Method for trajectory planning, method for operating a motor vehicle, computing device for carrying out such a method, control device with such a computing device and motor vehicle with such a control device
The proposed trajectory planning process for autonomous vehicles efficiently selects collision-free paths by integrating sensor data and cost calculations, addressing the limitations of existing methods in considering object impacts and reducing computational demands.
Patent Information
- Application Number
- DE102023004651
- Authority / Receiving Office
- DE · DE
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-11-16
- Publication Date
- 2025-01-23
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing trajectory planning for autonomous vehicles does not adequately consider the impact of the vehicle's trajectory on surrounding objects and requires high computational power and time, especially in predicting multiple scenarios.
A trajectory planning process that involves sensor recordings, scene predictions, and cost calculations for both ego and object trajectories, allowing simultaneous generation and coordination to select a planning trajectory that minimizes collision risks and computational effort.
Enables efficient and interpretable selection of a planning trajectory that considers interactions with surrounding objects, reducing computational requirements and ensuring collision-free navigation.
Smart Images
Figure 00000000_0000_ABST
Abstract
Description
[0001] The invention relates to a method for trajectory planning, a method for operating a motor vehicle, in particular a highly automated or autonomously operated motor vehicle, a computing device for carrying out such a method, a control device with such a computing device and a motor vehicle with such a control device.
[0002] It is known that motion planning and / or trajectory creation are necessary for the operation of autonomous vehicles. Dynamic objects in the vehicle's environment are taken into account during motion planning.
[0003] Published patent application DE 10 2022 001 207 A1 describes a method in which the movements of the dynamic objects are first predicted. Subsequently, during motion planning, a trajectory of the motor vehicle is sought that does not lead to a collision with the objects. For this purpose, it is checked whether the planned trajectory of the motor vehicle overlaps with the predicted trajectories of the objects. A disadvantage of this method is that a possible impact of the planned trajectory of the motor vehicle on the trajectories of the objects is not taken into account.
[0004] Furthermore, methods are known in which a plurality of trajectories of the motor vehicle are proposed. Subsequently, a conditioned prediction of the surrounding objects is performed for each of these trajectories. The disadvantage of this approach is that it requires high computing power and / or a considerable amount of time.
[0005] The invention is therefore based on the object of creating a method for trajectory planning, a method for operating a motor vehicle, in particular a highly automated or autonomously operated motor vehicle, a computing device for carrying out such a method, a control device with such a computing device and a motor vehicle with such a control device, wherein the aforementioned disadvantages are at least partially remedied, preferably avoided.
[0006] The object is achieved by providing the present technical teaching, in particular the teaching of the independent claims as well as the embodiments disclosed in the dependent claims and the description.
[0007] The problem is solved in particular by creating a method - also referred to as a planning method - for trajectory planning in a particularly highly automated or autonomously operated motor vehicle. At a first point in time, at least one sensor recording is created using at least one environmental sensor. A scene with at least one object is detected in the at least one sensor recording. Subsequently, a plurality of different scene predictions are created for a same prediction horizon. Each scene prediction has an ego trajectory of the motor vehicle for the prediction horizon. In addition, each scene prediction has an object trajectory for the at least one object for the prediction horizon. Furthermore, an ego cost value is calculated for each ego trajectory. In addition, an object cost value is calculated for each object trajectory.For each scene prediction of the various scene predictions, a scene prediction cost value is calculated based on the respective ego cost value and the respective object cost value. Based on the plurality of scene prediction cost values, an ego trajectory is selected as the planning trajectory. This advantageously makes it possible to select the planning trajectory taking into account a scene and an interaction with objects in the scene. Furthermore, the ego trajectory and the at least one object trajectory are advantageously generated simultaneously and in a coordinated manner, eliminating the need to generate suggestions for the ego trajectory and to generate object trajectories based on the suggestions. Furthermore, the simultaneous generation of the trajectories takes into account the effect of the respective ego trajectory on the respective object trajectory.This allows for a simple and efficient selection of a planning trajectory. Furthermore, the selection of the planning trajectory can be interpreted and / or traced based on the cost values.
[0008] In one embodiment, a plurality of sensor images are created at the first time, wherein the plurality of sensor images are combined, in particular using a sensor fusion method, to form an environmental image. Subsequently, alternatively or additionally, the scene containing the at least one object is detected in the environmental image.
