an embodiment of the method and the device includes that the predicted for the time t + .DELTA.t course angle γ _{Obj, 0} (t + .DELTA.t) of the environmental object by γ Obj, 0 (t + Δt) = γ obj 0 (t) + p · Δt is calculated when cornering has been determined as a driving maneuver, wherein the quantity γ _{Obj, 0} (t) is the course angle of the surrounding object at an observation time t, Δt is a time interval and p is a parameter.

The In particular, the invention is also suitable for the movement trajectory of the own motor vehicle. Therefore sees an embodiment of the method and apparatus that it is at the object is the motor vehicle, wherein the amount of movement means at least one vehicle sensor is detected.

A Design of the method and the device is characterized in that in that the driving maneuver describing the transverse movement of the motor vehicle is based on a yaw rate of the motor vehicle and / or on the basis of a steerable wheels the vehicle set steering angle is determined.

A embodiment of the method and the apparatus includes that a straight exit of the motor vehicle is determined when the yaw rate of the vehicle and / or the steering angle is smaller than are a threshold.

It has also shown that between the yaw rate of the motor vehicle and the steering angle, a correlation exists for the lateral dynamic maneuvers of the vehicle, in particular for a turn on the one hand and a lane change on the other hand, is characteristic. Based on the correlation, these maneuvers can already be differ in their initial phase.

A Configuration of the method and the device therefore provides that a cornering of the motor vehicle is determined when the Product of steering angle and yaw rate greater than a predetermined threshold is and / or a yaw angle change while of the driving maneuver amount greater than is a predetermined threshold.

A further embodiment of the method and the device provides that a lane change of the motor vehicle is determined, if no Straight ahead and no cornering can be determined.

A Further development of the method and the device includes that a lane change divided into at least three successive phases is, with the yaw rate of the motor vehicle in a first phase amount greater than a threshold, in an intermediate phase in terms of amount smaller as a threshold and in a second phase in terms of amount greater than is a threshold and wherein the yaw rate in the first and second Phase has a different sign.

Based these phases is also advantageous a distinction between a lane change and a cornering or a straight ahead possible.

at an embodiment of the method and the device, it is provided that for the first phase a maximum duration is given, and that after a previous one Detecting a lane change a cornering is determined when the duration of the first phase exceeds the maximum duration.

hereby let yourself the identification of the driving maneuver correct if initially a lane change has been determined instead of cornering.

A another embodiment of the method and the device is characterized in that for the first Phase a minimum duration is given, and that after a previous one Detection of a lane change a straight-ahead driving is determined if the duration of the first phase is less than the minimum duration.

hereby let yourself the identification of the driving maneuver correct if initially a lane change has been determined instead of a straight ahead.

moreover includes an embodiment of the method and the device, that for the intermediate phase of the lane change given a maximum duration and that after a previous detection of a lane change a straight-ahead driving is determined when the duration of the intermediate phase the maximum duration exceeds.

hereby let yourself also correct the identification of the driving maneuver, if initially a lane change has been determined instead of a straight ahead.

A embodiment of the method and the device is characterized in that the movement trajectory of the motor vehicle for the first phase of a lane change dependent on determined longitudinal and lateral velocities and / or depending on determined longitudinal and Transverse accelerations of the motor vehicle based on an Eulerian prediction predicted becomes.

Further provides an embodiment of the method and the device, that the movement trajectory of the motor vehicle during the first phase of the lane change is recorded and by a polynomial approximated and that the polynomial thus obtained becomes a prediction the movement trajectory of the motor vehicle for the second phase of the lane change is used.

hereby can the prediction the movement trajectory for the second phase of the lane change advantageous based on the trajectory while the first phase predicts become. In the prediction The second phase is thus the behavior of the driver at considered a lane change, which has been detected in the first phase of the lane change. This results in a particularly reliable and accurate prediction.

A Further development of the method and the device is characterized in that that the predicated Movement trajectory for the second phase of the lane change corresponds to the determined polynomial, that by 180 ° an end point of the first phase of the lane change is rotated and that to the movement trajectory of the motor vehicle in the intermediate phase the lane change is recognized.

Furthermore includes an embodiment of the method and the device, that the polynomial is a third degree polynomial.

On the one hand allows such a polynomial a sufficiently accurate approximation the path of the vehicle and on the other hand remains the computational effort to calculate the parameters of the polynomial sufficiently low.

at an embodiment of the method and the device, it is also provided that the movement trajectory of the motor vehicle based on a predicated Length of one Arc and a radius of the arc is determined when it is in the determined driving maneuver is a cornering.

Farther is a method for preventing a collision between a Vehicle and an environment object or for reducing a collision strength provided. In the method, movement trajectories of the motor vehicle become and / or the surrounding object, wherein the movement trajectory of the motor vehicle and / or the movement trajectory of the environment object determined by a method of the kind previously described. Based on an evaluation of the movement trajectories, a collision location is determined. Then become dependent from the distance from the collision location collision preventing and / or collision-sequence-reducing measures initiated.

When Collision avoiding measures For example, automatic braking or evasive maneuvers come into question here as well as issuing warnings from the driver that cause of these measures to prevent collision. To reduce the collision strength or the collision sequences can in particular passive safety systems such as airbags be controlled. To reduce the collision strength is also warning the driver suitable, due to the warning appropriate measure to reduce collision strength can initiate.

Furthermore The invention and its embodiments are also suitable for one Use in a so-called ACC system (ACC: Adaptive Cruise Control), that the speed of a motor vehicle regulates that one predetermined distance between the motor vehicle and one in the direction of travel not located in front of the motor vehicle another motor vehicle is fallen short of.

Therefore is also a method for regulating the speed of a Motor vehicle, in which the speed of the motor vehicle in such a way is set that a distance between the motor vehicle and a in the direction of travel of the motor vehicle located in front of the motor vehicle further vehicle does not fall below a predetermined distance value, provided that is characterized in that based on a Method of the type shown above, a movement trajectory the motor vehicle and / or the other vehicle is determined, that in dependence from the movement trajectory a distance between the motor vehicle and the other vehicle predicts is, and that the speed of the motor vehicle in dependence from the predicated Distance is set.

On This way can be determined in particular whether the other vehicle due to a driving maneuver moved in the lane of the motor vehicle. If this is the case, can the vehicle speed can be adjusted early.

Further a computer program product is provided that has an algorithm defined, which comprises a method of the type previously described.

The Invention involves the idea of a model-based prediction the movement trajectory of an object moving in traffic undertake, wherein it is the object in particular a motor vehicle or an object in the environment of the motor vehicle can act. It will first a driving maneuver of the object identified and the prediction below a model for this driving maneuver carried out. As a result, the path of the object, in particular when cornering and a lane change of the object can be predicted reliably, what with a Extrapolation of the object movement according to the prior art not possible is. The driving maneuver is advantageous based on a given criterion due to a Evaluation of at least one movement variable of the object from a set typically on the road executed maneuvers selected. On the basis of a characteristic amount of movement can in particular also a distinction between a cornering and a lane change already be made in the initial phase of the maneuver.

The prediction results can to be used, an impending collision between the vehicle and to predict the object. In case of imminent collision can activities for collision avoidance and / or for reducing collision consequences be taken.

These and other aspects of the invention will become apparent from the embodiments clear and with regard to the embodiments below described with reference to the figures.

