METHOD AND DEVICE FOR EXCHANGING MANEUVER INFORMATION BETWEEN VEHICLES

Abstract

A method for exchanging pieces of maneuver information between vehicles. A parameterizable third-party trajectory planner provided by a third-party vehicle and mapping future pieces of maneuver information of the third-party vehicle are parameterized and executed in an ego vehicle, using at least one time parameter, to obtain at least one future third-party trajectory of the third-party vehicle.

Claims

1-14. (canceled)

15. A method for exchanging pieces of maneuver information between vehicles, the method comprising: providing a parameterizable ego trajectory planner for at least one third-party vehicle, the ego trajectory planner being configured to map future pieces of maneuver information of an ego vehicle to plan at least one ego trajectory of the ego vehicle.

16. A method for exchanging pieces of maneuver information between vehicles, the method comprising: parameterizing and executing, in an ego vehicle, a parameterizable third-party trajectory planner provided by a third-party vehicle, using at least one time parameter, the parameterizable third party trajectory planner configured to map future pieces of maneuver information of the third-party vehicle to obtain at least one future third-party trajectory of the third-party vehicle.

17. A method for exchanging pieces of maneuver information between vehicles, the method comprising: providing a parameterizable ego trajectory planner for at least one third-party vehicle, the ego trajectory planner being configured to map future pieces of maneuver information of an ego vehicle to plan at least one ego trajectory of the ego vehicle; and executing, in the ego vehicle, at least one parameterizable third-party trajectory planner received from the third-party vehicle for planning at least one future third-party trajectory of the third-party vehicle.

18. The method as recited in claim 16, wherein the third-party trajectory planner is parameterized and executed in the ego vehicle using at least one ego parameter of the ego vehicle.

19. The method as recited in claim 18, wherein the ego parameter characterizes a planned future ego trajectory of the ego vehicle.

20. The method as recited in claim 16, wherein, after the execution of the third-party trajectory planner, the time parameter is updated, and the third-party trajectory planner is re-parameterized and re-executed in the ego vehicle, using the updated time parameter, to obtain an updated future third-party trajectory of the third-party vehicle.

21. The method as recited in claim 16, wherein the third-party trajectory planner is received from the third-party vehicle in coded form and is decoded prior to the parameterization and the execution in the ego vehicle.

22. The method as recited in claim 15, wherein the ego trajectory planner is provided by the ego vehicle in coded form.

23. The method as recited in claim 15, wherein a simplified surroundings representation of surroundings of the ego vehicle is mapped in the ego trajectory planner.

24. The method as recited in claim 15, wherein multiple possible future ego trajectories are evaluated in the ego vehicle, and an evaluation of the ego trajectories is mapped in the ego trajectory planner for the third-party vehicle.

25. The method as recited in claim 15, wherein the pieces of maneuver information mapped in the ego trajectory planner for the third-party vehicle are simplified.

26. A device for exchanging pieces of maneuver information between vehicles, the device configured to: provide a parameterizable ego trajectory planner for at least one third-party vehicle, the ego trajectory planner being configured to map future pieces of maneuver information of an ego vehicle to plan at least one ego trajectory of the ego vehicle.

27. A device for exchanging pieces of maneuver information between vehicles, the device configured to: parameterize and execute, in an ego vehicle, a parameterizable third-party trajectory planner provided by a third-party vehicle, using at least one time parameter, the parameterizable third party trajectory planner configured to map future pieces of maneuver information of the third-party vehicle to obtain at least one future third-party trajectory of the third-party vehicle

28. A non-transitory machine-readable storage medium on which is stored a computer program for exchanging pieces of maneuver information between vehicles, the computer program, when executed by a processor, causing the processor to perform the following: providing a parameterizable ego trajectory planner for at least one third-party vehicle, the ego trajectory planner being configured to map future pieces of maneuver information of an ego vehicle to plan at least one ego trajectory of the ego vehicle.

Description

BRIEF DESCRIPTION OF THE DRAWING

[0029] Specific embodiments of the present invention are described below with reference to the figure, with neither the figure nor the description herein to be interpreted as limiting the present invention.

[0030] FIG. 1 shows a representation of a vehicle, including a device, according to one exemplary embodiment of the present invention.

