WO2019120709A1

WO2019120709A1 - Method and control unit for controlling a function of a vehicle driven at least partially in an automated manner

Info

Publication number: WO2019120709A1
Application number: PCT/EP2018/079423
Authority: WO
Inventors: Stefan Erschen; Felix Klanner; Thomas Helmer; Horst KLÖDEN; Christopher Bach
Original assignee: Bayerische Motoren Werke Aktiengesellschaft
Priority date: 2017-12-21
Filing date: 2018-10-26
Publication date: 2019-06-27
Also published as: DE102017223621A1

Abstract

The invention relates to a control unit (200) for a vehicle (110) driven at least partially in an automated manner. The control unit (200) is designed to determine sensor data relating to the surroundings of the vehicle (110). The control unit (200) is also designed to determine, on the basis of the sensor data, whether or not there is an unstructured traffic situation (100). In addition, the control unit (200) is designed to control a function of the vehicle (110) according to whether or not there is an unstructured traffic situation (100).

Description

Method and control unit for controlling a function of an at least partially automated vehicle

The invention relates to a method and a control unit for controlling a function, in particular an output function and / or a guidance function, of a vehicle which is at least partially automated.

Automated vehicles typically rely on detailed digital map information and sensor data from environmental sensors to create a model of the environment of the vehicle and to determine and implement a driving strategy based thereon.

In densely populated cities, in particular in the absence of discipline of the individual road users, driving situations often occur that can no longer be mastered reliably by an automated moving vehicle. The present document addresses the technical problem of increasing the reliability and / or availability and / or comfort of at least partially automated vehicles in such traffic situations.

The object is solved by the independent claims. Advantageous embodiments are described i.a. in the dependent claims. It should be noted that additional features of one of

independent claim dependent patent claim without the features of the independent claim or only in combination with a subset of the features of the independent claim may form an independent and independent of the combination of all features of the independent claim invention, the subject of an independent claim, a divisional application or an N be made can. This applies equally to technical teachings described in the specification, which may form an independent invention of the features of the independent claims.

The term "automated driving" in the context of this document can be understood as driving with automated catches or transverse guidance or autonomous driving with automated catches and transverse guidance. The automated driving may be, for example, a time-prolonged driving, such as on the highway, or a time-limited driving in the context of parking or maneuvering. The term "automated driving" includes automated driving with any one

Degree of automation. Exemplary automation s are an assisted, semi-automated, highly automated or fully automated driving. These degrees of automation were defined by the Federal Highway Research Institute (BASt) (see BASt publication "Forschung kompakt", issue 11/2012).

In assisted driving, the driver performs the catching or transverse guidance permanently, while the system assumes the other function within certain limits. In semi-automated driving (TAF), the system takes over the crab and lateral guidance for a certain period of time and / or in specific situations, whereby the driver must permanently monitor the system as in assisted driving. In highly automated driving (HAF), the system takes over the catch and lateral guidance for a certain period of time, without the driver having to permanently monitor the system; However, the driver must be in a position to be in charge of taking over the vehicle. In fully automated driving (VAF), the system can automatically handle driving in all situations for a specific application; no driver is required for this application. The above four degrees of automation are in accordance with SAE Fevel 1 to 4 of SAE J3016 (SAE - Society of Automotive Engineering). For example, Highly Automated Driving (HAF) Fevel 3 complies with the SAE J3016 standard. Furthermore, in SAE J3016 the SAE level 5 is still the highest Automation level not included in the definition of BASt. SAE level 5 is driverless driving, which allows the system to automatically handle all situations like a human driver throughout the journey; a driver is generally no longer required.

According to one aspect, a control unit for an at least partially automated vehicle (in particular a road vehicle) is described. The vehicle may preferably be set up to drive at least highly automated.

The control unit may be configured to determine sensor data regarding an environment of the vehicle. The sensor data can be detected by means of one or more environment sensors of the vehicle. Exemplary sensor data include a radar sensor, an image sensor, an ultrasonic sensor, a Fidar sensor, etc. The sensor data may be repeatedly and, in particular, periodically acquired and provided for a sequence of times, for example, at a sampling rate of IHz, 10Hz or more. The control unit can thus be set up to determine a temporal sequence of sensor data for a corresponding sequence of times.

