Method for selecting an automated driving process by means of a driving assistance system

Abstract

The invention relates to a method for selecting an at least partially automated driving process by means of an autonomous driving assistance system of a vehicle.

Description

TECHNICAL FIELD

[0001] The invention relates to a method for selecting an at least partially automated driving process by means of a driving assistance system of a vehicle. BACKGROUND

[0002] Driving assistance systems are known which can calculate a driving trajectory for an automated movement of the vehicle, i.e. for example without a steering intervention by the driver and / or without actuation of the accelerator pedal or brake pedal by the driver, on the basis of a current driving scenario. In particular, driving assistance systems are known which comply with autonomous driving level 2 (SAE level 2), which have the ability to keep to a lane, but cannot change lanes without initiation by the driver. If the driver initiates a change of lane actively, the driving assistance system can then execute the change of lane automatically.

[0003] Depending on the respective driving scenario, different driving processes can be initiated. For example on a multi-lane road, the speed can be reduced at different points in time or at different distances from a slower other vehicle driving ahead when approaching this other vehicle. This depends, inter alia, on whether it can be assumed that the driver of the own vehicle with a driving assistance system which complies with autonomous driving level 2 or a lower level will initiate a change of lane actively.

[0004] In another driving scenario, the possible traffic behavior of other actors decides whether it is advantageous to initiate or not to initiate a certain driving process. That is to say, it is not a trigger which is not influenced by the driving assistance system which is triggered by the driver of the own vehicle, but rather by other traffic participants, such as other vehicles, pedestrians, cyclists, etc., or traffic regulating units, such as traffic lights, barriers, etc.

[0005] If it is expected that the driver will initiate a change of lane actively, it is advantageous not to reduce the speed too early, since this leads to an unnecessary speed reduction from the perspective of the human driver. In contrast, if it is expected that the driver will not initiate a change of lane, it is advantageous to reduce the speed too early in order to prevent a too close distance to the other vehicle driving ahead while driving and thus a harsh braking which is triggered thereby. The same applies analogously to driving scenarios in which the traffic behavior of other actors decides which driving process is advantageous in terms of safety and / or comfort, respectively. SUMMARY

[0006] Against this background, the object of the present invention is to provide a method for selecting an at least partially automated driving process by means of a driving assistance system, which enables a control of the vehicle in a manner which is as similar as possible to human driving behavior.

[0007] According to a first aspect, the invention relates to a method for selecting an at least partially automated driving process by means of a driving assistance system of a vehicle. The method comprises the following steps:

[0008] First, the current driving scenario (i.e., the current driving situation of the vehicle) is determined. This scenario could be, for example, "The vehicle is traveling in the right lane of a two-lane road, the left lane is empty, and there is another vehicle traveling at a slower speed 150 meters ahead." Based on this current driving scenario, the driver assistance system determines several possible future driving maneuvers. These maneuvers could be, for example, constant-speed lane changing, constant-speed lane keeping, and deceleration while keeping in lane. Of course, these maneuvers can be initiated at different times.

[0009] Next, a decision structure (hereinafter also referred to as a decision tree) is automatically created using the computing unit of the driver assistance system. Here, multiple decision levels (hereinafter also referred to as nodes) of the decision tree are determined, with each node correspondingly associated with a specific future point in time. Depending on the driving scenario, these nodes are at least partially assigned with at least one decision regarding lane changing or traffic behavior made by the vehicle driver or by other actors. Other actors can be, for example, other vehicles, pedestrians, cyclists, or other objects capable of performing traffic behaviors (e.g., lane changing, direction change, entering a lane, etc.).

[0010] Furthermore, these nodes are at least partially assigned at least two decision or action options to be made by the driver assistance system itself. Thus, according to the decision tree, multiple decisions can be made at the points in time associated with the nodes of the decision tree. These decisions can be initiated, in particular, by the driver of the vehicle (changing lanes), by other actors (e.g., changing lanes of other vehicles), or by the driver assistance system itself (e.g., slowing down to maintain lane, not slowing down to maintain lane, or braking hard to avoid a collision).

[0011] The sub-driving processes are defined by the decisions made at each node, meaning these sub-driving processes are components of a driving process with a relatively long duration. In other words, the driving process consists of multiple sub-driving processes assigned to branches of the decision tree. Therefore, the path through the decision tree (e.g., passing through multiple nodes) consists of multiple decisions and sub-driving processes associated with those decisions.

[0012] Then, driving trajectories are calculated, where each trajectory is assigned to a path defined by a decision tree, and each path contains one or more successive decisions to be made. The corresponding driving trajectory defines the vehicle's motion path, for example, in a two-dimensional coordinate system, and preferably also determines the vehicle's speed as it traverses that trajectory. When calculating the driving trajectory, longitudinal and lateral accelerations, or longitudinal and lateral jerk, can also be determined. Therefore, in this step, multiple driving trajectories that can be traversed in the current driving scenario are calculated, more precisely, depending on the corresponding decisions that can be made based on the decision tree.

[0013] In addition, evaluation indicators are calculated for the driving trajectories. Here, the evaluation indicators are measures of the comfort and / or safety of the respective driving trajectory. The evaluation indicators are preferably numerical, so that the driving trajectories can be compared with one another in terms of comfort and / or safety by comparing the evaluation indicators.

[0014] In addition, probability values are calculated which are assigned to the nodes of the decision tree. The probability values indicate the probability of the vehicle driver or other actor making a decision to change lane or another traffic action at the point in time assigned to the node, respectively. In other words, the probability values thus indicate how great the likelihood is that the vehicle driver or other actor makes a decision at the point in time assigned to the node, to which the driving assistance system of the host vehicle reacts by providing a driving trajectory.