Short description the figures

From the figures shows:

1 a schematic representation of a motor vehicle with an environmental sensor,

2 a schematic block diagram of a device for predicting the trajectory of an environment object,

three an illustration of axis crosses used for prediction,

4 an illustration of a characteristic axcross,

5 an illustration of a sigmoid function in the characteristic axbox,

6 a schematic block diagram of a section of the device for predicting the trajectory of an environment object

7 a schematic block diagram of a device for predicting the trajectory of the vehicle,

8th Diagrams in which the time profiles of the yaw angle of the vehicle and a steering angle during a lane change and a subdivision of the lane change are shown in different phases,

9 3 is a schematic block diagram of a section of a device for identifying a driving maneuver of the vehicle,

10 a state diagram for illustrating the identification of a driving maneuver of the vehicle,

11 a sketch illustrating the Bahn prediction for the vehicle at a recognized Cornering,

12a a representation of the relative frequency of different error classes in the prediction using a method according to the invention and

12b a representation of the relative frequency of different error classes in Euler extrapolation.

presentation of exemplary embodiments

In 1 is an example of a two-axle, four-wheeled motor vehicle 101 represented by an environment sensor 102 has, with the environment objects in the environment of the vehicle 101 can be detected, which are in particular other vehicles that move in the same or an adjacent lane. An environment sensor becomes exemplary 102 with a detection area 103 shown a solid angle in front of the vehicle 101 includes, in the example of an environment object 104 is shown. In the environment sensor 102 is preferably a LIDAR sensor (Light Detection and Ranging) which is known in the art per se; however, other environment sensors can also be used 102 be used. The environment sensor 102 measures the distances d to the detected points of an environment object 104 and the angles φ between the connecting straight lines to these points and the central longitudinal axis of the vehicle 101 like this in 1 exemplarily for a point P of the environment object 104 is illustrated. The the vehicle 101 facing fronts of the detected environment objects 104 are composed of several recorded points, with an in 2 shown object recognition unit 201 to which the sensor signals are transmitted, the correlations between points and the shape of an environment object 104 and a reference point for the environment object 104 certainly. As a reference point, for example, the center of the environment object 104 or the center of the detected points of the environment object 104 to get voted. The speeds of the detected points and thus the speed of the detected environment objects 104 can by means of the environment sensor 102 can not be measured directly.

They are the difference between the distances measured in successive scanning steps in the object recognition unit 201 calculated. The duration of a sampling step is understood as the duration of a time step in which a one-time sampling of the detection area 103 by means of the environment sensor 102 and evaluation of the sensor data takes place. In a similar way as the speed can basically also the acceleration of the environment objects 104 be determined by deriving its position twice. The measured distances of the detected points and the measured angles, however, have measurement errors that have such a great effect on the acceleration thus determined that sufficiently reliable information can not be obtained from the calculated acceleration and the acceleration signal is not used. In addition, based on a change in the detected position of the surrounding object 104 a direction of movement of the environment object 104 determined.

The using the environment sensor 102 recorded movement information about the environment object 104 initially refer to a vehicle-fixed reference system. A conversion to any fixed reference system can be based on the position, orientation and speed of the vehicle 101 take place in the fixed reference frame. The position and orientation of the vehicle 101 within a reference system initialized at a starting point, for example, successively starting from the starting point on the basis of the distance covered by the starting point, which can be determined on the basis of signals from wheel speed sensors, and on the basis of the yaw angle of the vehicle 101 due to integration of the yaw rate of the vehicle sensed by a yaw rate sensor 101 is ascertainable. Similarly, a determination of the position and orientation of the vehicle 101 in a fixed reference system, for example using a satellite-based location system.

The results of the evaluation of the sensor data of the environment sensor 102 become a predictor 202 supplied, whose basic structure is also in 2 is illustrated by a schematic block diagram. It contains a maneuver recognition device 203 in which one of the surrounding object 104 executed driving maneuver is identified. The maneuver recognition device 203 is with the block 204 in which, depending on the identified driving maneuver, one of N models predicts the trajectory of the reference point of the surrounding object 104 is selected. From the block 204 becomes one of the blocks according to the selected model 205 ₁ , ..., 205 _N the prediction device 207 activates the predicated path based on the selected model r → Obj (T) the reference point of the environment object 104 which is also referred to below as the path or trajectory of the environment object 104 referred to as. In the block 206 become the predicted positions of the environment object 104 with the measurements Nine positions of the environment object 104 compared. If the deviations are very large, it can be assumed that the driving maneuver has not been correctly identified. In this case, a signal is sent to the maneuver recognition device 203 sent, which leads to a reset of the stored data. The maneuver recognition device thus transitions to the state which is present, for example, also at the system start, and the identification of the current maneuver of the surrounding object begins again. For small deviations between the predicted and the measured position of the environment object 104 Parameters of the prediction model are adjusted to improve the prediction. In the object recognition unit 201 as well as at the prediction unit 202 or their components 203 - 207 these are preferably software components of a computer program that is stored in a microprocessor unit of the vehicle 101 is performed.

• A. Absolutes Achsenkreuz: Bei dem absoluten Achsenkreuz 301 handelt es sich um ein raumfestes Achsenkreuz. Auf das absolute Achsenkreuz 301 bezogene Größen werden mit dem Index "0" gekennzeichnet.
• B. Fahrzeugachsenkreuz: Bei dem Fahrzeugachsenkreuz 302 handelt es sich um ein ortsfestes Achsenkreuz, dessen Ursprung in jedem Abtastschritt in den Mittelpunkt M des Fahrzeugs 101 gelegt wird. Die positive x-Achse zeigt in Fahrzeuglängsrichtung nach vorne und die positive y-Achse in Fahrzeugquerrichtung nach links. Auf das Fahrzeugachsenkreuz 302 bezogene Größen werden mit dem Index "V" gekennzeichnet.
• C. Objektachsenkreuz: Bei dem Objektachsenkreuz 303 handelt es sich um ein ortsfestes Achsenkreuz, dessen Ursprung in jedem Abtastschritt in den Bezugspunkt Q des Umfeldobjektes 104 gelegt wird. Die positive x-Achse zeigt in die Bewegungsrichtung des Umfeldobjekts 104, d.h. in Richtung von dessen Geschwindigkeit ν →Obj und die positive y-Achse nach links in Bezug auf diese Richtung. Auf das Objektachsenkreuz 303 bezogene Größen werden mit dem Index "OA" gekennzeichnet.
• D. Charakteristisches Achsenkreuz: Bei einem Spurwechselmanöver oder einer Kurvenfahrt wird das Objektachsenkreuz 303 zu Beginn des Fahrmanövers innerhalb des absoluten Achsenkreuzes 301 für die Dauer des Fahrmanövers fixiert. Das fixierte Objektachsenkreuz wird als charakteristisches Achsenkreuz 401 bezeichnet. Auf das charakteristische Achsenkreuz 401 bezogene Größen werden mit dem Index "C" gekennzeichnet.

In the three and 4 are the Achsenkreuze illustrated, which are used in particular for the identification of the driving maneuver and for the prediction of the object trajectory:

• A. Absolute axbox: The absolute axbox 301 it is a space-fixed axbox. On the absolute axbox 301 Sizes referenced are marked with the index "0".
• B. Vehicle Axes Cross: In the vehicle axle cross 302 it is a stationary axbox whose origin in each sampling step in the center M of the vehicle 101 is placed. The positive x-axis points in the vehicle longitudinal direction to the front and the positive y-axis in the vehicle transverse direction to the left. On the vehicle axle cross 302 Sizes referenced are indicated by the index "V".
• C. Object axis cross: At the object axis cross 303 it is a stationary coordinate system whose origin in each sampling step in the reference point Q of the environment object 104 is placed. The positive x-axis points in the direction of movement of the environment object 104 ie in the direction of its speed ν → Obj and the positive y-axis to the left with respect to this direction. On the object axis cross 303 Sizes referenced are marked with the index "OA".
• D. Characteristic Axis Cross: In a lane change maneuver or cornering, the object axis cross becomes 303 at the beginning of the driving maneuver within the absolute axbox 301 fixed for the duration of the maneuver. The fixed object cross is called a characteristic axbox 401 designated. On the characteristic axbox 401 Sizes referenced are marked with the index "C".

