Limited agent model use in hybrid traffic simulation with real trajectories
Patent Information
- Application Number
- EP2024700960
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-02-27
- Filing Date
- 2024-01-15
- Publication Date
- 2026-01-07
Smart Images
Figure EP2024050780_06092024_PF_FP
Abstract
Description
[0001] LIMITED AGENT MODEL USE IN HYBRID TRAFFIC SIMULATION WITH REAL TRAJECTORIES
[0002] The invention relates to a method for carrying out a traffic simulation for testing a vehicle component.
[0003] While automated driving control systems can support a human driver in the sense of driver assistance systems, they will become increasingly important for highly automated and autonomous vehicles in the future. Depending on the degree of authority, such automated driving control systems intervene more or less in the driving control of a vehicle and, particularly in autonomous vehicles, completely take over vehicle control. This leads to high safety requirements, as vehicles participating in road traffic inherently pose a risk to other road users in the event of incorrect behavior.Driving control functions receive data from corresponding sensor units to process direct measured values such as the current vehicle speed, but also to detect and interpret the environment, particularly other road users and objects in the surrounding area, including traffic signs, using complex perception processes. Since highly complex software components are often required to execute the automated driving control functions, take over vehicle control, and make appropriate decisions in traffic, demonstrating the required safety is correspondingly difficult and complex.
[0004] However, testing a sufficient number of real-world scenarios for the development and final validation of automated driving control systems in practice would require an unacceptably large number of distances driven and a correspondingly large number of driving hours for a vehicle equipped with an automated driving control system. Covering billions of kilometers, for example, and the correspondingly delayed evaluation of the scenarios experienced there is disproportionately long for the development and validation of an automated driving control system and cannot be carried out in practice. A purely distance-based, statistical proof of the safety of the driving function before the market launch of such a vehicle is therefore technically not feasible. This is also explained in the following publication: "W. Wachenfeld and H. Winner, “The release of autonomous vehicles,” in Autonomous Driving: Technical, Legal and Social Aspects, M. Maurer, J.C. Gerdes, B.Lenz, and H. Winner, Eds. Springer, 2016, pp. 425-449." Future homologation, e.g., according to NCAP, will also require an increasingly complex test procedure that can only be successfully carried out with the help of realistic simulation environments.
[0005] Furthermore, due to the risk to third parties, it is not possible to test potentially critical scenarios in public traffic. However, it is precisely these potentially critical scenarios that provide the data that must be taken into account in the development of an automatic driving control system for a vehicle, as these are the ones that lead to accidents. These critical scenarios are also important for the validation of a fully developed automatic driving control system. In the development and validation of automatic driving control systems, simulation-based methods are therefore used and tested, particularly in the operating areas that are critical for the automated driving control function, such as when encountering the end of a traffic jam or in busy urban traffic areas.Scenario-based development and testing can be used to estimate how safely the future real vehicle will operate in the open world (possibly within a limited operating range) by using the digital twin of the automated vehicle during early development phases.
[0006] In automated vehicles, trajectories are necessarily continuously planned and adjusted by an automated driving control system. Trajectory planning is typically based, among other things, on the perception of the environment by sensors, communication data from vehicle-to-traffic element communication, and high-resolution map data. The automated driving control system constantly regulates the lateral and longitudinal movement of the vehicle to maintain these planned trajectories. The quality of adherence to the planned trajectory therefore also depends at all times on the function and quality of the information acquired by the driving control system.If an automated driving control system were to only follow a predetermined optimal trajectory without the ability to adapt it in near real time, dynamic changes in traffic could not be processed safely and the reactions and adaptations to current traffic conditions and the behavior of other road users necessary for driving safety and the functioning of traffic flow could not be carried out.
[0007] So-called "adaptive replay-to-sim" methods are known in the state of the art. In such methods, real-world scenarios are recorded and information about them is made available for simulation, allowing tests of driving control systems (or their sensors) to be conducted based on realistic data. The advantage of transferring the data to a simulation is the possibility of safe virtual testing, as well as the possibility of deliberately modifying the real-world data.
