WO2016124368A1

WO2016124368A1 - Method for operating a vehicle, computer program, machine-readable storage medium and electronic control unit

Info

Publication number: WO2016124368A1
Application number: PCT/EP2016/050732
Authority: WO
Inventors: Thomas Lich; Thomas Schlender
Original assignee: Robert Bosch Gmbh
Priority date: 2015-02-02
Filing date: 2016-01-15
Publication date: 2016-08-11
Also published as: DE102015201733A1

Abstract

Method for operating a vehicle, comprising the following steps: a. detecting (601-603) an impending collision with a collision object, in particular with a VRU, in particular with a pedestrian (FG) and estimating an impact region (NS, CA, FS) of the collision object on the vehicle (FZG); b. determining (605) a lateral displacement (dy) of the vehicle (FZG) to avoid the collision in accordance with the estimated impact region (NS, CA, FS); c. bringing about (607a-c) the lateral displacement (dy) to avoid the collision.

Description

Description title

Method for operating a vehicle, computer program,

machine-readable storage medium, electronic control unit

The invention relates to a method for operating a vehicle to avoid a collision with a collision object, in particular with a pedestrian, a corresponding computer program, a corresponding

machine-readable storage medium on which the computer program is stored and a corresponding electronic control unit, thus a control device for a vehicle.

State of the art

In Germany, about 10% of all road accidents involve pedestrians. In more than 60% of these accidents involving pedestrians, a car is involved. In a not inconsiderable proportion of these accidents, it is not always possible to have the

Accident with a pure braking intervention, i. by delaying the vehicle, to prevent.

However, to avoid a collision in these cases, evasion, i. a lateral offset of the vehicle, be helpful.

One difficulty that evasive systems have to deal with is predicting pedestrian behavior in relation to its course of motion. Known prediction models have not yet considered the behavior of a pedestrian in the pre-crash phase of a real accident.

In order to avoid that a collision with a pedestrian occurs despite avoidance, because, for example, the pedestrian himself tries to avoid the direction selected by the system, the necessary lateral offset is increased by a safety impact to avoid a collision.

This leads to the conflict of objectives that although the pedestrian can be protected by the evasive maneuver, but other road users, such as oncoming traffic, neighboring vehicles, other pedestrians or other road users, u.U. be placed in danger.

If Dodge is to be used for collision avoidance, it is therefore the goal to choose the lateral offset as small as possible and at the same time as large as necessary.

Disclosure of the invention

Against this background, the present invention proposes a method for operating a vehicle with the steps:

a. Detecting an imminent collision of the vehicle with a

Collision object, especially with an increased risk

Road users (VRU), in particular with a pedestrian and

Estimating an impact area of the collision object on the vehicle; b. Determining a lateral offset of the vehicle to avoid the collision depending on the estimated impact area;

c. To induce lateral misalignment to avoid collision.

A method for operating a vehicle can be understood as methods that support the driver in coping with the driving task (driver assistance systems), partially assume the driving task (automated driving functions) or completely assume the driving task (fully automatic driving). In the present case, a collision object can be understood as any object with which a collision is possible. In particular, the present method is for avoiding collisions with increased risk

Road users, also called Vulnerable Road User (VRU). This means any road users who are in an inferior situation in the event of a collision with a vehicle. These include u. a. Two cyclists, pedestrians and especially children.

In the present case, a lateral offset can be understood to mean that the vehicle laterally deviates from its original trajectory.

In the present case, a collision area can be understood to mean a lateral area in the area of the vehicle front.

The invention is based on the recognition that a moving

Collision object, such as a VRU or pedestrian, in all directions within certain limits accelerate (see Fig. 1) could.

However, this "radius of action" is not exhausted, since - depending on the impact area - certain directions of acceleration of the collision object, such as a VRU or pedestrian, can not be selected or selected.

From the knowledge of the reaction behavior of the collision object, such as. A VRU or pedestrian, there is the advantage that the determination of the maximum necessary lateral offset to avoid a collision with a crossing pedestrian can be made more precisely or precisely.

