Method and device for estimating at least one object state of a moving object in the environment of a vehicle

Abstract

Method for estimating at least one object state (30) of a moving object (20) in the environment of a vehicle (50), wherein environmental data (10) of the environment are recorded by means of at least one sensor (51), wherein, based on the recorded environmental data (10), moving objects (20) in the environment are detected, wherein at least one object state (30) is estimated for at least some of the detected moving objects (20), wherein the estimation of the at least one object state (30) takes place taking into account provided swarm trajectories (11-y), and wherein the estimated at least one object state (30) is provided, wherein a detected moving object (20) is assigned to at least one provided swarm trajectory (11-y), wherein an assignment (21) of the detected moving object (20) to the at least one provided swarm trajectory (11-y) is carried out as a function of cost values (23-x / y) which are calculated by means of a predetermined cost function, and wherein the estimation of the at least one object state (30) of the detected moving object (20) is carried out taking into account the assignment (21).

Description

[0001] The invention relates to a method and a device for estimating at least one object state of a moving object in the environment of a vehicle.

[0002] The basis of current and future driver assistance systems is the perception of the vehicle's surroundings with the help of sensors. In the environmental data collected by the sensors, other objects are detected and their states, such as object position and speed, are estimated. Based on these estimated object states, the driver assistance system makes certain decisions and implements measures using the vehicle's actuators. For example, during a driver-initiated automated lane change, the system checks whether the area behind the vehicle in the lane to which the vehicle is to change is clear.

[0003] However, sensors have limitations and inaccuracies due to their design and / or physical nature. For example, it may happen that another vehicle is detected as a moving object, but the vehicle is assigned to the wrong lane or its direction of movement cannot be correctly determined.

[0004] DE 10 2012 009 297 A1 discloses a method for assisting a driver when driving a vehicle. Driver instructions are issued depending on a predicted future potential collision risk and / or subsequent collision risk between the vehicle and other road users within the vehicle. Maneuver options of the vehicle and the other road users are predicted based on their probabilistically detected intentions. Based on the maneuver options, several competing situation hypotheses between the vehicle and all relevant other road users are determined. For each situation hypothesis, a risk assessment of a respective potential or actual collision risk and / or subsequent collision risk is performed using logical context rules.In addition, motion hypothesis trajectory bundles are predicted and used to determine the probability of a real collision, along with the margin of movement between the vehicle and other road users. The situation hypotheses are hierarchically ranked in a priority list based on the intentions, depending on the respective real or acute collision risk and / or subsequent collision risk, with or without margin of movement to defuse the situation.Furthermore, depending on the priority list and a determined driver status of the driver and / or driver preferences, the driver of the vehicle is informed, warned and / or supported by means of an automatic intervention in a longitudinal and / or lateral control of the vehicle in several escalation levels by means of the driver instructions, whereby the information, warning and automatic intervention are carried out for the situation which has a risk of collision and / or consequential collision risk with the highest priority in the priority list and a resulting highest criticality.

[0005] DE 10 2018 123 896 A1 discloses a method for operating an at least partially automated vehicle. Here, at least one object in the vehicle's surroundings is detected from environmental sensor data, and an environmental model of the vehicle is created. An object position and at least one object movement feature are determined for the detected object and transferred to the environmental model. Using an optimization method, a trajectory for the vehicle is calculated from the environmental model, and the vehicle is controlled according to the calculated trajectory.

[0006] From DE 10 2017 212 629 A1 a prediction device for a motor vehicle for predicting a future behavior of a road user in the environment of the motor vehicle is known, wherein the prediction device is configured to determine a road user class of the road user, to receive a movement model dependent on the road user class of the road user from a vehicle-external learning device, to adapt the received movement model depending on an observed current behavior of the road user, and to predict the future behavior of the road user depending on the adapted movement model.

