Swarm-based trajectories for motor vehicles

Abstract

Method for generating swarm trajectories for a given lane (LS) of a roadway section, wherein - a large number of vehicles (F1, ..., Fn) traveling in the specified lane transmit their respective trajectories (T1, ... Tn) to a backend computer (BE), where a swarm trajectory for the specified lane is determined and stored from the transmitted trajectories, - determine the specified boundary conditions for the vehicles (F1, ..., Fn) while driving along the lane (FS) and transfer the determined boundary conditions together with the respective trajectory (T1, ..., Tn) to the backend computer (BE), - the backend computer (BE) determines and stores at least one swarm trajectory (ST) for the specified lane (FS) as a function of at least one boundary condition from the transmitted trajectories (T1, ... Tn) and associated boundary conditions, - at least the travel time is added as a boundary condition to the trajectories (T1, .., Tn) to be transmitted, whereby the travel time is realized by a travel time interval, - the date of travel, the day of the week, the weather, the traffic situation, the parking situation, the road surface condition and / or the lighting situation with regard to the specified lane (FS) are determined as further boundary conditions and added to the respective trajectories (T1, ... Tn), and - the boundary conditions are weighted differently when creating the swarm trajectories.

Description

[0001] The invention relates to a method for generating swarm trajectories for motor vehicles and their use in an ego vehicle.

[0002] The view of road traffic as a swarm of motor vehicles is now widespread in traffic research. For example, swarm-based simulation of road traffic has been successfully used to optimize traffic light phases at busy intersections.

[0003] If the trajectories of a large number of vehicles, i.e., a swarm of vehicles, are considered on a section of road, it turns out that the trajectories of the individual vehicles on that section are usually different. It is therefore possible to define an average path for the swarm of vehicles under consideration, which is called the swarm trajectory for the specific section of road.

[0004] For example, if a vehicle is driving autonomously on a section of road, such as a highway, its trajectory is typically centered between the lane markings. The trajectories driven on a given section of road are fed into a backend computer, which uses a multitude of different vehicles' trajectories to determine a swarm trajectory for that section. This swarm trajectory can then be retrieved from the backend computer by another vehicle on the same section of road to compare its own trajectory with the swarm trajectory and adjust it if necessary. This can be helpful, for example, when using assisted driving in construction zones or areas without lane markings, allowing drivers to plan or optimize their own trajectory.

[0005] The swarm trajectory described above is stored by the backend computer independently of time and made available to the user for the requested road section.

[0006] However, in certain areas of a roadway, especially in urban areas, an "ideal" trajectory depends on boundary conditions, since an ideal trajectory depends to a large extent on the state of traffic at a given time on the roadway area under consideration, on possible parked vehicles at the edge of the roadway, etc.

[0007] This leads to the following problems. If the swarm trajectory is averaged in the usual way, a mixture of trajectories with different boundary conditions results. This means that, in the worst case, the swarm trajectory no longer adequately satisfies any of the boundary conditions.

[0008] Furthermore, the driver of a vehicle expects a trajectory adapted to the boundary conditions. Therefore, even if a trajectory is correctly driven under one boundary condition, it may be perceived as less than ideal by the driver under a different boundary condition.

[0009] The German patent application DE 10 2016 212 292 A1 relates to a method for providing a plan trajectory for a motor vehicle, comprising the following steps: receiving provided environmental data of the motor vehicle's environment, providing a state space for a given planning horizon, wherein the state space comprises several states of the motor vehicle in the environment described by the environmental data, selecting a subset of the several states from the state space based on at least one selection criterion, creating a search graph, wherein the selected states each represent nodes of the search graph, and decomposing the search graph into at least two temporally or spatially successive sub-areas.wherein, subsequently, for each of the at least two temporally or spatially successive sub-areas, several trajectory candidates are calculated in parallel and evaluated according to at least one evaluation criterion in each of the sub-areas, and subsequently: compilation of a planned trajectory for the given planning horizon from individual calculated trajectory candidates, whereby those trajectory candidates connected across the sub-areas are selected whose evaluations result in the highest sum.