[0008] DE 102019206 908 B4 relates to a method for training at least one algorithm for a control unit of a motor vehicle, wherein the control unit is provided for implementing an automated or autonomous driving function by intervening in assemblies of the motor vehicle on the basis of input data using the at least one algorithm, wherein the algorithm is trained by a self-learning neural network, comprising the following steps: a) providing a computer program product module for the automated or autonomous driving function, wherein the computer program product module contains the algorithm to be trained and the self-learning neural network, b) embedding the trained computer program product module in the control unit of the motor vehicle, c) driving the motor vehicle in a real traffic environment by a human driver, wherein the driving determines a driven trajectory,d) feeding data from an environmental sensor system and a motor vehicle sensor system to the control unit and calculating a virtual trajectory using the algorithm, e) deriving a metric from a comparison of the driven trajectory and the virtual trajectory and storing the data from the environmental sensor system and the motor vehicle sensor system in a memory if certain metric criteria are met for a traffic situation, f) providing information regarding the traffic situation to a traffic simulation, g) analyzing the traffic situation using the traffic simulation, whereby a virtual image of the traffic situation is created using the map data, the environmental sensor system and the motor vehicle sensor system, whereby the data of the traffic situation are varied using the traffic simulation,and h) training the algorithm by varying the traffic situation. The traffic simulation is carried out as a simulation of the traffic situation from the perspective of the traffic environment. Behavioral and technological models are provided for each individual road user in the traffic simulation, so that a future traffic situation is simulated based on a variation of the provided behavior and technological models. Furthermore, past environmental and vehicle data are also varied in the traffic simulation.
[0009] For testing and / or validating driver assistance systems that support manual driving of vehicles such as passenger cars or trucks, as well as automation systems for highly automated and even autonomous vehicles (collectively referred to as "automated driving control functions"; English: "Advanced Driver Assistance Systems" or "Autonomous Driving Systems," collectively abbreviated to ADAS / AD), the use of scenario libraries is also known. These ideally contain descriptions of a multitude of (ideally as widely as possible) varying scenarios, so that a vehicle does not encounter fundamentally untested situations during subsequent normal operation. In such scenario libraries, scenario descriptions are typically stored in a machine-readable and automatically storable form.Scenario description languages, which employ specialized documentation and programming languages, are typically used for this purpose. One example is the established scenario library of the "ASAM eV" association called "OpenSCENARIO®." There are also scenario libraries with data in machine-readable formats that also provide human-understandable context, such as "ADScene" and "SafetyPool."
[0010] In order to be able to investigate the behavior of a driving control system or its sensor unit in a reactive manner to the environment of a road user under consideration, in particular an automated vehicle (typically called "vehicle under test"), with the help of such a scenario library, the journey of this selected real road user under consideration can be simulated, while the behavior of the other real road users as well as the environmental parameters can be taken from the scenario library and, in this case, are based on real data.
[0011] However, because the data on the behavior of other real road users and on the environmental parameters reflect facts that have come about as a result of the other road users adapting their reactions to the behavior of the real road user under consideration (which is replaced in the simulation by the "vehicle under test" as a virtual counterpart), the behavior of the other road users recorded in the simulation according to their stored data is only valid if the original behavior of the real road user under consideration and that of its virtual counterpart in the simulation (the simulated "vehicle under test") are essentially the same.If the simulated behavior of the road user as a virtual "vehicle under test" deviates too much from the behavior of the real road user under consideration, the integrity of the degree of realism of the simulation environment is limited and, in some cases, is therefore unsuitable for testing a driving control system or its sensors or a sensor unit for the simulated and thus virtual "vehicle under test".
[0012] By using recorded, real trajectories, a realistic simulation can be conducted, which is more suitable for the evaluation of automated driving systems than the use of purely artificial scenarios. However, by substituting a road user for an automated driving system under test, the recorded data becomes unnatural as soon as the automated driving system deviates from the recorded trajectory. This problem is addressed in the publication "A Needle in a Haystack - How to Derive Relevant Scenarios for Testing Automated Driving Systems in Urban Areas" by Nico Weber, Dr.-lng. Christoph Thiem, and Prof. Dr.-lng. Ulrich Konigorski, part of the preprint for the "30th Aachen Colloquium Sustainable Mobility 2021" as follows: A simulation-based toolchain for the development and testing of vehicles equipped with automated driving control systems in urban environments is presented.Furthermore, multimodal interactions between different road users are considered and provided for a scenario library. If the above-described case occurs, where recorded data becomes unnatural when used in the simulation as soon as the automated driving control system detects trajectories that deviate from the recorded ones, the aforementioned publication proposes taking over control of the road users by software agents. A disadvantage of this solution, however, is that real data is no longer used after the switch from recorded data to artificial software agents. This weakens the argument of replicating real behavior in the simulation and the associated credibility of real data.