[0009] Preferably, the at least one object is selected from a group consisting of a motor vehicle, in particular a passenger car, a lorry, or a motorcycle, a person, in particular a pedestrian, a means of public transport, in particular a bus, a tram, an underground train, a suburban train, or a train, a bicycle, and a combination of at least two of the objects mentioned.
[0010] Particularly preferably, each scene prediction of the plurality of scene predictions is created such that the respective scene prediction is scene consistent.
[0011] In the context of the present technical teaching, a scene prediction is scene-consistent in particular when a first object trajectory of a first object of the scene prediction and a second object trajectory of a second object of the same scene prediction result in a consistent traffic scene and, in particular, do not collide and / or contradict each other. Thus, the object trajectories of a scene-consistent scene prediction are preferably collision- and contradiction-free. In particular, a scene-consistent scene prediction is also referred to as a joint prediction.
[0012] In one embodiment, the scene predictions are created using a method according to the work ScePT.- Scene-consistent, Policy-based Trajectory Predictions for Planning by Yuxiao Chen et al., published at https: / / arxiv.org / abs / 2206.13387. Alternatively or additionally, the scene predictions are created using a method according to the work Latent Variable Sequential Set Transformers For Joint Multi-Agent Motion Prediction by Roger Girgis et al., published at https: / / arxiv.org / abs / 2104.00563. Alternatively or additionally, the scene predictions are created using a method according to the work Multi-future Transformer: Learning diverse interaction modes for behavior prediction in autonomous driving by Baotian He and Yibing Li, published at https: / / ietresearch.onlinelibrary.wiley.com / doi / pdfdirect / 10.1049 / itr2.12207.
[0013] Preferably, the plurality of scene predictions is created using at least one, in particular exactly one, scene-consistent prediction. Alternatively, the plurality of scene predictions is created using a plurality of scene-consistent predictions. A first number of scene predictions is identical to a second number of scene-consistent predictions, or the first number is greater than the second number.
[0014] In the context of the present technical teaching, the prediction horizon is in particular a period from the first point in time to a second point in time following the first point in time.
[0015] Preferably, the ego trajectory includes information about a path, in particular x and y coordinates, of the motor vehicle. Additionally, the ego trajectory includes information about the motor vehicle's dynamics, in particular its speed and / or acceleration.
[0016] Preferably, the object trajectory includes information about a path, in particular x and y coordinates, of the at least one object. Additionally, the object trajectory includes information about object dynamics, in particular a speed and / or acceleration, of the at least one object.
[0017] According to a further development of the invention, the planning trajectory is selected by minimizing the scene prediction cost values. This advantageously makes selecting the planning trajectory easy.
[0018] In particular, the scene prediction ego trajectory with the lowest scene prediction cost is selected as the planning trajectory.
[0019] According to a further development of the invention, the ego cost value is determined by means of a first cost function based on at least one ego parameter selected from a group consisting of ego driving comfort, ego efficiency, ego collision risk, overall trajectory progress, and a combination of at least two of the aforementioned ego parameters. Advantageously, the determination of the ego cost value is thus interpretable and comprehensible. Furthermore, the determination of the ego cost value is human-interpretable.
[0020] According to a further development of the invention, the object cost value is determined by means of a second cost function based on at least one object parameter selected from a group consisting of object ride comfort, object efficiency, object collision risk, and a combination of at least two of the aforementioned object parameters. Advantageously, the determination of the object cost value is thus interpretable and comprehensible. Furthermore, the determination of the object cost value is human-interpretable.
[0021] In the context of the present technical teaching, a cost value—this applies to the ego cost value and the object cost value—is understood, in particular, as a quantification of the respective trajectory. The cost value preferably increases when the respective driving comfort is reduced. In particular, a first cost value of a first trajectory is greater than a second cost value of a second trajectory if a first driving comfort of the first trajectory is lower than a second driving comfort of the second trajectory. Alternatively or additionally, the cost value increases when the respective efficiency is reduced. In particular, a first cost value of a first trajectory is greater than a second cost value of a second trajectory if a first efficiency of the first trajectory is lower than a second efficiency of the second trajectory.Alternatively or additionally, the cost value increases if the respective collision risk, in particular with the at least one object, the motor vehicle, or another object, increases. In particular, a first cost value of a first trajectory is greater than a second cost value of a second trajectory if a first collision risk of the first trajectory is greater than a second collision risk of the second trajectory. Alternatively or additionally, the ego cost value increases if the overall trajectory progress decreases. In particular, a first ego cost value of a first ego trajectory is greater than a second ego cost value of a second ego trajectory if a first overall trajectory progress of the first ego trajectory is smaller than a second overall trajectory progress of the second ego trajectory.
