Traffic-rule-compliant decision-making in dynamic traffic scenarios

Abstract

The present disclosure relates to a method for determining a driving maneuver of a vehicle by a control device, the method comprising: receiving measurement data relating to a traffic situation from at least one sensor; determining a current detection area by evaluating the received measurement data; creating a sensor model based on the received measurement data; modeling an estimated detection area of ​​the at least one sensor from the received measurement data by forward simulation based on a vehicle position of the vehicle; using the created sensor model to determine a change in the detection area as a result of the at least one driving maneuver; and determining a driving maneuver that results in an increase in the simulated detection area as a result of the sensor model. Further disclosed are a control device, a computer program, and a machine-readable storage medium.

Description

[Technical field]

[0001] The present invention relates to a method for determining a driving operation of a vehicle, a control device, a computer program, and a machine-readable storage medium.

[0002] <Related Applications> This patent application claims priority from German patent application No. 102020111486.9, the disclosure content of which is incorporated herein by reference. [Background technology]

[0003] Addressing dynamic traffic situations is often further hindered by limited visibility and partial obscuration of the detection area of ​​the environmental sensor system, which may be, for example, partially obscured by other road users, dynamic objects, or stationary obstacles such as trees, buildings, parked vehicles, etc.

[0004] To achieve autonomous driving functions, such occluding objects that limit the detection area of ​​the environmental sensor system must be taken into account, especially the subsequent changes of the detection area due to the movement of the vehicle performing the autonomous driving function and the movement of dynamic objects, e.g. trucks.

[0005] The detection range of the vehicle's environmental sensor system may change if road users move out of or into the scanning range of the environmental sensor system, even if the vehicle is stationary.

[0006] There is a known method of executing a partially observable Markov decision process. However, this type of method requires a high processing load and cannot be implemented in a typical vehicle-mounted control device. Summary of the Invention [Problem to be solved by the invention]

[0007] An object of the invention can be seen as proposing a method to counteract impairments in the detection field of a sensor. [Means for solving the problem]

[0008] This object is achieved by the subject matter of the respective independent claims. Advantageous designs of the invention form the subject matter of the respective dependent claims.

[0009] According to one aspect of the invention, there is provided a method for determining a driving maneuver of a vehicle by a control device, in one step, measurement data relating to a traffic situation are received from at least one sensor.

[0010] By evaluating the received measurement data, a current detection area or a current degree of visibility of the at least one sensor is determined. The degree of visibility can be defined as a measure of the coverage or concealment of the scanning area of ​​the at least one sensor. In an open area without obstacles, the degree of visibility of the sensor, which represents the detection area of ​​the sensor, can be e.g. 100%. An increase in the number of obstacles leads to a decrease in the degree of visibility or a smaller detection area. The current degree of visibility represents the degree of visibility at the time when the measurement data is received.

[0011] The detection region may be designed as an area or volume that can be scanned or registered by a sensor system to determine measurement data.

[0012] In a further step, a sensor model is created based on the received measurement data, and an estimated detection area of ​​at least one sensor is modeled from the received measurement data by forward simulation based on the vehicle position of the vehicle. The created sensor model can preferably be compared with the actual traffic situation at the time when the measurement data was created or received. Through this measure, the quality of the sensor model can be determined and optionally improved.

[0013] The created sensor model is used to determine a change in the detection area of ​​at least one sensor due to at least one driving maneuver. For this purpose, a predefined list of possible driving maneuvers can be checked by the sensor model. Stopping the vehicle can also be implemented as a driving maneuver.

[0014] A maneuver is then determined, as a result of which the simulated degree of disability according to the sensor model is increased or the simulated detection area of ​​at least one sensor is increased. For this purpose, the results of the maneuvers used can be compared with each other and / or with the current traffic situation or the current degree of visibility.

[0015] The method allows the creation of a sensor-based or measurement-database sensor model that can be used to optimize the detection area or the degree of visibility. In particular, the subsequent detection area of ​​the vehicle's sensor system can be increased and a decision regarding the driving maneuver to be performed is made by the control device depending on the improved degree of visibility. The degree of visibility can be improved if the detection area of ​​the sensor is increased. This increase can be designed in the form of an increase in angle and / or in the form of an increase in the range of the detection area. The detection area can be increased if the detection area is partially hidden by fewer obstacles, by obstacles at a greater distance from the sensor system, or by smaller obstacles, or is not hidden by obstacles.