In addition, the control unit may be configured to determine on the basis of the sensor data (in particular on the basis of the temporal sequence of sensor data) whether there is an unstructured traffic situation or not. In this case, an unstructured traffic situation may be characterized by one or more properties and / or caused by one or more causes. The control unit may be configured to determine on the basis of the sensor data (in particular on the basis of the temporal sequence of sensor data) whether there is a traffic situation in the surroundings of the vehicle which has the one or more properties of an unstructured traffic situation, and / or Whether there are one or more possible causes of an unstructured traffic situation in the vicinity of the vehicle. A particularly reliable feature for the detection of an unstructured traffic situation is that, in the case of an unstructured traffic situation, at least one traffic rule valid in the surroundings of the vehicle in accordance with a highway code is provided by at least one other traffic participant, in particular by a plurality of other traffic participants. is not followed. The control unit can thus be set up on the basis of (sequence of)

Sensor data to determine if a traffic situation exists, this

Has property.

Machine-learned classification methods can be used to detect an unstructured traffic situation. The control unit may be configured to determine at least one value of a feature vector having a plurality of features based on the (sequence of) sensor data. Particularly reliable features for detecting an unstructured traffic situation are: the driving speed of the vehicle; a distance measure with respect to a distance of the vehicle to one or more others

Driver in the vicinity of the vehicle (e.g., the weighted inverse distance D described in this document); and / or a

Lane measurement that indicates to what extent one or more others

Driving participants in the environment of the vehicle within at least one defined lane (for example, the one described in this document

Lane measurement F).

The control unit may be further configured based on the value of

Feature vector and using a machine-learned classifier to determine whether or not there is an unstructured traffic situation. In this case, the classifier can be set up to identify at least one subspace for the existence of an unstructured traffic situation within a total of possible values of the feature vector. The

Classifier may preferably have a decision tree (DT) and / or a random Forest (RF) classifier include. By using a machine-learned classifier, it can be reliably determined whether or not there is an unstructured traffic situation. The control unit may be further configured to control a function of the vehicle depending on whether there is an unstructured traffic situation or not. The function may include an output function for outputting information relating to the unstructured traffic situation to a user of the vehicle and / or an output function for sending information relating to the unstructured traffic situation to a receiver outside the vehicle. For example, in the presence of an unstructured traffic situation, an issue can be made to the driver of the vehicle, in which the driver is informed that there is an unstructured traffic situation (and if necessary a takeover by the driver should take place to at least partially cover the vehicle to lead manually). In another example, a central processing unit may be aware of the presence of an unstructured

Be informed of the situation. This information can then be used to guide and / or route other participants in the journey. Alternatively or additionally, the function may include a guidance function for at least partially automated driving of the vehicle.

For example, the mode of operation (in particular the driving strategy) of the at least partially automated vehicle may depend on whether there is an unstructured traffic situation or not.

By automatically recognizing an unstructured traffic situation, the reliability and comfort of an at least partially automated vehicle, in particular of a vehicle driving at least highly automated, can be increased. The control unit may be configured to determine a cause from a plurality of possible causes for the unstructured traffic situation on the basis of the (temporal sequence of) sensor data. This can be done, for example, by using a classifier. For this purpose, the subspace for the presence of an unstructured traffic situation in a plurality of

Subspaces may be subdivided for the corresponding plurality of possible causes.

The function of the vehicle can then be controlled as a function of the determined cause for the unstructured traffic situation.

For example, a user and / or a central processing unit can be informed about the cause of the unstructured traffic situation. Alternatively or additionally, the ascertained cause can be taken into account when driving the at least partially automated vehicle. So can the

Reliability and comfort of an at least partially automated moving vehicle can be further increased.

Furthermore, if it has been determined that there is an unstructured traffic situation, the control unit can be set up to determine a traffic rule which is effective within the unstructured traffic situation on the basis of the (temporal sequence of) sensor data. The effective traffic rule may deviate from a traffic regulation in force under the Highway Code. The effective traffic rule can be based on a

Pattern recognition method for detecting at least one pattern of the sensor data are determined. The effective traffic rule can indicate how the one or more traffic participants in the vicinity of the vehicle behave differently from the one or more traffic regulations in force in accordance with the Highway Code.