[0015] Next, a plurality of decision indicators is calculated, wherein each decision indicator is assigned a node. The calculation of the decision indicators is based at least in part on the probability values of the respective node which indicate the probability of the vehicle driver or other actor making a decision to change lane or a traffic action at the point in time assigned to the node. In addition, the calculation of the decision indicators is based at least on the evaluation indicators of the driving trajectories determined by the decisions made at the point in time of the respective node.

[0016] Finally, based on the decision tree, the decision indicators and depending on whether the driver or other actor of the vehicle initiates a change of lane or a traffic action at the point in time assigned to the node, an at least partially automated driving process is selected by the autonomous driving assistance system. In particular, if the decision of the host vehicle driver or the decision of the other actor has not yet forced the start of another driving process in advance, the decision tree will run to the node with the lowest decision indicator and at this point the decision to stay in the lane while slightly decelerating (for example at least 80% or 90% less than in the case of full braking) is made automatically by the driving assistance system.

[0017] The technical advantage of the method according to the application is that by taking into account the probability of initiating a change of lane or a traffic action and the trajectory characteristics in terms of comfort and / or safety, a driving process is automatically selected which enables a driving experience as natural as possible, which tends towards human driving behavior. In other words, in other words, a driving process is selected which corresponds very well to the driving behavior of a human driver.

[0018] According to an embodiment, for selecting the at least partially automated driving process, the values of the decision indicators of the nodes of the decision tree are compared. The selected driving process is based on a driving trajectory which initiates a lane change or a deceleration maneuver at the point in time assigned to the node with the smallest value of the decision indicator. In other words, the decision indicators of the nodes are compared and it is determined which node is assigned the smallest value of the decision indicator. The smallest value of the decision indicator indicates that, taking into account the probability of the driver initiating a lane change at the point in time of the node and the evaluation indicator of the driving trajectory of the node, it is advantageous to initiate a lane change at this point in time (which must be initiated actively by the driver of the vehicle or another vehicle) or to keep the lane and to reduce the vehicle speed (which is initiated alone by the driving assistance system).

[0019] It should be noted that a lane change is preferably initiated actively only by the driver. The lane change itself can be performed automatically by the driving assistance system.

[0020] According to an embodiment, the decision indicators of the nodes are calculated at least partially by adding the result of a first multiplication and the result of a second multiplication. In the first multiplication, the value of the probability of the driver of the vehicle or the other actor initiating a lane change or a traffic maneuver at the point in time assigned to the node is multiplied by the evaluation indicator of the driving trajectory by means of which the lane change is initiated. In the second multiplication, the value of the probability of the driver of the vehicle or the other actor not initiating a lane change or a traffic maneuver at the point in time assigned to the node is multiplied by the evaluation indicator of the driving trajectory by means of which the lane is kept and the vehicle is decelerated. Thus, for calculating the decision indicator assigned to a node and thus to the respective point in time of the decision, the evaluation indicator of the driving trajectory selected at the point in time and the probability of selecting the driving trajectory based on the action of the driver of the vehicle or the other actor should both be taken into account.

[0021] According to an embodiment, the decision made by the driver of the vehicle is detected by switching on a turn signal and / or taking a steering intervention which leads to a lane change. Thus, the driving assistance system can recognize when the driver wants to change lanes.

[0022] According to an embodiment, the decision made by the other actor is detected by sensing a turn signal and / or a lane change of the other vehicle, a turn of the other vehicle, a start and cut-in of the other vehicle into the lane of the vehicle, or a person entering the lane. This can be detected, for example, by a sensor system suitable therefor, for example with a camera, a radar sensor and / or a lidar sensor. Thus, the driving assistance system can recognize whether the other vehicle is changing lanes or whether the other actor is performing a traffic maneuver which can require a driving process to be performed by the driving assistance system.

[0023] According to one embodiment, the calculation of the driving trajectory comprises calculating the multi-dimensional coordinates the vehicle moves through when moving over the driving trajectory, calculating longitudinal and lateral acceleration values, and / or calculating longitudinal and lateral jerk values. Thereby, it can be determined whether the trajectory is collision-free and which forces act on the vehicle's occupants when the vehicle moves over the driving trajectory. These are important parameters for the comfort and safety impression of the driving trajectory.

[0024] According to one embodiment, the calculation of the evaluation measure of the driving trajectory involves the distance of the vehicle to other vehicles and / or objects, the difference to a set speed, and / or the frequency of lane changes. These are also important parameters for the comfort and safety impression of the driving trajectory.

[0025] According to one embodiment, the driving trajectory corresponds to a path through the decision tree, wherein the driving trajectory is based on the decisions assigned to the path of the decision tree. The driving trajectory can consist of several sub-trajectories, for example initially keeping the lane at constant speed and only later changing the lane.

[0026] According to one embodiment, the probability value of the driver initiating a lane change at a future point in time assigned to the node is calculated based on a neural network trained by means of data obtained from the driver's past driving. Thereby, the probability of a lane change can be estimated or calculated from statistical relationships obtained in the past.

[0027] According to one embodiment, the probability value of the driver initiating a lane change at a future point in time assigned to the node is calculated based on an evaluation measure of the driving trajectory initiated by means of the decision made at the respective node's point in time. When a driving trajectory of a lane change is available that is sufficiently good in terms of comfort and / or safety, a human driver will typically initiate a lane change. The evaluation measure of the driving trajectory can thus be used to determine or estimate the probability of a lane change being initiated by the driver.

