Method for assisted or automatic vehicle steering

The method employs attraction and repulsion rules to define voids for improved trajectory planning, addressing precision and computational efficiency issues in driver assistance systems, enhancing vehicle maneuvering accuracy.

JP2024540945AInactive Publication Date: 2024-11-06コンチネンタル·オートナマス·モビリティ·ジャーマニー·ゲゼルシャフト·ミト·ベシュレンクテル·ハフツング
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2024523683
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2021-11-15
Publication Date
2024-11-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing driver assistance systems lack precision in trajectory estimation due to inaccurate reliability information and error models, leading to high computational load and low scalability, particularly in selecting objects of interest over long distances.

Method used

Implement a method using attraction and repulsion rules to define voids based on captured objects, integrating them into a coordinate system for improved trajectory planning, allowing flexible use of measurement data and reducing calculation time.

Benefits of technology

Enhances measurement accuracy and reduces computational complexity by leveraging voids to improve trajectory estimation, enabling precise vehicle maneuvering.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024540945000001_ABST
    Figure 2024540945000001_ABST
Patent Text Reader

Abstract

A method for assisted or automatic vehicle steering of an ego vehicle (1), comprising an ego vehicle (1) comprising a control means (2) and at least one sensor for capturing surroundings and objects, wherein a trajectory plan is performed based on the captured surroundings and captured objects in order to steer the ego vehicle (1), for which voids (10, 11, 12) are generated, for which objects attraction-repulsion rules are defined, and wherein the trajectory plan is performed based on the voids (10, 11, 12), the method comprising the following method steps: initialization, i.e. the voids (10, 11, 12) are brought into a coordinate system, a step of using the attraction-repulsion rules, and a step of simulating according to the defined behavior model.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a method, particularly preferably a computer-implemented method, for assisted or automatic vehicle steering of an ego vehicle, as well as a driver assist system for an ego vehicle for assisted or automatic vehicle steering of an ego vehicle. [Background technology]

[0002] Appropriate vehicles, for example passenger cars (Personnenkraftfahrzeuge), motorized goods vehicles (Lastkraftwaegen) and motorcycles, are increasingly being equipped with driver assistance systems which, with the aid of sensor systems, capture the surroundings or the situation around, recognize the traffic situation and support the driver, for example by braking or steering interventions or by optical, haptic or audible warnings. As sensor systems for capturing the surroundings, radar, lidar, camera, ultrasonic sensors, etc. are usually used. From the sensor data determined by the sensors, inductive inferences can then be made about the surroundings, so that, for example, a so-called surroundings model can be created. On the basis of this, warnings / information for the driver or commands for carrying out controlled steering, braking and acceleration can then be output. The assistant functions thus processing sensor and surrounding data can support or, in some cases, completely take over (semi-automated or fully automated) the driving task or vehicle operation, for example to avoid accidents with other traffic participants or to facilitate complex maneuvers, i.e. to use an emergency brake assistant (EBA, Emergency Brake Assist) to perform autonomous emergency braking (AEB, Automatic Emergency Brake), or to use distance control or adaptive cruise control (ACC) to perform speed adjustment and following control.

[0003] Furthermore, it is possible to determine the path to be traveled and the path along which the vehicle will move. Using sensors, static targets or objects can be detected, which allows, for example, the distance to the preceding vehicle, the road progression or the route to be estimated. The detection or recognition of the objects and, in particular, their validation is important, for example, in order to recognize whether a preceding vehicle is of importance for the respective assistant function or engine control. A criterion for this is, for example, "an object traveling in the same lane as the own vehicle (ego vehicle) recognized as a target vehicle (target)". Known driver assistance systems, for example, with the help of sensors, estimate the lane progression in order to determine whether the target vehicle is in the own lane or not. For this purpose, information on lane markings, surrounding buildings, but also the paths traveled by other vehicles are used. Furthermore, suitable algorithms (for example curve fitting algorithms) are used to predict the future path or trajectory of the ego vehicle. Furthermore, deviations from this path from other traffic participants can also be taken into account in order to determine which path the respective traffic participant is traveling.