[0010] The publication DE 10 2011 083 677 A1 proposes a method for predicting a traffic situation for a vehicle, in which at least one possible future for the traffic situation of the vehicle is determined based on a current state of the vehicle and on historical data.

[0011] The German patent application DE 10 2015 014 651 A1 relates to a method for providing lane information for a motor vehicle's lane, in which, while the motor vehicle is moving on the lane, a trajectory of the motor vehicle is determined with respect to an external reference system, the determined trajectory is transmitted to an external central lane information collection device, and the lane information determined at least on the basis of the transmitted trajectory is provided by the lane information collection device to at least one other motor vehicle.

[0012] The publication DE 10 2013 019 112 A1 relates to a method for operating a driver assistance system for a motor vehicle. The driver assistance system comprises a monitoring device designed to receive data packets containing position data from other vehicles traveling on a road, and, based on the position data contained in the data packets, to determine a lane path for at least one lane of the road, and, based on the lane path, to determine at least one geometric property of the road and, depending on the at least one determined geometric property, to output a control signal to at least one device of the motor vehicle provided for assistance in driving the motor vehicle.

[0013] For current considerations and intended applications, the determination of a swarm trajectory for given road sections is carried out statically, so that the use of a swarm trajectory once determined may lead to unsatisfactory results.

[0014] The invention is therefore based on the objective of enabling the determination of improved swarm trajectories, so that the use of a swarm trajectory in an ego vehicle leads to a more satisfactory driving result for the driver.

[0015] This problem is solved by a method for determining a swarm trajectory for a road segment with the features of claim 1, and by a method for using a swarm trajectory determined for the road segment with the features of claim 3. Preferred embodiments of the invention are the subject of the dependent claims.

[0016] In the inventive method for generating swarm trajectories for a predetermined lane of a roadway section, - A large number of vehicles traveling in the designated lane transmit their respective trajectories to a backend computer, where a swarm trajectory for the designated lane is determined and stored from the transmitted trajectories. - the vehicles determine predefined boundary conditions while driving along the lane and transmit the determined boundary conditions together with the respective trajectory to the backend computer, and - the backend computer determines at least one swarm trajectory (ST) for the specified lane (FS) as a function of at least one boundary condition from the transmitted trajectories and associated boundary conditions and stores it, whereby - at least the travel time is added as a boundary condition to the trajectories to be transmitted, whereby the travel time is realized by a travel time interval.

[0017] By determining the boundary conditions prevailing while driving in the lane, parallel to the trajectory, boundary-condition-specific swarm trajectories can be determined and stored, so that these boundary-condition-specific swarm trajectories can be made available to other drivers in the corresponding lane. Driving time is, in effect, the basic boundary condition that is always determined and transmitted. This is based on the consideration that some traffic situations, such as the traffic density in a lane of a given road section, are in most cases dependent on driving time.

[0018] Furthermore, the date of travel, the day of the week, the weather, the traffic situation, the parking situation at the edge of the lane, the road surface condition, and / or the lighting conditions with respect to the specified lane are defined as additional boundary conditions and are added to the respective trajectories to be transmitted. This list of possible boundary conditions is not exhaustive.

[0019] To generate the boundary-condition-specific swarm trajectories, the boundary conditions are weighted differently. This can be particularly important when the swarm trajectory is a function of a combination of boundary conditions, such as date, day of the week, travel time, and parking situation. If the date is a public holiday, the day of the week boundary condition might be less significant.

[0020] Preferably, the swarm trajectories determined from the trajectories in conjunction with the boundary conditions are stored in a database, whereby swarm trajectories are determined for given boundary conditions and / or possible combinations of given boundary conditions for the given lane.

[0021] The inventive method for using a swarm trajectory in an ego-vehicle, wherein the swarm trajectory is determined according to the method described above, comprises the following steps: - Driving in the designated lane by the ego vehicle, - Determining predefined boundary conditions on the part of the ego vehicle while driving along the specified lane, whereby at least the driving time forms a boundary condition that is realized by a driving time interval, - Transmitting the determined boundary conditions to a backend computer that includes a database of swarm trajectories for the currently driven lane, in order to find at least one swarm trajectory relating to the transmitted boundary conditions, - Sending the at least one detected swarm trajectory for the specified lane to the ego vehicle, and - Comparing the current driving trajectory with the swarm trajectory.