[0028] According to one embodiment, the probability value of the driver initiating a lane change at a future point in time assigned to the node is calculated based on driver behavior sensed by at least one sensor. For example, the sensor can sense whether the driver has looked into the side mirror, whether a Schulterblick has been made, whether the driver has been looking only forward into his own lane for a long time, etc. Thereby, the probability of a lane change can be calculated or estimated based on the sensor.

[0029] According to one embodiment, for at least a part of the nodes, a plurality of decision indicators is calculated accordingly, which are based on different probability values for the driver of the vehicle or for another actor to make a decision for a lane change or a traffic action at a point in time assigned to the node; and for each node, the maximum decision indicator is used accordingly to select the at least partially automated driving procedure. The probability values related to the lane change or the traffic action have a fuzziness, the further in the future the point in time to which the probability value relates, the greater the fuzziness. By calculating the decision indicators based on the different probability values, a driving procedure can be selected which achieves an optimized comfort and / or safety even in the case of inaccurate probability values.

[0030] According to one embodiment, the decisions made by the driver of the vehicle include a lane change from the right lane to the left lane or a lane change from the middle lane to the left lane or to the right lane.

[0031] According to another aspect, the invention relates to a driving assistance system for a vehicle, which is designed to select an at least partially automated driving procedure from a plurality of possible driving procedures. The driving assistance system is designed to perform the following steps:

[0032] a) determining a current driving scenario by means of the driving assistance system and determining a plurality of different driving procedures possible in the future based on the current driving scenario;

[0033] b) creating a decision tree automatically by means of a computing unit of the driving assistance system by determining a plurality of nodes of the decision tree, wherein the nodes are associated with points in time in the future, wherein the nodes are at least partially assigned at least one decision for a lane change or a traffic action made by the driver of the vehicle or by another actor depending on the driving scenario, and wherein the nodes are at least partially assigned at least two decision or action options made by the driving assistance system itself;

[0034] c) calculating driving trajectories by means of an automated trajectory planner, wherein each driving trajectory is assigned to a path defined by the decision tree and each path contains one or more decisions made successively;

[0035] d) calculating evaluation indicators for the driving trajectories automatically, wherein the evaluation indicators are a measure of the comfort and / or safety of the respective driving trajectory;

[0036] e) calculating probability values, wherein the probability values are assigned to the nodes of the decision tree and indicate the probability for the driver of the vehicle or for another actor to make a decision for a lane change or a traffic action at a point in time assigned to the node;

[0037] f) calculating a plurality of decision indicators, wherein a decision indicator is assigned to a node, wherein the calculation of the decision indicator is based at least partially on:

[0038] - a probability value of the respective node indicating that the driver or other actor of the vehicle makes a decision to change lane or traffic behavior at the point in time assigned to the node, and

[0039] - at least one evaluation indicator of the driving trajectory determined by the decision made at the point in time of the respective node;

[0040] g) selecting, by the autonomous driving assistance system, an at least partially automated driving process based on the decision tree, the decision indicators and depending on whether the driver or other actor of the vehicle initiates a change of lane or traffic behavior at the point in time assigned to the node.

[0041] According to another aspect, the invention relates to a vehicle having a driving assistance system according to one of the preceding embodiments.

[0042] In the sense of the invention, a "decision tree" is to be understood as a decision structure having a plurality of decision levels, also referred to as nodes. These decision levels are assigned to different points in time in the future. At these decision levels, decisions are made which can be made by the driving assistance system of the own vehicle itself or by the driver, or by the driving assistance system or driver of other vehicles.

[0043] In the sense of the invention, an "other actor" is to be understood as any actor who is able to perform an action which can influence the driving process of the own vehicle. This is, for example, another vehicle, a bus departing from a bus stop, a pedestrian who can enter the lane, a cyclist who can change lane from a bicycle lane to a motor vehicle lane, a car which can turn off at a junction, however also a traffic light or a barrier.

[0044] In the sense of the invention, a "traffic behavior" is to be understood as any behavior of the driver or other actor of the own vehicle which restricts the action options or decision freedom of the driving assistance system. For example, a person entering the lane restricts the action options of the driving assistance system in such a way that a driving trajectory of "constant speed lane keeping" can no longer be implemented, but rather a change of lane or braking is attempted.

[0045] In the sense of the invention, the terms "approximately", "essentially" or "about" mean a deviation of + / - 10%, preferably of + / - 5% from the respective exact value, and / or a form of change which is unimportant for the function.

[0046] The refinements, advantages and application possibilities of the invention are also given by the following description of embodiments and the figures. Here, all described and / or illustrated features are in principle the subject of the invention, both individually and in any combination. BRIEF DESCRIPTION OF DRAWINGS

[0047] The application will be explained in detail below with reference to the drawings. In the drawings:

[0048] Figure 1 An exemplary and schematic illustration of a vehicle in a driving scenario in which a decision about changing lane or decelerating to stay in lane is to be made;

[0049] Figure 2 An exemplary illustration of a decision tree which can select a plurality of possible driving processes for Figure 1 the driving scenario shown;

[0050] Figure 3 An exemplary illustration of a schematic diagram of a driving process according to Figure 2 the decision tree in

[0051] Figure 4 An exemplary illustration of a decision tree according to Figure 2 , with probability values assigned to the decisions;

[0052] Figure 5 An exemplary illustration of a decision tree according to Figure 4 , with evaluation metrics for the driving trajectories and decision metrics assigned to the nodes;

[0053] Figure 6 An exemplary and schematic illustration of a block diagram of a driving assistance system for performing the selection method; and

[0054] Figure 7 An exemplary and schematic illustration of a flow chart illustrating the steps of a method for selecting an at least partially automated driving process.

