METHOD FOR PLANNING THE BEHAVIOR OF A VEHICLE

Abstract

A method for planning a behavior of a vehicle with respect to one or more occluded area(s) along a navigation path of the vehicle, wherein the method comprises an occluded area identification step, during which the occluded area(s) is/are identified, and a phantom object generation step, during which at least one phantom object is generated for at least one of the occluded areas, the occluded area(s) is/are defined based on information from a predefined occlusion scenario catalog during the occluded area identification step.

Claims

1. A method for planning a behavior of a vehicle with respect to one or more occluded areas along a navigation path of the vehicle, the method comprising: identifying one or more occluded areas; and generating at least one phantom object for at least one of the one or more occluded areas, wherein the one or more occluded areas are defined based on information from a predefined occlusion scenario catalog.

2. The method according to claim 1, wherein the occlusion scenario catalog comprises different occlusion scenarios and scenario information for each occlusion scenario.

3. The method according to claim 1, wherein generating the at least one phantom object comprises calculating an appearance probability for the at least one phantom object, wherein the appearance probability describes a probability for the phantom object to emerge from its occluded area into a field of view of the vehicle.

4. The method according to claim 3, wherein the appearance probability comprises a static component wherein the static component takes into account a map and/or road topology information, and/or wherein the static component depends on an initial environmental probability and/or on a phantom object distance and/or on a distance threshold.

5. The method according to claim 3, wherein the appearance probability comprises a dynamic component, wherein the dynamic component at least indirectly takes into account a geometric modification of the occluded area between two moments in time, and/or wherein the dynamic component depends on a one-phantom-object-length, wherein the one-phantom-object-length is defined as a length inside an occluded area in which exactly one phantom object is expected, wherein the one-phantom-object-length is preferably measured in a direction which is perpendicular to a longitudinal axis of the vehicle, and/or wherein the dynamic component depends on a field of view increase, wherein the field of view increase is a length measured in a direction which is perpendicular to a longitudinal axis of the vehicle, wherein the one-phantom-object-length and the field of view increase are typically directed in a parallel manner.

6. The method according to claim 5, wherein: the static component is calculated according to the equation $P_{env}\left(d\right)=\max \left(\left(K_{env}\frac{D_S-d}{D_S}\right),0\right)$ and/or the dynamic component is calculated according to the equation $P_{FoV}\left(u\right)=\{\begin{array}{cc}0,& \mathrm{for}u?0\\ \frac{u}{L},& \mathrm{for}u>0.Math.u<L\\ 1,& \mathrm{for}u?L\end{array}$ and/or the appearance probability is preferably calculated according to the equation
P.sub.a(d, u)=min((P.sub.env(d)+P.sub.FoV(u)),1)).

7. A system for planning a behavior of a vehicle with respect to one or more occluded areas along a navigation path of the vehicle, comprising: at least one processing device configured to: identify the one or more occluded areas; generate at least one phantom object for at least one of the one or more occluded areas; and define the one or more occluded areas based on information from a predefined occlusion scenario catalog.

8. A vehicle comprising: the system according to claim 7.

9. (canceled)

10. A non-transitory computer-readable medium having stored thereon computer program code that, when executed by a processing device, cause the processing device to execute a method comprising: identifying one or more occluded areas; and generating at least one phantom object for at least one of the one or more occluded areas wherein the one or more occluded areas are defined based on information from a predefined occlusion scenario catalog.

11. The system according to claim 7, wherein the occlusion scenario catalog comprises different occlusion scenarios and scenario information for each occlusion scenario.

12. The system according to claim 7, wherein the at least one processing device is configured to: generate the at least one phantom object by calculating an appearance probability for the at least one phantom object, wherein the appearance probability describes a probability for the phantom object to emerge from its occluded area into a field of view of the vehicle.

13. The system according to claim 12, wherein the appearance probability comprises a static component, wherein the static component takes into account a map and/or road topology information, and/or wherein the static component depends on an initial environmental probability and/or on a phantom object distance and/or on a distance threshold.