[0022] In one embodiment, ride comfort—this applies to both ego ride comfort and object ride comfort—is quantified using a ride comfort parameter value. The ride comfort parameter value is determined as a function of positive and / or negative acceleration. In particular, the ride comfort parameter value increases as a function of the absolute value of the acceleration. If the cost value of the trajectory is obtained by simply summing the various parameter values, the ride comfort parameter value ensures that the cost value of the trajectory increases when ride comfort is reduced.
[0023] In a further embodiment, efficiency—this applies to both ego efficiency and object efficiency—is quantified with an efficiency parameter value. The efficiency parameter value is determined as a function of the speed of the motor vehicle. In particular, the efficiency parameter value increases as a function of the absolute value difference between the speed of the motor vehicle and a legally permitted maximum speed. If the cost value of the trajectory is obtained by simply summing the various parameter values, the efficiency parameter value ensures that the cost value of the trajectory increases when efficiency is reduced.
[0024] In a further embodiment, the collision risk—this applies to the ego collision risk and the object collision risk—is quantified using a collision risk parameter value. The collision risk parameter value is determined depending on the collision risk with at least one object, the motor vehicle, or another object. In particular, the collision risk parameter value increases depending on the collision risk.
[0025] In a further embodiment, the overall trajectory progress is quantified using a progress parameter value. The progress parameter value is determined depending on the duration required to reach a navigation destination and / or the distance to the navigation destination. In particular, the progress parameter value increases depending on an increase in the duration and / or distance.
[0026] Particularly preferably, the cost value—this applies to both the ego cost value and the object cost value—is the driving comfort parameter value. Alternatively, the cost value is the efficiency parameter value. Alternatively, the cost value is the collision risk parameter value. Alternatively, the ego cost value is the progress parameter value. Alternatively, the cost value is the sum of at least two values selected from the driving comfort parameter value, the efficiency parameter value, the collision risk parameter value, and the progress parameter value.
[0027] According to a further development of the invention, it is provided that the planning trajectory is optimized taking into account at least one optimization parameter.
[0028] In one embodiment, the at least one optimization parameter is selected from a group consisting of ego driving comfort, ego efficiency, ego collision risk, a safety distance, and a combination of at least two of the aforementioned optimization parameters.
[0029] Particularly preferably, the planning trajectory is optimized taking into account maximum ego driving comfort—in particular, corresponding to a minimum ego driving comfort parameter value. Alternatively or additionally, the planning trajectory is optimized taking into account maximum efficiency—in particular, a correspondingly minimum ego efficiency parameter value. Alternatively or additionally, the planning trajectory is optimized taking into account a minimum ego collision risk—in particular, corresponding to a minimum ego collision risk parameter value. Alternatively or additionally, the planning trajectory is optimized taking into account maintaining the safety distance from the at least one object.
[0030] According to a further development of the invention, the majority of scene predictions are created depending on the navigation destination of the motor vehicle. In particular, the respective ego trajectories of the scene predictions are created depending on the navigation destination of the motor vehicle. This ensures that only ego trajectories that are consistent with the navigation destination are predicted, thereby reducing the computational effort of the method.
[0031] The object is also achieved by creating a method—also referred to as an operating method—for operating a motor vehicle, in particular a highly automated or autonomously operated one. A planning trajectory is obtained using a planning method according to the invention or a planning method according to one or more of the previously explained embodiments. The motor vehicle is then moved along the planning trajectory. In connection with the operating method, the advantages already explained in connection with the planning method arise in particular.
[0032] The object is also achieved by providing a computing device configured to execute a method according to the invention or a method according to one or more of the previously explained embodiments, in particular the planning method and / or the operating method. In connection with the computing device, the advantages already explained in connection with the planning method and the operating method arise in particular.
[0033] The computing device is in particular configured to be operatively connected to at least one environmental sensor and configured to control it.