[0007] DE 10 2018 202 712 A1 discloses a method for generating swarm trajectories for a given lane of a road section. A plurality of vehicles traveling in the given lane transmit their respective trajectories to a backend computer, which determines and stores a swarm trajectory from the transmitted trajectories for the given lane. Furthermore, the vehicles determine specified boundary conditions while traveling in the lane, and the determined boundary conditions are transmitted to the backend computer along with the respective trajectory. The backend computer uses the transmitted trajectories and the boundary conditions to determine and store at least one swarm trajectory for the given lane as a function of at least one boundary condition.

[0008] The invention is based on the object of creating a method and a device for estimating at least one object state of a moving object in the environment of a vehicle, with which the at least one object state can be estimated in an improved manner.

[0009] The object is achieved according to the invention by a method having the features of patent claim 1 and a device having the features of patent claim 9. Advantageous embodiments of the invention emerge from the subclaims.

[0010] In particular, a method is provided for estimating at least one object state of a moving object in the environment of a vehicle, wherein environmental data of the environment are recorded by means of at least one sensor, wherein moving objects in the environment are detected on the basis of the recorded environmental data, wherein at least one object state is estimated for at least some of the detected moving objects, wherein the estimation of the at least one object state takes place taking into account provided swarm trajectories, and wherein the estimated at least one object state is provided.

[0011] Furthermore, in particular, a device for a vehicle for estimating at least one object state of a moving object in the environment of a vehicle is created, comprising a computing device, wherein the computing device is configured to receive environmental data detected by means of at least one sensor, to detect moving objects in the environment on the basis of the detected environmental data, to estimate at least one object state for at least some of the detected moving objects, wherein the estimation of the at least one object state takes place taking into account provided swarm trajectories, and to provide the estimated at least one object state.

[0012] The method and device enable an improved estimation of the object state of a moving object detected in the environment. This is achieved by taking swarm trajectories into account when estimating the object state, in addition to the recorded environmental data. The idea behind this is that moving objects, in particular other vehicles, predominantly move along the same paths, paths, and lanes. The paths, paths, and lanes are mapped using the swarm trajectories. By taking the swarm trajectories into account, a trajectory followed with a high probability by a moving object can be taken into account when estimating at least one object state. In particular, an object position and / or object movement (e.g. in the form of an object speed and a direction of the speed) can be estimated more accurately.

[0013] A moving object is, in particular, a (different) vehicle. However, in principle, a moving object can also be a pedestrian or other road user.

[0014] An object state of a moving object includes, in particular, an object position. Furthermore, an object state of a moving object can alternatively or additionally also include a speed and / or a speed direction of the moving object. It can also be provided that an object state merely includes an assignment to a lane of a road.

[0015] A swarm trajectory is a trajectory completed, in particular driven, in the past by a moving object, in particular by one or more vehicles. In particular, a swarm trajectory comprises several trajectories completed by several moving objects. Several such trajectories are then combined, in particular merged, to form a swarm trajectory, provided that a predetermined similarity measure between the several trajectories is met. For example, trajectories driven by several vehicles in the same lane can be combined, in particular merged, to form a single swarm trajectory. Properties can be assigned to the swarm trajectory. In particular, such a property can include a number of vehicles and / or a frequency with which the swarm trajectory was used. This can, for example, determine a vehicle density orTraffic density can be mapped along a swarm trajectory. The swarm trajectories are generated, for example, on the basis of historical trajectories recorded by a fleet of vehicles, in a central backend server. The swarm trajectories can be stored, for example, in an environment map.

[0016] A sensor is, for example, a camera, a radar sensor, a lidar sensor, or an ultrasonic sensor. The device may include at least one sensor.

[0017] Parts of the device, in particular the computing device, can be implemented individually or collectively as a combination of hardware and software, for example as program code executed on a microcontroller or microprocessor. However, it can also be provided that parts are implemented individually or collectively as an application-specific integrated circuit (ASIC).

[0018] It is provided that a detected moving object is assigned to at least one provided swarm trajectory, wherein an assignment of the detected moving object to the at least one provided swarm trajectory occurs depending on cost values calculated using a predetermined cost function, and wherein the estimation of the at least one object state of the detected moving object occurs taking the assignment into account. By assigning, an object state can be estimated more effectively. In particular, it is assumed that a detected moving object is highly likely to be located on one of the provided swarm trajectories. By determining a respective cost value, the most plausible swarm trajectories can be assigned to each moving object.