[0016] Following a determination of a driving maneuver that increases the detection region, a control command may be generated by the controller that can be used to control the vehicle, thus instructing the vehicle to perform the determined or selected driving maneuver.

[0017] The created sensor model preferably takes into account road users, static and dynamic objects detected within the scanning area of ​​the sensor and is able to predict their subsequent behavior, thereby determining in particular the static and dynamic factors influencing the detection area.

[0018] In addition to the detection area, priority rules, for example relating to unobserved or undetectable road users, collision risk, comfort factors, possible violations of traffic rules can also be taken into account by the created sensor model.

[0019] The method can improve traffic safety and comfort during automated driving operations: it can further prevent automated vehicles from disrupting traffic, and in particular it can enable a smoother traffic flow.

[0020] The decision regarding the maneuver can be made, for example, in the form of a tree search algorithm. Alternatively or additionally, a sequence of maneuvers can be advantageously selected in relation to the detection area of ​​the sensor.

[0021] The tree search can span several seconds into the future and is therefore designed to be discrete: possible maneuvers can be continuously recalculated, preferably as soon as new measurement data becomes available from at least one sensor.

[0022] When new measurement data and / or environmental information occurs, an existing or active tree search can be interrupted.

[0023] Alternatively, or in addition, when new measurement data and / or environmental information occurs, existing or active tree searches can continue to run and at least one new parallel tree search based on the new measurement data is initiated.

[0024] The results of the converged tree search can be used as soon as the performed tree search converges. The results can preferably be generated in the form of a driving maneuver that causes an increase in the simulated detection area by the sensor model.

[0025] As a result of the increased detection area, the degree to which the detection area is obscured decreases.

[0026] Based on the method, a vehicle can, for example, perform a series of multiple consecutive driving maneuvers, such as approaching a stop line, slowing down a specified distance to the intersection, and then crossing the intersection, in the presence of static objects in the vehicle environment of a road intersection.

[0027] If dynamic objects are present in the vehicle environment, the vehicle may, for example, first stop at the stop line and then cross the intersection if the detection area is sufficient.

[0028] According to a further aspect of the invention, a control device is provided, which is configured to perform the method. The control device can for example be a control device on board the vehicle, a control device external to the vehicle or a server unit external to the vehicle, for example a cloud system. The control device is preferably capable of receiving and evaluating measurement data determined from at least one sensor on board the vehicle.

[0029] According to a further aspect of the invention there is provided a computer program comprising instructions which, when executed by a computer or control device, are capable of instructing the computer to perform the method according to the invention. According to a further aspect of the invention there is provided a machine readable storage medium having stored thereon the computer program according to the invention.

[0030] As used herein, a vehicle may be operable as assisted, partially automated, highly automated and / or fully automated, or driverless in accordance with the BASt standard. A vehicle may be designed as, for example, a passenger car, a robot, a drone, a personal watercraft, a rail vehicle, a robotaxi, an industrial robot, a utility vehicle, a bus, an airplane, a helicopter, etc.

[0031] According to an exemplary embodiment, at least one driving maneuver is designed as driving forward at low speed, entering an intersection, turning, stopping, or keeping stopped, and is used by the created sensor model to determine the change in the detection area. As a result, a list of possible driving maneuvers that can be used by the sensor model can be provided. Each driving maneuver can also be used as a series of driving maneuvers that can be executed consecutively.

[0032] In a further exemplary embodiment, the current detection area is determined by identifying dynamic and / or static objects based on the received measurement data. This measure allows estimating the influence of the environment on the detection area or the degree of concealment of the detection area. In particular, parked vehicles, containers, buildings, vegetation, etc. can be assigned to static objects, which are unlikely to change over time and therefore do not affect the detection area if the position of the sensor or the vehicle does not change. In the case of dynamic objects, the detection area may change even if the vehicle equipped with the sensor for detecting the environment does not move.