[0055] LIST OF REFERENCE SIGNS:

[0056] 1 vehicle

[0057] 2, 2' other vehicle

[0058] 3 decision tree

[0059] 3a trunk

[0060] 3b branch

[0061] 3.1 node

[0062] 3.2 decision

[0063] 3.3 driving trajectory

[0064] 10 driving assistance system

[0065] 11 control unit

[0066] 12 environment model unit

[0067] 13 sensor

[0068] 14 driving process planner

[0069] 15 vehicle motion controller

[0070] d distance DETAILED DESCRIPTION

[0071] Figure 1 A driving scenario is exemplarily shown, in which a vehicle 1 (also referred to as ego vehicle) is driving on a multi-lane road on which one or more other vehicles 2, 2’ are also driving.

[0072] The driving scenario is characterized, inter alia, in that the other vehicle 2’ is driving on the same lane in front of the vehicle 1 but at a lower speed, so that the distance d between the vehicles 1, 2’ is getting smaller and smaller.

[0073] The driving situation can also be characterized in that a further other vehicle 2 is approaching from behind on the left lane of the road.

[0074] In such a driving scenario, a plurality of driving processes can be envisaged. First, the vehicle 1 can initiate a lane change, as indicated by the arrow, or the vehicle can not change lanes but reduce the driving speed and follow the other vehicle 2 driving in front.

[0075] The two basic driving processes can be initiated at different points in time. Thus, the vehicle 1 can for example already slightly reduce the speed when there is still a larger distance d to the other vehicle 2’, so as to slowly drive towards the other vehicle 2’ driving in front. Alternatively, the deceleration process can be initiated later, which requires a more drastic deceleration. In the same way, the lane change of the vehicle 1 can be initiated at different points in time and thus at different distances d to the other vehicle 2’.

[0076] In a vehicle equipped with a driving assistance system in accordance with autonomous driving level 2 (SAE-2 level), the lane change must be initiated by the driver actively, i.e. the autonomous driving assistance system cannot initiate the lane change independently, but only if the driver has actively initiated the lane change, for example by switching on the turn signal or by taking a steering intervention that leads to the lane change. After such an initiation of the lane change, the lane change can then be executed automatically.

[0077] The method described hereinafter is able to execute a driving process that is as comfortable as possible and adapted to the natural driving behavior of a human driver depending on the probability that the driver initiates a lane change at a certain point in time. This means, inter alia, that driving processes that require sudden braking or sudden lane changes can be effectively avoided.

[0078] Figure 2 A decision tree 3 is shown, based on which an at least partially automated driving process can be selected.

[0079] The decision tree 3 comprises a trunk 3a along which a plurality of nodes 3.1 are arranged. From the nodes 3.1 branches 3b branch off. Here, each branch 3b corresponds to a decision 3.2. Each branch 3b of the decision tree ends in a so-called leaf, which corresponds to a driving trajectory 3.3. In other words, one or more decisions 3.2 thus made result in a specific driving trajectory 3.3. The driving trajectory is automatically calculated by a trajectory planner and indicates a trajectory in which the vehicle 1 is to be moved automatically when the decisions 3.2 resulting in the driving trajectory 3.3 have been made successively.

[0080] The nodes 3.1 are respectively associated with time points to to t3, which successively occur in chronological order starting from the current time point (“now”). Thus, the trunk 3a of the decision tree 3 forms a time axis with its nodes 3.1. The time points to to t3 associated with the nodes 3.1 indicate at which time points which decisions associated with the nodes 3.1 can be made.

[0081] The decision tree 3 thus depicts the decisions 3.2 made at different time points and the driving processes resulting therefrom together with the driving trajectories 3.3. The entire decision tree can cover a time period of, for example, 5 to 15 seconds in the future, the time period between two nodes can be, for example, between 1 and 5 seconds, preferably 1, 2 or 3 seconds.

[0082] In the decision tree 3 according to Figure 2 at the time point to, for example, the decisions “change lane and maintain speed”, “maintain lane and maintain speed” and “maintain lane and reduce speed” can be made.

[0083] A part of the decisions 3.2 made at the nodes must be initiated, for example, by the driver himself. Another part of the decisions 3.2 made can be initiated by the autonomous driving assistance system itself. For example, the decision 3.2 “change lane” must be initiated by the driver of the vehicle 1, while the decisions 3.2 to maintain the lane can be initiated by the autonomous driving assistance system itself, i.e. without the driver having to initiate this in particular (e.g. switching on the turn signal).

[0084] Figure 3 The possible driving processes exhibited in the decision tree according to Figure 2 are shown graphically. When the vehicle 1 moves towards the other vehicle 2’, the change of lane can be initiated at different time points and thus at different distances between the vehicles. For example, if the change of lane is initiated at the time points to, ti or t2, the driving trajectories Trji, Trj3, Trj5 correspond to the change of lane trajectory.

[0085] It should be mentioned that a driving trajectory does not only relate to one decision, but rather corresponds to the complete path through the decision tree 3 from "now" to the respective "leaf" and thus comprises a succession of a plurality of decisions 3.2 made successively. For example, the prerequisite for applying the driving trajectory Trj5 is that the decision "keep lane and keep speed" was made at the time points t0 and t1 respectively and that the driver actively initiated a lane change at the time point t2.

[0086] The driving trajectories Trj2, Trj4, Trj6 relate to the fact that the driver did not actively initiate a lane change and initiated a deceleration at different time points t0 to t2.

[0087] If the driver does not initiate a lane change and does not previously reduce the driving speed, then finally an emergency braking of the vehicle 1 is automated by an emergency braking assistant of the vehicle 1.

