WO2011092229A1

WO2011092229A1 - Method for automatically decelerating a vehicle to prevent a collision or reduce the consequences of a collision

Info

Publication number: WO2011092229A1
Application number: PCT/EP2011/051103
Authority: WO
Inventors: Philipp Reinisch; Peter Zahn
Original assignee: Bayerische Motoren Werke Aktiengesellschaft
Priority date: 2010-01-29
Filing date: 2011-01-27
Publication date: 2011-08-04
Also published as: DE102010006215A1

Abstract

The invention relates to a method for automatically decelerating a vehicle to prevent a collision or reduce the consequences of a collision with a detected collision object, wherein at a determined intervention time a brake system of the vehicle is automatically actuated such that a collision with the detected collision object can be avoided or at least the consequences of a collision can be reduced, wherein the intervention time is established depending on the detection or assumption of an evasive or passing maneuver performed by the driver. The invention is characterized in that an evasive or passing maneuver is detected or assumed depending on a degree of the relative movement of the vehicle to the detected collision object in the longitudinal direction and a degree of the relative movement in the transverse direction and a comparison of the two degrees.

Description

Method for automatic braking of a vehicle for collision avoidance or collision following reduction

The invention relates to a method for the automatic braking of a vehicle for collision avoidance or Kollisionsfolgenmeidung according to the preamble of claim 1.

Modern driver assistance systems are able to completely avoid an imminent collision by an autonomously initiated full braking or at least to minimize the collision consequences. These systems detect the vehicle environment using suitable sensors (radar, lidar, image processing) or by evaluating vehicle-vehicle communication and determine possible collision objects. If a collision is imminent, a full deceleration is initiated at a certain intervention time.

As the relative speed between the ego vehicle and the potential collision partner increases, the path required for this braking increases quadratically. In comparison, the distance for a close Overtaking maneuvers or evasive maneuvers only increases linearly with speed, overtaking or avoidance is still possible at higher differential speeds after the time for collision-avoiding braking has already been exceeded. This so - called intervention dilemma leads to a conflict of objectives in the interpretation of the

Emergency brake assist. If the assistant engages with a full deceleration on reaching the last possible braking time, the collision is avoided - however, a driver who was planning a close overtaking or avoidance maneuver will be surprised by the intervention. In order to avoid these subjective perceived as defective braking and to meet the product liability requirements, the currently marketed systems are designed so that braking takes place only at the time when the driver collision neither by braking nor by overtaking or dodging However, at high differential speeds, this also means that autonomous emergency braking makes it possible to avoid collision avoidance - only the collision sequences are reduced.

From DE 601 26 398 T2, a brake control system with system intervention in object recognition is already known, which suppresses an otherwise generated automatic brake intervention to prevent a collision with a detected collision object at a detected alternative intention of the driver.

The object of the invention is to specify an improved method for automatic braking of a vehicle for collision avoidance or collision sequence reduction taking into account a driver overtaking or avoidance procedure, wherein the overtaking or avoidance procedure is to be recognized quickly and clearly. This object is achieved by a method according to claim 1. Advantageous developments emerge from the dependent claims.

The invention is based on a conventional method for automatic braking of a vehicle for collision avoidance or collision sequence reduction with a detected collision object, wherein at a determined intervention time, a braking system of the vehicle is automatically controlled such that a collision with the detected collision object avoided or at least the collision sequences are reduced can. In order to avoid the problem in particular at high differential speeds that a braking intervention already takes place, even though the driver intends to avoid the collision object or to overtake the collision object, the intervention time is determined taking into account the recognition or presumption of a driver's avoidance or passing process.

In order to be able to determine the intervention time in a situation-adaptive manner, it is again necessary to recognize a (probable) avoidance or overtaking procedure.

For the avoidance or overtaking detection, the movement of the vehicle in the longitudinal direction and the movement of the vehicle in the transverse direction are compared with each other, d. H. an evasive or overtaking maneuver is detected or assumed as a function of a measure of the (relative) movement of the vehicle to the detected collision object in the longitudinal direction and a measure of the (relative) movement in the transverse direction and a comparison of the two dimensions.

Advantageously, both measures for the relative movement in the longitudinal direction and in the transverse direction are each expressed in terms of normalized numbers. This allows a simple and quick comparison of the two values. Advantageously, as a measure of the relative movement in the longitudinal direction, the duration until the collision object collides - the so-called "time-to-collision (TTC)" - is determined, which represents a speed-normalized numerical value The time taken for the vehicle to collide with the preceding vehicle while maintaining the instantaneous speed and acceleration., In addition to the relative speed and the distance between the two vehicles in the longitudinal direction, the relative acceleration can also be evaluated.