[0031] The figure is only schematic and not true to scale. The same reference numerals designate the same or functionally equivalent features in the figure.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

[0032] FIG. 1 shows a representation of an ego vehicle 100a, including a device 102, according to one exemplary embodiment. Device 102 is designed to exchange pieces of maneuver information 104 between ego vehicle 100a and at least one third-party vehicle 100b. Host vehicle 100a may also be referred to as ego vehicle 100a. Third-party vehicle 100b may also be referred to as traffic vehicle 100b. Third-party vehicle 100b also includes a corresponding device, which is not illustrated here, for exchanging pieces of maneuver information.

[0033] Devices 102 each include a mapping unit 106. Mapping units 106 each map pieces of maneuver information 104 in a parameterizable algorithm referred to as a trajectory planner 108. Pieces of maneuver information 104 each include possible future trajectories 110 of vehicles 100. A future possible trajectory 110 of ego vehicle 100 may be referred to as ego trajectory 110a. A future possible trajectory 110 of third-party vehicle 100b may be referred to as third-party trajectory 110b. The different trajectories 110 of a piece of maneuver information 104 all essentially begin at same starting point 112 and may lead to different destinations 114.

[0034] Trajectories 110 of a piece of maneuver information 104 may be referred to as a trajectory group 116. Trajectory planner 108 of ego vehicle 100a may be referred to as ego trajectory planner 108a. Trajectory planner 108 generated in third-party vehicle 100b may be referred to as third-party trajectory planner 108b.

[0035] Ego trajectory planner 108a is transferred wirelessly to third-party vehicle 100b. Ego trajectory planner 108a may also be sent from ego vehicle 100a to third-party vehicle 100b, for example, via a higher-level data processing unit 118, such as a cloud server. The data transfer for third-party trajectory planner 108b of third-party vehicle 100b correspondingly takes place in the opposite direction.

[0036] Third-party trajectory planner 108b is received by device 102. Third-party trajectory planner 108b is parameterized in a parameterizing unit 120 of device 102, using at least one parameter 122. Parameter 122 may be, in particular, a time parameter 124. Time parameter 124 defines a point in time of interest. For example, time parameter 124 may map a future point in time.

[0037] Parameterized third-party trajectory planner 108b is an executable computer program product. Parameterized third-party trajectory planner 108b is executed in an execution unit 126 of device 102. Execution unit 126 outputs at least one third-party trajectory 110b as a function of parameter 122.

[0038] In one exemplary embodiment, third-party trajectory planner 108b is parameterized as parameter 122, using an ego parameter 128. Ego parameter 128 may map, for example, an ego velocity of ego vehicle 100a. Likewise, ego parameter 128 may map a planned future ego trajectory 110a. Ego parameter 128 may represent, for example, a sequence of reference points to be approached by ego vehicle 100a in the future. Third-party trajectory planner 108b parameterized using ego parameter 128 then outputs third-party trajectories 110b in the area of ego trajectory 110a.

[0039] In one exemplary embodiment, third-party trajectory planner 108b is executed multiple times consecutively. Time parameter 124 is updated in each case according to a movement of ego vehicle 100a. Exact third-party trajectory 110b may thus continue to be obtained within a tolerance range, using third-party trajectory planner 108b created at an earlier time.

[0040] In other words, a system and a V2X protocol are provided for coordinating cooperative driving maneuvers via V2X, based on a transfer of maneuver planners.

[0041] Thanks to V2X (vehicle to everything) communication, vehicles may receive surroundings data from other vehicles or infrastructure equipment or even obtain a complete global surroundings model, to thereby enrich their own local surroundings model.

[0042] In addition to surroundings data, vehicles may exchange maneuver data to be able to plan and implement, e.g., cooperative driving maneuvers based thereon. Various approaches exist for coordinating driving maneuvers. For example, vehicles may exchange trajectories. Depending on the approach, the receiving vehicle may infer therefrom how the sending vehicle will behave and whether it wishes to carry out a particular maneuver.

[0043] Based on this information, the receiving vehicle may plan its maneuver and, if necessary, communicate its desired maneuver to other vehicles, which confirm it by adapting their own plans. Further methods may be based on so-called distributed state machines, via which cooperative maneuvers may be coordinated.