[0088] It is described below how a driving procedure can be selected from a large number of possible driving procedures which is improved in terms of comfort and safety, at least partially automated, depending on the probability with which the driver actively initiates a lane change in the respective driving situation.

[0089] Figure 4 It is shown that a decision tree 3 according to Figure 2 is shown, wherein probability values are additionally assigned to the branches 3b of the decision tree 3. The branches 3b which relate to the decisions 3.2 regarding the lane change initiated by the driver are in particular assigned the probability values p0, p1, p2. Here, the probability values are associated with the nodes respectively and thus indicate with which probability p0 to p2 the driver makes the decision to change lane at the time points t0 to t2.

[0090] Thereby, in the shown embodiment, the driver assistance system can make one of the two further decisions, i.e. keep lane and decelerate or keep lane and keep speed unchanged, at the respective time points t0 to t2 with the probability 1-p0, 1-p1, 1-p2.

[0091] The probabilities p0 to p2 with which the driver initiates a lane change are pre-calculated or estimated for all nodes 3.1 or time points t0 to t2.

[0092] Here, the probabilities p0 to p2 to change lane at the time points t0 to t2 can be calculated in different ways. Some examples of calculating these probabilities are listed below. Of course, there are also different possibilities of calculating the probabilities.

[0093] Information from the driving history of the respective driver, i.e. in particular information from past lane change events, can be used to determine the probability of a lane change at a specific point in time. In particular, these information from the driving history of the respective driver can be used as training data for a machine learning system, in particular a neural network, which thus learns the statistical relationship of when a driver initiates a lane change in a pre-given driving scenario. After the training, the machine learning system can be used to calculate the probability of a lane change at future points in time in a current driving scenario.

[0094] According to another alternative, the evaluation indicators assigned to the individual driving trajectories are used to estimate the probabilities po to p2 of a lane change at the points in time to to t2. The evaluation indicators assigned to a driving trajectory are indicative of the comfort (e.g. longitudinal and lateral accelerations below a pre-given threshold, longitudinal and lateral jerk below a pre-given threshold, frequency of lane changes) and safety (e.g. distance to surrounding other vehicles and / or objects) of this driving trajectory. The calculation of the probabilities po to p2 of a lane change at the points in time to to t2 based on these evaluation indicators is done in the background that a human driver usually performs a lane change when he can do so with high comfort and high safety. Thus, the probability of a lane change at a point in time at which a driving trajectory with the highest comfort or the highest safety is available is the greatest of all possibilities.

[0095] According to another alternative, other available information can be used to calculate the probability of a lane change. For example, information that can be used is:

[0096] - whether the turn signal has been switched on;

[0097] - whether it can be determined by means of a driver camera that the driver is looking at the side mirror pointing to the target lane;

[0098] - whether it can be determined by means of an environmental recognition device (i.e. one or more sensors) that the target lane is free;

[0099] - whether it can be determined by means of a driver camera that the driver is performing a craning look in the direction of the target lane;

[0100] - whether a navigation system suggests a lane change to be advantageous for reaching the navigation destination.

[0101] According to another alternative, other available information can be used to calculate the probability of a lane change. For example, information that can be used is:

[0102] - whether it can be recognized by means of a driver monitoring camera that the driver is looking only at the current lane for a long time;

[0103] - whether it can be recognized by means of a driver monitoring camera that the driver is looking only at an obstacle on the current lane for a long time;

[0104] - whether the navigation system suggests a route that keeps to the lane for reaching the navigation destination.

[0105] It is also possible to combine these listed examples to calculate the probability po to p2 of changing lane at the time points to to t2.

[0106] As mentioned above, for the decision made according to the decision tree 3, at least partially a driving trajectory is calculated. On the one hand, this driving trajectory indicates the movement trajectory of the vehicle in space, for example based on two-dimensional coordinates. On the other hand, also information such as the velocity in the x-direction and the y-direction, the longitudinal acceleration and the lateral acceleration, or the longitudinal jerk and the lateral jerk, is assigned to the driving trajectory. Thereby, it is possible to determine the comfort of the driver driving on this trajectory.

[0107] Furthermore, also other information can be taken into account when calculating the driving trajectory, for example information containing the distance of the vehicle to surrounding third party vehicles or objects.

[0108] It should be noted that in order to be considered in the decision tree 3, the calculated driving trajectories must be collision-free.

[0109] The planning of the driving trajectories can be performed by a trajectory planner. This trajectory planner can calculate trajectories for different driving maneuvers. These driving maneuvers can be, among others:

[0110] - keeping to the lane;

[0111] - changing lane;

[0112] - decelerating while keeping to the lane;

[0113] - keeping to the lane without deceleration for x seconds and then decelerating, wherein x is an integer.

[0114] Furthermore, there is also provided a unit that is able to evaluate the planned driving trajectories in terms of comfort and safety. Preferably, an evaluation indicator is calculated that is assigned to the respective driving trajectory and that indicates the comfort and safety of the driving trajectory in a numerical form. Thus, it is possible to determine, by means of the evaluation indicator, how the driver of the vehicle perceives the comfort and safety when driving over the respective driving trajectory.

[0115] For example, a first driving trajectory that is driven over with low longitudinal and lateral acceleration has a lower evaluation indicator than a second driving trajectory that occurs with high longitudinal and lateral acceleration.

[0116] Likewise, a driving trajectory in which the vehicle is decelerated slowly has a lower evaluation indicator than a driving trajectory in which the vehicle is decelerated suddenly.

[0117] In other words, the evaluation index of the driving trajectory thus measures the comfort and safety perception of the human driver with respect to the driving trajectory.