Furthermore, based on the possible reaction pattern of the collision object, such as, for example, a VRU or pedestrian, a lateral offset to avoid the collision is determined.

Specifically, a maximum necessary lateral offset to avoid collision can be reduced if it is shown that it is very likely that the collision object, such as a VRU or pedestrian, will not move into this area.

This reduces the risk of endangering other road users, for example, oncoming traffic or other objects or people on the roadside udgl.

The method of the present invention is designed with special reference to the reaction to collisions with pedestrians. Since pedestrians have a probalistic reaction to impending collisions whose knowledge of this reaction behavior can advantageously be used to determine a lateral offset in order to avoid a collision with pedestrians.

Advantageous embodiments of the method emerge from the

Subclaims and the description below.

In an advantageous embodiment, the time taken for the estimated collision to arrive at the lateral offset is used to determine the lateral offset. By adding the time until the collision arrives, the lateral offset can be more accurately determined and can be smaller. This leads to a reduced risk of endangering other road users by the laterally offset vehicle.

In a further advantageous embodiment, at least one movement parameter of the collision object is used to determine the lateral offset. Under motion parameters can present u. a. to be understood:

a moving speed of the collision object;

a direction of movement of the collision object;

a probability of changes of one of the motion parameter of the collision object;

a degree of change of one of the motion parameter of

Collision object. These movement parameters (for example movement speed,

Movement direction) can be partially detected by suitable sensors (eg., Video and / or radar and / or lidar sensors) and can be partially predetermined as a parameter of the method (eg.

Probability of changes of one of the motion parameters).

The probabilities of changing a motion parameter and the extent of the change of a motion parameter can be determined, for example, from empirical investigations of real or trailing accident events.

By adding at least one motion parameter of the

For a potential collision object, the lateral offset necessary to avoid the collision can be determined more accurately and can be smaller. This leads to a reduced risk of endangering other road users by the laterally offset vehicle.

In a further advantageous embodiment of the method, a minimal lateral offset for avoiding the collision and / or a maximum lateral offset for avoiding the collision is determined in the step of determining and in the step of bringing about a lateral offset which is greater than or equal to the minimum lateral offset Offset is and / or is less than or equal to the maximum lateral offset.

The minimum lateral offset can be determined to parameters that result in a lateral offset that is as large as necessary to avoid collision. The maximum lateral offset can be determined to parameters that result in a lateral offset that is as small as possible to avoid collision.

Furthermore, the invention provides a decision criterion for a possible avoidance strategy in order to avoid an imminent collision with a collision object, such as, for example, a VRU or pedestrian. Taking into account the direction of movement of the collision object, such as a VRU or

Pedestrian, and the environment, an evasive strategy is set. For this purpose, in the step of recognizing an environment of the vehicle is detected and determined in the step of determining the lateral offset depending on the detected environment.

In the present context, a detected environment is to be understood as meaning suitable information that originates from a sensor system that is designed to detect an environment around the vehicle. For example. it is possible by using the recorded environment in the

Deciding whether to make a lateral shift to the left or to the right, to estimate in which of the possible directions a lower risk of endangering other road users. In a further advantageous embodiment, the method of the present invention can be carried out with a simultaneous deceleration of the vehicle, thus in combination with a braking intervention. In particular, in that before and / or after and / or during the step of

Inducing a step of decelerating takes place, wherein in the step of

Delaying the vehicle is delayed.

In a specific embodiment of this embodiment, the vehicle is maximally decelerated in the deceleration step. In a further advantageous embodiment of the method takes place before

Step of determining and / or before the step of bringing about a step of checking, wherein in the step of checking is checked whether the collision can be avoided only by decelerating the vehicle, in particular by maximum deceleration of the vehicle, and the steps of determining and / or only if the test is negative.