In 4 is the characteristic axbox 401 shown at three different times t ₁ , t ₂ and t ₃ . The dashed line shows a path of the environment object 104 , At time t ₁ , the environment object moves 104 straight and the characteristic axbox 401 corresponds to the object axis cross 303 ie, its origin becomes the reference point Q of the environment object in each sampling step 104 set. At time t ₂ , it is determined that the environment object 104 starts to go through a bend. Therefore, the characteristic axbox becomes 401 Fixed at time t ₂ until it is determined at the time t ₃ , that the environment object 104 moving in a straight line again. Furthermore, in 4 the decomposition of the object speed ν → Obj into the components ν _{x, Obj, C} and ν _{y, Obj, C} within the characteristic axioms 401 illustrated.

The maneuver recognition in the maneuver recognition device is divided into a separate recognition with regard to the longitudinal and lateral movement behavior of the environment object 104 , The information about the detected longitudinal motion behavior is used to estimate the velocity of the environment object 104 and the information about the detected lateral movement behavior becomes the prediction of the course angle of the surrounding object 104 used. There is also the special case of a stationary environment object 104 and an unidentifiable behavior. The maneuver recognition is based on the evaluation of the course angle γ _{Obj, 0} in the absolute axbox. Neglecting the slip angle, this course angle can be calculated from the longitudinal velocity ν _{x, Obj, 0} and the lateral velocity ν _{y, Obj, 0} in the absolute axbox 301 be calculated:

The longitudinal and transverse velocities of the environment object 104 in absolute axbox 301 can take into account the position, orientation and speed of the vehicle 101 within the absolute axbox 301 from the signals of the environment sensor 102 be won.

With regard to the longitudinal movement behavior, only two behaviors are distinguished within the scope of the maneuver recognition as a consequence of the above-described inaccuracies of the acceleration signal. If Δν _{Obj, 0} <-7 m / s ² , an emergency braking of the surrounding object takes place 104 where Δν _{Obj, 0 is} the difference between the amounts ν _{Obj, 0} (k) and ν _{Obj, 0} (k-1) of the velocity of the surrounding object 104 in the down solute axbox 301 in the current sampling step k and in the preceding sampling step k-1 is: Δν Obj, 0 = ν Obj, 0 (k) - ν Obj, 0 (k - 1)

If no emergency braking of the environment object 104 is determined, a constant speed ride is detected.

• a) Geradeausfahrt
• b) Kurvenfahrt
• c) Spurwechsel

With regard to the lateral movement behavior, three typical driving situations are distinguished:

• a) straight ahead
• b) cornering
• c) lane change

For this purpose, based on the absolute course angle γ _{Obj, 0,} the characteristic course angle γ _{Obj, Ch, 0 of} the surrounding object 104 in absolute axbox 301 If the difference between the absolute course angle γ _{Obj, 0} (k) in a sampling step k and the absolute course angle γ _{Obj, 0} (k-1) remains below a threshold value σ ₁ in the preceding sampling step k- ₁ , then the characteristic course angle is the same in the time step k the absolute course angle, ie, we have γ _{Obj, Ch, 0} (k) = γ _{Obj, 0} (k). However, if the difference is greater in magnitude than the threshold value σ ₁ , the characteristic course angle does not change in the time step k and γ _{Obj, Ch, 0} (k) = γ _{Obj, Ch, 0} (k-1). The definition of the characteristic price angle is thus summarized:

If the heading does not change within a given limit, especially if | Δγ (k) | ≡ | γ Obj, 0 (k) - γ Obj, Ch, 0 (K) | ≤ σ 2 (4) with a threshold value σ ₂ , a straight-ahead travel is detected. A cornering or lane change of the environment object 104 is determined if: | Δγ (k) | > σ 2 (5)

A distinction between a cornering and a lane change is based on an evaluation of the change in the longitudinal and transverse velocities of the environment object 104 in the characteristic axbox 401 , When the characteristic longitudinal velocity decreases and the characteristic lateral velocity increases, ie when Δν x, Obj, C (k) = ν x, Obj, C (k) - ν x, Obj, C (k - 1) <-σ three and (6) Δν y, Obj, C (k) = ν y, Obj, C (k) - ν y, Obj, C (k - 1)> -σ 4 (7) with predetermined positive threshold values σ ₃ and σ ₄ , cornering is detected. The sign of Δγ indicates whether it is a left-hand or right-hand curve, whereby a left-hand bend at Δγ (k)> 0 and a right-hand curve at Δγ (k) <0 is detected.

When changing lanes, the course angle of the environment object changes 104 fast, with the rate of change of the velocity of the environment object 104 (it is immediately clear that the rate of change of the heading angle during a lane change at 30 m / s is smaller than the rate of change during a lane change at a speed of 10 m / s). Therefore, to detect the lane change, the size becomes K LC (k) = Δγ (k) · ν Obj, 0 (k) (8) evaluated, where with ν _{Obj, 0} (k), the amount of the object velocity in the time step k is designated. A lane change is detected when | K _LC | is within a predetermined range; otherwise a cornering is detected. Preferably, a lane change is detected in particular if | K _LC | between 0.9 m / s and 4 m / s (where Δγ is given in radians). The sign of K _LC or Δγ again determines the direction in which the lane change takes place.

The following table summarizes the previously explained criteria represented by a tripod (0: approximately zero, +: positive value, -: negative value, ++: high value):

wird die Position (x_Obj,0, y_Obj,0) berechnet, wobei n = 1, ..., N eine positive Zahl und Δt_p ein Prädiktionszeitintervall ist. Mit T ist die Dauer eines Abtastschrittes bezeichnet. Die Prädiktion erfolgt rekursiv anhand folgender Beziehungen: xObj,0(kpn ) = xObj,0(kpn–1 ) + νx,Obj,0(kpn )·Δtp (10) yObj,0(kpn ) = yObj,0(kpn–1 ) + νy,Obj,0(kpn )·Δtp (11) After detection of the driving maneuver in the maneuvering device 203 the result is sent to the block 204 in which a model for calculating the trajectory is selected as a function of the determined driving maneuver. Based on the selected model, the positions of the environment object become 104 or its reference point. For every prediction step

the position (x _{Obj, 0} , y _{Obj, 0} ) is calculated, where n = 1, ..., N is a positive number and Δt _{p is} a prediction time interval. T denotes the duration of a scanning step. The prediction is done recursively using the following relationships: x Obj, 0 (k p n ) = x Obj, 0 (k p n-1 ) + ν x, Obj, 0 (k p n ) · .DELTA.t p (10) y Obj, 0 (k p n ) = y Obj, 0 (k p n-1 ) + ν y, Obj, 0 (k p n ) · .DELTA.t p (11)

Based on these equations, starting from the in the time step k = k p 0 from the signals of the environment sensor 102 determined position (x _{Obj, 0} (k), y _{Obj, 0} (k)) at the beginning of the detected maneuver successively calculated the positions in the time steps kp / l to kp / N, wherein the time period N · .DELTA.t _{p is a} few seconds. The longitudinal and lateral speed of the environment object 104 in equations (10) and (11) are calculated in the following manner: ν x, Obj, 0 (k p n ) = ν Obj, 0 (k p n ) * Cos (γ Obj, 0 (k p n )) (12) ν y, Obj, 0 (k p n ) = ν Obj, 0 (k p n ) * Sin (γ Obj, 0 (k p n )) (13)

The prediction of the object speed and the course angle of the environment object 104 takes place as a function of the recognized driving maneuver and is described below for the different driving maneuvers. Longitudinal and transverse movement are again considered separately.