The function of the vehicle can then be controlled or operated as a function of the determined, effectively valid traffic rule. For example the vehicle may continue to be at least partially, in particular at least highly automated, taking into account the effective traffic rule. Thus, it can be achieved that an autonomously driving vehicle can reliably master an unstructured traffic situation.

Alternatively or additionally, a takeover request can be issued to a user of the vehicle to guide the vehicle at least partially manually via a user interface of the vehicle. Furthermore, information relating to the determined, effective traffic rule can be output to the user via the user interface of the vehicle. The user can then consider the effective traffic rule when manually guiding the vehicle. Thus, the comfort for a driver of a vehicle can be increased within an unstructured traffic situation.

According to a further aspect, a method for controlling or operating a function of an at least partially automated vehicle is described. The method includes determining sensor data relating to an environment of the vehicle. In addition, the method includes determining, based on the sensor data, whether an unstructured traffic situation exists or not. Further, the method includes controlling a function of the vehicle depending on whether there is an unstructured traffic situation or not.

According to a further aspect, a road vehicle (in particular a passenger car or a truck or a bus) is described which comprises the control unit described in this document.

In another aspect, a software (SW) program is described. The SW program can be set up to run on a processor (eg a control unit of a vehicle) and thereby carry out the method described in this document.

In another aspect, a storage medium is described. The storage medium may include an SW program that is set up to be executed on a processor, and thereby perform the same in this

Document described method.

It should be understood that the methods, devices and systems described herein may be used alone as well as in combination with other methods, devices and systems described in this document. Furthermore, any aspects of the methods, devices, and systems described herein may be combined in a variety of ways. In particular, the features of the claims can be combined in a variety of ways.

Furthermore, the invention will be described in more detail with reference to exemplary embodiments. Show

FIG. 1 shows an exemplary unstructured traffic situation;

FIG. 2 shows a block diagram with exemplary components of a vehicle; and

FIG. 3 shows a flow chart of an exemplary method for controlling an at least partially automated vehicle.

As stated at the outset, the present document deals with the reliable and comfortable control of an at least partial

Automated moving vehicle in complex traffic situations. 1 shows an exemplary unstructured traffic situation 100, in which it is provided by a traffic rule stipulated in the prevailing road traffic regulations that the vehicles 112, 113 travel a first traffic lane 101 into a second traffic lane 102 and from there left onto a third traffic lane 103 turn off. The Traffic rule is given in the example shown by a traffic sign 104 and by lane markers.

Particularly in the case of a high traffic volume and in the case of a lack of discipline on the part of the individual traffic participants, it may happen that the current state of health

Traffic rules are no longer observed. In the illustrated example, the vehicles 112, 113 do not thread from the first lane 101 into the second lane 102, but remain on an improvised lane 103 adjacent to the second lane 102 to turn onto the third lane 105. This results in the problem for a vehicle 110 in the second lane 102 that when turning to the third lane 105 not only vehicles 111 on the third lane 105 but also vehicles 112, 113 must be considered on the improvised lane 103.

The unstructured traffic situation 100 illustrated by way of example in FIG. 1 is typically associated with a relatively high level of stress for a human driver, but can usually be mastered by the driver. On the other hand, an at least partially automated vehicle 110 usually does not cope with such a traffic situation 100, since such a traffic situation 100 requires an adapted driving strategy.

2 shows a block diagram of a vehicle 110 that is configured to drive at least partially automated. In particular, the vehicle 110 may be capable of at least highly automated driving. The vehicle 110 includes one or more environmental sensors 201 configured to detect sensor data relating to an environment of the vehicle 110. exemplary

Environment sensors 201 are a radar sensor, an image camera, a

Ultrasonic sensor, a lidar sensor, etc. The vehicle 110 further includes a storage unit 202 on which digital map information related to the

Road network in the environment of the vehicle 110 may be stored. Further, the vehicle 110 includes a control unit 200 that is configured to Vehicle 110 based on the sensor data and / or based on the digital

Map information at least partially automated to lead. In particular, for this purpose, the control unit 200 may determine control data for one or more longitudinal and / or lateral guide actuators 203 of the vehicle 110 (e.g., a steering, a drive motor, and / or a braking device of the vehicle)

Vehicle 110).