[0033] According to a further embodiment, different driving maneuvers are used in parallel or successively by the created sensor model to determine the change in the detection area. This measure allows possible driving maneuvers to be checked in parallel, as a result of which the method can be executed more quickly. If driving maneuvers are checked in succession, for example, a predefined checking step can be performed.

[0034] Preferably, driving maneuvers are compared or selected with a view to optimizing the degree of visibility.

[0035] Map information may also be received by the control device to determine the detection area or the degree of obscuration of the detection area.

[0036] According to a further exemplary embodiment, the current detection area is determined based on traffic lane intervals measurable by a sensor, which approximates the lanes determined based on the measurement data by rectangles of different length and / or width. Thus, in the method, the visible area, and therefore also the detection area, can be defined and measured in the form of a traffic lane interval coverage. The lane interval coverage can preferably be approximated by a rectangle that overlaps the lane portions that are visible or measurable by the sensor system. For each detectable lane portion, a rectangle, a polygon, a triangle, a circle, etc. can be used. The method can consequently be limited to areas relevant to the automated driving function.

[0037] As a result, the comparability of the detection areas can further be implemented: the size of the area detectable by the sensor can be determined in a technically simple manner and used for comparison purposes.

[0038] In a further exemplary embodiment, the detection area increases if the total length and / or total area of ​​the lane sections approximated by the triangle increases. This measure can provide a particularly effective parameter for selecting a driving maneuver. In particular, the total length or total area can be determined in a technically simple manner and used for comparison purposes.

[0039] According to a further embodiment, data is received via a communication connection and a sensor model is created based on the received measurement data and based on the data received via the communication connection. This measure allows the sensor model to be designed to be more accurate, since additional measurement data is received from road users or infrastructure devices. The data can preferably be received via a vehicle-infrastructure communication connection or a vehicle-vehicle communication connection. As a result, in particular accurate predictions regarding the future behavior of road users can be made.

[0040] In the following, preferred exemplary embodiments of the invention are described in detail with reference to highly simplified schematic drawings. [Brief description of the drawings]

[0041] [Figure 1] 1 is a schematic flow chart illustrating a method according to an embodiment. [Diagram 2] FIG. 2 shows a top view of an example of a traffic situation. [Diagram 3] 3 shows a top view of a traffic situation following the traffic situation shown in FIG. 2. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0042] 1 shows a schematic flow chart illustrating a method 1 according to an embodiment, which serves to determine a driving maneuver of a driveable vehicle 2 in an automated manner. The method 1 can be preferably executed by a control device 4.

[0043] 2, the vehicle 2 includes, in addition to the control device 4, an environmental sensor system 6. The environmental sensor system 6 may consist of one or more sensors. The environmental sensor system 6 may include, for example, a radar sensor, a LIDAR sensor, an ultrasonic sensor, a camera sensor, etc.

[0044] In step 20, measurement data relating to a traffic situation 8 is received from at least one sensor 6. The method 1 may begin with an identified traffic situation 8, such as an intersection, or an approach to a highway.

[0045] The current detection area is then determined (21) by evaluating the received measurement data.

[0046] A sensor model is created based on the received measurement data and an estimated detection area of ​​at least one sensor 6 is modeled (22) from the received measurement data by a forward simulation based on the vehicle position of the vehicle 2. The forward simulation is performed by the sensor model and makes it possible to estimate a change in the detection area that would occur due to a change in the position of the sensor 6.

[0047] The created sensor model is used to determine the change in the detection area due to at least one driving maneuver (23). For this purpose, for example, a number of different driving maneuvers are checked to determine whether the driving maneuvers increase or decrease the detection area. The change in the detection area due to the execution of one or more driving maneuvers is determined by the sensor model.

[0048] In the illustrated exemplary embodiment, multiple example driving maneuvers are performed or reviewed in series. Alternatively, possible driving maneuvers can be reviewed in parallel.

[0049] In a further step 24, the entry of the vehicle 2 into a road fork or into an intersection is simulated by the sensor model and the change in the detection area is determined.

[0050] If the detection area increases compared to the initial detection area (25) or if the safe execution of the driving maneuver is confirmed, the execution of the driving maneuver can be initiated (26) by the control device 4. If this request 25 is rejected, the next driving maneuver can be confirmed (27).