[0004] Known systems do not use reliability information or ideal reliability information, such as the standard deviation of the measured values. However, error models for such sensors are not accurate enough to use reliability information, such as directly as a weighting factor. Two error sources are of primary importance here: incorrect prediction of the own lane movement and incorrect or unreliable measurement of the position of the observed traffic participant or other traffic participants, which can cause significant deviations of the predicted path. Both error sources cannot be corrected without knowledge of the correct reliability information. As a result, the computational load is high and the scalability to multiple objects or new object types is low. In particular for driver assistance functions such as ACC, the object of interest must be preselected at long distances in many cases. Object acquisition is usually performed by radar sensors, which have a sufficient sensor range and detection certainty. Nevertheless, at the start of the measurement, the geometric or kinematic estimates are often still of too low quality or too few measurements or too few measurement points are made. The variations in the filters used are often so large that radar objects located, for example, 200 meters away cannot be assigned to a lane with sufficient reliability.

[0005] DE 10 2015 205 135 A1 discloses a method for representing relevant objects of a scene (e.g. guard rails, lane centerlines, traffic participants) as objects in a swarm. These objects are recognized by external sensors and represented in an object constellation, which contains two or more objects, i.e. measurements / objects are merged to save computation time and improve the accuracy of the estimation. Similarly, the merger of different measurements of the same object is also a technically necessary constellation to save computation time, but it is not a semantic constellation, since it is different measurements of the same object and not different objects. The data of the external sensors is, for example, sensor raw data or preprocessed and / or sensor data selected on the basis of predefined criteria. For example, it can be image data, laser scanner data or object lists or so-called "point clouds" (representing, for example, certain object parts or object edges). [Prior art documents] [Patent documents]

[0006] [Patent Document 1] DE 10 2015 205 135 A1 Summary of the Invention [Problem to be solved by the invention]

[0007] The problem to be solved by the present invention, which starts from the prior art, is to provide a method which allows increasing the accuracy of the estimation while keeping the calculation time advantageous. [Means for solving the problem]

[0008] This problem is solved by the general teaching of claim 1 and the parallel independent claims. Expedient embodiments of the invention are claimed in the dependent claims.

[0009] In the method for assisted or automatic vehicle steering of an ego vehicle according to the invention, the ego vehicle includes a control means and at least one sensor, preferably a plurality of sensors for capturing the surroundings, which capture objects in the surroundings of the ego vehicle. Furthermore, the vehicle steering of the ego vehicle is performed based on a trajectory plan, which is performed based on the captured surroundings, in which the surrounding objects are referenced. For these objects, voids are created, which are determined based on attraction-repulsion rules. Then, the trajectory plan is performed based on these voids. As a result, the advantage is obtained that the estimation accuracy is increased and the required calculation time can be significantly reduced. Furthermore, the method according to the invention includes a method step for initializing, in which the objects or the created voids (OSBs - Object-Selection-Boids) are reduced in a coordinate system (or ego coordinate system) using a simulation based on attraction-repulsion rules and a configurable motion model. This results in a simple and flexible use of various measurement data, without having to consider whether or not there is any measurement data at all, or which measurement data is present. If further measurement data are present, they can also be used to obtain better results. Even if some measurement data are missing, a sufficiently good result can be achieved from the remaining measurement data.

[0010] The concept of "trajectory planning" in the context of the present invention explicitly includes purely spatial planning (path planning) as well as planning in space and time (trajectory planning). Similarly, voids can be used only for parts of the system, for example for velocity adaptation or for the selection of specific objects ("Object-of-Interest Selection").

[0011] Preferably, the attraction-repulsion rules are determined by defining objects that are adjacent to each other and arranged in parallel as attracting voids, while objects that are arranged in parallel to each other but with a large gap between them as repulsive voids.