[0019] The cost function can, for example, take into account one or more of the following parameters: an angle between a direction of movement of the moving object and a course of a swarm trajectory under consideration, a rate of change of this angle, a distance (determined from the environmental data on the object position) between the moving object and the swarm trajectory under consideration and a historical correspondence between object positions and the swarm trajectory under consideration, etc. Individual parameters can be taken into account in a weighted manner.

[0020] In one embodiment, the cost function specified for calculating the cost values is specified depending on at least one selection criterion. This allows a cost function to be specifically selected for a current situation and / or a current context. In particular, the parameters considered in the cost function and their weighting can be selected depending on the at least one selection criterion. Such a selection criterion can, for example, be a road type or a road characteristic (e.g., cobblestone, asphalt, country road, highway, number of lanes, type and number of road markings, etc.). Using the respective selection criteria, which can be provided, for example, in the form of a table or a database, a weighting of the aforementioned parameters (angle, distance, history, etc.) can be determined for different roads.) and / or other signs can be used within the cost function. This allows a cost function to be selected depending on the quality criteria required, for example, for different road characteristics.

[0021] In an alternative embodiment, a utility function can be used instead of a cost function, which allows an evaluation and a selection based on a utility value, whereby the procedure is essentially analogous.

[0022] In one embodiment, the swarm trajectories are provided depending on at least one context parameter. Such a context parameter can be, for example, the following: a current time of day, a current day of the week, a current month, a season, a weather condition or the current weather, etc. For example, this can take into account that vehicles drive differently in rush hour traffic than in normal traffic, for example at a reduced speed due to an increased number of vehicles and / or increased traffic density. This changed behavior is then mapped using the respective associated swarm trajectories. Using the weather, for example, it can be taken into account that vehicles drive differently in poor visibility due to snowfall, fog, or rain than in sunshine, in particular at a reduced speed.Another example of a moving object is that pedestrians tend to walk less closely to the roadside during rain to maintain a greater distance from puddles on the road that could be driven through and sprayed by vehicles. The associated pedestrian swarm trajectories change accordingly.

[0023] In one embodiment, a weighting value is determined and provided for the estimated at least one object state depending on at least one property assigned to the swarm trajectories. This makes it possible, for example, to assign a probability to the estimated object state or to provide an object state as the sum of individual object states, each weighted with the weightings.

[0024] In one embodiment, it is provided that, in order to provide the swarm trajectories, an environment map is at least partially stored in the vehicle or is stored and / or transmitted to the vehicle, wherein the swarm trajectories are retrieved from the environment map and provided depending on a current position of the vehicle and / or a current context. The environment map is generated in particular by a vehicle fleet from recorded environment data. Sensors installed in vehicles of the vehicle fleet each continuously record an environment while driving. Objects are detected in this environment using methods known per se. The respectively detected objects are merged with their associated positions to form an environment map. Trajectories driven by the vehicles of the vehicle fleet are also stored in this environment map and summarized or merged into swarm trajectories using a similarity measure.The environment map can be created depending on at least one context parameter, so that the environment map can be provided in a context-dependent manner, as described above. A vehicle can then locate itself within the environment map using acquired environment data and / or a geographical position and retrieve swarm trajectories stored therein, in particular for a specific radius, and use them in the method described in this disclosure.

[0025] In one embodiment, it is provided that acquired environmental data from the at least one sensor corresponding to a detected object is checked for plausibility based on the estimated at least one object state. This makes it possible to estimate the quality of the acquired environmental data. For example, a plausibility value can be calculated and provided for the acquired environmental data. Such a plausibility value can, for example, be selected to be greater the smaller the distance and / or the smaller the angular deviation of a movement direction of a detected moving object from a swarm trajectory assigned to it. In particular, such a plausibility value can be transmitted to a driver assistance system or a vehicle control system alongside the acquired environmental data.In this way, the recorded environmental data can be provided with a measure of their plausibility and processed in a driver assistance system and / or vehicle control system, taking this additional information into account.

