EP3439932A1

EP3439932A1 - Method for operating a vehicle

Info

Publication number: EP3439932A1
Application number: EP17715082.8A
Authority: EP
Inventors: Oliver Pink; Christoph Schroeder
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 2016-04-07
Filing date: 2017-03-29
Publication date: 2019-02-13
Also published as: US20190084562A1; JP6661789B2; CN109070881A; CN109070881B; US10787171B2; JP2019513617A; DE102016205761A1; WO2017174408A1

Abstract

The invention relates to a method for operating a vehicle, wherein said vehicle has a surroundings sensor system and at least one at least partially automated driving function, comprising the steps of: detecting (102) an object state by means of the surroundings sensor system at a first point in time; detecting (103) an object state by means of the surroundings sensor system at a later second point in time; adjusting (104) an activation state of the driving function on the basis of the detected object states.

Description

description

title

Method for operating a vehicle Prior art

The invention relates to a method for operating a vehicle.

Different driver assistance functions for vehicles are known from the prior art. These can usually be activated or deactivated by a driver.

EP 2 562 060 B1 discloses a method for computer-based prediction of motion patterns of at least one target object, such as a vehicle.

In DE 10 2012 215 093 AI a method for operating a

Driver assistance system for a driving on a road vehicle described. In this case, depending on a predicted trajectory of an oncoming traffic vehicle, an anticipated meeting point with the moving vehicle is determined and, if necessary, a safety device is activated.

Disclosure of the invention

The invention describes a method for operating a vehicle. The vehicle here comprises an environment sensor system and at least one at least partially automated driving function. By means of the environmental sensor system is at least a first time and a later second time detects an object state and adjusted based on the detected object states an activation state of the driving function.

The inventive method has the advantage that the security in

Road traffic is increased. By means of the detection of object states too

At different times can be determined whether a driving function at a time of detection is able to guide the vehicle safely. By adjusting the activation state, it is possible to prevent a driving function from being used outside of areas where it is not full

Functioning and can ensure a certain level of security.

The environment sensors can be conventional sensors, such as cameras, lidar, radar or ultrasonic sensors or corresponding sensor arrangements. The at least partially automated driving function can be, for example, systems for driver assistance or for automated guidance of the vehicle up to systems for highly automated or even autonomous driving. This includes, for example, lane keeping systems, Adaptive Cruise Control, congestion assistants with transverse and / or longitudinal guidance or systems for at least temporarily complete adoption of the vehicle guidance. A driving function can also be understood as meaning a single maneuver that can be carried out by the driving function or a driving state that can be reached by means of the driving function.

All conceivable properties of detected objects can be understood as the detected object states, with all environment details or objects that can be detected by means of the environmental sensor system being referred to as objects.

This includes, for example, other road users, vehicles,

Vehicle occupants or drivers, road markings, miscellaneous

Markers, landmarks such as trees or signs, infrastructure facilities such as tunnels, construction sites or crash barriers. The properties of the objects may include, for example, position, size, speed, distance, color, direction of movement, structure and / or further details.

In an advantageous embodiment of the method, a prediction of at least one takes place by means of the object state detected at the first and / or second time Object state at a later time. The adaptation of the

Activation state is based on the at least one predicated

Object state.

This embodiment offers the advantage that in the adaptation of

Activation state, an object state that was calculated based on an estimation of the object behavior in an environment of the vehicle. Thus, depending on the assessment, it can be judged to what extent the activation state should be adjusted.

The prediction can be carried out on the basis of an object model, which in the

Vehicle can be deposited. If, for example, the speed, the direction of travel and the current position of an object are detected, as well as the

Considering the speed and position of the vehicle, the position of the object can be predicted for a later time, for example assuming a constant direction of travel and speed. It is also conceivable that for different object classes, such as, for example, other vehicles, motorcycles,

Cyclists, pedestrians or fixed, non-moving

Infrastructures, different object models are deposited. The

Classification of the objects can take place by means of known classification methods, for example with the aid of captured camera images. Depending on the assigned

Object class can thus be made different predictions.