[0022] Preferably, the ego-vehicle determines for which boundary condition or combination of boundary conditions the swarm trajectory is determined in the database. This can be achieved, for example, by transmitting only selected boundary conditions to the backend computer, so that it can only perform a database search for the transmitted boundary conditions. It is also conceivable that multiple swarm trajectories are transmitted to the ego-vehicle, corresponding to the number of boundary conditions and / or the desired combinations of boundary conditions.

[0023] Preferably, the current trajectory of the ego vehicle is corrected or influenced by the transmitted swarm trajectory, for example by determining the difference between the current trajectory and the swarm trajectory and using the difference to readjust the current trajectory of the ego vehicle.

[0024] A preferred embodiment of the invention is explained below with reference to the drawings. Fig. 1. The procedure for determining swarm trajectories in schematic representation, and Fig. 2. The use of such a determined swarm trajectory in a motor vehicle in a schematic representation.

[0025] Fig. Figure 1 schematically illustrates the determination of boundary-condition-dependent swarm trajectories for a road or track segment, where a fixed track segment is specified in the illustrative example. For further track segments, the procedure must be extended accordingly, and the respective track segment is included as a parameter in the analysis.

[0026] A number of vehicles F1, F2, ..., Fn-1, Fn travel along a predefined section of track, with each vehicle F1, F2, ..., Fn-1 and Fn generating an individual trajectory T1, T2, ..., Tn-1 and Tn along that section. Suitable devices installed in vehicles F1, F2, ..., Fn-1, Fn assign individual boundary conditions as parameter sets P1, P2, ..., Pn-1, Pn to each trajectory T1, T2, ..., Tn-1, Tn of the considered track section. Each trajectory T1, T2, ..., Tn-1 and Tn is assigned at least the parameter "time traveled along the predefined track section." In addition to the travel time, other boundary conditions or parameters include the current traffic situation, the parking situation at the edge of the track section, the current weather conditions, and / or the lighting conditions, which can be determined using appropriate environmental monitoring devices.The boundary conditions are then combined into a respective parameter set P1, P2, ..., Pn-1, Pn.

[0027] Formally, there exists a given set of boundary conditions Rk, where the index k can take the values ​​1 to m, i.e., k ∈ {1, ..., m}, from which the parameter sets P1 to Pn of the trajectories T1 to Tn are formed. Consequently, a parameter set Pi, with i ∈ (1, ..., n), can be written as the following tuple: ∈ Pi ⊆{Rj |j ∈ 1, …, k}

[0028] In other words, there exists a finite set of boundary conditions Rj, j∈ {1, ... k} and the parameter set Pi comprises a subset of boundary conditions Rj from the set of all subsets, i.e., the power set of the set of all boundary conditions Rj.

[0029] In this specific application, the boundary conditions Rj could be defined as follows: R1: Date and day of the week Mon to Sun, R2: Time in time intervals, for example in hours, R3: Weather, for example Sunny, Dry, Drizzle, Rain, Snow, Fog, Frost, R4: Parking situation on the right-hand side of the road with the values ​​empty, medium, full, R5: Traffic situation with the values ​​low, medium, heavy, congested.

[0030] In the example with the five boundary conditions R1 to R5 mentioned, the parameter set Pi would belong to the trajectory Ti, where the parameter set Pi includes the boundary conditions R1 to R5, thus forming the following tuple: Pi = (R1,R2,R3,R4,R5), for example R1 = (Weekday: Monday) R2 = (Time interval: 14:00 to 15:00) R3 = (Weather: Sunny and dry), R4 = (Parking situation right side of road: empty), and R5 = (Traffic: low).

[0031] The boundary conditions R1 to R5 mentioned above and their values ​​are only examples and serve for illustrative purposes. In the specific case, the boundary conditions and their specifications are adapted in detail to the application.