[0026] In one embodiment, it is provided that, based on the estimated at least one object state of a moving object detected in the environment and the swarm trajectory(s) assigned to the detected moving object, at least one future object trajectory of the detected moving object is estimated and provided. This makes it possible to provide one or more possible future trajectories for detected moving objects. The future object trajectories can be transmitted, for example, to a driver assistance system or to a vehicle control system, which supports and / or controls future behavior of the vehicle based on the estimated future trajectories of the moving object. In particular, a future object trajectory is selected such that it corresponds to the further course of one of the assigned swarm trajectories.It may be provided that a kinematic object model, for example of a vehicle, is taken into account when selecting the future object trajectory.

[0027] In a further embodiment, it is provided that a probability value is assigned to the estimated at least one future object trajectory depending on at least one property of the associated swarm trajectory(s). This allows a probability of occurrence to be taken into account in future planning. The property can, for example, comprise one or more of the following: a frequency with which the swarm trajectory was traversed or an associated (average) traffic density with which the swarm trajectory was traversed, or a number of vehicles that have traversed the swarm trajectory.

[0028] Further features of the device design will become apparent from the description of embodiments of the method. The advantages of the device are the same as those of the embodiments of the method.

[0029] The invention will be explained in more detail below using preferred embodiments with reference to the figures. Fig. 1 a schematic representation of an embodiment of the device for a vehicle for estimating at least one object state of a moving object in the environment of a vehicle; Fig. 2 a schematic representation to illustrate individual measures of an embodiment of the method; Fig. 3 a schematic representation to illustrate individual measures of an embodiment of the method; Fig. 4 a schematic representation to illustrate the assignment of a detected moving object to a swarm trajectory; Fig. 5a-5c schematic representations to illustrate the estimation of at least one future trajectory; Fig. 6 a schematic representation to illustrate an application scenario of an embodiment of the method and the device; Fig. 7 a schematic representation to illustrate a further application scenario of an embodiment of the method and the device; Fig. 8 a schematic representation to illustrate a further application scenario of an embodiment of the method and the device.

[0030] In Fig. Figure 1 shows a schematic representation of an embodiment of the device 1 for a vehicle 50 for estimating at least one object state 30 of a moving object in the environment of a vehicle 50. The device 1 comprises a computing device 2 and a memory device 3. The computing device 2 can access data stored in the memory device 3 and perform computing operations on this data. In particular, the computing device 2 executes the method described in this disclosure.

[0031] Environmental data 10 acquired from the surroundings of the vehicle 50 by means of at least one sensor 51 are fed to the computing device 2. The computing device 2 receives the acquired environmental data 10. The at least one sensor 51 is, for example, a camera, a radar sensor, a lidar sensor, or an ultrasonic sensor of the vehicle 50, etc. It can be provided here that the device 1 comprises the at least one sensor 51. Furthermore, swarm trajectories 11 are fed to the computing device 2. The swarm trajectories 11 are provided, for example, by means of a backend server 90 and transmitted to the computing device 2, for example via an interface set up for this purpose (not shown).

[0032] In this case, it can be provided that the swarm trajectories 11 are provided depending on at least one context parameter 13. For this purpose, it can be provided that the computing device 2 receives or determines the context parameter 13 and queries and receives the swarm trajectories 11 depending on this context parameter 13, for example, from the backend server 90. The context parameter 13 includes, for example, a current time of day, a current day of the week, a current month, a current season, a current atmospheric condition or the current weather, etc. The backend server 90 then selects swarm trajectories 11 in the vicinity of a current vehicle position of the vehicle 50 depending on the at least one context parameter 13 and transmits them to the computing device 2.

[0033] The necessary measures of the method are carried out, for example, by means of individual modules 100, 101 within the computing device 2, as shown schematically in the Fig. 2 is shown.