[0034] The object is also achieved by providing a control device that has at least one environmental sensor and a computing device according to the invention or a computing device according to one or more of the previously explained embodiments. In connection with the control device, the advantages already explained in connection with the planning method, the operating method, and the computing device arise in particular.
[0035] The computing device is in particular operatively connected to the at least one environmental sensor and configured to control it.
[0036] In one embodiment, the control device has a plurality of environmental sensors, wherein the plurality of environmental sensors are preferably synchronized with one another.
[0037] The object is also achieved by providing a motor vehicle having a control device according to the invention or a control device according to one or more of the previously explained embodiments. In connection with the motor vehicle, the advantages already explained in connection with the planning method, the operating method, the computing device, and the control device are particularly advantageous.
[0038] In one embodiment, the motor vehicle is designed as a passenger car. However, it is also possible for the motor vehicle to be a truck, a commercial vehicle, or another motor vehicle. Alternatively or additionally, the motor vehicle is designed as a highly automated or autonomously operated motor vehicle.
[0039] The invention is explained in more detail below with reference to the drawing.
[0040] Showing: Fig. 1 a schematic representation of an embodiment of a motor vehicle, Fig. 2 a flowchart of a first embodiment of a method for trajectory planning, and Fig. 3 a flowchart of a second embodiment of a method for trajectory planning.
[0041] Fig. 1 shows a schematic representation of an embodiment of a motor vehicle 1 with a control device 3. The control device 3 has an environmental sensor 5 and a computing device 7.
[0042] Motor vehicle 1 is particularly preferably a highly automated or autonomously operated motor vehicle. A highly automated or autonomously operated vehicle is understood to be an automated vehicle whose degree of automation corresponds to at least Level 3 according to the SAE J3016 standard.
[0043] The environmental sensor 5 is configured to create at least one sensor recording 9 of an environment of the motor vehicle 1.
[0044] The computing device 7 is shown only schematically here and is operatively connected to the environmental sensor 5 in a manner not explicitly shown and is configured to control it. The computing device 7 is configured, in particular, to carry out a planning method of an operating method. The methods are based on Fig. 2 and Fig. 3 is explained in more detail.
[0045] Fig. 2 shows a flowchart of a first embodiment of the method for trajectory planning and / or for operating the motor vehicle 1.
[0046] Identical and functionally identical elements are provided with the same reference numerals in all figures, so that reference is made to the previous description.
[0047] In a first step a), at least one sensor recording 9 is created at a first point in time by means of the at least one environmental sensor 5.
[0048] In a second step b), a scene with an object 11 is detected in the at least one sensor recording 9.
[0049] In a third step c), a plurality of different scene predictions 13—in particular, a first scene prediction 13.1, a second scene prediction 13.2, and a third scene prediction 13.3—are created for a same prediction horizon. Each scene prediction 13 has an ego trajectory 15 of the motor vehicle 1 for the prediction horizon. In addition, each scene prediction 13 has an object trajectory 17 for the object 11 for the prediction horizon. In particular, the first scene prediction 13.1 has a first ego trajectory 15.1 and a first object trajectory 17.1. Furthermore, the second scene prediction 13.2 has a second ego trajectory 15.2 and a second object trajectory 17.2. Furthermore, the third scene prediction 13.3 has a third ego trajectory 15.3 and a third object trajectory 17.3.Particularly preferably, the plurality of scene predictions 13 are created as a function of a navigation destination of the motor vehicle 1. In particular, the respective ego trajectories 15 of the scene predictions 13 are created as a function of the navigation destination of the motor vehicle 1.
[0050] In a first fourth step d1), a first ego cost value 19.1 of the first ego trajectory 15.1 is calculated. In addition, in a second fourth step d2), a second ego cost value 19.2 of the second ego trajectory 15.2 is calculated. In addition, in a third fourth step d3), a third ego cost value 19.3 of the third ego trajectory 15.3 is calculated. Particularly preferably, the ego cost value 19 is determined by means of a first cost function based on at least one ego parameter selected from a group consisting of ego driving comfort, ego efficiency, ego collision risk, overall trajectory progress, and a combination of at least two of the aforementioned ego parameters.