[0032] The trajectories T1 to Tn, including the associated parameter sets P1 to Pn, of each vehicle F1 to Fn are transmitted independently of one another via a transmission link ÜB, for example a radio link such as mobile communication, to a backend computer BE, preferably wirelessly. The information can also be stored in the "test vehicles" F1 to Fn and, after completion of the journey on the specified route, transmitted to the backend computer BE in any desired manner.

[0033] In a memory SP of the backend computer BE, the wirelessly transmitted information tuples (Ti, Pi) consisting of the trajectory Ti of the vehicle Fi and the associated parameter set Pi BE are stored for further processing and summarization.

[0034] In a processing and correlation facility (VK) for the stored information tuples (Ti, Pi), the trajectories Ti are linked together according to their correlation to form corresponding swarm trajectories ST, taking into account their associated parameters Pi. In other words, for specific combinations of boundary conditions Rj, i.e., for specific parameter sets, a swarm trajectory ST is determined, for example, as the mean of the trajectories Ti calculated for this parameter set.

[0035] Assuming that for a given section of road under consideration, a total of m trajectories Tm, m ∈ {1, ... , n}, were determined during the measurement period for a given parameter set Pi ⊆ {Rj | j ∈ 1, ..., k}, then there are m information tuples (Tm, Pi). Consequently, the m trajectories can be considered quasi as a function of the considered parameter set Pi, i.e., Tm(Pi), and using a simple averaging, the swarm trajectory ST(Pi) is obtained as a function of the m information tuples: ST(Pi)=∑1mTi(Pi) / m, where the parameter set Pi could look like the example above. In other words, there is a swarm trajectory ST(Pi) for the parameter values ​​Pi: R1 = (Weekday: Monday) R2 = (Time interval: 14:00 to 15:00) R3 = (Weather: Sunny and dry), R4 = (Parking situation right side of road: empty), and R5 = (Traffic: low).

[0036] Furthermore, the calculation of the mean value here is only an example. Other methods and procedures for combining and correlating the determined trajectories for a combination of boundary conditions are conceivable.

[0037] The swarm trajectories ST(Pi) resulting from the correlations for a given parameter set Pi and the considered road section are then stored in a knowledge database DB to enable retrieval from external sources.

[0038] It is therefore not only possible to generate swarm trajectories for a so-called "complete" parameter set Pi(R1, ... Rm) that includes all boundary conditions, but also swarm trajectories for individual boundary conditions Ri, i ∈ {1, ..., m} and for desired combinations of boundary conditions, for example Ri,j with i, j ∈ {1, ..., m}, where the combinations include all subsets of the set of all boundary conditions Ri, i ∈ {1, ..., m}.

[0039] In this way, it is also possible to determine further swarm trajectories ST(Ri, .., Rj) and store them in the backend calculator BE, which, for example, satisfy the conditions R1 = "Tuesday" and R2 = "2:00 PM - 3:00 PM" or R1 = "Tuesday", R2 = "2:00 PM - 3:00 PM" and R4 = "empty". This provides the user with information about the swarm trajectory for the specified day of the week and the desired time interval, or for day of the week R1, time interval R2, and parking situation R4, without considering further constraints such as weather R3, parking situation R4, and traffic R5 in the first example, or weather R3 and traffic R5 in the second example.

[0040] It is also important to consider region-specific effects. For example, drivers react differently to traffic conditions and utilize available degrees of freedom more or less effectively. Furthermore, the DB knowledge database contains both the swarm trajectories for the various combinations of boundary conditions and the historical data on the different boundary conditions for the corresponding road segments.

[0041] Fig. Figure 2 shows, schematically, the use of a swarm trajectory in an ego-vehicle traveling a specific section of road. For example, an ego-vehicle Fego travels along a lane FS of an inner-city roadway, passing a parking lane PS located to the right of lane FS, at a speed v. The parking lane PS is partially occupied, specifically in this case in the Fig.2 with two parked vehicles FP1 and FP2, where the second parked vehicle FP2 is located very close to the driving lane FS. Therefore, it is to be expected that the trajectory of the ego vehicle Fego, when passing the second parked vehicle FP2, will not run in the middle of the driving lane FS, but will instead run to the left towards the center of the driving lane FB due to the second parked vehicle FP2.