The control unit 200 may be configured to determine that an unstructured traffic situation 100 exists on the basis of the sensor data and / or on the basis of the digital map information. An unstructured traffic situation 100 can have one or more properties:

• one or more other traffic participants 112, 113 in the vicinity of the vehicle 110 do not obey at least one traffic rule valid for the environment in accordance with a highway code.

A relatively high level of attentiveness is required by a user of the vehicle 110 and / or a relatively high level of stress is caused to a user.

• The traffic situation 100 contradicts and / or exceeds that

(machine-learned) state of knowledge of the control unit 200;

• Traffic is primarily geared to free and less open spaces

Road markings; and or

• There may be relatively low speeds.

An unstructured traffic situation 100 may be caused by one or more causes:

• an undisciplined and / or threatening behavior of one or

a plurality of other traffic participants 112, 113, in particular a behavior that violates a traffic rule in accordance with a highway code; • a special road configuration, such as a roundabout

Merging lanes 101, 102, a construction site, a

Toll booth, etc .;

• a relatively high volume of traffic or relatively dense traffic; and or

• pedestrians and / or cyclists.

The control unit 200 may be configured to determine feature values for one or more features on the basis of the sensor data and / or on the basis of the digital map information. The one or more features can be considered

Dimensions of a feature vector are summarized. The value of a feature vector may then describe a present traffic situation 100. Exemplary features for detecting an unstructured traffic situation 100 are:

The driving speed of the ego vehicle 110, possibly a sliding one

Average of the driving speed and / or a standard deviation from an average of the driving speed.

• a weighted mean of the inverse distance to others

Participants 112, 113, 111, and possibly the standard deviation from the mean. In particular, a mean distance can be determined based on the following formula:

where di is the distance of the vehicle 110 to another

It is the traffic participant 112, 113, 111.

The total number of objects or passengers in the vicinity of the ego vehicle 110; while moving objects or

Travel participants incl. The direction of movement, stopped and / or parked objects or Vkehr participants are distinguished; furthermore, the moving average and / or the standard deviation of the total number can be considered. The number of lanes 101, 102.

• a measure of the quality or visibility of road markings.

• the number of two-wheelers and / or the number of two-man overtaking maneuvers.

A measure of how reliably the individual road users 112, 113 remain within their respective lane 102. In this case, the proportion of the width of a return participant 112, 113, which is located outside a lane 102, with the inverse of the distance d of this

Traffic participant 112, 113 are weighted. An exemplary measure is

• Time to collision with another party (Time to Collision (TTC)).

• emergency distance required by ego vehicle 1 10 for emergency braking; In this case, the initial distance to another traffic participant can be taken into account. Furthermore, it is possible to take into account the route that the other passenger travels during emergency braking. The control unit 200 may be configured to repeatedly and / or periodically determine a value of a feature vector having one or more of the above-mentioned. Features includes. It can thus be a temporal sequence of the values of the

Feature vector can be determined. Furthermore, the control unit 200 may be configured to determine, based on one or more values of the feature vector and on the basis of a classifier, whether an unstructured

If the situation is 100 or not. The classifier may be arranged to divide the space from possible values of the feature vector into subspaces for different classes. Possible classes are:

• the existence of a structured traffic situation; and

· The existence of an unstructured traffic situation 100. The class for the presence of an unstructured traffic situation 100 may possibly be subdivided into several subclasses, in particular depending on the cause of the unstructured traffic situation 100. Exemplary subclasses are:

The merging of several lanes 101, 102 as the cause of the unstructured traffic situation 100;

• a multi-lane roundabout as the cause of the unstructured

Traffic situation 100;

A disturbing object on one or both sides of a lane 102 as the cause of the unstructured traffic situation 100; and or

• a construction site as the cause of the unstructured traffic situation 100.

The classifier may have been learned during a learning phase. For this purpose, training data with a variety of

Training records are provided. A training data set comprises a value of the feature vector and a corresponding class. The classes may have been assigned manually if necessary. The classifier may include a neural network, a support vector machine, a decision tree (DT) and / or a random forest (RF) classifier.