[0051] In a first fifth step e1), a first object cost value 21.1 of the first object trajectory 17.1 is calculated. In addition, in a second fifth step e2), a second object cost value 21.2 of the second object trajectory 17.2 is calculated. In addition, in a third fifth step e3), a third object cost value 21.3 of the third object trajectory 17.3 is calculated. Particularly preferably, the object cost value 21 is determined by means of a second cost function based on at least one object parameter selected from a group consisting of an object's driving comfort, an object's efficiency, an object's collision risk, and a combination of at least two of the aforementioned object parameters.
[0052] In a first sixth step f1), a first scene prediction cost value 23.1 is calculated for the first scene prediction 13.1 based on the first ego cost value 19.1 and the first object cost value 21.1. In a second sixth step f2), a second scene prediction cost value 23.2 is calculated for the second scene prediction 13.2 based on the second ego cost value 19.2 and the second object cost value 21.2. In a third sixth step f3), a third scene prediction cost value 23.3 is calculated for the third scene prediction 13.3 based on the third ego cost value 19.3 and the third object cost value 21.3.
[0053] In a seventh step g), an ego trajectory 15 is selected as the planning trajectory 25 based on the plurality of scene prediction cost values 23. Preferably, the planning trajectory 25 is selected by minimizing the scene prediction cost values 23. In particular, the ego trajectory 15 of the scene prediction 13 with the lowest scene prediction cost value 23 is selected as the planning trajectory 25.
[0054] In an optional eighth step h), the planning trajectory 25 is optimized taking into account at least one optimization parameter. Particularly preferably, the at least one optimization parameter is selected from a group consisting of ego driving comfort, ego efficiency, ego collision risk, a safety distance, and a combination of at least two of the aforementioned optimization parameters.
[0055] In an optional ninth step i), the motor vehicle 1 is moved along the particularly optimized planning trajectory 25.
[0056] Particularly preferably, the method is carried out repeatedly, in particular cyclically with a predetermined frequency.
[0057] Fig. 3 shows a flowchart of a second embodiment of the method for trajectory planning and / or for operating the motor vehicle 1.
[0058] The second embodiment according to Fig. 3 is an extension of the first embodiment according to Fig. 2, so that only the differences will be discussed below.
[0059] In the second step b), a scene with a first object 11.1 and a second object 11.2 is detected in the at least one sensor recording 9.
[0060] In the third step c), the plurality of different scene predictions 13—in particular, the first scene prediction 13.1, the second scene prediction 13.2, and the third scene prediction 13.3—are created for a same prediction horizon. The first scene prediction 13.1 comprises the first ego trajectory 15.1, a first object trajectory 17.1.1, and a first second object trajectory 17.1.2. Furthermore, the second scene prediction 13.2 comprises the second ego trajectory 15.2, a second object trajectory 17.2.1, and a second object trajectory 17.2.2. Furthermore, the third scene prediction 13.3 includes the third ego trajectory 15.3, a third first object trajectory 17.3.1, and a third second object trajectory 17.3.2. Particularly preferably, the plurality of scene predictions 13 are created depending on the navigation destination of the motor vehicle 1.In particular, the respective ego trajectories 15 of the scene predictions 13 are created depending on the navigation destination of the motor vehicle 1.
[0061] In the first fifth step e1), a first object cost value 21.1.1 of the first object trajectory 17.1.1 and a first second object cost value 21.1.2 of the first object trajectory 17.1.2 are calculated. In addition, in the second fifth step e2), a second object cost value 21.2.1 of the second object trajectory 17.2.1 and a second object cost value 21.2.2 of the second object trajectory 17.2.2 are calculated. In addition, in the third fifth step e3), a third object cost value 21.3.1 of the third object trajectory 17.3.1 and a third second object cost value 21.3.2 of the third object trajectory 17.3.2 are calculated.Particularly preferably, the object cost value 21 is determined by means of the second cost function based on at least one object parameter selected from a group consisting of an object driving comfort, an object efficiency, an object collision risk, and a combination of at least two of the said object parameters.