14. The system according to claim 12, wherein the appearance probability comprises a dynamic component, wherein the dynamic component at least indirectly takes into account a geometric modification of the occluded area between two moments in time, and/or wherein the dynamic component depends on a one-phantom-object-length, wherein the one-phantom-object-length is defined as a length inside an occluded area in which exactly one phantom object is expected, wherein the one-phantom-object-length is preferably measured in a direction which is perpendicular to a longitudinal axis of the vehicle, and/or wherein the dynamic component depends on a field of view increase, wherein the field of view increase is a length measured in a direction which is perpendicular to a longitudinal axis of the vehicle, wherein the one-phantom-object-length and the field of view increase are typically directed in a parallel manner.

15. The system according to claim 14, wherein the at least one processing device is configured to: calculate the static component according to the equation $P_{env}\left(d\right)=\max \left(\left(K_{env}\frac{D_S-d}{D_S}\right),0\right)$ and/or calculate the dynamic component according to the equation $P_{FoV}\left(u\right)=\{\begin{array}{cc}0,& \mathrm{for}u?0\\ \frac{u}{L},& \mathrm{for}u>0.Math.u<L\\ 1,& \mathrm{for}u?L\end{array}$ and/or calculate the appearance probability according to the equation
P.sub.a(d, u)=min((P.sub.env(d)+P.sub.FoV(u)), 1).

Description

[0025] In the following, the invention is explained by means of Figures, wherein show:

[0026] FIG. 1 shows a flow diagram of one embodiment of a method according to the invention,

[0027] FIG. 2 a schematic drawing explaining the static component of the appearance probability,

[0028] FIG. 3 a first schematic drawing explaining the dynamic component of the appearance probability,

[0029] FIG. 4 a second schematic drawing explaining the dynamic component of the appearance probability, and

[0030] FIG. 5 a schematic drawing visualizing the one-phantom-object-length L.

[0031] FIG. 1 shows a flow diagram of one embodiment of a method according to the invention. The method comprises an occluded area identification step S1 and a phantom object generation step S2. During operation of an autonomous vehicle (not shown in FIG. 1), the steps S1, S2 are being carried out continuously, typically by means of an infinite loop. In certain embodiments, however, it is not necessary for the method to be carried out in an infinite loop. For example, it is also possible that the method with the steps S1 and S2 is only carried out on demand at certain moments in time. During the occluded area identification step S1, occluded areas along a navigation path of the autonomous vehicle are identified, in particular by means of analyzing map data and by analyzing sensor data supplied by sensors present in the autonomous vehicle. Furthermore, this map data and sensor data is matched with content from an occlusion scenario catalog, such that the detected occluded areas along the navigation path of the vehicle are being matched to generalized occluded area types which are present in the occlusion scenario catalog. The occluded area identification step S1 also comprises setting typical variables and constants for the different identified occluded areas based on information present in the occlusion scenario catalog.

[0032] In the phantom object generation step S2, phantom objects are generated for at least one of the identified occluded areas, typically for all of the occluded areas. The generation of the phantom objects is typically at least partly based on information taken from the occlusion scenario catalog and/or on the map data and/or on the sensor data. In typical embodiments, the phantom object generation step S2 comprises the calculation of the appearance probability/probabilities for the phantom object(s).

[0033] FIG. 2 shows a schematic drawing explaining the static component of the appearance probability. In particular, FIG. 2 shows a vehicle 1 on a road 8. A zebra crossing 7 crosses road 8. The vehicle 1 is next to an obstacle 2 which is also referred to as occluding object. Because of the obstacle 2, the vehicle 1 is not able to detect its entire surrounding: In particular, the obstacle 2 creates an occluded area 3 which cannot be seen by the vehicle 1. FIG. 2 furthermore shows nine pedestrians 4 out of which only one is equipped with a reference sign for the purpose of simplicity. Each pedestrian 4 has a hypothetic walking path 6 which crosses the road 8 in a direction which is perpendicular to a longitudinal axis 5 of the vehicle 1. FIG. 1 also shows a representation of the static component Penv(d) for the traffic situation depicted in FIG. 2. The zebra crossing 7 has a width Wz. Over this entire width Wz, the static component Penv(d) takes a constant value of Kenv, namely the initial environmental probability. This initial environmental probability Kenv is for example determined based on information from the occlusion scenario catalog and/or on the map data. The graph at the bottom of FIG. 2 shows the static component Penv(d) of the appearance probability as a function of the phantom object distance d. It can be seen that d is measured from the start of the zebra crossing 7 (the zebra crossing 7 itself is a high risk area). Start means that edge of the zebra crossing 7 that is closest to the vehicle 1. FIG. 2 also shows a distance threshold DS which is located outside the actual high risk area/zebra crossing 7. Between the zebra crossing 7 and the distance threshold DS, the static component Penv(d) of the appearance probability linearly decreases from the initial environmental probability Kenv down to 0. At points which are further away from the zebra crossing 7 than the distance threshold DS, the static component Penv(d) of the appearance probability equals 0.