With regard to the longitudinal movement, a constant speed ride and an emergency braking of the surrounding object 104 distinguished. For a uniform motion at a constant speed: ν Obj, 0 (k p n ) = ν Obj, 0 (k) (14)

In emergency braking of the environment object 104 It is assumed that the speed of the environment object 104 with a given acceleration α _{Obj, Ref} (k) <0 reduces until the environment object 104 brought to a standstill. Therefore applies here:

The stopping distance during the emergency braking maneuver of the environment object 104 is according to this calculation

With regard to the transverse movement of the environment object 104 A distinction is made between straight-ahead driving, cornering and lane change. When driving straight ahead, the course angle does not change, ie: γ Obj, 0 (k p n ) = γ Obj, 0 (k) (16)

If an environment object 104 when passing through a curve, it quickly gets out of the detection range of the environment sensor 102 out. Therefore, it is not necessary to determine the parameters of the curve, such as radius and angle section. For prediction, therefore, the following simple approach is used: γ Obj, 0 (k p n ) = γ Obj, 0 (k) + p · n · Δt p (17)

Of the Parameter p has an arbitrarily chosen initial value, the by adapted to be described correction algorithm can.

beschrieben wird. Mit b = d/B und c = 1/B·exp(d·C) kann dies auch geschrieben werden als:

When changing lanes, it is assumed that the trajectory of the environment object 104 within the characteristic axbox 401 by a sigmoid function of the shape

is described. With b = d / B and c = 1 / B * exp (d * C) this can also be written as:

The function, which is schematically in 5 is shown, has a turning point at x _c = C. The function value at the turning point is B / 2. The curve is also point-symmetrical about the point of inflection (c, B / 2) and the following applies: y Obj, C (x Obj, C ) → 0 for x Obj, C → -∞ such as y Obj, C (x Obj, C ) → B for x Obj, C → + ∞

so dass für den Parameter B_left der Sigmoide gilt:

B is thus the lateral transverse offset of the environment object 104 in the maneuver, which is also referred to here as maneuver width, and in a lane change of the lane width D _Lane corresponds to a lane. In a lane change to the left, ie, in the direction of the positive y-axis, is therefore

so that for the parameter B _{left of} the sigmoid applies:

Furthermore, a duration T _change and a start time t _{detect are} assumed for the lane change, wherein a typical value is specified for the duration T _change , which, however, can likewise be adapted by the correction algorithm to be described later. Since the total lateral offset is achieved only in infinite duration of the maneuver and the sigmoid not exactly by the origin of the characteristic axbox 401 a small lateral tolerance y _{Tol is} introduced so that: y Obj, C (x Obj, C (t detect )) = y Tol (22) y Obj, C (x Obj, C (t detect - T Change ) = D Lane - y Tol (23)

Due to the construction of the characteristic axbox 401 is x _{Obj, C} (t _detect ) = 0. Thus, using equation (19),

Furthermore, taking into account that the amounts of the object speed in the absolute axbox 301 and in the characteristic axbox 401 are equal and neglecting the speed of the environment object 104 in the y _C direction: x Obj, C (t detect + T Change ) - x Obj, C (t detect ) + ν Obj, 0 (t detect ) * T Change = ν Obj, 0 (t detect ) * T Change (25)

Using equation (20), this yields the parameter d of the sigmoid function:

Using the equations (18), (21) and (22), the sigmoid function, which describes a lane change to the left, is completely defined within the characteristic axbox:

For a lane change to the right, ie in the direction of the negative y _C axis, the following applies accordingly:

The parameters are given here by: b right = -B left c right = -C left d right = d left

The course angle of the environment object 104 in the characteristic axbox 401 is given by:

The course angle to be calculated in the time step kp / n in the absolute axbox is given by: γ Obj, 0 (k p n ) = γ Obj, 0 (k) + γ Obj, C (x Obj, C (k p n )), (30)

Again, neglecting the lateral velocity of the surrounding object 104 in the characteristic axbox 401 approximately x Obj, C (k p n ) = ν Obj, 0 (K) * n * .DELTA.t p be set so that: γ Obj, 0 (k p n ) = γ Obj, 0 (k) + γ Obj, C (ν Obj, 0 (K) * n * .DELTA.t p ) (31)

In summary, the prediction algorithm is now based on 6 described. In the maneuvering device 203 becomes that of the environment object 104 executed driving maneuvers with regard to the longitudinal and lateral dynamics of the environment object 104 identified. In terms of lateral dynamics, in the block 601 In this case, a distinction is made between straight-ahead driving, lane change and cornering, and with regard to the longitudinal dynamics occurring in the block 602 is investigated between a movement with constant speed and an emergency braking maneuver. Furthermore, if necessary, it is also determined that the environment object 104 not moved (block 603 ). Depending on the identified maneuver is then in the prediction 205 the movement of the environment object 104 predicted. With regard to the lateral dynamics, this is done in the block 604 in which the course angle of the environment object 104 for the prediction steps kp / l, ..., kp / N. These are in particular the blocks 605 . 606 and 607 provided that the course angles are determined on the basis of a model according to the identified lateral dynamic maneuver. If a straight-ahead drive has been identified, the calculation is done in the block 605 in case of a lane change in the block 606 and in the case of a curve in the block 607 , With regard to the longitudinal dynamics, the prediction of the object movement takes place in the block 608 , also based on a model selected according to the identified longitudinal dynamic maneuver. In particular, the velocities of the environment object become 104 in the prediction steps kp / l, ..., kp / N, which in the case of constant velocity motion in the block 609 happens and in case of emergency braking in the block 610 , Was found that the environment object 104 not moving, the prediction of the "motion" in the block is done 611 ,

The course angles {γ Obj, 0 (k p i )} , i = 1, ..., N, in the block 604 be calculated as well as those within the block 608 calculated speeds {ν Obj, 0 (k p i )} , i = 1, ..., N, become the block 612 to hand over. In this block, the longitudinal and lateral velocities are calculated from equations (12) and (13) {ν x, Obj, 0 (k p i )} and {ν y, Obj, 0 (k p i )} , i = 1, ..., N, of the environment object 104 certainly. Then in the block 613 successively using equations (10) and (11) the positions (x Obj, 0 (k p l ), y Obj, 0 (k p l )) to (x Obj, 0 (k p N ), y Obj, 0 (k p N )) of the environment object 104 in the prediction steps kp / l, ..., kp / N and thus the trajectory of the environment object 104 determined. In the case of a standing environment object 104 applies (x Obj, 0 (k p l ) = x Obj, 0 (k p N ) = x Obj, 0 (k) (28) y Obj, 0 (k p l ) = ... = y Obj, 0 (k p N ) = y Obj, 0 (k) (29)

Based on the procedure described above, a prediction of the object trajectory is possible. However, it has been shown that the results, in particular inaccuracies due to the measurement errors of the environmental sensor 102 can be improved by the fact that the prediction in each sampling step k based on the detected information about the environment object 104 is adjusted. The adaptation preferably takes place according to the principle of fault diagnosis and troubleshooting. First, the predicted position of the previous prediction step is compared with the measured position in the current sampling step. If a significant deviation is detected, the driving maneuver of the environment object is determined 104 not correctly identified or ended the identified maneuver. In this case, a reset and the identification of the current driving maneuver of the environment object takes place 104 starts again regardless of the previously saved maneuver data. Until the correct maneuver has been identified, the orbit becomes the surrounding object 104 predicated on Euler's extrapolation. If the deviation between the predicted position and the measured position is small, then the prediction is adjusted. In particular, here takes place when cornering of the environment object 104 an adaptation of the curve parameter p in equation (17). When changing lanes of the maneuver, the parameters D _Lane and T _{Change of} the sigmoid function are adjusted.