[0118] Figure 5 A decision tree according to Figure 4 is shown, wherein the evaluation indices C sw0 , C sw1 , C sw2 , C sh0 , C sh1 , C sh2 are assigned to the individual trajectories Trj1 to Trj6, respectively. Here, the subscript "sw" stands for a lane change, and the subscript "sh" stands for a lane keeping. Thus, the evaluation index C sw0 forms, for example, an evaluation index for the trajectory Trj1, the evaluation index C sh0 forms, for example, an evaluation index for the trajectory Trj2, and so on.

[0119] Based on the probabilities of a lane change (p0 to p2) and the evaluation indices of the driving trajectories, a decision index can be calculated. Here, each decision index is preferably assigned to one node 3.1, respectively. In Figure 5 , the decision index of the first node is labeled Cost_0, the decision index of the second node is labeled Cost_1, and the decision index of the third node is labeled Cost_2.

[0120] Based on a comparison of the decision indices of the different nodes 3.1, the driver assistance system can judge, for example, that, for a situation in which the driver does not initiate a lane change at a certain node, it is more advantageous in terms of the comfort and safety perception of the driver to keep the lane and initiate a deceleration or to keep the lane without deceleration.

[0121] The calculation of the decision index Cost_x (x stands for a node index or a time point index) of the different nodes 3.1 can be performed, for example, according to the following formula:

[0122] Cost_x = p x *C swx +(1-p x )*C shx

[0123] Here:

[0124] Cost_x: decision index of the respective node;

[0125] p x : probability of the driver initiating a lane change at the time point tx;

[0126] C swx : evaluation index of the lane change trajectory at the node x;

[0127] C shx: evaluation measure of the keep-lane trajectory at node x;

[0128] Furthermore, the emergency braking option is assigned the highest possible evaluation measure or decision measure value C NB (NB: emergency braking).

[0129] After the decision measure Cost_x has been calculated for all nodes 3.1, the node with the lowest decision measure value is determined first. For example, if the decision measure Cost_0 of node 0 is 40, the decision measure Cost_1 of node 1 is 30, the decision measure Cost_2 of node 2 is 50, and the decision measure Cost_3 of node 3 is 100, then node 1 is selected, since this node has the lowest decision measure value. NB

[0130] Here, the lowest decision measure value means that the costs of the decisions or the costs of the driving trajectories assigned to the decisions are lowest, taking into account the probability that the driver initiates a lane change at the node assigned to this lowest decision measure value.

[0131] Alternatively, the lowest decision measure value can mean that the expected value of the costs at this node is lowest. This can be due to the fact that all leaves of the branch of this node have trajectories with low costs, or due to the fact that trajectories with low costs have a higher probability than trajectories with high costs.

[0132] The decision tree 3 is then run in accordance with the driving behavior of the driver and the decision measures of the nodes 3.1. For example, at node 0 (time point to) with a decision measure value higher than node 1 (time point ti), if the driver does not initiate a lane change, the decision tree continues to node 1 (i.e. not driving over trajectory Trj2), since at node 1 a driving operation can be provided which is more advantageous in terms of comfort and safety.

[0133] If the driver now also does not initiate a lane change at node 1, a decision is made in accordance with the lowest decision measure value: the automated driving assistant system initiates trajectory Trj4, i.e. keep lane and decelerate, at time point ti.

[0134] In other words, if the driver does not initiate a lane change at time point to and thus drives over trajectory Trj1, at time point ti either trajectory Trj3 is initiated (in particular if the driver initiates a lane change), or trajectory Trj4, i.e. keep lane and decelerate, is initiated.

[0135] The above-described method for selecting an at least partially automated driving procedure can be modified as required as follows:

[0136] ​As described above, the decision metrics (Cost_0 to Cost_2) assigned to the node 3.1 are influenced by the probability values (p0 to p2) that the driver initiates a lane change at the specific time point assigned to the node 3.1. These probability values have inaccuracies. The further in the future a node or a time point assigned with a probability value is, the greater the inaccuracy of this probability value is.

[0137] It is therefore proposed to determine the decision metrics for different probability biases of the calculated original probability values, for example for a first scenario without bias, for a second scenario with a +20% bias, for a third scenario with a -20% bias, for a fourth scenario with a +30% bias, for a fourth scenario with a -30% bias.

[0138] Then, the decision metrics of the nodes can be calculated for all these scenarios, and the decision metric with the maximum value can be selected for each node. Then, the at least partially automated driving process is selected on the basis of the decision metrics of the nodes with these maximum values.

[0139] According to a further embodiment of the application, it can be decided at a node when running the decision tree 3 whether a trajectory planning should be carried out again on the basis of the current environmental information and thus the decision tree 3 is regenerated, and whether possible trajectories, probabilities, evaluation metrics of the driving trajectories, and decision metrics should be calculated. As a condition, a collision-free trajectory should be provided for the respective decision at least at two time points / nodes in the future.

[0140] The application is described above by way of example, in which the vehicle changes lane to the left in order to start an overtaking process. Of course, the application can also be applied to other driving scenarios in which the vehicle changes lane to the right. It is also conceivable to apply the application to driving scenarios in which both a change of lane to the left and to the right is possible.

[0141] Furthermore, the application can also be applied to driving scenarios in which an automated driving process is carried out depending on whether it can be expected that other vehicles will cut into the own lane or not. In other words, it is not the probability of the driver of the own vehicle (the ego vehicle) initiating a lane change that is predicted, but the probability of other vehicles initiating a lane change.

[0142] The probability values associated with the nodes of the decision tree therefore indicate the likelihood of other vehicles initiating a lane change into the lane of the ego vehicle. The decision tree can be used to decide whether to initiate a driving process when other vehicles change lanes or whether it is better to slow down moderately in order to prepare for a lane change of other vehicles or to maintain the speed and initiate a more drastic braking later when changing lanes.