[0034] Based on the recorded environmental data 10, the computing device 2 detects moving objects 20 in the environment (module 100 in Fig. 2). Known pattern recognition methods can be used here. As a result of object recognition, for example, the position of a detected object 20 and the type of the detected object 20 are determined and provided. In addition, the detected moving object 20 is assigned a unique identifier for unambiguous marking.

[0035] For at least some of the detected moving objects 20, the computing device 2 estimates at least one object state 30, wherein the estimation of the at least one object state 30 takes place taking into account provided swarm trajectories 11 (module 101 in Fig. 2). Here, at least one object state 30 is estimated, in particular for moving objects 20 arranged in the vicinity of the vehicle 50. The at least one object state 30 includes, for example, an object position and / or a direction of movement and a speed of the respective moving object 20.

[0036] The estimated at least one object state 30 is provided, for example, by outputting the estimated at least one object state 30 in the form of a digital data packet. The at least one object state 30 can be supplied, for example, to a driver assistance system or a vehicle control system 52.

[0037] The one in the Fig. The measures 200, 201 carried out for this purpose in the module 101 shown in Figure 2 are shown schematically in the Fig. 3 shown.

[0038] In this case, in measure 200, a detected moving object 20 is assigned to at least one swarm trajectory 11. In particular, the detected moving object 20 is assigned to one of the provided swarm trajectories. Measure 200 provides at least one assignment 21 to one or more swarm trajectories 11 for the observed moving object 20.

[0039] It is provided that an assignment 21 of the detected moving object 20 to the at least one provided swarm trajectory 11 takes place as a function of cost values 23, which are calculated by means of a predetermined cost function 22 (cf. Fig. 1). For the cost function 22, for example, the Kuhn-Munkers algorithm, also known as the Hungarian algorithm, can be used. The cost function 22 can include one or more of the following metrics: a Euclidean distance between an object position of the detected moving object 20 and points on a respective swarm trajectory 11 under consideration, an angular deviation between an orientation and / or direction of movement of the moving object 20 and a respective swarm trajectory 11 under consideration, and / or probabilistic (weighting) methods, for example in the form of a Mahalanobis distance.

[0040] It can be provided that the cost function 22 specified for calculating the cost values 23 is specified depending on at least one selection criterion 24. Such a selection criterion 23 can, for example, be a road type or a road characteristic (e.g., cobblestone, asphalt, country road, motorway, number of lanes, type and number of road markings, etc.). Depending on the road type, a different cost function 22 is then selected, in which, for example, influencing factors are selected and / or weighted differently.

[0041] In measure 201, a probabilistic hypothesis is created based on the assignment(s) 21 made in measure 200. As a result, one or more estimates of the object state 30 with an associated probability are provided.

[0042] The estimated object state 30 can include an object position, a direction of movement and / or a speed and / or acceleration, as well as, in particular, probabilistic hypotheses regarding a future movement path with the respective probabilities of occurrence. The future movement path is estimated, in particular, based on the swarm trajectory 11 associated with the moving object 20 under consideration. Based on this, at least one future trajectory 40 of the detected moving object 20, i.e., future object positions, can be estimated in measure 201. Therefore, the method can be used to generate both a movement hypothesis for a detected moving object 20 and a movement prediction for the detected moving object 20.

[0043] The estimated future trajectory 40 can then be transmitted, for example, to a driver assistance system or a vehicle control system of the vehicle 50. Subsequently, based on the estimated future trajectory 40, it can be checked, for example, whether an assistance function (e.g., lane change or traffic jam assistance function) is enabled or not.

[0044] In Fig. Figure 4 shows a schematic representation to illustrate the assignment of a detected moving object 20 (shown in the form of a rectangle) to a swarm trajectory 11. Only a single swarm trajectory 11 is shown as an example. However, a plurality of swarm trajectories 11 are considered during the assignment process.

[0045] An object position P0 of the moving object 20 is determined based on the acquired environmental data and a known sensor position. The sensor position is determined, for example, from a vehicle position, whereby the vehicle position is determined, for example, via a global positioning system or via a localization of the vehicle within an environmental map.