According to the principle of model-based prediction described above, the movement trajectory of the vehicle also becomes 101 determined yourself. A means for performing the prediction is in 7 schematically illustrated by a block diagram. The state of motion of the motor vehicle 101 is using the vehicle sensors 701 which includes, for example, wheel speed sensors, a longitudinal and lateral acceleration sensor, a yaw rate sensor, and a steering angle sensor. The sensor signals are in the block 702 evaluated. Further, motion quantities inaccessible to a direct measurement may be in the block 702 can be estimated from the sensor signals. The determined movement quantities of the vehicle 101 become a prediction unit 703 supplied, whose basic structure in 7 is also shown. It contains a driving maneuvering device 704 in which, based on the quantities of movement, driving maneuvers of the vehicle 101 be identified. The driving maneuver recognition device 704 is signaled with the block 705 in which, depending on the identified driving maneuver, one of N models predicts the trajectory of the vehicle 101 is selected. According to the selected vehicle model is determined by the block 705 one of the blocks 706 ₁ , ..., 706 _N the prediction device 708 activated, in which then the predicated path r → veh (T) the reference point of the vehicle 101 is calculated, which is also referred to below as a train or trajectory vehicle 101 referred to as. In the block 707 become the predicted positions vehicle 101 with the help of vehicle sensors 701 determined positions of the vehicle 101 compared. If the deviations are very large, it can be assumed that the driving maneuver has not been correctly identified. In this case, a signal is sent to the driving maneuver recognition device 704 sent, which leads to a reset. After the reset, a new identification of the current maneuver without access to the previously stored data on the maneuver. For small deviations between the predicted and the measured position parameters of the prediction used model adapted to improve the prediction. The parts 702 - 708 of the system are also formed as software components of a computer program stored in a microprocessor unit of the vehicle 101 is performed.

As with the prediction of the object trajectory longitudinal and lateral dynamics of the vehicle 101 in terms of maneuver identification considered separately. With respect to the longitudinal dynamics, a distinction is made, for example based on the signals of the wheel speed sensors, the signals of a longitudinal acceleration sensor and / or based on information about the operating state of the accelerator pedal and the vehicle brake between an acceleration maneuver, a uniform movement and a braking maneuver. With regard to the lateral dynamics, a distinction is made between straight-ahead driving, cornering and lane change.

nicht überschreiten, d.h. wenn gilt:

A straight ahead is detected when the yaw rate ψ. or the steering angle δ _H magnitude thresholds k _{dψ / dt} or

do not exceed, ie if:

gilt, wobei

und k_ψ vorgegebene Schwellenwerte sind und ψ_Start den Gierwinkel des Fahrzeugs 101 zu Beginn des Fahrmanövers bezeichnet, der in einem Speicher hinterlegt wird, wenn das Manöver anhand der ersten Bedingung in (31) erkannt worden ist. Ein Spurwechsel wird erkannt, wenn weder eine Geradeausfahrt noch eine Kurvenfahrt festgestellt worden ist. Um die Robustheit der Erkennung zu verbessern, werden Spurwechsel zudem in drei Phasen unterteilt. In Phase 1 (SW1) wird das Fahrzeug 101 ausgehend von der momentanen Spur in Richtung der gewünschten Spur gelenkt. Während einer Zwischenphase (SWZ) bewegt sich das Fahrzeug 101 im Wesentlichen geradlinig. Die Zwischenphase ist in der Regel wesentlich kürzer als die übrigen Phasen und geht über in die Phase 2 (SW2) des Spurwechsels. In der Phase 2 wird das Fahrzeug 101 wieder parallel zu der Ausgangsrichtung vor Beginn des Spurwechsels gedreht. Der Gierwinkel des Fahrzeugs 101 vergrößert sich betragsmäßig in Phase 1, bleibt in der Zwischenphase im Wesentlichen konstant und geht in Phase 2 wieder auf den ursprünglichen Wert zurück. Die zeitlichen Verläufe des Gierwinkels ψ und des Lenkwinkels δ_H sowie die Zuordnung zu den Phasen des Spurwechsels sind in 8 für einen beispielhaft untersuchten Spurwechsel dargestellt.With regard to cornering and lane change, it has been found that an evaluation of the product from the steering angle at the steerable wheels of the vehicle 101 and the yaw rate of the vehicle 101 to allow a distinction between these driving maneuvers already in their initial phase. It has been found that cornering can be detected when

applies, where

and k _{ψ are} predetermined threshold values and ψ _start the yaw angle of the vehicle 101 at the beginning of the driving maneuver, which is stored in a memory when the maneuver has been recognized by the first condition in (31). A lane change is detected if neither straight-ahead driving nor cornering has been detected. To improve the robustness of the recognition, lane changes are also divided into three phases. In phase 1 (SW1) the vehicle becomes 101 starting from the current lane in the direction of the desired track. During an intermediate phase (SWZ) the vehicle moves 101 essentially straightforward. The intermediate phase is usually much shorter than the other phases and goes into phase 2 (SW2) of the lane change. In phase 2, the vehicle is 101 again rotated parallel to the output direction before the start of the lane change. The yaw angle of the vehicle 101 increases in magnitude in phase 1, remains essentially constant in the intermediate phase and returns to its original value in phase 2. The time profiles of the yaw angle ψ and the steering angle δ _H and the assignment to the phases of the lane change are in 8th shown for an exemplary examined lane change.

Phase 1 of the lane change is recognized when no straight-ahead driving or cornering has been detected on the basis of conditions (30) and (31). After the detection of the lane change, the rate of change of the yaw rate, that is, the yaw rate of the vehicle 101 , as well as the steering angle monitors. If the yaw rate and the steering angle fall below predetermined threshold values, a transition to the intermediate phase is detected. After a sign change of the yaw rate and after the yaw rate and steering angle have again exceeded the threshold values, a transition from the intermediate phase to phase 2 is detected. Since at the beginning of the maneuver often no clear distinction between a cornering and a lane change can be made, additional demolition conditions are given in terms of the duration of the individual phases of the lane change. If these demolition conditions are met, cornering or straight ahead driving is recognized instead of the lane change. A first termination condition is fulfilled if the duration of phase 1 is less than a threshold value T _{SW1, min} . t - t SW1, Start <T SW1, min (32) t _{SW1, start} is the time at which the beginning of lane change has been recognized, and the size t - t _{SW1, start} corresponds to the time period of the phase 1 of the lane change at the observation time t. The second termination condition is met if the duration of phase 1 of the lane change is greater than a threshold T _{SW1, max} : t - t SW1, Start <T SW1, max (33)

A third termination condition is met if the duration of the intermediate phase of the lane change is greater than a threshold T _{SWZ, max} t - t SWZ, Start <T SWZ, max (34) t _{SWZ, start} is the time in which the beginning of the intermediate phase of the lane change has been recognized, and the size t - t _{SWZ, start} corresponds to the duration of the intermediate phase of the lane change at the observation time t.

G:

In Block 901 wurde eine Geradeausfahrt erkannt.

G -:

In Block 901 wurde keine Geradeausfahrt erkannt.

K:

In Block 902 wurde eine Kurvenfahrt erkannt.

K -:

In Block 902 wurde keine Kurvenfahrt erkannt.

A_1K:

Die Abbruchbedingung (32) ist erfüllt.

A -_1K:

Die Abbruchbedingung (32) ist nicht erfüllt.

A_1L:

Die Abbruchbedingung (33) ist erfüllt.

A_ZL:

Die Abbruchbedingung (34) ist erfüllt.

As in 9 on the basis of a schematic block diagram of the driving maneuver recognition device 704 is illustrated, is used to identify the driving maneuver of the vehicle 101 in the block 901 or the block 902 Determines whether the conditions for straight ahead driving (G) or cornering (K) are fulfilled. In the block 903 the driving maneuver is determined using these results and using the demolition conditions. This is done on the basis of a state diagram, which is shown in 10 is shown. In the diagram, the transition conditions between the detection of a straight-ahead driving (GAF), a cornering (KUF) and the three phases of a lane change (SW1, SWZ, SW2) are shown. The transition conditions are indicated in the figure in the following manner:

G:

In block 901 a straight-ahead drive was detected.