[0042] The ego vehicle Fego uses its sensors (not shown) to measure the necessary boundary conditions, such as the boundary conditions R1 to R5 mentioned above for the road section FB on which Fego is currently located. These measurements are combined into a parameter set Pego. This parameter set Pego is transmitted via mobile network using the schematically depicted Internet IN to the responsible backend computer BE. Based on the transmitted parameter set Pego, the backend computer BE determines a corresponding swarm trajectory ST(Pego) in its database. This trajectory is then transmitted via the Internet using the appropriate mobile connection to Fego, where it is used to plan and / or correct the vehicle's own driving trajectory. Reference symbol list F1 Car 1 F2 Vehicle 2 Fi vehicle i Fn-1 Vehicle n-1 Fn vehicle n T1 Trajectory of vehicle 1 T2 Trajectory of vehicle 2 Tn-1 Trajectory of vehicle n-1 Vehicle trajectory n P1 parameter set for trajectory T1 P2 parameter set for trajectory T2 Pn-1 parameter set for trajectory Tn-1 Pn parameter set for trajectory Tn ÜB transmission BE Backend Calculator SP storage VK processing and correlation DB database Fego Ego vehicle FP1 first parked vehicle FP2 second parked vehicle PS Parking lane FB Roadway FS lane Pego parameter set ST swarm trajectory ON the Internet v speed

Claims

[1] Method for generating swarm trajectories for a given lane (FS) of a road section, wherein - a plurality of vehicles (F1, ..., Fn) traveling in the specified lane transmit their respective travel trajectories (T1, ... Tn) to a backend computer (BE), where a swarm trajectory is determined and stored from the transmitted trajectories for the specified lane, - the vehicles (F1, ..., Fn) determine specified boundary conditions while driving along the lane (FS) and transmit the determined boundary conditions together with the respective trajectory (T1, ..., Tn) to the backend computer (BE), - the backend computer (BE) determines and stores at least one swarm trajectory (ST) for the specified lane (FS) from the transmitted trajectories (T1, ... Tn) and associated boundary conditions as a function of at least one boundary condition, - at least the travel time is added as a boundary condition to the trajectories (T1, .., Tn) to be transmitted, whereby the travel time is realized by a travel time interval, - the date of travel, the day of the week, the weather, the traffic situation, the parking situation, the road condition and / or the lighting situation with regard to the given lane (FS) are determined as further boundary conditions and added to the respective trajectories (T1, ... Tn), and - when creating the swarm trajectories, the boundary conditions are weighted differently. [2] Method according to claim 1, characterized by that the determined swarm trajectories are stored in a database (DB), whereby swarm trajectories for given boundary conditions and / or possible combinations of given boundary conditions for the given lane (FS) are determined. [3] Method for using a swarm trajectory (ST) in an ego vehicle (Fego), wherein the determination of the swarm trajectory is carried out by means of the method according to one of the preceding claims, comprising the steps: - Driving in the specified lane (FS) by the ego vehicle (Fego), - Determination of specified boundary conditions (Pego) by the ego vehicle (Fego) while driving along the specified lane (FS), whereby at least the driving time realized by a driving time interval represents a boundary condition, - transmitting the determined boundary conditions (Pego) to a backend computer (BE) which comprises a database (DB) with swarm trajectories for the currently travelled lane (FS), in order to find at least one swarm trajectory (ST(Pego)) relating to the transmitted boundary conditions, - Sending the detected swarm trajectory (ST(Pego)) for the given lane (FS) to the ego vehicle (Fego), and - Compare the current trajectory with the swarm trajectory (ST(Pego)). [4] Method according to claim 3, characterized by that the ego vehicle (Fego) determines for which boundary condition (Pego) or combination of boundary conditions (Pego) the swarm trajectory (ST(Pego)) is determined in the database (DB). [5] Method according to one of claims 3 or 4, characterized by that the current trajectory of the ego vehicle (Fego) is corrected or influenced by the transmitted swarm trajectory (ST(Pego)). [6] Method according to claim 5, characterized by that the difference between the current driving trajectory and the swarm trajectory (ST(Pego)) is determined and the difference is used to adjust the current trajectory.

Images