[0012] It also makes sense to define static objects as repelling voids and movable objects as attracting voids. The movable objects can then be observed over time to create their transition history and attract (or repel) voids based on their transition history.

[0013] According to a further advantageous aspect of the invention, the moveable object can be observed (tracked) over time so as to create a transition history and, based on said transition history, attractive voids can be defined.

[0014] Additionally, the captured objects and / or voids can be stored in an object list where all captured objects are stored along with all captured data (position, speed, signal strength, magnitude, elevation and the like).

[0015] It is reasonable to define the feature space based on the position and direction of the ego vehicle, but all the void matching rules can converge to a point in the feature space, which can additionally improve the measurement accuracy.

[0016] The feature space is defined based on the clothoid parameters of the trajectory planning.

[0017] The feature space can also be advantageously extended to other traffic participants, which can greatly increase the measurement accuracy and in addition greatly improve the perimeter capture.

[0018] It also makes sense to equip the vehicle with at least one camera and / or a lidar sensor and / or a radar sensor and / or an ultrasonic sensor and / or other sensors for capturing the surroundings known from the prior art as sensors for capturing the surroundings.

[0019] Furthermore, a behavioral model or attraction-repulsion rules can be expressed as a combination of simple behavioral models, which can be realized geometrically, by feedback control, or by logic.

[0020] Furthermore, it makes sense for the code of conduct or attraction / repulsion rules to be implemented based on priorities, weightings, and / or averaging.

[0021] In addition, a swarm model (based on the behavior of schools of fish or birds) can be used, whereby collision avoidance is performed by the swarm members by adjusting their orientation, or by avoiding collisions with other swarm members. Furthermore, the speed of the neighboring swarm members is adjusted to allow them to follow their neighbors and to favor collision avoidance, not just to stay together. Then, a centering of the swarm is performed, where it is assumed that the boyd adjusts its orientation to stay close to the swarm. This is achieved by centering only on its direct neighbors. For example, if the boyd is on the periphery of the swarm, there are more neighbors in the direction of the swarm center, and therefore the center of the direct neighbors is also in the direction of the swarm center.

[0022] As a parallel independent claim, the invention also encompasses a driver assist system for an ego vehicle for assisted or automatic vehicle steering of the ego vehicle, in which the ego vehicle comprises a control means and at least one sensor, preferably a plurality of sensors, for capturing a periphery, which capture objects in the periphery of the ego vehicle. The control means performs a trajectory planning based on the captured periphery, on the basis of which vehicle steering of the ego vehicle is performed. Sensors for the periphery and object capture can be, for example, radar, lidar, camera or ultrasonic sensors. Objects are referenced for the trajectory planning, for which voids are created, which are defined on the basis of attraction-repulsion rules, thereby enabling trajectory planning taking the voids into account.

[0023] Furthermore, the driver assistance system may be a system which, in addition to sensors for capturing the surroundings, also includes a computer, a processor, a controller, a calculator or the like for carrying out the method of the invention. A computer program having a program code for carrying out the method of the invention when the computer program is executed in a computer or in another programmable computing means known in the prior art is also conceivable. Likewise, the method may be implemented or subsequently introduced as a computer-implemented method in an existing system. The concept of a "computer-implemented method" in the context of the present invention represents a flow chart or a procedure which is realized or carried out by a computer, whereby the computer processes data using programmable calculation rules. However, an important feature of the method is that it may also be implemented later, for example by a new program, a number of new programs, an algorithm or the like. The computer may then be configured as a control means or as part of a control means (for example an (Integrated Circuit)-module, a microcontroller or a system-on-chip (SoC)).