G -:

In block 901 no straight-ahead driving was detected.

K:

In block 902 a cornering was detected.

K -:

In block 902 no cornering was detected.

A _1K :

The termination condition (32) is fulfilled.

A - _1K :

The termination condition (32) is not fulfilled.

A _1L :

The termination condition (33) is fulfilled.

A _ZL :

The termination condition (34) is fulfilled.

If it is not indicated by a v-sign in the figure, must be individual Conditions for a state transition cumulatively fulfilled be. The star at the transition from the intermediate phase to phase 2 of the lane change means that the sign of the yaw rate opposite to the sign of Yaw rate at the transition must be from a straight exit in phase 1 of the lane change.

In contrast to the prediction of the object trajectory are used in the prediction of the trajectory of the vehicle 101 the x and y coordinates of the reference point M of the vehicle 101 calculated directly. The prediction is initially based on the vehicle coordinate system 302 , Then the prediction becomes the absolute coordinate system 301 transformed. If a straight ahead of the vehicle 101 has been detected, so is a vanishing lateral speed of the vehicle 101 accepted, so that y V (t + T p ) = 0 (35) where T _{p is} the prediction time, which in turn is a few seconds. For the x-coordinate of the reference point M of the vehicle 101 applies:

Preferably, the longitudinal acceleration α _{x of} the vehicle measured, for example, with a longitudinal acceleration sensor or determined from the signals of the wheel speed sensors 101 low pass-filtered if acceleration or deceleration has been detected. If this is not the case, the longitudinal acceleration signal is preferably strongly low-pass filtered. This ensures that the noise of the acceleration signal is suppressed when driving at almost constant speed, with speed changes, however, the entire dynamics of the signal is taken into account. The longitudinal speed v _{x, V} (t) of the vehicle 101 in the vehicle axle cross 302 at the time of observation t corresponds to the longitudinal velocity of the vehicle in the absolute axbox 301 because it is at the vehicle cross 302 also a stationary axbox acts.

For the first phase of a lane change and for the intermediate phases, the prediction of the position takes place on the basis of an Eulerian prediction of 2nd order, so that the following applies:

In addition, the trajectory of the vehicle 101 recorded during the first phase of lane change and approximated by a third degree polynomial. The polynomial is specified in a lane change coordinate system, which is analogous to the characteristic axbox 401 for the environment object 104 is defined. It is thus a stationary axbox whose origin is the position of the reference point of the vehicle 101 at the beginning of the lane change, and that for the duration of the lane change within the absolute axbox 301 is fixed. Sizes that refer to the lane change axis are hereafter marked with an index "lane". The polynomial thus has the form y track (t) = c 0 + c 1 .x track (t) + c 2 .x 2 track (t) + c three .x three track (t) (38)

The parameters c _i , i = 1,..., 3 are determined by a parameter estimation method, such as the recursive DSFI (Discrete Square Root Filter in the Information Form) method. In order to fully consider the first phase of the lane change, the "forgetting factor" of the method is set to λ = 1, so that all measured values are weighted equally.

mit XHH(t + Tp) = xSpur,End1 – (xSpur(t + Tp) – xSpur,EndZ) (41) It is assumed that the second phase of the lane change is mirrored to the first phase. Therefore, the second phase of the lane change can be predicted by the polynomial (38) rotated 180 ° at the end point of the first phase and applied to the end point of the intermediate phase. It thus applies: y track (t + T p ) = y Track Endz + y Track End1 - [c 0 + c 1 .x H (t + T p ) + c 2 .x 2 H (t + T p ) + c three .x three H (t + T p )] (39) where the point (x _{lane, End1} , y _{lane, End1} ) is the position of the reference point of the vehicle 101 at the completion of the first phases of the lane change and the point (x _{lane, EndZ} , y _{lane, EndZ} ) is the position of the reference point of the vehicle 101 indicates the completion of the intermediate phase of the lane change. The size x _H is given by

With X HH (t + T p ) = x Track End1 - (x track (t + T p ) - x Track Endz ) (41)

The x-coordinate x _track (t + T _p ) is calculated using equation (37a), wherein the result calculated from the equation is transformed into the lane change coordinate system.

For the prediction of the vehicle trajectory in a recognized cornering of the vehicle 101 the length of the driven circular arc and its radius are considered. It has been shown that the circular arc can be approximated as a straight line to determine its length. For the length of the circular arc therefore applies here:

Neglecting the slip angle applies to a ride of the vehicle 101 at a constant speed on a circle of radius R, based on a single-track model of the vehicle 101 :

wie sich aus der Skizze in 11 erkennen lässt. Der Radius und die Kurvenlänge werden anhand der Gleichungen (42) und (43) ermittelt.The speed of the vehicle 101 can be determined, for example, in turn based on the signals of the wheel speed sensors and the lateral acceleration α _{y, V} can be determined for example by means of a lateral acceleration sensor. The predicted coordinates result when cornering

as can be seen from the sketch in 11 lets recognize. The radius and the curve length are determined by equations (42) and (43).

Based on the predicted movement trajectory of the environment of the motor vehicle 101 captured environment object 104 and the movement trajectory of the motor vehicle 101 can now be determined whether a collision between the motor vehicle 101 and an environment object 104 threatening. For this purpose, the predicted trajectory of the environment object 104 with regard to the trajectory of the reference point of the vehicle 101 rated. A collision course is assumed when the trajectories are approaching to a distance equal to the width of the vehicle 101 and the environment object 104 considered. Furthermore, the so-called collision time (TTC, Time To Collision) is determined, ie the time until the collision with the surrounding object that has been determined 104 , If a collision course exists and the collision time falls below a certain value, in one embodiment, a driver warning is initially triggered, which may be, for example, a visual, audible and / or haptic warning. On the basis of the warning, the driver is made aware of the dangerous situation, so that he can initiate timely measures for collision avoidance or collision consequence reduction. In particular, if the collision time falls short of another smaller value, in one embodiment, a trigger signal is transmitted to a safety means control system that controls passive safety means, such as airbags or belt tensioners. By appropriate control of the passive safety means, the vehicle 101 preconditioned for a collision, so that the consequences of the collision are reduced for the vehicle occupants. In a further embodiment, it is checked in a decision direction as a function of the present driving situation whether the collision can be prevented by a braking maneuver or an evasive maneuver is to be carried out. If it is determined that the collision can be prevented by a braking maneuver, a distance to the collision location is determined in which the braking maneuver must be started. Reach the vehicle 101 the determined distance is automatically built up a brake pressure in the wheel brakes to perform the maneuver. If it is determined that the collision can be prevented by an evasive maneuver, a trigger signal is transmitted to a web-setting device. The trigger signal leads in this embodiment to the fact that initially within the path setting device an avoidance path y (x) is calculated to the collision with the environment object 104 to prevent. Then, in dependence on the determined avoidance path, the starting point at which the avoidance maneuver must be started is determined for the environment object 104 just to dodge. The aforementioned steps are preferably repeated in time steps until there is no danger of collision due to course changes of the environment object 104 or the vehicle 101 exists more or until the vehicle 101 reached the starting point for an evasive maneuver. If so, the egress path or parameters representing that lane are communicated to a steering actuator controller. This then controls by means of a control method in a preferred embodiment, a steering actuator such that at the steerable wheels of the motor vehicle 101 Steering angle can be adjusted, which is the motor vehicle 101 follow the avoidance path. In a further embodiment, by means of the steering actuator in accordance with the calculated path, a torque is impressed on a steering handle operated by the driver, whereby the driver receives a steering recommendation for an evasive maneuver.