[0046] The moving object 20 moves at a speed in a direction of movement, symbolized by a speed vector V1. A speed vector V2 also corresponds to the swarm trajectory 11. In particular, the speed vector V2 represents an averaged or fused speed and an averaged or fused direction of movement of other vehicles that have traveled this swarm trajectory 11 in the past.

[0047] Positions of the moving object 20 and the swarm trajectory 11 are transformed into a common coordinate system so that an assignment 21 can take place.

[0048] Three metrics A, B, and C are shown as examples. Metric A considers an angular difference W between the velocity vectors V1 and V2. A change in velocity over time can also be taken into account. Metric B considers a shortest distance D between the velocity vectors V1 and V2, or between the object position P0 and the swarm trajectory 11. Metric C can be used to consider a history, for example, by considering (weighted) the shortest distances D (alternatively or additionally the respective angular differences W) to the object position P0 at past times.

[0049] A cost function is defined using one or more of the metrics A, B, C described as examples. For the provided swarm trajectories 11-y, a cost value 23-x / y is determined using such a cost function for each combination of a moving object 20-x or an associated object position P0-x and a swarm trajectory 11-y. It is particularly provided that a preselection of swarm trajectories 11-y has already been carried out, for example, to limit a number of swarm trajectories 11-y around a vehicle position, in order to reduce the computational effort when finding assignments 21. A moving object 20-x is then assigned the swarm trajectory 11-y with the smallest cost value 23-x / y. This is shown in a Fig. The table shown in Figure 4 is shown schematically for three moving objects 20-x and three swarm trajectories 11-y. It can also be provided that several swarm trajectories 11-y are assigned to a moving object 20-x, for example, the two or three, etc., with the lowest cost values 23-x / y, etc., i.e., the most plausible swarm trajectories with regard to a movement pattern.

[0050] In the Fig. 5a, Fig. 5b and Fig. 5c shows schematic representations to illustrate the estimation of at least one future trajectory 40. The estimation of the at least one future trajectory 40 is carried out based on the estimated at least one object state of the detected moving object 20, which, for example, includes at least one current object position P0, and the swarm trajectory(s) 11-x assigned to the detected moving object 20. In particular, object positions of the moving object 20 at future points in time are estimated (i.e., predicted). For each of the swarm trajectories 11-x assigned to the moving object 20, a hypothesis for a future travel path is estimated. Each of the hypotheses can be assigned a probability of occurrence, wherein the probability is determined in particular based on at least one property of the respective swarm trajectory 11-x.Such a property can, for example, be a frequency with which the considered swarm trajectory 11-x was followed in the past.

[0051] The Fig. The example shown in Figure 5a corresponds to a two-lane expressway from which an exit branches off. Swarm trajectories 11-1, 11-2 represent the lanes on the expressway, while swarm trajectory 11-3 represents a trajectory along the exit from the expressway. Since most vehicles, i.e., most moving objects 20 in the vicinity of the vehicle, continue straight ahead on the expressway and only a few take the exit, a frequency distribution results for the individual swarm trajectories 11-x, for example due to a right-hand driving requirement, in which most vehicles drive on swarm trajectory 11-2 (right lane), fewer vehicles on swarm trajectory 11-1 (left lane), and the fewest vehicles on swarm trajectory 11-3 (exit). Such a distribution is shown schematically in the Fig. 5c, where the representation shows the probability 60 with which vehicles on the expressway in the past, starting from the current object position P0, have used the swarm trajectories 11-x shown.

[0052] Furthermore, a kinematic object model can be taken into account in order to estimate a future trajectory 40 of the moving object 20, taking into account physical laws of the moving object 20, such as inertia, speed, speed change and / or a possible steering angle, etc.

[0053] Based on this, the future trajectory 40 of the moving object 20 (ie, the other vehicle) is estimated, i.e., object positions for future times T+n (n is an integer with n = 1,..., N) are estimated. The future trajectory 40 is shown schematically in Fig. 5b, which results from a direction of movement of the moving object 20 taking into account the probability 60 or is the most plausible taking into account all influencing factors.