Since the effect of causing a lateral offset is associated with the risk of posing a hazard to other road users, it is advantageous if the method of the present invention is used only if a less risky method than hopeless or less effective is recognized. Thus, as a last resort yet a protection of the

Collision object, such as, for example, a VRU or pedestrian, can be achieved, which would otherwise have been unreachable by a method that would have provided only a braking intervention.

It is understood that the advantageous embodiments of the method of the present invention can be implemented both individually and in combination with each other. Another aspect of this invention relates to a computer program configured to carry out all steps of the method of the present invention.

Another aspect of this invention relates to a machine-readable one

Storage medium is stored on the above computer program.

A further aspect of this invention relates to an electronic control unit, that is to say a control unit, in particular for a vehicle, which is set up to carry out all steps of the method of the present invention.

Another aspect of this invention relates to a driver assistance system for a vehicle, wherein the driver assistance system is adapted to at least partially automatically cause a lateral offset of the vehicle comprising means for detecting an imminent collision and means for determining a lateral offset to avoid the collision and means for

To bring about the lateral offset to avoid the collision and which is adapted to perform all the steps of the method of the present invention.

The approach presented here will be described below with reference to the attached

Illustrated drawings by way of example. Show it:

Figure 1 shows a range of action of a walking pedestrian. FIG. 2 shows a definition of lateral collision areas; FIG.

3 shows an evaluation of the reactions of pedestrians as a function of an impact area;

4 shows an alternative definition of lateral collision areas;

5 shows an example for the determination of a lateral offset to avoid a collision;

Fig. 6 is a flowchart of the method of the present invention;

7 is a block diagram of an electronic control unit

In the following description of favorable embodiments of the present invention are for the in the various figures

represented and similar elements acting the same or similar

Reference numeral used, wherein a repeated description of these elements is omitted.

Fig. 1 shows acceleration possibilities of a walking pedestrian FG. Direction A shows the current direction of walking of the pedestrian FG. If the pedestrian recognizes an imminent collision, the pedestrian will, in order to avoid a collision, in the event that the pedestrian FG reacts, switch to one of the acceleration possibilities B1 to B4.

Fig. 2 shows a definition of lateral collision areas. Depending on the direction of travel A of a pedestrian FG in relation to a vehicle FZG, the vehicle front FF can be divided into three areas: Near Side NS, Center Area CA and Far Side FS. In the illustrated variant, the near side NS and the far side FS each occupy about 30% of the vehicle front FF. For the Center Area CA, the remaining approx. 40% remain. 3 shows the reaction of the pedestrians as a function of the impact area (Near Side NS, Center Area CA, Far Side FS). The results are based on an analysis of video-documented accidents involving pedestrian participation.

The reactions were evaluated:

the pedestrian accelerates;

the pedestrian delays or stops;

the pedestrian goes back;

the reaction of the pedestrian could not be assigned;

the pedestrian showed no reaction.

From the illustration of the evaluation in FIG. 3 it can be shown that

(1) pedestrians mostly unresponsive

(2) Do not accelerate or decelerate pedestrians to the center of the vehicle (Center Area, CA);

(3) When pedestrian acceleration occurs, it is almost exclusively away from the center (CA).

FIG. 4 shows an alternative division of the vehicle front into near-side NS, center-area CA and far-side FS areas. In this alternative allocation, the shares of the near-side NS and far-side FS areas are much larger than the center area area. In this alternative division, there is the advantage of responding to the pedestrian response, according to which the pedestrian, if he responds, accelerates away from the center of the vehicle.

A portion of the delaying FG in the near-side area and a portion of the accelerating FG in the far-side area would have been taken without FG reaction from the middle area of the car front.

Therefore, it may be useful for an algorithm that makes the decision on the direction of the deviation and the necessary lateral offset, the

Outside areas are enlarged.