In addition, the prediction of the trajectories of the environment object 104 and the vehicle 101 within an ACC system (ACC: Adpative Cruise Control). This system regulates the vehicle speed so that a certain distance to a preceding vehicle is not undershot. Based on the trajectory prediction for the vehicle and the surrounding object 104 and your own vehicle 101 For example, you can also change the lane of the environment object 104 and / or the vehicle 101 taken into account in the ACC regime, with the environment object 104 is about another vehicle. Moves such a vehicle, for example, due to a lane change in the lane of the motor vehicle 101 , its speed can be adjusted early.

For the applications described above, a sufficiently accurate prediction of the object trajectory is required. In order to demonstrate the improved predictive quality that can be achieved using the prediction method presented above, several driving situations have been simulated. Since the results of prediction only at great expense with a sufficiently accurately measured position of an environment object 104 a different approach was used. First, some test drives were with a motor vehicle 101 carried out. During these test drives were the positions of the vehicle 101 recorded in relation to the starting point; the position of starting was determined by the driver. Yaw rate and speed were also measured. In an offline simulation, the recorded data sets were used as input variables for a sensor simulation model whose output variables were the format of the environment sensor data 102 to have. The simulated sensor data were then used as inputs to a prediction algorithm, using firstly the prediction algorithm presented above and secondly Euler's prediction (equation (1)). Then the data measured during the test runs was compared with the predicted data for both prediction algorithms.

• 1) 0 < Δ < 0,5 m: Sehr gute Prädiktion, mit der Kollisionen zuverlässig vorhergesagt werden können.
• 2) 0, 5 m < Δ < 1 m: Noch gute Prädiktion, insbesondere für große Prädiktionszeiten. Es ist möglich, Kollisionen mit reduzierter Sicherheit vorherzusagen.
• 3) 1 m < Δ < 2 m: Die Abweichung entspricht ungefähr der Breite eines PKW bzw. der halben Länge eines PKW und der halben Breite einer Fahrspur einer Autobahn. Es besteht nur noch eine begrenzte Möglichkeit, um Kollisionen vorherzusagen.
• 4) 2 m < Δ < 4 m: Die Abweichung entspricht in etwa der Länge eines PKW und der Breite einer Fahrspur einer Autobahn. Eine Kollisionsvorhersage ist nicht hinreichend genau möglich, um Maßnahmen zur Kollisionsvermeidung einzuleiten; wird eine Kollision vorhergesagt, kann jedoch eine Fahrerwarnung erfolgen.
• 5) 4 m < Δ < 8 m: Die Abweichung entspricht ungefähr der doppelten Länge eines PKW oder der doppelten Breite einer Autobahnspur. Die Prädiktion ist hier nur sinnvoll, um eine Vorstellung von der Situation zu erhalten. Maß nahmen zur Kollisionsvermeidung können jedoch nicht sinnvoll eingeleitet werden.
• 6) Δ > 8 m: Die Abweichungen sind so groß, dass die Prädiktionsergebnisse nicht sinnvoll genutzt werden können.

The results of the comparison between the data are for the prediction algorithm according to the invention in the 12a shown. The error classes of the histogram correspond to intervals for the deviation Δ between the predicted position and the measured position. The classes were chosen according to the following qualitative classification:

• 1) 0 <Δ <0.5 m: Very good prediction, with which collisions can be reliably predicted.
• 2) 0, 5 m <Δ <1 m: still good prediction, especially for large prediction times. It is possible to predict collisions with reduced security.
• 3) 1 m <Δ <2 m: The deviation corresponds approximately to the width of a car or half the length of a car and half the width of a lane of a highway. There is only a limited possibility to predict collisions.
• 4) 2 m <Δ <4 m: The deviation corresponds approximately to the length of a car and the width of a lane of a motorway. A collision prediction is not possible with sufficient accuracy to initiate collision avoidance measures; if a collision is predicted, however, a driver warning can occur.
• 5) 4 m <Δ <8 m: The deviation is approximately twice the length of a car or twice the width of a motorway lane. The prediction is only useful here to get an idea of the situation. However, measures to avoid collisions can not be usefully initiated.
• 6) Δ> 8 m: The deviations are so great that the prediction results can not be meaningfully used.

12a shows in particular the results of the comparison in a lane change of the environment object 104 in the environment object 104 and vehicle 101 each move at a speed of 18 m / s. The environment object 104 performs a lane change with a lateral offset of 3.5 m. Since the velocities during the maneuver are nearly constant, only the prediction error of the y-component becomes the trajectory of the surrounding object 104 rated. The prediction interval was 3 seconds. As can be seen from the figure, more than 40% of the prediction results in the error class 0 <Δ <0.5 m, which has the greatest benefit for a collision avoidance system. Furthermore, the deviation Δ is less than 1 m for more than 55% of the prediction results.

12b shows the deviations between the prediction results and the measured positions for the Euler extrapolation. Here, only about 21% of the deviations are smaller than 0.5 m and less than half of the deviations are smaller than 1 m. From a comparison of 12a and 12b the improvement achieved with the invention will become apparent.

Claims (39)