[0044] In the conventional procedure, redundant pieces of information may possibly be exchanged if trajectories greatly resemble each other in certain temporal sections. In addition, the communication complexity increases along with the length of the prediction horizon and the quantity of exchanged trajectories. In the case of restrictions in the communication channel, the latter may result in multiple communication rounds taking place before maneuvers may be successfully coordinated. If state machines are used, each possible maneuver is mapped in a state machine. Only a few situation classes may thus be implemented.

[0045] In the approach presented here, a new type of exchanged maneuver information between vehicles is described, with the aid of which cooperative maneuvers may be efficiently planned.

[0046] Instead of trajectories or pieces of synchronization information for a distributed state machine, software libraries (hereinafter referred to as reference planners) are exchanged, which, based on a (standardized) surroundings representation and a map, generate a set of possible and, if necessary, evaluated trajectories (e.g., with the aid of costs), which the vehicle is able to drive, due to the surroundings representation, as well as to drive in a preferential manner according to the costs. Instead of sharing a limited number of trajectories per V2X, the calculation rule is shared.

[0047] In the case of IP concerns relating to the sharing of the calculation rule, an encryption method may be used, or a neural network may be distributed, which ensures the functionality without specifically disclosing the calculation rule.

[0048] In one simple design, the reference planner may be event-driven, i.e., communicated only once to vehicles which have not yet received it (e.g., vehicles which have come into communication range of the sending vehicle for the first time). Otherwise, the cooperation may be subsequently confirmed or rejected on a one-time basis. A cancellation of the cooperation or an update of the reference planner due to an unfavorable change in the situation is furthermore possible.

[0049] With the aid of the reference planner, the receiving vehicle may generate an arbitrary number of trajectories for the vehicle where the reference planner originated to be able to infer the behavior of this vehicle for its own given maneuvers and vice versa. The quantity of trajectories for vehicles, for which a reference planner is present, is not limited by the communication, but only by the computing power. The latter is much more favorable than communication.

[0050] The described method may be combined, in that a reference planner is used for vehicles, for which such a planner exists, while, for vehicles for which no reference planner exists, the trajectories sent by these vehicles are taken into account in the vehicle's own maneuver planning.

[0051] DA (driver assistance) and AD (automated driving) systems are generally structured according to the sense/plan/act principle. This means that the vehicle first plans its behavior and then carries it out, based on the sensing of the surroundings. Once vehicles in addition to the ego vehicle (so-called traffic vehicles) are in the surroundings of the ego vehicle, it is necessary to infer the future behavior of these vehicles upon a given behavior of the ego vehicle and vice versa.

[0052] In classic systems, this behavior is inferred according to a special behavior and maneuver planner, which is generally designed to be significantly simpler than the planner of the ego vehicle. For a planning and coordination of cooperative maneuvers, the maneuvers of the traffic vehicles generated in this way are generally not the maneuvers which the traffic vehicle would plan for itself, because neither the destination of the traffic vehicle (in particular, in urban scenarios) nor its dynamic preferences are known.

[0053] This problem may be solved by exchanging trajectories between the vehicles. In contrast thereto, in the approach presented here, each vehicle generates a planner for the instantaneous situation and its instantaneous destination and distributes it to vehicles in its surroundings, to thereby enable the receiving vehicles to correctly predict/calculate the trajectories of the third-party vehicles. The planner may and should be provided with a simple design (e.g., planning only along the middle of the lanes or planning on a grid) to enable a rapid calculation of the trajectories and a low capacity utilization of the V2X channel for transferring the reference planner. Moreover, the interface to the planner may be defined and/or standardized (representation of the surroundings made available to the planner, as well as representation of the trajectories returned by the planner). Alternatively, the sending vehicle may transfer the interface definition together with the planner, e.g., in the form of a JSON file. The more the AD systems and their modules become a commodity, the less need will there be for IP protection and concealment of the algorithms used, by which the standardization thereof and the implementation of the method presented here will be made easier.

[0054] Finally, it should be noted that terms such as “having,” “including,” etc. do not exclude other elements or steps, and terms such as “a” or “one” do not exclude a plurality.