[0062] The first embodiment from Fig. 2 and the second embodiment from Fig. 3 can be extended to a plurality of detected objects 11 and / or a plurality of scene predictions 13. QUOTES CONTAINED IN THE DESCRIPTION
[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature
[0000] DE 10 2022 001 207 A1
[0003] Cited non-patent literature
[0000] -based Trajectory Predictions for Planning by Yuxiao Chen et al., published at https: / / arxiv.org / abs / 2206.13387
[0012] Variable Sequential Set Transformers For Joint Multi-Agent Motion Prediction by Roger Girgis et al., published at https: / / arxiv.org / abs / 2104.00563
[0012] Work Multi-future Transformer: Learning diverse interaction modes for behavior prediction in autonomous driving by Baotian He and Yibing Li, published at https: / / ietresearch.onlinelibrary.wiley.com / doi / pdfdirect / 10.1049 / itr2.12207
[0012]
Claims
[1] Method for trajectory planning in a particularly highly automated or autonomously operated motor vehicle (1), wherein - at a first time, at least one sensor recording (9) is created by means of at least one environmental sensor (5), wherein - in the at least one sensor recording (9) a scene with at least one object (11, 11.1, 11.2) is detected, wherein - a plurality of different scene predictions (13, 13.1, 13.2, 13.3) are created for a same prediction horizon, whereby - each scene prediction (13, 13.1, 13.2, 13.3) has an ego trajectory (15, 15.1, 15.2, 15.3) of the motor vehicle for the prediction horizon, wherein - each scene prediction (13, 13.1, 13.2, 13.3) has an object trajectory (17, 17.1, 17.2, 17.3, 17.1.1, 17.1.2, 17.2.1, 17.2.2, 17.3.1, 17.3.2) for the at least one object (11, 11.1, 11.2) for the prediction horizon, wherein - for each ego trajectory (15, 15.1, 15.2, 15.3) an ego cost value (19, 19.1, 19.2, 19.3) is calculated, where - for each object trajectory (17, 17.1, 17.2, 17.3, 17.1.1, 17.1.2, 17.2.1, 17.2.2, 17.3.1, 17.3.2) an object cost value (21, 21.1, 21.2, 21.3, 21.1.1, 21.1.2, 21.2.1, 21.2.2, 21.3.1, 21.3.2) is calculated, where - for each scene prediction (13, 13.1, 13.2, 13.3) of the various scene predictions (13, 13.1, 13.2, 13.3), a scene prediction cost value (23, 23.1, 23.2, 23.3) is calculated based on the respective ego cost value (19, 19.1, 19.2, 19.3) and the respective object cost value (21, 21.1, 21.2, 21.3, 21.1.1, 21.1.2, 21.2.1, 21.2.2, 21.3.1, 21.3.2), where - based on the plurality of scene prediction cost values (23, 23.1, 23.2, 23.3) an ego trajectory (15, 15.1, 15.2, 15.3) is selected as the planning trajectory (25). [2] Method according to one of the preceding claims, wherein the planning trajectory (25) is determined by minimizing the scene prediction cost values (23, 23.1, 23.2, 23.3). [3] Method according to one of the preceding claims, wherein the ego cost value (19, 19.1, 19.2, 19.3) is determined by means of a first cost function based on at least one ego parameter selected from a group consisting of an ego driving comfort, an ego efficiency, an ego collision risk, an overall trajectory progress, and a combination of at least two of said ego parameters. [4] Method according to one of the preceding claims, wherein the object cost value (21, 21.1, 21.2, 21.3, 21.1.1, 21.1.2, 21.2.1, 21.2.2, 21.3.1, 21.3.2) is determined by means of a second cost function based on at least one object parameter selected from a group consisting of an object driving comfort, an object efficiency, an object collision risk, and a combination of at least two of said object parameters. [5] Method according to one of the preceding claims, wherein the planning trajectory (25) is optimized taking into account at least one optimization parameter. [6] Method according to one of the preceding claims, wherein the plurality of scene predictions (13, 13.1, 13.2, 13.3) are created as a function of a navigation destination of the motor vehicle (1). [7] Method for operating a particularly highly automated or autonomously operated motor vehicle (1), wherein a planning trajectory (25) is obtained by means of a planning method according to one of claims 1 to 6, and wherein the motor vehicle (1) is moved along the planning trajectory (25). [8] Computing device (7) arranged to carry out a method according to one of claims 1 to 7. [9] Control device (3) with at least one environmental sensor (5) and a computing device (7) according to claim 8. [10] Motor vehicle (1) with a control device (3) according to claim 9.
Citation Information
Patent Citations
Methods for predicting the trajectories of objects
DE102022001207A1
Object interaction prediction systems and methods for autonomous vehicles
EP3714345B1
Cost-based path determination
US20210046924A1