[0054] The estimated future trajectory 40 of the moving object 20 can then be provided as a trajectory signal, for example in the form of a digital data packet, and transmitted, for example, to a vehicle control system of the vehicle, so that the future trajectory 40 of the moving object 20 can be taken into account in trajectory planning for the vehicle and / or future environmental data of the at least one sensor can be checked for plausibility.

[0055] In Fig. Figure 6 shows a schematic representation to illustrate an application scenario of an embodiment of the method and device. It shows an intersection 70 toward which the vehicle 50 is approaching. Moving objects 20 in the form of other vehicles are approaching from the opposite direction and from the right, which are also moving toward the center of the intersection. After the moving objects 20 are detected in the acquired environmental data, at least one swarm trajectory 11-y is assigned to each of the moving objects 20. In the example shown, each of the moving objects 20, i.e., the other vehicles, is assigned three swarm trajectories 11-y, which map the possible outflows from the intersection 70 and, for example, describe the three strongest hypotheses for the respective driving path.

[0056] If, for example, the vehicle 50 is equipped with a camera as a sensor 51 for detecting an environment, it may be that the camera can detect the moving objects 20, but not—or at least not correctly—an exact intersection topology in the form of, for example, lane markings of individual lanes at the intersection 70. Using the respective assignments 21 of the swarm trajectories 11-y to the moving objects 20, an object state 30 of the moving objects 20 can be better estimated, since the moving objects 20, i.e., the other vehicles, can be located in relation to lanes of the intersection 70. Based on the swarm trajectories 11-y, both past object states and possible future object states in the environment of the vehicle 50 are known or can be estimated. The respectively estimated object states 30 include, for example, a direction of movement and a speed of the respective moving object 20.In addition, future trajectories (not shown) of the moving objects 20 can be estimated based on the respectively assigned swarm trajectories 11-y.

[0057] In Fig. Figure 7 shows a schematic representation to illustrate another application scenario of an embodiment of the method and device. It shows a highway 71 on which the vehicle 50 and several moving objects 20 in the form of other vehicles are stuck in a traffic jam. Particularly in traffic jam situations, camera-based detection of lanes 72 is only possible to a very limited extent due to obscuration by other vehicles. Radar sensors are capable of detecting and recognizing vehicles traveling ahead. However, lane assignment of the detected vehicles traveling ahead is not reliably possible due to inadequate lane recognition. In such traffic jam situations, vehicles swerving from a convoy pose a particular problem.

[0058] However, using the method and device, convoy lanes can be estimated more effectively via the associated swarm trajectories 11-y, and an assignment 21 of vehicles (moving objects 20) to lanes 72 can be performed more reliably, so that, in particular, vehicles pulling out can be better detected. This can increase safety in traffic jam situations, for example, where the vehicle 50 is required to follow a vehicle in front.

[0059] In Fig.Figure 8 shows a schematic representation to illustrate another application scenario of an embodiment of the method and device. Shown is a road curve 73, which the vehicle 50 travels through in the right lane 72-2. Two moving objects 20 in the form of other vehicles are detected at a greater distance behind the vehicle 50, for example, in environmental data acquired by a sensor configured as a rear radar. However, lane assignment is not readily possible due to the pronounced curvature of the road curve 73, particularly if measurement errors and / or measurement noise increase due to a greater distance at which moving objects 20 are detected. Lane assignment is then performed via an assignment 21 of the respectively detected moving objects 20 to one of the swarm trajectories 11-1, 11-2 provided for the current environment.In this way, the vehicle directly following vehicle 50 can be assigned to the right lane 72-2 via the assignment 21 to the swarm trajectory 11-2, and the vehicle subsequently following can be assigned to the left lane 72-1 via the assignment 21 to the swarm trajectory 11-1.

[0060] Furthermore, a future driving course can be estimated in the form of future trajectories (not shown) based on the swarm trajectories 11-1, 11-2 assigned to the moving objects 20.