[0013] It is therefore an object of the invention to remedy this disadvantage and, in particular, to carry out such a traffic simulation that can ensure plausible behavior of road users and, at the same time, can maximize the use of actually recorded data.
[0014] The invention is based on the features of the independent claims. Advantageous developments and refinements are the subject of the dependent claims.
[0015] A first aspect of the invention relates to a method, in particular a computer-implemented method, for carrying out a traffic simulation, comprising the steps:
[0016] - Providing a data set with information about a real traffic scenario from a scenario library, wherein the data set comprises movement data of real road users in the traffic scenario, at least one of which is a vehicle,
[0017] - Starting the execution of the traffic simulation with simulated counterparts of the real road users in such a way that the movement data of the simulated road users are equal to the movement data of the real road users according to the data set, with the exception of the movement data of a vehicle to be replaced from the real road users, which is simulated as a virtual test vehicle and whose movement data is recalculated in the simulation,
[0018] - sequentially checking whether a metric relating to a deviation of the movement data of a respective non-replaced road user according to the data set from planned movement data calculated for him in the simulation in response to the behavior of the virtual test vehicle exceeds a predetermined first limit value, and retaining the movement data according to the data set for that respective non-replaced road user if the predetermined first limit value is not exceeded, and otherwise:
[0019] - Replacing or mixing the movement data of this respective non-replaced road user, which has been adopted according to the data set, with movement data calculated by an agent model assigned to it and executed in the simulation, and executing a return trajectory to the movement data according to the data set by this respective non-replaced road user and thus terminating the execution of the associated agent model if a predefined termination condition is met.
[0020] In one step of the method, preferably continuously during the traffic simulation, the vehicle component to be tested is tested by executing the traffic simulation. The vehicle component to be tested comprises, in particular, a corresponding sensor and / or a trajectory planner such as a driving control computer.
[0021] The vehicle component to be tested is either integrated into the simulation as a physical device using a hardware-in-the-loop method, or is represented virtually and also simulated. This results in improved testing of the vehicle component to be tested compared to conventional testing methods.
[0022] The dataset contains information about a real traffic scenario, for which an intersection or a large area was captured, in particular from a bird's eye view, for example using an unmanned aerial vehicle, and in particular the movement data of the road users present there was recorded. All movement data, except for at least one vehicle, whose perception system and / or automated driving control system is to be tested (as a simulated system or via hardware-in-the-loop), are initially retained in the simulation. One or more replaced road users, however, are simulated as respective virtual test vehicles, and their movement data, in particular their trajectories, are recalculated during the simulation runtime.
[0023] A metric is now used to determine a measure for the following deviation: While predefined and / or recorded movement data of the non-replaced road users, stored in the dataset, serve as a reference, currently planned movement data, in particular planned trajectories including the trajectory and speed (and optionally also the acceleration), of the respective non-replaced road user are determined in relation to this during the simulation runtime and compared with the data in the dataset. A simple metric is the area over the differences between the respective trajectories; furthermore, speed differences, acceleration differences, or a combination of these differences and areas / integrals can also be used.
[0024] The result of calculating the value of this metric is compared with the first specified threshold. If the value of the metric is below the first specified threshold, the specified or recorded motion data from the dataset can be considered sufficiently consistent with the current simulation situation, particularly the behavior of the virtual test vehicle.
[0025] However, if this metric exceeds the first specified limit, significant deviations are to be expected and it would be tantamount to continue using the movement data of the non-replaced road users in the simulation that were specified in advance or recorded and stored by observing reality, as this could lead to inconsistent situations with the virtual test vehicle.
[0026] Accordingly, for a limited period of time, the recorded or specified movement data of the non-replaced road users can be replaced or merged with (or with) movement data calculated by agent models for the respective non-replaced road users. In particular, precisely those non-replaced road users are simulated by agent models for whom the aforementioned metric exceeds the first specified threshold. Therefore, advantageously, not all non-replaced road users need to be simulated by agent models at the same time; for non-replaced road users for whom the metric is below the first specified threshold, the specified or recorded movement data according to the dataset can continue to be used in the traffic simulation.