The alternative division of the vehicle front, as shown in Fig. 4, takes this fact into account. 5 shows an example of the determination of a lateral offset to avoid a collision. A pedestrian moves at a speed of movement vFG = 2 m / s in a walking direction A. A vehicle moves with one

Driving speed vego = 12 m / s on its trajectory. At the time shown, the lateral distance between the vehicle and the

Pedestrian 0.5 m. If there is no change in the motion parameters, the present system estimates a time to collision on the near side NS of the vehicle FG of 0.42 s. During this time, the pedestrian would cover a distance of 0.83 m. From this, the method of the present invention may determine that a lateral offset dy = 0.33 m would be necessary to avoid the collision. Based on the finding that a pedestrian will most likely not react, regardless of the expected impact area, i. his movement parameters will not change, or that a pedestrian, who must expect an impact on the near side NS of the vehicle FG, will delay its movement speed or stop completely, can be determined by the method of the lateral displacement dy = 0.33 m , This lateral offset is lower than in previous avoidance strategy, so the conflict of objectives can be solved to avoid a collision by an evasive maneuver, without further

Endanger road users through the maneuver.

Fig. 6 shows a flowchart of the method of the present invention. In steps 601 to 603, an impending collision is detected by, pedestrian detection and classification, prediction of pedestrian movement 602, for example, assuming uniform pedestrian movement, and decision 603 as to whether a collision is imminent or pedestrian on a collision course located. If the decision is denied, that will

Detecting a collision 601 - 603 performed. We affirm the decision becomes dependent on at least one motion parameter 604 of the

Pedestrian a determination 605 of a collision area (Near Side NS, Center Area CA, Far Side FS) performed. If a collision with the near side is determined, then the size of the lateral offset is determined 606a. This is followed in step 607a by the examination of, for example, data about the surroundings around the vehicle FZG, as to whether a lateral offset is possible. Is the offset possible, a lateral offset in the direction of the far side FS is brought about 608a. If a collision with the center area is determined, then the size of the lateral offset is determined 606b. This is followed in step 607b by the examination of, for example, data about the surroundings around the vehicle FZG, as to whether a lateral offset is possible. If the offset is possible, then a lateral offset in

Direction away from the Center Area CA induced 608b. The direction can be predetermined. If a collision with the Far Side FS is determined, then the size of the lateral offset is determined 606c. This is followed in step 607c by the examination of, for example, data about the surroundings around the vehicle FZG, as to whether a lateral offset is possible. If the offset is possible, then and a lateral offset in the direction of the near side NS is brought about 608c.

In the illustrated embodiment, the direction of the lateral offset may be predetermined.

In the following table, for example, an evasion strategy is listed which, depending on the impact area (Near Side NS, Center Area CA, Far Side), includes deflection directions and exemplary (maximum or minimum) variables for the lateral offset.

In an alternative embodiment of the method of the present invention, the direction of the lateral offset may depend on determined environment information.

The method of the present invention has been described above with the assumption that right-hand traffic is involved, while the method may also apply the left-hand traffic. The evasion strategy depending on the pedestrian movement direction and the pedestrian response remains the same.

7 shows a block diagram of a driver assistance system 7 comprising an electronic control unit 70, that is to say a control device, according to the present invention.

Based on the probabilistic response behavior of the collision object described above, such as a VRU or pedestrian FG, the information available, e.g. stored in an E EP ROM 74, used by the lateral offset dy correspondingly

parameterize or determine, in order to bring about the lateral offset dy.

The above-described method 600 according to the present invention may be performed in an electronic control unit 70 provided therewith, that is, a control unit. Ideally, the method of the present invention runs

Invention in an existing controller 70 from.