The control unit 200 may thus be set up to determine whether an unstructured traffic situation 100 exists or not. Furthermore, the control unit 200 may be configured to assign an unstructured traffic situation 100 to a subclass and / or determine the cause of the unstructured traffic situation 100.

In the presence of an unstructured traffic situation 100, if necessary in

Dependence on the identified cause for the unstructured

In the event of a situation 100, at least one effective traffic rule will be established, which will result in the context of the unstructured traffic situation 100 has emerged. The effective traffic rule may differ from the one or more traffic regulations in force in accordance with the Highway Code. For example, in the unstructured traffic situation 100 shown in FIG. 1, it can be recognized that alternately a vehicle 112, 113 from the improvised lane 103 and a vehicle 110 from the second lane 102 turn onto the third roadway 105. Such an effective traffic rule can then be taken into account in the control of the (at least partially automated) ego vehicle 110. For example, an at least partially automated ego vehicle 110 may use the determined effective traffic rule to continue to maintain the ego vehicle 110 at least partially automated within the unstructured traffic situation 100.

An effective traffic rule can be determined on the basis of the sensor data and / or the digital map information. For this purpose can

Pattern recognition method can be used to determine patterns for the behavior of the vehicle participants 112 based on the time sequence of the sensor data.

113 within the unstructured traffic situation 100. An exemplary method for pattern recognition is described in Mehdiyev et al.

"Determination of Rule Patterns in Complex Event Processing Using Machine Leaming Techniques", Procedia Computer Science, Volume 61, 2015, pages 395-401. The contents of this document are incorporated by reference into this document. Alternatively or additionally, Bayes

Network models and / or deep leaming methods may be used to determine effective traffic rules within an unstructured traffic situation 100 based on the sensor data.

In determining an effective traffic rule, the identified cause of the unstructured traffic situation can be 100, the location of the unstructured traffic situation 100, the traffic density, the traffic speed, the Time of day, the day of the week and / or the weather are taken into account to determine the effective traffic rule with increased accuracy.

The vehicle 110 may be configured to communicate information regarding a detected unstructured via a communication unit of the vehicle 110

Traffic situation 100 and / or with respect to an effective traffic rule to a central processing unit, e.g. to a backend server. The information relating to the unstructured traffic situation 100 may then be e.g. be taken into account in the determination of a driving route, e.g. in order to be able to offer a user a route that is as stress-free as possible (in which the location of the unstructured traffic situation 100 is bypassed).

FIG. 3 shows a flowchart of an exemplary method 300 for

Control or for operating a function of an at least partially automated moving vehicle 110. The vehicle 110 may in particular be configured to drive at least highly automated. The function may include an output and / or a guidance function of the vehicle 110. The method 300 may be performed by a control unit 200 of the vehicle 110.

The method 300 includes determining 301 sensor data with respect to an environment of the vehicle 110. In particular, a temporal sequence of sensor data for a corresponding sequence of times can be determined. For example, sensor data with a time frequency of 1 Hz, 10 Hz or more can be detected and provided for further processing. On the basis of the temporal sequence of sensor data, the (temporal) behavior of one or more other vehicle participants 112, 113 in the environment of the vehicle 110 can be determined.

In addition, method 300 includes determining 302 based on

Sensor data, whether an unstructured traffic situation 100 is present or not. In this case, an unstructured traffic situation 100 may be present if one or more traffic participants 112, 113 in the vicinity of the vehicle 110 do not comply with the one or more traffic regulations for the surroundings of the vehicle 110 that are valid in accordance with a highway code. An unstructured

Income situation 100 may be based on this property and / or the o.g.

Properties are identified. The detection of an unstructured

Income situation 100 may be based on a machine-learned classifier.

The method 300 further comprises controlling 303 a function of the vehicle 110 depending on whether an unstructured

If the situation is 100 or not. For example, a

Guiding function of the vehicle 110 (for at least partially automated driving of the vehicle 110) are controlled or operated on the basis of the traffic regulations in force, if it has been determined that there is no unstructured traffic situation 100. On the other hand, at least one may be unstructured within a detected one

In the event of a situation 100 an existing traffic rule has to be identified and taken into account as part of the management function, if it has been determined that an unstructured traffic situation 100 exists.