In an advantageous embodiment of the method, the adaptation of the activation state of the driving function is based on a comparison of at least two predicted object states and / or at least one predicted for a later time and one detected at the later time object state.

This embodiment of the method offers the advantage that the adaptation of the activation state on the basis of a comparison of an assessment of the

Object behavior in an environment of the vehicle with a further assessment at a later time or the detected object state can be done at a later date. This makes it possible to validate the assessment by means of at least partially independent object states at a time and in Dependence of the comparison of the object states on the activation state

adapt.

In an advantageous embodiment of the method, a quality of at least one predicated object state is determined and the activation state is adjusted based on the quality.

This embodiment of the method offers the advantage that the

Activation state based on an accuracy of the prediction, and thus the ability of the vehicle to estimate the environment is adjusted.

In an advantageous embodiment of the method, the quality is rated too low if the quality is below a threshold value for the quality. Consequently, with a quality below the threshold, the at least one at least partially automated driving function is deactivated or an activation of the driving function is prevented.

This embodiment of the method offers the advantage that the security is increased. If the prediction of the predicted object state is too inaccurate, for example if the predictions of object states detected at different times deviate too far from one another, this is an indication that the object behavior in the surroundings of the vehicle can not be sufficiently well estimated by the vehicle. Consequently, if it was active at the time of determining the quality, the driving function is deactivated. Deactivation may occur immediately after detection, delayed or incremental, for example, by predefined fallback levels, which may include speed reduction or automated stopping or reaching a safe state. It can also issued a warning to the driver and this will be prompted to take over the vehicle control. If the driving function is not active at the time when the insufficient quality is determined, activation of the driving function is prevented. For this purpose, for example, the power-on function of the driving function or a corresponding switch is deactivated. The limitation of the driving function can also be displayed to the driver, for example in the form of light symbols in the valve. In an advantageous embodiment of the method, a mean deviation of at least two predicted object states at the second time point and / or at least one predicted object state and the object state detected at the second time point is determined as a measure of the quality.

This embodiment offers the advantage that the quality of at least one predicted object state can be determined very quickly and very simply, and consequently a rapid adaptation of the activation state of the drive function can take place.

To determine the quality, either at least two object states predicted from different object states at different times at a later common time or at least one object state predicted from a first time can be used at a second time with the object state detected at the second time. The determination can always be carried out at the time of the second detection, regardless of whether the object state detected at the second time or an object state predicted by means of the object state detected at the second time are used to determine the quality.

For the determination of the mean deviation, all already mentioned

Object states and properties are used. For example, distances of the predicted or detected object positions, deviations in the

Object speeds, deviations of the object shape or the

Object movement direction or object color (lighting, tunnel entrance, etc.).

Instead of calculating the average deviation of the predicted or detected object states, a determination of the variance or a formation of the sum of the distances is also conceivable. The smaller the deviations / sum, the better the quality of the prediction.

In an advantageous embodiment of the method is in the adaptation of the

Activation state, the driving function is deactivated or the driving function remains activatable, or the driving function remains active or it is prevented activation of the driving function. This embodiment offers the advantage that the driving function can only be executed or executed if the estimation of the object conditions in the surroundings of the vehicle is possible with sufficiently high accuracy. Consequently, activation of the driving function is prevented when the safe execution of the driving function is not ensured. If the driving function has already been activated, the vehicle control is returned to a driver or a corresponding driver

Security maneuvers performed to increase the safety of all road users. Since a driving function can also be understood as meaning a single maneuver that can be carried out by the driving function or a driving state that can be reached by means of the driving function, an adaptation of the driving function can likewise signify a restriction of the driving function. For example, that by means of

Driving function a maximum speed must not be exceeded.

The invention further comprises an electronic control unit which is set up to carry out all the steps of the method described in this application, as well as a corresponding computer program. Likewise, the invention comprises a machine-readable storage medium on which the computer program is stored.

drawings

FIG. 1 shows a process diagram.