[0027] In contrast to the fixed movement data from the data set, an agent model continuously calculates new predetermined trajectories for the corresponding road user over the course of the simulation time, i.e. the agent model, in contrast to the movement data according to the data set, is able to react to its environment as determined in the simulation, in particular to react to the behavior and the presence of the virtual test vehicle, whose trajectories are by definition not known a priori before the simulation is carried out, since they are only determined during the simulation.
[0028] However, this (replacement or mixing) of the specified or recorded movement data according to the data set with the movement data currently determined from the (one or more) agent models with the simulation time only occurs for as long as necessary, which is expressed by a termination condition. This termination condition checks, in particular, whether a changed behavior of the respective non-replaced road user compared to the movement data according to the data set is still currently necessary, or whether the behavior calculated according to the respective agent models is not already sufficiently close to the movement data of the data set, which is preferably done by comparing a metric for determining a measure of this deviation with a second specified limit value, wherein the first limit value is preferably equal to the second limit value.
[0029] It is therefore an advantageous effect of the invention that traffic simulations do not unnecessarily lose realism when motion data calculated by agent models are used for an unnecessarily long time instead of using motion data from real traffic scenarios, since the latter typically and naturally exhibit a high degree of realism and thus contribute to a high degree of realism and, in turn, high reliability of the simulation. In particular, long-term scenarios can thus be used realistically in a traffic simulation to test vehicle components such as sensors or driving control systems, and agent models are only used as necessary. However, the use of agent models ensures consistent and plausible scenario generation in the traffic simulation.
[0030] According to an advantageous embodiment, the data set comprises, in addition to real movement data, manually specified movement data of non-replaced road users.
[0031] According to a further advantageous embodiment, a sequentially repeated determination of a possible return trajectory to the movement data according to the data set takes place, wherein the termination condition is only fulfilled if a metric regarding a deviation of the currently determined possible return trajectory and its planned movement data of the non-replaced road user calculated in the simulation in response to the behavior of the virtual test vehicle exceeds a predetermined second limit value.
[0032] According to a further advantageous embodiment, the sequentially repeated determination of a possible return trajectory takes place within constant time periods related to the simulation time.
[0033] According to a further advantageous embodiment, the termination condition is only fulfilled if, when calculating the planned movement data of the non-replaced road user in response to the behavior of the virtual test vehicle, a restriction that must be taken into account in the traffic situation influenced by the behavior of the virtual test vehicle has ended in the simulation.
[0034] Such a restriction that must be taken into account occurs, for example, when the virtual test vehicle stops unexpectedly and blocks a traffic route for another road user following behind, so that the movement data determined for the other road user from the real traffic scenario can no longer be executed without the following road user causing a rear-end collision with the virtual test vehicle in the simulation. The same applies to evasive situations or at intersections where the other road user must yield the right-of-way to the virtual test vehicle. This may have been irrelevant in the real traffic scenario due to changes in the time of the encounter between the virtual test vehicle and another road user, but in the simulation the other road user must observe the right-of-way and therefore adjust their planned trajectory.
[0035] According to a further advantageous embodiment, the movement data of the non-replaced road user adopted according to the data set are mixed with the planned movement data calculated by the associated agent model in such a way that a proportion of the calculated planned movement data with a larger deviation is reduced.
[0036] This concerns the above-mentioned deviation of the movement data of a real but non-replaced road user depicted in the simulation according to the data set and its planned movement data calculated in the simulation in response to the behavior of the virtual test vehicle.
[0037] According to a further advantageous embodiment, the movement data adopted according to the data set are mixed with the planned movement data calculated by the associated agent model for the respective non-replaced road user if the metric lies between the first predetermined limit value and a first threshold value, and the movement data adopted according to the data set are replaced by the planned movement data calculated by the associated agent model if the metric lies between the first threshold value and a second threshold value that is higher than the first.
[0038] According to a further advantageous embodiment, the return trajectory is carried out by mixing the movement data acquired according to the data set with the planned movement data calculated by the associated agent model with the proportion of the movement data acquired according to the data set increasing over time.
[0039] According to a further advantageous embodiment, the sequentially repeated checking of the metric takes place within respective constant time periods related to the simulation time.
[0040] According to a further advantageous embodiment, the respective movement data comprise trajectory information, wherein the trajectory information comprises one or more of the following information: geometric trajectory curve, speed assumed on the geometric trajectory curve, acceleration assumed on the geometric trajectory curve.