[0051] For example, stopping of the vehicle 2 in the initial position can be confirmed by the sensor model as the next driving maneuver.

[0052] In the check, the change in the detection area is determined (28). If safety is guaranteed when the vehicle is stopped, and the detection area increases as a result of the vehicle stopping, for example because a road user leaves the detection area of ​​the environmental sensor system 6, then stopping is initiated (29) by the control device 4. If this is not the case, the next operation is checked (30).

[0053] As a next maneuver, a slow and careful approach to the intersection (30) can be considered by the sensor model. If this type of maneuver is classified as safe and allows an increase in the detection area (31), the control device 4 initiates a slow and careful approach of the vehicle 2 to the intersection (32). If this confirmation 31 is rejected or classified negatively, the vehicle 2 can remain in its initial position (33).

[0054] Figure 2 shows a top view of an example of a traffic situation 8. There are dynamic objects 10 or obstacles and static objects 12 or obstacles.

[0055] A dynamic object 10 is for example a road user obscuring part of the scanning area A of the vehicle 2. Static objects 12 considered can be buildings or parked road users.

[0056] The arrow P indicates the planned trajectory of the road user 10 and the vehicle 2 .

[0057] In the illustrated traffic situation 8, the vehicle 2 is stopped at a stop line 14 before a T-junction. To make the decision, the method 1 is executed by the control device 4. For this purpose, the current or initial detection area of ​​the environmental sensor system 6 is determined.

[0058] The detection area, and in particular the dimensions or size of the detection area, is determined based on lane segments 16 detected by the sensor. Lane segments 16 are portions of lanes identified in the scan area A and are approximated by rectangles.

[0059] The rectangles may have different lengths or widths and may be used to determine the size of the detection area. The total length or area of ​​the determined lane segments 16 may be used as a measure of the detection area. Figure 3 shows a top view of a traffic situation subsequent to the traffic situation 8 shown in Figure 2. The vehicle 2 is approaching an intersection or T-junction slowly and carefully, driving past the stop line 14, and the positions of the static objects 12 and the dynamic objects 10 relative to the vehicle 2 are changing.

[0060] Due to the performed driving maneuver, the scanning area A changes and the detection area increases, as a result of which additional dynamic road users 11 can be detected by the sensor and the total area or length of the lane section 16 increases compared to the situation in FIG.

Claims

1. A method (1) for determining a driving operation of a vehicle (2) by a control device (4), comprising: Measurement data relating to traffic conditions (8) is received from at least one sensor (6), by evaluating the received measurement data a current detection area of ​​the at least one sensor (6) is determined, a sensor model is created based on the received measurement data, and an estimated detection area of ​​the at least one sensor (6) is modeled from the received measurement data by a forward simulation based on a vehicle position of the vehicle (2); Using the created sensor model, a change in the detection area of ​​the at least one sensor (6) due to at least one driving maneuver is determined; A driving maneuver that causes an increase in the simulated detection area by the sensor model is determined; the current detection area is determined based on lane sections (16) which can be determined by a sensor approximating the lane determined based on the measurement data by rectangles of different length and / or width; method.

2. The method of claim 1 , wherein the at least one driving maneuver is designed as driving forward at a low speed, entering an intersection, turning, stopping, or staying stopped, and is used to determine changes in the detection area by the created sensor model.

3. 3. The method according to claim 1 or 2, wherein the current detection area is determined by identifying dynamic objects (12) and / or static objects (10) based on the received measurement data.

4. 4. The method according to claim 1, wherein different driving maneuvers are used in parallel or successively by the created sensor model to determine the changes in the detection area.

5. 2. The method of claim 1, wherein the detection area increases if the total length and / or the total area of ​​the lane sections (16) approximated by a triangle increases.

6. 6. The method according to claim 1, wherein data is received via a communication connection and the sensor model is created based on the received measurement data and based on the data received via the communication connection.

7. A control device (4), configured to carry out the method according to one of claims 1 to 6.

8. A computer program comprising commands which, when the computer program is executed by a computer or a control device (4), cause the computer or the control device (4) to perform a method according to any one of claims 1 to 6.

9. A machine-readable storage medium having stored thereon the computer program of claim 8.

Images