FIG. 2 shows a further process diagram.

FIG. 3 shows object states at different times.

embodiment

In Fig. 1, an exemplary method for operating a vehicle is shown. The vehicle is equipped with an environment sensor system, at least one at least

semi-automatic driving function and a control unit for adapting a Activation state of the driving function equipped. As in this example, the method can be executed on a separate control unit, but alternatively it can also be integrated into the driving function and the corresponding control unit. The process starts in step 101.

In step 102, by means of the environmental sensor system at a first time

Object state detected. In this example the position, direction of travel and

Speed of another vehicle.

In step 103, another object state is detected at a later second time. In this case again the position, direction of travel and the speed of the other vehicle.

In step 104, based on the detected object states by means of

Control unit an activation state of at least one at least

partially automated driving function adapted. For this purpose, the at least two object states are evaluated by means of the control unit, for example, stored and then compared.

The driving function may be, for example, an overtaking assistant, which can automatically initiate overtaking maneuvers on two-lane roads, for example by pressing a button. Is determined in step 104 in an evaluation of the detected object states that, for example, a detected overtaking another vehicle has such a high speed that a safe execution of an overtaking on the currently traveled section by means of the driving function does not seem safe, for example

Environment sensor has no correspondingly large detection range, so activation of this driving function is prevented. For this purpose, for example, the button for activating the driving function can be deactivated. In addition, the driver can be shown that the function can not be activated due to the current traffic situation. The method ends in step 105.

In Fig. 2, another method is shown, which can be performed on the same vehicle. The method starts in step 201. In step 202, at least one object state is detected at a first time t_l. The detected object may be, for example, a cyclist whose object states, position, speed and viewing direction are detected at the time t_l.

In step 203, at least one object state of the detected object is predicted by the control unit at a later second time t_2. In this example, the driver's position and speed at time t_2. It may but need not necessarily all detected at time t_l

Object states are predicted at a later date. For example, it is also conceivable that from the position, speed and direction of the

Cycle rider only a prediction of the position of the cyclist at time t_2 is performed or all three states are predicted.

In step 204, an object state of the same object is detected again at the later second time t_2. In this example, again, the position

Speed and direction of the driver.

In step 205, the object states, which were detected at the second time t_2, are compared with the object states which were predicted from the object states at the time t_1 for the time t_2 by means of the control unit. For example, a quality of the prediction can be determined by means of the comparison. For example, the distance of one predicted object state or an average deviation of all predicted object states for the time t_2 to the one or more actual object state detected at the time t_2 can determine the quality.

Object states are calculated. Instead of calculating the mean

Deviation of the object states is also a determination of the variance or formation of the sum of the distances conceivable. The smaller the deviations, the better the quality of the prediction.

Based on the performed comparison of the object states is a

Decision on the adaptation of the activation state of the at least one driving function. If the determined quality falls below a predetermined limit, then For example, the mean deviation greater than a predetermined threshold, this may indicate that the vehicle can not predict the current traffic situation with sufficient accuracy and assess accordingly. For example, goodness can be seen as a measure of the complexity of a specific traffic situation that the function has to master. Since the sensors used to detect the object states can also be used for the driving function, and are also used in this example, there is a direct correlation between the quality and the functionality of the driving function.

If the quality determined in step 205 falls below a predetermined threshold value, the method continues in step 206, in which an adaptation of the

Activation state of the driving function takes place. If the driving function is active at the time t_2 and / or at the time of determining the prediction quality, a

Deactivation of the function. This can be done directly or with a customized delay. For example, the driver could first be informed that he must take over the vehicle. Alternatively, the vehicle can also work by means of implemented fallback levels, so that successively individual driving functions are deactivated or properties of the driving function are limited. For example, a reduction in speed or control of a safe could

Holding area done.