[0041] A further aspect of the invention relates to a data processing device comprising means for carrying out the steps of the method as described above and below.
[0042] A further aspect of the invention relates to a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method as described above and below.
[0043] Advantages and preferred developments of the proposed device or the proposed computer program product result from an analogous and analogous transfer of the statements made above in connection with the proposed method.
[0044] Further advantages, features, and details will become apparent from the following description, which – where appropriate with reference to the drawings – describes at least one embodiment in detail. Identical, similar, and / or functionally equivalent parts are provided with the same reference numerals.
[0045] They show:
[0046] Fig. 1 : A method for performing a traffic simulation according to an embodiment of the invention.
[0047] Fig. 2: An example scenario of the traffic simulation according to Fig. 1.
[0048] The representations in the figures are schematic and not to scale.
[0049] Fig. 1 shows a computer-implemented method for conducting a traffic simulation. Before the method is implemented, a real traffic scenario is observed. In the traffic scenario, several road users occur within a limited area, and depending on their distance from each other, at least some of them interact through action-reaction behavior. In this example, the traffic scenario is captured from a bird's-eye view by an unmanned aerial vehicle or with corresponding camera systems on masts or tall buildings. Furthermore, access to sensor information is available.which were recorded from the ego perspective of at least one vehicle in the traffic scenario. All recordings thus obtained are stored in a scenario library. In a first step of the method, a data set containing information about this real traffic scenario is provided S1 from the scenario library. The data set includes movement data of the real road users in the observed traffic scenario. This is followed by the initial execution S2 of the traffic simulation with simulated counterparts of the real road users such that the movement data of the simulated road users are equal to the movement data of the real road users according to the data set, with the exception of the movement data of a vehicle to be replaced from the real road users.which is simulated as a virtual test vehicle and whose movement data is recalculated in the simulation. Finally, at regular intervals related to the simulation time, a check S3 is repeatedly carried out to determine whether a metric relating to a deviation of the movement data of a respective non-replaced road user according to the data set from the planned movement data calculated for him in the simulation in response to the behavior of the virtual test vehicle exceeds a predetermined first limit value, and the movement data is retained according to the data set for this respective non-replaced road user if the predetermined first limit value is not exceeded,and otherwise: replacing S4 the movement data of this respective non-replaced road user, which has been transferred according to the data set, with movement data calculated by an agent model assigned to it and executed in the simulation, and executing a return trajectory to the movement data according to the data set by this respective non-replaced road user, thus terminating the execution of the associated agent model if a predefined termination condition is met.
[0050] Fig. 2 shows an example sequence in a Cartesian coordinate system with the coordinate axes x and y from a bird's eye view. The trajectory specified according to the data set for the non-replaced road user using the specified movement data is shown as a dash-dot-dash sequence. The trajectory planned according to the planned movement data calculated in the simulation is shown as a solid curve. Up to point A, the other road user follows the specified trajectory and is in state A. From point A, the metric of the deviation of the trajectories exceeds the first specified limit value. The non-replaced road user then follows the calculated planned trajectory in state B. The maximum possible deviation from the currently traveled position (taking into account the maximum acceleration and speed values) is shown in a densely dashed curve and marked with C.Using this curve (C), an intersection point with the trajectory specified in the data set is now calculated. The intersection point serves as a reference point for trajectory E (green) to calculate a return trajectory for the vehicle to approach the trajectory specified in the data set (dash-dot-dash sequence). Trajectory E is fed into the non-replaced trajectory as the target trajectory. This trajectory is exited at point F. The deviation between the calculated trajectory and the trajectory specified in the data set is again below the first specified limit, and the other road user again follows the trajectory specified in the data set. This leaves point B, the beginning of the return to the trajectory specified in the data set. This point can only be reached when the hazardous situation, or more generally the necessity, is no longer present.One approach for detecting this state is to utilize the aforementioned metric to switch from the trajectory specified in the data set to the trajectory calculated in the simulation. As soon as the other road user at point A (state B) follows the trajectory calculated in the simulation, specific return trajectories (E) are continuously calculated (every x seconds to reduce computational effort). The metric then refers to the difference between the return trajectory and the calculated trajectory. If the second specified threshold is exceeded, the return trajectory is set as a supplement to the specified trajectory, and the other road user follows the return trajectory towards the trajectory specified in the data set. If the second threshold is not exceeded (the automated driving system detects a danger zone), the non-replaced road user continues to follow the calculated trajectory of the agent model.