[0143] Figure 6 A schematic diagram of a driving assistance system for carrying out a driving process implemented at least partially automatically is shown.

[0144] The driving assistance system preferably has a central control unit 11 which provides the computing and storage means for performing the processes described below. Alternatively, a distributed computer architecture can also be provided, or the processes can be performed or computed on the computing units of the sensors, as long as the sensors provide sufficient computing power.

[0145] The control unit 11 has an environment model unit 12 which provides an environment model of the vehicle environment. This environment model unit 12 receives information from one or preferably multiple sensors 13. The sensors 13 can be, for example, cameras, radar sensors, lidar sensors, ultrasonic sensors, etc. The environment model unit 12 generates an environment model from the information of the sensors 13, which is provided to a driving process planner 14. The driving process planner 14 is a planner which can calculate one or multiple driving trajectories on the basis of the environment model, more precisely preferably which can calculate multiple driving trajectories which can be applied in the current driving scenario and which can be executed without a collision from the current driving scenario.

[0146] Preferably, the driving process planner 14 also calculates evaluation indices for the respective driving trajectories. The aforementioned method for selecting a driving process can also be performed in the driving process planner 14.

[0147] If a driving trajectory on which the vehicle should move is determined, the corresponding information is then transmitted to a vehicle movement controller 15. This vehicle movement controller 15 can be provided in the control unit 11 or can also be provided in a distributed manner (i.e. provided in a separate control unit). The vehicle movement controller 15 generates output signals for the steering device, the brake system and the powertrain.

[0148] Figure 7 A flowchart of a method for selecting at least partially automated driving processes by means of a driving assistance system of a vehicle is schematically shown.

[0149] First, the current driving scenario is determined (S10) by the driving assistance system and on the basis of this current driving scenario a plurality of different driving processes possible in the future are determined.

[0150] Then, a decision tree is automatically created (S11). The decision tree has a plurality of nodes, wherein the nodes are associated with future points in time. Depending on the driving scenario, the nodes are at least partially assigned to at least one decision made by the driver of the vehicle or of another vehicle with regard to a lane change, respectively. In addition, the nodes are at least partially assigned to at least two decisions made by the driving assistance system itself, respectively.

[0151] Furthermore, driving trajectories are calculated (S12), wherein each driving trajectory is assigned to a path defined by the decision tree, and each path comprises one or more decisions made successively.

[0152] Next, evaluation indicators are calculated (S13) for the driving trajectories, wherein the evaluation indicators are a measure of the comfort and / or safety of the respective driving trajectory.

[0153] Furthermore, probability values are calculated (S14), wherein the probability values are assigned to the nodes of the decision tree, and the probability values respectively indicate a probability of the driver of the vehicle or of the other vehicle making a decision to change lane at a time point assigned to the node.

[0154] Next, a plurality of decision indicators is calculated (S15). Here, each decision indicator is assigned to a node. The calculation of the decision indicator is based at least partly on the probability value assigned to the respective node, which indicates a probability of the driver of the vehicle or of the other vehicle making a decision to change lane at a time point assigned to the node. Furthermore, the calculation of the decision indicator is based on at least one evaluation indicator of a driving trajectory determined by the decision made at the time point of the respective node.

[0155] Finally, the at least partially automated driving procedure is selected (S16) by the autonomous driving assistance system, more precisely based on the decision tree, the decision indicators, and depending on whether the driver of the vehicle or of the other vehicle actively initiated a change of lane at a time point assigned to the node.

[0156] The application has been described above by way of example. Of course, many variations and modifications are possible without departing from the scope of the disclosure.

Claims

1. A method for selecting an at least partially automated driving procedure by means of a driving assistance system of a vehicle (1), the method comprising the following steps: a) determining (S10) a current driving scenario by the driving assistance system, based on which a plurality of different future possible driving procedures are determined; b) automatically creating (S11) a decision structure (3) by means of a computing unit of the driving assistance system by determining a plurality of decision levels (3.1) of the decision structure (3), wherein different decision levels (3.1) are associated with different future points in time, wherein the decision levels (3.1) are at least partially respectively assigned at least one decision (3.2) on a traffic action made by a driver of the vehicle (1) or by another actor (2) in accordance with the current driving scenario, wherein the decision levels (3.1) are at least partially respectively assigned at least two decisions (3.2) made by the driving assistance system itself; c) calculating (S12) driving trajectories (3.3), wherein each driving trajectory is assigned to a path defined by the decision structure (3), each path containing one or more decisions (3.2) made successively; d) calculating (S13) evaluation indicators for the driving trajectories (3.3), wherein the evaluation indicators are measures of the comfort and / or safety of the respective driving trajectory (3.3); e) calculating (S14) probability values, wherein the probability values are assigned to the decision levels (3.1) of the decision structure (3), the probability values respectively indicating a probability of a decision (3.2) on a traffic action made by the driver of the vehicle (1) or the other actor (2) at a point in time assigned to the decision level (3.1); f) calculating (S15) a plurality of decision indicators, wherein a decision indicator is respectively assigned to a decision level (3.1), wherein the calculation of the decision indicator is based at least partially on: - the probability value of the respective decision level (3.1) indicating a probability of a decision (3.2) on a traffic action made by the driver of the vehicle (1) or the other actor (2) at a point in time assigned to the decision level (3.1), and - at least one evaluation indicator of a driving trajectory (3.3) determined by the decisions made at the point in time of the respective decision level (3.1); g) selecting (S16) an at least partially automated driving procedure by the driving assistance system based on the decision structure (3), the decision indicators and depending on whether the driver of the vehicle (1) or the other actor (2) initiates a traffic action at the point in time assigned to the decision level (3.1) or not.