[0034] FIG. 3 shows a first schematic drawing explaining the dynamic component of the appearance probability. In particular, FIG. 3 shows a vehicle 1, but at two different moments in time. The vehicle 1t1 is the vehicle at the current moment in time t1 and the vehicle 1t0 is the vehicle at one time step before, namely at the time step t0. Just as in FIG. 2, the vehicle 1t0, the vehicle 1t1 is on a road 8 which is crossed by a zebra crossing 7. The vehicle 1t0, 1t1 has a longitudinal axis 5. The vehicle 1t0, 1t1 is driving by the obstacle 2. At the time t1 the obstacle 2 creates an occluded area 3 for the vehicle 1t1. FIG. 3 also shows two edges of a previous occluded area 9.1, 9.2, namely the occluded area as it was for the vehicle 1t0. FIG. 3 also shows a pedestrian 4 and a field of view increase u1. The field of view increase u1 is a distance (typically measured in meters) which is measured in the direction perpendicular to the longitudinal axis 5 of the vehicle 1t1. In the example shown in FIG. 3, the field of view increase u1 is also measured in the direction of the hypothetic walking path 6 of the pedestrian 4. The field of view increase u1 corresponds to the increase of the field of view at the height of the pedestrian 4. In other words: The dynamic component of the appearance probability for the pedestrian 4 (which is a phantom object) after the time step between the moments in time t0 and t1 is calculated with a field of view increase u1 as shown in FIG. 3. One can also say that u1 is the length of the hypothetic walking path 6 of the pedestrian 4 that is being uncovered from the occluded area 3 between the time points t0 and t1. In the case of FIG. 3, u1 is large enough for the pedestrian 4 to actually move out of the occluded area 3. In other words, the dynamic component corresponding to the situation shown in FIG. 3 takes the value 1.

[0035] FIG. 4 shows a situation similar to the situation shown in FIG. 3. However, in FIG. 4 the vehicle 1t0, 1t1 has moved a shorter distance along the obstacle 2 than in FIG. 3. Therefore, the field of view increase u2 is smaller than the field of view increase u1 shown In FIG. 3. In particular, the field of view increase u2 shown in FIG. 4 is such that the appearance probability of the pedestrian 4 is still so small that the pedestrian 4 remains in the occluded area 3.

[0036] FIG. 5 shows a schematic drawing visualizing the one-phantom-object-length L. In particular, FIG. 5 shows how the one-phantom-object-length L is determined at a certain location in the horizontal direction of FIG. 5. The one-phantom-object-length L is the length measured in the direction of the hypothetic walking path 6 of the pedestrian 4 which contains exactly one pedestrian 4.

[0037] The invention is not limited to the preferred embodiments described here. The scope of protection is defined by the claims.

[0038] It is further to be noted that methods disclosed in the specification or in the claims may be implemented by a device having means for performing each of the respective acts of these methods.

LIST OF REFERENCE SIGNS

[0039] 1, 1t.sub.0, 1.sub.t1 vehicle

[0040] 2 obstacle (also referred to as occluding object)

[0041] 3 occluded area

[0042] 4 pedestrian

[0043] 5 longitudinal axis (of vehicle)

[0044] 6 hypothetic walking path

[0045] 7 zebra crossing

[0046] 8 road

[0047] 9.1, 9.2 edges of previous occluded area

[0048] S1 occluded area identification step

[0049] S2 phantom object generation step

[0050] u.sub.1, u.sub.2 field of view increase

[0051] P.sub.env(d) static component (of appearance probability)

[0052] K.sub.env initial environmental probability

[0053] D.sub.sdistance threshold

[0054] d phantom object distance

[0055] W.sub.z zebra crossing width

[0056] L one-phantom-object-length