High-resolution sensors allow analog viewing in the transverse direction. As a measure of the relative movement in the transverse direction is the duration until the occurrence of a zero overlap in the transverse direction of the vehicle with the collision object, ie the time until the two vehicles have no cross overlap, determined. This duration can be calculated analogously to the above by an evaluation of the relative lateral velocity and the distance to the preceding colliding object in the transverse direction. This time period can also be referred to as a so-called "time-to-no-overlap (TNO)." Again, the assumption is made that the instantaneous lateral velocity is assumed to be constant at the time of the calculation.

As already mentioned above, an avoidance or overtaking operation can be established by comparing the two determined speed-normalized time periods TTC and TNO. If the TNO is smaller than the TTC, an overtaking maneuver is more likely than the imminent collision, ie an evasion or overtaking process is detected or suspected if the determined time to collision occurs at least not less than the determined time until a zero overlap between the vehicle and the collision object in the transverse direction. In this case, the - possibly necessary - intervention in the brake system should be made only at a time, to which no evasive maneuver is possible, since in principle it is suspected that the driver would like to drive past the collision object and thus would interfere too early intervention of the brake system. If, however, the determined time duration until the occurrence of the collision is less than the determined time duration until a zero overlap occurs, no avoidance or overtaking maneuver appears to be planned. A collision-avoiding braking intervention at the last possible braking time is necessary.

Under certain circumstances, an avoidance or overtaking procedure can be assumed even if the TTC is smaller than the TNO. This is the case, for example, when the TNO is still larger than the TTC, but tends to decrease or falls below a predetermined threshold. In this case, the vehicle shears with previously insufficient lateral velocity behind the collision object, but it is expected that the driver continues to break and intends to avoid or overtake. Thus, an avoidance or overtaking maneuver may be suspected even if the time to onset of the collision is not more than a predetermined limit less than the duration until the zero overlap occurs in the transverse direction and / or the gradient of the duration is Occurrence of the zero overlap is strongly negative. On the other hand, as soon as the driver withdraws the deceleration movement, the TNO rises again and an evasive maneuver is unlikely.

If, as already mentioned above, a high-resolution sensor system is used to determine the transverse and longitudinal movement, a comparison of longitudinal and lateral movement is very well possible and an overtaking maneuver can be well recognized on the basis of the proposed method.

Subsequently, the concrete determination of the intervention time for automatic braking of the vehicle will now be explained in detail. In addition to the consideration of an alternative or overtaking procedure to be carried out by the driver, the determination of the intervention time also includes the Driver reaction time or the end time of the driver reaction time and a determined last possible braking time to be considered.

In real traffic, the driver is constantly experiencing events to which he must respond in order to avoid a collision - including those relevant for autonomous emergency braking:

• A continuous approach to a front vehicle,

• a sudden braking maneuver of a vehicle in front as well

• a neighboring vehicle shattering into its own lane.

Although there is no information on subjective driver perception, the vehicle sensors can be used to create a model of these reaction triggers. For this purpose, driving psychological threshold values are used which describe the transition from a perceived safe driving to a reaction-requiring approximation.

The continuous approach to a vehicle within the self-trajectory, which in many situations also takes place at high relative speed, can be described independently of distance by the time to collision (TTC). In the state of the art, a TTC value of 5 s is regarded as the threshold below which the driver perceives an acute need for action, to increase the gap between the first and the foreign vehicle again or to avoid the obstacle by means of an overtaking maneuver.

This threshold value can also be specified as a distance value in a limited differential speed range. It is composed of a constant safety distance at standstill and a speed-dependent value. With this minimum following distance, the remaining applications "sudden braking" and "shearing" can now be detected. Delays a preceding vehicle in the following traffic so much that the driver of the ego vehicle, while maintaining his would fall below the safety distance wegungszustandes, there is a need for action to restore the safe driving condition. This need for action also arises when a vehicle on the adjacent lane within the minimum following distance starts a Einschermanöver.

The reaction time, which begins after the occurrence of the corresponding reaction trigger and extends to the time at which the driver reaction can be measured on the vehicle bus system, can be modeled using the so-called OODA feedback loop. This represents the individual components of human decision making: O (bserve, Observe), O (rient, Arrange), D (ecide, Decide) and A (ct, Act).

In both the observation and the action process, on average, all drivers can assume approximately constant values. The duration for the pure perception of the situation is about 0.2s, the execution of the action - ie the foot movement to the pedal operation - due to the frequent exercise at about 0.3s.