[0051] Although the invention has been illustrated and explained in detail by preferred embodiments, the invention is not limited by the disclosed examples, and other variations may be derived therefrom by those skilled in the art without departing from the scope of the invention. It is therefore clear that a multitude of variations exist. It is also clear that exemplary embodiments are truly only examples and should not be construed as limiting the scope, possible applications, or configuration of the invention in any way.Rather, the preceding description and the description of the figures enable the person skilled in the art to implement the exemplary embodiments in concrete terms. With knowledge of the disclosed inventive concept, the person skilled in the art can make various changes, for example, with regard to the function or arrangement of individual elements mentioned in an exemplary embodiment, without departing from the scope of protection defined by the claims and their legal equivalents, such as further explanations in the description. List of reference symbols.
[0052] 51 Providing a data set
[0053] 52 Starting the traffic simulation
[0054] 53 Checking a metric S4 Replacing or mixing
Claims
Patent claims 1. A method for performing a traffic simulation for testing a vehicle component, comprising the steps: - Providing (S1) a data set with information about a real traffic scenario from a scenario library, wherein the data set comprises movement data of real road users in the traffic scenario, at least one of which is a vehicle, - Starting (S2) the execution of the traffic simulation with simulated counterparts of the real road users in such a way that the movement data of the simulated road users are equal to the movement data of the real road users according to the data set, with the exception of the movement data of a vehicle to be replaced from the real road users, which is simulated as a virtual test vehicle with a vehicle component to be tested and whose movement data is recalculated in the simulation, and testing the vehicle component to be tested, - sequentially checking (S3) whether a metric relating to a deviation of the movement data of a respective non-replaced road user according to the data set of planned movement data calculated for him in the simulation in response to the behavior of the virtual test vehicle exceeds a predetermined first limit value, and retaining the movement data according to the data set for this respective non-replaced road user if the predetermined first limit value is not exceeded, and otherwise: - Replacing or mixing (S4) the movement data of this respective non-replaced road user, which has been taken over according to the data set, with movement data calculated by an agent model assigned to it and executed in the simulation, and executing a return trajectory to the movement data according to the data set by this respective non-replaced road user and thus terminating the execution of the associated agent model if a predefined termination condition is present.
2. The method according to claim 1, wherein the data set comprises, in addition to real movement data, manually specified movement data of non-replaced road users.
3. Method according to one of the preceding claims, wherein a sequentially repeated determination of a possible return trajectory towards to the movement data according to the data set, whereby the termination condition is only fulfilled if a metric regarding a deviation of the currently determined possible return trajectory from the planned movement data of the non-replaced road user calculated in the simulation in response to the behavior of the virtual test vehicle exceeds a predetermined second limit value.
4. The method according to claim 3, wherein the sequentially repeated determination of a possible return trajectory takes place within constant time periods related to the simulation time.
5. Method according to one of the preceding claims, wherein the termination condition is only fulfilled if, in the calculation of the planned movement data of the non-replaced road user in response to the behavior of the virtual test vehicle, a restriction which must be taken into account in the traffic situation influenced by the behavior of the virtual test vehicle in the simulation has ended.
6. Method according to one of the preceding claims, wherein the mixing of the movement data of the non-replaced road user adopted according to the data set with the planned movement data calculated by the associated agent model is carried out in such a way that a proportion of the calculated planned movement data with a greater deviation is reduced.
7. Method according to one of the preceding claims, wherein the movement data adopted according to the data set are mixed with the planned movement data calculated by the associated agent model for the respective non-replaced road user if the metric lies between the first predetermined limit value and a first threshold value, and wherein the movement data adopted according to the data set are replaced by the planned movement data calculated by the associated agent model if the metric lies between the first threshold value and a second threshold value higher than the first.
8. Method according to one of the preceding claims, wherein the return trajectory is created by mixing the movement data taken over according to the data set with the data calculated by the associated agent model. planned movement data with an increasing proportion of the movement data transferred according to the data set over time.
9. Method according to one of the preceding claims, wherein the sequentially repeated checking of the metric takes place within respective constant time periods related to the simulation time.
10. Method according to one of the preceding claims, wherein the respective movement data comprise trajectory information, wherein the trajectory information comprises one or more of the following information: geometric trajectory curve, speed assumed on the geometric trajectory curve, acceleration assumed on the geometric trajectory curve.