[0061] In the example shown, the disclosed method and device can, in particular, improve overtaking detection by a lane change assistant. In particular, when an enable signal must be provided to enable an automated lane change, this enable signal can be provided in an improved manner based on the described method and device. The availability and safety of a lane change assistant in the vehicle 50 can be increased. List of reference symbols 1 device 2 computing device 3 Storage device 10 Environmental data 11, 11-y swarm trajectory 13 context parameters 20, 20-x detected moving object 21 Assignment 22 Cost function 23, 23-x / y cost values 24 Selection criteria 30 Object condition 40 future object trajectory 50 vehicles 51 Sensor 52 Vehicle control 60 Probability 70 intersection 71 Autobahn 72, 72-x lane 73 road curve 90 backend servers 100 modules 101 Module 200-201 Measures P0 Object position V1 velocity vector (object) V2 velocity vector (swarm trajectory) A metric (“angle”) B Metric (“distance”) C Metric (“History”) D Distance W Angle difference

Claims

[1] Method for estimating at least one object state (30) of a moving object (20) in the environment of a vehicle (50), wherein environmental data (10) of the environment are recorded by means of at least one sensor (51), wherein, based on the recorded environmental data (10), moving objects (20) in the environment are detected, wherein at least one object state (30) is estimated for at least some of the detected moving objects (20), wherein the estimation of the at least one object state (30) takes place taking into account provided swarm trajectories (11-y), and wherein the estimated at least one object state (30) is provided, wherein a detected moving object (20) is assigned to at least one provided swarm trajectory (11-y), wherein an assignment (21) of the detected moving object (20) to the at least one provided swarm trajectory (11-y) is carried out as a function of cost values (23-x / y) which are calculated by means of a predetermined cost function, and wherein the estimation of the at least one object state (30) of the detected moving object (20) is carried out taking into account the assignment (21). [2] Method according to claim 1, characterized by that the cost function specified for calculating the cost values (23-x / y) is specified as a function of at least one selection criterion (24). [3] Method according to one of the preceding claims, characterized by that the swarm trajectories (11-y) are provided as a function of at least one context parameter (13). [4] Method according to one of the preceding claims, characterized by that a weighting value is determined and provided for the estimated at least one object state (30) depending on at least one property assigned to the swarm trajectories (11-y). [5] Method according to one of the preceding claims, characterized by in that, in order to provide the swarm trajectories (11-y), an environment map is at least partially stored in the vehicle (50) or is stored and / or transmitted to the vehicle (50), wherein the swarm trajectories (11-y) are retrieved and provided from the environment map depending on a current position of the vehicle (50) and / or a current context. [6] Method according to one of the preceding claims, characterized by that detected environmental data (10) of the at least one sensor (51) corresponding to a detected object (20) are checked for plausibility on the basis of the estimated at least one object state (30). [7] Method according to one of the preceding claims, characterized by that, based on the estimated at least one object state (30) of a moving object (20) detected in the environment and the swarm trajectory(s) (11-y) associated with the detected moving object (20), at least one future object trajectory (40) of the detected moving object (20) is estimated and provided. [8] Method according to claim 7, characterized by that the estimated at least one future object trajectory (40) is or will be assigned a probability value depending on at least one property of the associated swarm trajectory(s) (11-y). [9] Device (1) for a vehicle (50) for estimating at least one object state (30) of a moving object (20) in the environment of a vehicle (50), comprising: a computing device (2), wherein the computing device (2) is configured to receive environmental data (10) acquired by means of at least one sensor (51), to detect moving objects (20) in the environment based on the detected environmental data (10), to estimate at least one object state (30) for at least some of the detected moving objects (20), wherein the estimation of the at least one object state (30) takes place taking into account provided swarm trajectories (11-y), and to provide the estimated at least one object state (30), wherein a detected moving object (20) is assigned to at least one provided swarm trajectory (11-y), wherein an assignment (21) of the detected moving object (20) to the at least one provided swarm trajectory (11-y) is carried out as a function of cost values (23-x / y) which are calculated by means of a predetermined cost function, and wherein the estimation of the at least one object state (30) of the detected moving object (20) is carried out taking into account the assignment (21).

Images