Information about the vehicle _speed v _eg0 can thereby

Vehicle communication system 85, for example. Via a CAN bus, are transmitted to the controller 70. Movement parameters (v _FG , A, Bl - B4) of pedestrians FG can be determined via suitable environment sensor systems 83, such as, for example, video. Lidar- or radar or ultrasound systems are detected. It would also be conceivable for the movement parameters (v _FG , A, Bl - B4) to be transmitted by pedestrians FG by means of Car-2-X communication means. Of the

Environment sensors 83 or communication means can

Movement parameters (v _FG , A, Bl - B4) of the pedestrian FG are also transmitted via a vehicle communication system 85 to the control unit 70. The electronic control device 70 has suitable means 75 for recognizing from the transmitted data an imminent collision with a pedestrian FG, suitable means 75, depending on at least one

Movement parameters (v _FG , A, Bl - B4) of the pedestrian FG to determine a lateral displacement dy to avoid the collision and suitable means 91 to bring about a corresponding lateral displacement dy in the vehicle FZG. These means 91 for inducing may also take the form of Interfaces to suitable actuators of the vehicle FZG are present, d the interfaces receive suitable signals to cause the determined lateral displacement dy. Among other things, the actuators can be steering systems or brake systems of the vehicle FZG.

Claims

claims

A method (600) for operating a vehicle (FZG), comprising the steps of: a. Recognizing (601 - 603) an impending collision with a collision object, in particular with one with an increased vulnerable road user (VRU), in particular with a

Pedestrian (FG) and estimating an impact area (NS, CA, FS) of the collision object on the vehicle (FZG);

b. Determining (605) a lateral offset (dy) of the vehicle (FZG) to avoid the collision depending on the estimated impact area (NS, CA, FS);

c. Inducing (607a - c) the lateral displacement (dy) to

Avoiding the collision.

2. The method (600) according to claim 1, wherein in the step of

Detecting (601 - 603) a time to impact (TTC) is estimated and wherein the impact area (NS, CA, FS) is determined as a function of the estimated time (TTC) and / or in the step of

Determining (605) the lateral offset (dy) is determined as a function of the estimated time (TTC).

3. The method according to claim 1, wherein at least one motion parameter of the collision object, in particular a movement speed of the collision object and / or a movement direction of the collision object, are detected in step (601-603) of the recognition

Collision object and / or a probability of changes of one of the motion parameter of the collision object and / or an extent of the change of one of the motion parameters of the collision object

In the step of determining (605), the lateral offset (dy) is determined as a function of the at least one detected motion parameter. Method (600) according to one of the preceding claims, wherein in the step of determining (605) as a lateral offset (dy) a minimum lateral offset for avoiding the collision and / or a maximum lateral offset for avoiding the collision is determined and wherein in the step of Inducing (607a-c) a lateral offset (dy) greater than or equal to the minimum lateral offset and / or less than or equal to the maximum lateral offset.

Method (600) according to one of the preceding claims, wherein in the step of detecting (601-603) an environment of the vehicle (FZG) is detected and in the step of determining (605) the lateral offset (dy) depending on the detected environment, in particular depending on the region of the environment in which the lateral offset is to lead being free.

Method (600) according to one of the preceding claims, wherein before and / or after and / or during the step of

Inducing (607a-c) a step of decelerating, wherein in the step of decelerating the vehicle (FZG) is decelerated,

especially delayed maximum.

Method (600) according to one of the preceding claims, wherein before the step of determining (605) a step of checking takes place, wherein in the step of checking it is checked whether the collision is only by decelerating the vehicle (FZG), in particular maximum deceleration of the vehicle (FZG) can be avoided and the steps of determining (605) and / or causing (607a-c) to occur only if the test is negative.

Computer program, which is set up, all steps of the

Method (600) according to one of Claims 1 to 7.

9. A machine-readable storage medium on which the computer program according to claim 8 is stored.

An electronic control unit (70) arranged to perform all the steps of the method (600) of any one of claims 1 to 7.

11. Driver assistance system (7) for a vehicle (FZG), wherein the

Driver assistance system (7) is adapted to at least partially automatically cause a lateral offset (dy) of the vehicle (FZG) comprising means for detecting an imminent collision and means (75) for determining a lateral offset (dy) to avoid the collision and means (91) for effecting the lateral offset (dy) for avoiding the collision and which is arranged to carry out all the steps of the method (600) according to one of claims 1 to 7.