Method for predicting a movement trajectory of an object moving on the road, comprising the following steps: - determining at least one movement variable of the object ( 101 ; 104 ), - determining a driving maneuver that is caused by the object ( 101 ; 104 ) is executed, based on an evaluation of the determined amount of movement, - selecting a model of the movement of the object ( 101 ; 104 ) depending on the determined driving maneuver and - calculating the movement trajectory of the object ( 101 ; 104 ) based on the selected model. Method according to claim 1, characterized in that the driving maneuver of the object ( 101 ; 104 ) is selected from a set of predetermined driving maneuvers. A method according to claim 1 or 2, characterized in that the determined driving maneuver from a the longitudinal movement of the object ( 101 ; 104 ) descriptive driving maneuvers and a transverse race the object ( 101 ; 104 ) describing driving maneuvers. Method according to one of the preceding claims, characterized in that the transverse movement of the object ( 101 ; 104 ) descriptive driving maneuver is selected from an amount that includes at least one straight ahead, a cornering and a lane change. Method according to one of the preceding claims, characterized in that a distinction between a cornering and a lane change is made on the basis of a characteristic amount of movement, which consists of a first amount of movement which changes the direction of movement of the object ( 101 ; 104 ) and at least a second amount of movement of the object ( 101 ; 104 ) is formed. Method according to claim 5, characterized in that, in the case of the second movement variable of the object ( 101 ; 104 ) by a speed of the object ( 101 ; 104 ). Method according to claim 5, characterized in that the second amount of movement is a steering angle of steerable wheels of the object ( 101 ; 104 ). Method according to one of the preceding claims, characterized characterized in that the movement trajectory based on the model dependent on is determined by at least one parameter. Method according to one of the preceding claims, characterized in that a position predicted on the basis of the movement trajectory of the object ( 101 ; 104 ) with a detected position of the object ( 101 ; 104 ) and that the parameter is adjusted depending on the result of the comparison. Method according to one of the preceding claims, characterized in that the object ( 101 ; 104 ) around a motor vehicle ( 101 ) environment object ( 104 ), where the motion quantity of the surrounding object ( 104 ) by means of an environment sensor ( 102 ) of the motor vehicle ( 101 ) is determined. Method according to one of the preceding claims, characterized in that the longitudinal movement of the environment object ( 104 ) describing driving maneuvers as a function of a comparison of speeds of the object present in the successive times ( 104 ) is determined from an amount comprising at least one uniform movement and emergency braking. Method according to one of the preceding claims, characterized in that a movement direction of the environment object ( 104 ) is determined at successive points in time, and that the transverse movement of the surrounding object ( 104 ) descriptive driving maneuver is determined on the basis of a comparison of determined from the determined Be wegungsrichtungen course angles. Method according to one of the preceding claims, characterized in that the transverse movement of the environment object ( 104 ) describing driving maneuvers based on a comparison of the size (γ Obj, 0 (t) - γ Obj, 0 (t 0 )) · Ν Obj, 0 with at least one threshold value, where γ _{Obj, 0} (t) is the heading angle in a stationary axbox at an observation time, γ _{Obj, 0} (t ₀ ) is the heading angle in the stationary axbox at the beginning of a change in heading angle and ν _{Obj, 0} the speed of the environment object ( 104 ) designated. Method according to one of the preceding claims, characterized in that a lane change of the environment object ( 104 ) is determined when the course angle of the environment object ( 104 ) changes to a predetermined size and if the size (γ Obj, 0 (t) - γ Obj, 0 (t 0 )) · Ν Obj, 0 Takes values within a given interval. Method according to one of the preceding claims, characterized in that a cornering of the environment object ( 104 ) is determined when the course angle of the environment object ( 104 ) in one given dimensions changes and if the size (γ Obj, 0 (t) - γ Obj, 0 (t 0 )) · Ν Obj, 0 Takes values outside the specified interval. Method according to one of the preceding claims, characterized in that the movement trajectory of the environment object ( 104 ) based on a predicted velocity of the environment object ( 104 ) and a predicted course angle of the environment object ( 104 ) is determined. Method according to one of the preceding claims, characterized in that the movement trajectory of the environment object ( 104 ) is calculated recursively, whereby a position of the environment object ( 104 ) is determined in a prediction step on the basis of the predicted speed and the predicted course angle from the position determined in a preceding prediction step. Method according to one of the preceding claims, characterized in that the predicated velocity of the environment object ( 104 ) based on the selected the longitudinal movement of the environment object ( 104 ) descriptive model is calculated. Method according to one of the preceding claims, characterized in that the predicted course angle of the environment object ( 104 ) based on the selected the transverse movement of the environment object ( 104 ) descriptive model is calculated. Method according to one of the preceding claims, characterized in that the predicted course angle of the environment object ( 104 ) is calculated according to a sigmoid function if a driving maneuver has been determined to be a lane change, wherein the shape of the sigmoid function is determined by at least one parameter which depends on the speed of the environment object ( 104 ) is determined. Method according to one of the preceding claims, characterized in that the predicted for the time t + .DELTA.t course angle γ _{Obj, 0} (t + .DELTA.t) of the environment object ( 104 ) by γ Obj, 0 (t + Δt) = γ obj, 0 (t) + p · Δt is calculated when cornering has been determined as a driving maneuver, wherein the variable γ _{Obj, 0} (t) is the course angle of the environment object ( 104 ) is at an observation time t, Δt is a time interval and p is a parameter. Method according to one of the preceding claims, characterized in that the object ( 101 ; 104 ) around the motor vehicle ( 101 ), wherein the amount of movement is detected by means of at least one vehicle sensor. Method according to one of the preceding claims, characterized in that the transverse movement of the motor vehicle ( 101 ) descriptive driving maneuver is determined based on a yaw rate of the motor vehicle and / or on the basis of a steerable wheels of the vehicle steering angle. Method according to one of the preceding claims, characterized in that a straight exit (GAF) of the motor vehicle ( 101 ) is averaged when the yaw rate of the vehicle and / or the steering angle are smaller in magnitude than a threshold value. Method according to one of the preceding claims, characterized in that a cornering (KUF) of the motor vehicle is determined is when the product of steering angle and yaw rate greater than is a predetermined threshold and / or a yaw angle change while of the driving maneuver amount greater than is a predetermined threshold. Method according to one of the preceding claims, characterized in that a lane change of the motor vehicle ( 101 ) is determined when no straight-ahead driving (GAF) and no turning (KUF) are detected. Method according to one of the preceding claims, characterized in that a track is divided into at least three successive phases, wherein the yaw rate of the motor vehicle ( 101 ) is greater than a threshold value in a first phase (SW1), less than a threshold value in an intermediate phase (SWZ) and greater than a threshold value in a second phase (SW2), and the yaw rate in the first phase (SW1) and the second phase (SW2) has a different sign. Method according to one of the preceding claims, characterized in that for the first phase (SW1) a maximum duration is given ben PRE, and that after a previous detection of a lane change cornering (KUF) of the motor vehicle ( 101 ) is determined when the duration of the first phase exceeds the maximum duration. Method according to one of the preceding claims, characterized marked that for the first phase (SW1) is given a minimum duration, and that after a previous detection of a lane change a straight ahead (GAF) is determined when the duration of the first phase (SW1) the minimum duration falls short. Method according to one of the preceding claims, characterized marked that for the intermediate phase (SWZ) of the lane change predefined a maximum duration and that after a previous detection of a lane change a straight-ahead (GAF) is determined when the duration of the intermediate phase (SWZ) exceeds the maximum duration. Method according to one of the preceding claims, characterized in that the movement trajectory of the motor vehicle ( 101 ) for the first phase of a lane change as a function of determined longitudinal and transverse speeds and / or as a function of determined longitudinal and lateral accelerations of the motor vehicle ( 101 ) is predicated on the basis of Euler's prediction. Method according to one of the preceding claims, characterized in that the movement trajectory of the motor vehicle ( 101 ) is recorded during the first phase (SW1) of the lane change and is approximated by a polynomial, and in that the polynomial thereby determined is used for the prediction of the movement trajectory of the motor vehicle ( 101 ) is used for the second phase (SW2) of the lane change. Method according to Claim 32, characterized in that the predicted movement trajectory for the second phase (SW2) of the lane change corresponds to the determined polynomial which is rotated through 180 ° about an end point of the first phase of the lane change and which matches the movement trajectory of the motor vehicle ( 101 ) in the intermediate phase (SWZ) of the lane change. A method according to claim 32 or 33, characterized that the polynomial is a third degree polynomial. Method according to one of the preceding claims, characterized in that the movement trajectory of the motor vehicle ( 101 ) is determined on the basis of a predicated length of a circular arc and a radius of the circular arc, if the determined driving maneuver is a cornering (KUF). Method for preventing a collision between a motor vehicle and an environment object in the vicinity of the motor vehicle or for reducing collision sequences, in which movement trajectories of the motor vehicle and / or the surrounding field object are determined and a collision location is determined on the basis of an evaluation of the movement trajectories, and in dependence initiated by the distance from the collision location collision-preventing and / or collision-sequence-reducing measures, characterized in that the movement trajectory of the environment object ( 104 ) and / or the movement trajectory of the motor vehicle ( 101 ) are determined by a method according to any one of the preceding claims. Method for regulating the speed of a motor vehicle, in which the speed of the motor vehicle is set such that a distance between the motor vehicle and a further vehicle located in front of the motor vehicle in the direction of travel of the motor vehicle does not fall below a predetermined distance value, characterized in that a method according to one of claims 1 to 35 a movement trajectory of the motor vehicle ( 101 ) and / or the other vehicle ( 104 ) it is determined that, as a function of the movement trajectory, a distance between the motor vehicle ( 104 ) and the other vehicle ( 101 ) is predicated, and that the speed of the motor vehicle ( 101 ) is set depending on the predicted distance. Computer program product, characterized that it defines an algorithm that is a procedure after a of the preceding claims includes. Device for predicting a movement trajectory of an object moving in traffic, comprising - at least one sensor ( 102 ), from whose signals a motion quantity of the object ( 101 ; 104 ), - a driving maneuver recognition device ( 203 ; 704 ), with the aid of an evaluation of the amount of movement, a driving maneuver of the object ( 101 ; 104 ) is determinable and - a prediction device ( 205 ₁ ; ...; 205 _N ; 706 ₁ ; ...; 706 _N ), which uses a model to describe the movement trajectory of the object ( 101 ; 104 ), wherein the model is selectable from a set of models depending on the determined driving maneuver.