In contrast, the remaining process steps for ranking and decision-making are dependent on the initiating event, i. H. the driver reaction time can be determined as a function of the current vehicle environment and / or the type of occurrence of the collision object. Unexpected triggers (for example, unpredictable heavy braking of the vehicle ahead) or low-probability events lead to a longer decision-making phase than expected or frequent events. Together, these two process steps can be modeled as gamma-distributed.

In order to be able to derive a reaction time from the gamma distribution, the risk is considered to exceed a certain duration. For this purpose, the cumulative gamma distribution of one is deducted. A risk of 20% Accordingly, in the study under consideration, 80% of drivers needed less or exactly the resulting duration to respond to the triggering event.

In addition to the dependence on the triggering event or on the type of occurrence of the collision object, the reaction time resulting from the OODA active circuit is extended, as long as the driver has several alternative courses of action available. Since the individual occurrence probabilities of the action alternatives are a priori unknown, they are assumed to be equally distributed.

For the application of active emergency braking, this means that in addition to the possibility of braking and the feasibility of an evasive or overtaking maneuver must be checked. For this purpose, for example, the area next to the vehicle is divided into three sub-areas: behind the ego vehicle, laterally at the level of the ego vehicle and in front of the ego vehicle. For all areas, a lane change-relevant value is calculated and standardized in a comparison function to a factor between 0 (lane change not feasible) and 1 (lane change safely feasible). The minimum of the three factors determines the global lane change feasibility. If a defined threshold for the minimum is exceeded, the feasibility of a lane change in the corresponding direction is assumed.

In the rear area, the necessary deceleration for an approaching vehicle in the event of a lane change of the ego vehicle is calculated as a lane change relevant value. In the front area the necessary delay for the ego vehicle is analyzed, it should carry out a lane change and the destination lane already be occupied by another vehicle. In the page area, the space required to carry out the lane change is evaluated. The lane change feasibility then results from the ratio between the necessary and the reasonable deceleration (rear and front area) or between the required and the available space in the side area.

Since the last possible alternative time must be taken into account when determining the engagement time, in particular on the condition that the driver can still overtake or avoid a detected collision object, this too must be taken into account when determining the

To be considered at the time of intervention.

To determine the last possible escape time or the necessary distance for a close overtaking maneuver, for example, an overtaking parabola with an assumed lateral acceleration is placed in the free area between the emergency vehicle and the potential collision object. The necessary for the overtaking or evasive maneuver distance in the longitudinal direction then results in dependence on the instantaneous relative speed, as well as the lateral distance to be overcome. This consists of the object widths of the ego and other vehicles as well as the current object position. Since a maximum lateral acceleration can not be assumed at low speeds, the lateral acceleration is adapted as a function of the speed.

Taking into account the end time of the determined driver reaction, a determined last possible braking time and the determined last possible avoidance time, the situation can be evaluated at the end of the reaction time and thus at least determine a provisional intervention time for an autonomous braking intervention. For this purpose, the corresponding reaction time is first assigned to the three possible reaction triggers. A continuous approach to a front vehicle is considered as expected, a sudden braking or a collapsing vehicle as an unexpected event. If one subtracts the corresponding reaction time from the time to the collision (TTC) at the time of the occurrence of the reaction requirement, the determined value - ie the value of the Collision avoidance remaining time after the reaction - with the

Compare thresholds for braking or overtaking maneuvers.

In the case of low differential speeds (i.e., the last possible avoidance time in terms of time before the last possible braking time), the last possible braking time can always be used as the intervention time since the driver does not have any other alternative to collision avoidance.

However, in the case of high differential speeds where overtaking or dodging is still possible after the last possible braking time has been reached, there are three possible situations.

In the event that the determined end time of the driver reaction time in terms of time before the determined last possible braking time, therefore, the last possible braking time can be selected as the provisional intervention time, as in the case of an independent collision avoidance by the driver, a corresponding reaction is already present. In this case, accident prevention is always possible.

In the second case, it is assumed that the determined end time of the driver reaction time occurs before the determined last possible avoidance time and after the determined last possible braking time. Since the driver's likelihood of autonomous control of the situation is much higher than that of a rear-end collision, braking intervention may be initiated (ie, if necessary) only at the end of the reaction time. In the worst case, however, this may mean that the emergency braking is no longer sufficient for complete collision avoidance. In order to partially resolve this conflict, the normalized value is used to evaluate the lane change possibility from the reaction time estimation, ie the provisional intervention time in this case is dependent on the Feasibility of an evasive maneuver set. In this case, the last possible braking time is determined as a provisional intervention time when an evasive maneuver is not feasible, and only the endpoint of the driver reaction time when an evasive maneuver is safely feasible. If no lane change is possible, a completely collision-avoiding braking takes place.