If the driving function is not activated at the time t_2 or the time at which the insufficient quality was determined, then in step 206 an adaptation of the

Activation state of the driving function in the form of preventing the activation of the driving function. This can be done, for example, by deactivating the activating function so that a corresponding switch can no longer be actuated or its actuation does not activate the driving function. At the same time, the information that the activation of the driving function is currently not possible, the driver can be displayed.

If the quality is too low, the method ends in step 208 and can

then be started again in step 201. Alternatively, object states can be detected permanently and, based on the detected object states, predicted object states can be determined at later times, in each case be compared with each other or with detected at later times object states.

If the quality determined in step 205 does not fall below the predetermined threshold value, then a traffic situation controllable for the respective driving function is present. The method continues in this case in step 207, in which an adjustment of the activation state of the driving function takes place. If the drive function is already active, the adjustment confirms the activation state and the drive function remains active. An adaptation in this case can thus also be understood as meaning that nothing changes in the state of the driving function and no change is necessary.

If the driving function is not activated at time t_2 or at the time the quality is determined, an adaptation of the activation state of the driving function takes place in step 206 in the form of a release of the activability of the driving function. If the driving function could not be activated before the time t_2 or the time of determining the quality, the activation state is set to activatable and the driving function can be activated. If the driving function was already activatable, nothing changes in the activation state and the driving function remains activatable.

The process ends in this case in step 209 and can then be run again.

In Fig. 3 is an exemplary representation of the detection and prediction of

Object states shown. In this example, object states are called

Object positions detected 301 and object positions predicts 302. At different times t_l to t_6 object positions 301 are detected by means of the environment sensor. For each object position 301 detected at a time t_x, xeM, a plurality of object positions 302 are predicted for later times t_x + i, ieM. These can be predicted at any given time t_y, yeM with further

Object positions 302 and / or the actually detected at the corresponding time object position 301 are compared.

Only predicted object positions 302 (or predicated

Object states). It does not necessarily have a recorded Object position 301 (a detected object state). For example, if an object state predicted for the time t_3 from the object state detected at the time t_1 differs too much from an object state detected from the object state detected at the time t_2 for the time t_3

Object state, can already before the time t_3 an adjustment of the

Activation state of the driving function done. The object state actually detected at time t_3 thus does not flow into the adaptation in this case.

A comparison of object states can take place at any time t_x, xeM. For example, in FIG. 4, the detected 301 and predicted 302 object positions (plotted along an x-axis) are compared and, as a quality for the prediction, a deviation 303 of the predicted object positions 302 from the actually detected object position 301 is determined.

Claims

claims

1. A method for operating a vehicle, wherein the vehicle is a

Environment sensor and at least one at least partially automated driving function, wherein by means of the environment sensor at least a first time and a later second time an object state is detected and based on the detected object states an activation state of the driving function is adjusted.

2. Method according to claim 1, characterized in that by means of the object state detected at the first and / or second time a prediction of at least one object state takes place at a later time and the adaptation of the activation state takes place on the basis of the at least one predicted object state.

3. The method according to claim 2, characterized in that the adaptation of the activation state of the driving function based on a comparison of

at least two predicted object states and / or

at least one predicted for a later date and one detected at the later time object state takes place.

4. The method according to any one of the preceding claims, characterized in that a quality of at least one predicated object state is determined and the activation state is adjusted based on the quality.

5. The method of claim 4, wherein the quality is assessed as being too low,

if the quality is below a threshold value for the quality, in which case the at least one at least partially automated driving function is deactivated or activation of the driving function is prevented.

6. The method according to any one of claims 4 or 5, characterized in that as a measure of the quality, a mean deviation of predicted object states at the second time and / or at least one predicted object state for second time and the object state detected at the second time is determined

7. The method according to any one of the preceding claims, characterized in that during the adaptation of the activation state, the driving function is deactivated or remains activatable or remains active or activation of the

Driving function is prevented.

8. Electronic control unit, which is configured to perform all the steps of the method according to one of claims 1 to 7.

A computer program configured to perform all the steps of the method of any one of claims 1 to 7.

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