By the measures described in this document, the control of an at least partially automated vehicle 110 can be improved. In particular, even in unstructured traffic situations 100, an at least partially automated driving of the vehicle 110 can be made possible. In this case, one or more effective traffic rules can be determined and used to control the vehicle 110. Possibly. For example, in an unstructured traffic situation 100, a transfer to a driver of the vehicle 110 may occur, wherein the driver may be notified of the one or more effective traffic rules in use (eg via a user interface of the vehicle 110). Furthermore, information regarding the existence of an unstructured Traffic situation 100 and / or transmitted in relation to a determined effective effective traffic rule to a central unit to take this information in the context of the control of another vehicle to take into account (eg for planning a route).

The present invention is not limited to the embodiments shown. In particular, it should be noted that the description and figures are intended to illustrate only the principle of the proposed methods, apparatus and systems.

Claims

claims

1) control unit (200) for an at least partially automated driving

Vehicle (110); wherein the control unit (200) is set up,

Sensor data relating to an environment of the vehicle (110)

determine;

- based on the sensor data to determine if an unstructured

The situation (100) is present or not; and

to control a function of the vehicle (110) depending on whether or not there is an unstructured traffic situation (100).

2) control unit (200) according to claim 1, wherein at an unstructured

In the traffic situation (100), at least one traffic rule valid in the environment of the vehicle (110) according to a highway code is not followed by at least one other traffic participant (112, 113), in particular by a plurality of other traffic participants (112, 113) becomes.

3) control unit (200) according to one of the preceding claims, wherein the control unit (200) is set up,

based on the sensor data at least one value of

To determine feature vector with a plurality of features; and

determine whether or not there is an unstructured traffic situation (100) based on the value of the feature vector and using a machine-learned classifier; wherein the classifier is arranged to identify, within a total of possible values of the feature vector, at least one subspace for the presence of an unstructured traffic situation (100). 4) control unit (200) according to claim 3, wherein the plurality of features comprises one or more,

a travel speed of the vehicle (110);

a distance with respect to a distance of the vehicle 110 to one or more other traffic participants 112, 113 in the vicinity of the vehicle 110; and or

a lane gauge indicating the extent to which one or more other road users (112, 113) travel in the vicinity of the vehicle (110) within at least one defined traffic lane (101, 102).

5) Control unit (200) according to one of the preceding claims, wherein the control unit (200) is set up,

based on the sensor data one cause of a plurality of

possible causes for the unstructured traffic situation (100); and

- to control the function of the vehicle (110) as a function of the determined cause for the unstructured traffic situation (100).

6) control unit (200) according to one of the preceding claims, wherein the control unit (200) is set up when it has been determined that an unstructured traffic situation (100) is present,

- based on the sensor data one within the unstructured

Transient situation (100) to determine effective traffic rules; and

- To control the function of the vehicle (110) in dependence on the determined, effective traffic rule.

7) Control unit (200) according to claim 6, wherein the control unit (200)

is set up when it has been determined that an unstructured

Situation (100), via a user interface of the vehicle (110) output a takeover request to a user of the vehicle (110) to at least partially manually guide the vehicle (110); and

- Output to the user via the user interface of the vehicle (110) information relating to the determined effective effective traffic rule.

8) Control unit (200) according to one of the preceding claims, wherein the control unit (200) is set up,

a temporal sequence of sensor data for a corresponding one

To determine the sequence of times; and

Determine whether an unstructured traffic situation (100) exists or not on the basis of the temporal sequence of sensor data.

9) Control unit (200) according to one of the preceding claims, wherein

the function comprises an output function for outputting information relating to the unstructured traffic situation (100) to a user of the vehicle (110); and or

- The function of a leadership function to at least partially

automated guidance of the vehicle (110); and or

- The function comprises an output function for sending information relating to the unstructured traffic situation (100) to a receiver outside the vehicle (110).

10) Method (300) for controlling a function of at least partially

automated moving vehicle (110); the method (300) comprising

Determining (301) sensor data relating to an environment of the

Vehicle (110);

Determining (302), based on the sensor data, whether or not there is an unstructured traffic situation (100); and - controlling (303) a function of the vehicle (110) depending on whether or not there is an unstructured traffic situation (100).