In the third case, it is assumed that the determined end time of the driver reaction time is after the determined last possible braking time and after the determined last possible avoidance time, with the determined last possible avoidance time occurring after the determined last possible braking time. In this case, the automatic brake intervention can be made already at the last possible braking time, since a timely response of the driver is very unlikely.

For an even better definition of the intervention time of the emergency brake assistants, it is also important, in addition to the above, to check and take into account the actual driver reaction. In particular, the incoming driver reaction (eg braking or overtaking) during and / or after the determined driver response time should be taken into account. If a reaction is present (or at least suspected) and is sufficient for independent collision avoidance or if the avoidance or overtaking maneuver is feasible, the provisional engagement time determined in the case distinction becomes the so-called "point of no return", ie the last possible avoidance time Although this should never be achieved due to the driver response that has been recognized as adequate, it may still be maintained as a fallback level, otherwise the intervening intervention time is determined as the intervening intervention time.

Indicators can be determined for the two action options "braking" and "overtaking". A braking effect can be achieved both by the original operation of the brake pedal as well as by the construction of drag torque by releasing the accelerator pedal are applied. The overtaking reaction can be detected by the method according to claim 1 or for example by detecting a decreasing overlap to the front vehicle or by a corresponding steering angle course. In order to be able to include the avoidance or overtaking request at an early stage, the use of lane change motivational models is obvious. A simplified approach for this is the detection of the turn signal operation for lane change.

In order to check whether the detected reaction is also timely and adequate, in the case of braking the applied deceleration is compared with the deceleration necessary to avoid the collision. This is dependent on the instantaneous relative speed between ego vehicle and collision object, the instantaneous distance between the two vehicles and the delay of the other vehicle.

In the case of a present overtaking intention or reaction, it must also be checked, as already mentioned above, whether the lane change or the avoidance maneuver can be carried out. For this purpose, it is possible to make use of the assessment of the lane change possibility calculated within the framework of the reaction time estimation.

Taking into account all aspects, a quick and unambiguous determination of the intervention time for an automatic emergency braking is thus possible.

Claims

Method for automatic braking of a vehicle for collision avoidance or collision reduction claims

A method for automatically braking a vehicle for collision avoidance or Kollisionsfolgenminderung with a detected collision object, wherein at a determined intervention time, a braking system of the vehicle is automatically controlled such that a collision with the detected collision object avoided or at least the collision sequences can be reduced, the engagement time is determined as a function of detecting or suspecting a driver made avoidance or overtaking, characterized in that an evasive or overtaking operation in dependence on a measure of the relative movement of the vehicle to the detected collision object in the longitudinal direction and a measure of the relative movement in the transverse direction and Comparison of the two measures is recognized or suspected.

2. The method according to claim 1, characterized in that the measure of the relative movement in the longitudinal direction and in the transverse direction is expressed in each case in a speed normalized time value.

3. The method according to any one of the preceding claims, characterized in that as a measure of the relative movement in longitudinal Direction the duration is determined until the collision occurs with the collision object.

4. The method according to any one of the preceding claims, characterized in that as a measure of the relative movement in the transverse direction, the duration is determined until the occurrence of a zero overlap in the transverse direction of the vehicle with the collision object.

5. The method according to claim 3 and 4, characterized in that an avoidance or passing process is detected or suspected depending on the determined duration until the collision and the duration until the onset of zero overlap in the transverse direction, in particular such that an avoidance or overtaking operation is detected or suspected if the duration until the onset of the collision is at least not less than the time until the zero overlap occurs in the transverse direction.

6. The method according to any one of claims 3-5, characterized in that an evasive or overtaking process is detected or suspected, if the time until the onset of the collision by no more than a predetermined limit is less than the duration until the occurrence of zero Overlap in the transverse direction, and / or the time gradient of the duration until the onset of zero overlap is negative).

7. The method according to any one of the preceding claims, characterized in that when detecting or suspecting an evasive or overtaking process as the intervention time a determined last possible fallback time is determined.

8. The method according to any one of the preceding claims, characterized in that in an unlikely evasion or overtaking the engagement time in dependence on the end time of a determined driver reaction time and / or a determined last possible braking time and / or a determined last possible fallback time is determined.

9. The method according to claim 8, characterized in that the intervention time is determined such that at a determined end time of the driver reaction time, which occurs before the determined last possible braking time, the provisional engagement time of the last possible braking time is determined and / or that at a determined end time of Driver reaction time, which occurs before the determined last possible escape time and after the determined last possible braking time is set as the provisional engagement time of the last possible braking time or the end time of the driver reaction time and / or that at a determined last possible escape time, after the determined last possible braking time and before the determined end time the driver reaction time occurs, as the provisional engagement time of the last possible braking time is set.