2. The method of claim 1, wherein, For selecting an at least partially automated driving procedure, the values of the decision indicators of the decision levels (3.1) of the decision structure (3) are compared; the selected driving procedure is based on a driving trajectory (3.3) which initiates a lane change or a deceleration operation at the point in time assigned to the decision level (3.1) having the smallest value of the decision indicator.

3. The method according to claim 1 or 2, characterized in that, The decision indicators of these decision levels (3.1) are calculated at least partially by adding the result of a first multiplication to the result of a second multiplication, wherein in the first multiplication a probability value of the driver of the vehicle (1) or of the other actor (2) to initiate a lane change at the point in time assigned to the decision level (3.1) is multiplied by an evaluation indicator of a driving trajectory by means of which the lane change is made; in the second multiplication a probability value of the driver of the vehicle (1) or of the other actor (2) not to initiate a lane change at the point in time assigned to the decision level (3.1) is multiplied by an evaluation indicator of a driving trajectory by means of which the lane is maintained and the vehicle (1) is decelerated.

4. The method according to claim 1 or 2, characterized in that, The decision made by the driver of the vehicle (1) is detected by the activation of a turn signal and / or by a steering intervention leading to a lane change.

5. The method according to claim 1 or 2, characterized in that, The decision made by the other actor (2) is detected by sensing a turn signal, another vehicle turning and / or changing lanes, or a person (2) entering the lane.

6. The method of claim 1 or 2, wherein, The calculation of the driving trajectory (3.3) includes the calculation of a multi-dimensional coordinate along which the vehicle (1) moves when driving the driving trajectory, the calculation of longitudinal and lateral acceleration values, and / or the calculation of longitudinal and lateral jerk values.

7. The method according to claim 1 or 2, characterized in that, The calculation of the evaluation indicator of the driving trajectory (3.3) involves the distance of the vehicle (1) to other vehicles (2, 2') and / or to non-vehicle objects, the difference to a set speed, and / or the frequency of lane changes.

8. The method of claim 1 or 2, wherein, The probability value of the driver initiating a lane change at a future point in time assigned to a decision level (3.1) is calculated on the basis of a neural network trained by means of data obtained from past driving processes of the driver.

9. The method of claim 1 or 2, wherein, The probability value of the driver initiating a lane change at a future point in time assigned to a decision level (3.1) is calculated on the basis of an evaluation indicator of a driving trajectory (3.3) initiated by a decision made at the point in time of the respective decision level (3.1).

10. The method of claim 1 or 2, wherein, The probability value of the driver initiating a lane change at a future point in time assigned to a decision level (3.1) is calculated on the basis of driver behavior sensed by at least one sensor.

11. The method of claim 1 or 2, wherein, A plurality of decision indicators is calculated at least for a part of these decision levels (3.1) on the basis of different probability values of the driver of the vehicle (1) or of the other actor (2) making a decision of a traffic behavior at a point in time assigned to the decision level (3.1); for each decision level (3.1) the decision indicator with the largest value is used to select an at least partially automated driving process.

12. The method of claim 1 or 2, wherein, The decision made by the driver of the vehicle (1) includes a lane change from the right lane to the left lane or from the middle lane to the left lane or to the right lane.

13. The method of claim 1, wherein, The traffic behavior includes a lane change.

14. A driving assistance system for a vehicle (1) designed to select an at least partially automated driving process from a plurality of possible driving processes, wherein the driving assistance system is designed to perform the following steps: a) determining a current driving scenario by the driving assistance system, on the basis of which a plurality of different possible future driving processes is determined; b) automating the creation of the decision structure (3) by means of a computing unit of the driving assistance system by determining a plurality of decision levels (3.1) of the decision structure (3), wherein different decision levels (3.1) are associated with different points in time in the future, wherein, depending on the current driving scenario, these decision levels (3.1) are at least partially respectively assigned at least one decision (3.2) about a traffic action made by the driver of the vehicle (1) or by the other actor (2), wherein these decision levels (3.1) are at least partially respectively assigned at least two decisions (3.2) made by the driving assistance system itself; c) calculating driving trajectories (3.3) by means of an automated trajectory planner, wherein each driving trajectory is assigned to a path defined by the decision structure (3) and each path contains one or more decisions (3.2) made in succession; d) automatically calculating evaluation criteria for the driving trajectories (3.3), wherein the evaluation criteria are a measure of the comfort and / or safety of the respective driving trajectory (3.3); e) calculating probability values, wherein the probability values are assigned to the decision levels (3.1) of the decision structure (3), which respectively indicate the probability of the driver of the vehicle (1) or the other actor (2) making a decision about a traffic action at the point in time assigned to the decision level (3.1); f) calculating a plurality of decision criteria, wherein a decision criterion is respectively assigned to a decision level (3.1), wherein the calculation of the decision criterion is based at least partially on: - the probability value of the respective decision level (3.1) indicating the probability of the driver of the vehicle (1) or the other actor (2) making a decision (3.2) about a traffic action at the point in time assigned to the decision level (3.1), and - at least one evaluation criterion of a driving trajectory (3.3) determined by the decision (3.2) made at the point in time of the respective decision level (3.1); g) selecting, by means of the driving assistance system, an at least partially automated driving process based on the decision structure (3), the decision criteria and depending on whether the driver of the vehicle (1) or the other actor (2) initiates a traffic action at the point in time assigned to the decision level (3.1) actively.

15. A vehicle comprising a driving assistance system according to claim 14.

Images