[0024] In summary, the method describes a relationship between an object captured by a sensor and a set of semantically defined voids, which is "shifted" between the voids and the captured object based on attraction-repulsion rules. Thus, trajectory planning can be performed based on the sensor data of the objects selected by the voids. According to a special embodiment of the invention, the shifting of the voids can be performed anew from the same starting point in each processing cycle. Furthermore, the shifting of the voids can also be performed in each processing cycle based on the position of the voids in the previous cycle.

[0025] The invention is explained in more detail below by means of illustrative examples. [Brief description of the drawings]

[0026] [Figure 1] FIG. 1 is a highly simplified schematic representation of an ego-vehicle equipped with an assistance system according to the invention; [Diagram 2] Figure 2 is a simplified depiction of a traffic scene in which an ego vehicle is negotiating a curve where several other vehicles are already traveling; [Diagram 3] FIG. 3 is a simplified representation of the traffic scene of FIG. 2, in which the measurement principle according to the invention is illustrated on the basis of various measurement points; and [Figure 4] Figure 4 is a simplified schematic depiction of the design of the operation method of the navigation module for the individual OSBs. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] Reference number 1 in FIG. 1 denotes an ego vehicle, which comprises a control means 2 (ECU, Electronic Control Unit or ADCU, Assisted and Automated Driving Control Unit), various actuators (steering system 3, engine 4, brakes 5) and sensors for surroundings acquisition (camera 6, lidar sensor 7, radar sensor 8 and ultrasonic sensors 9a-9d). Here, the ego vehicle 1 can be (partially) automatically driven, as the control means 2 has access to the actuators and sensors and their sensor data. In the field of assisted or (partially) automated driving, the sensor data for surroundings and object recognition can be used to enable various assistant functions, such as distance control (ACC, Adaptive Cruise Control), emergency braking assistant (EBA, Electronic Brake Assist), lane keeping control or lane keeping assistant (LKA, Lane Keep Assist), parking assistant or similar functions, by the control means 2 or algorithms installed therein.

[0028] Fig. 2 depicts a typical traffic scene where the ego vehicle 1 enters a curve where there are already several vehicles 10a, 10b, 10c, 10d ahead of it. The ego vehicle 1 detects surrounding objects (vehicles 10a-10d ahead of it, road markings, road edge structures, and similar objects) using its surrounding sensors and creates its own path or trajectory to be traveled based on this information. Furthermore, the movements of other traffic participants can be predicted and used for trajectory planning. For example, the trajectory created based on the detection location and movement prediction of vehicle 11d (indicated by the black arrow) is not optimal or is incorrect because the trajectory based on the movement prediction of vehicle 10d does not follow the lane transition and involves undesired lane changes in the curve area.

[0029] In the method according to the invention, important objects in a scene (guard rails, lane centerlines, traffic participants and similar objects) are then represented as objects in a group (i.e. a so-called flock or association of objects). In contrast to the known simple object integration (simple clusters), the captured objects are not only integrated but also kept individually and influence each other, i.e. interact with each other. The behavior of these objects is then defined by simple rules based on the sensor data and their relationships with each other, i.e. the interaction of objects similar to so-called voids (interacting objects for simulating swarm behavior). The voids correspond to the measured objects and are not constellations of integrated objects, i.e. they semantically represent individual objects and are not simple constellations. In the case of void-based models, the complexity of the model is the result of the interactions of individual objects or voids that follow simple rules, such as separation (selection of a movement or direction that opposes the collection of voids), adjustment (selection of a movement or direction that corresponds to the average direction of adjacent voids), and coherence (selection of a movement or direction that corresponds to the average position of adjacent voids).

[0030] FIG. 3 illustrates the principle of measuring road markings and vehicles, or their travel paths, using voids 11, 12, 13, using the road scene of FIG. 2 as an example. For example, for each cycle, road markings (modeled as straight lines for each section, for example) are measured again and added to the existing list of road marking objects. Then, attraction-repulsion rules are calculated (for example, objects that are adjacent and parallel attract each other, and parallel objects with a large distance between them repel each other). Thus, a repulsive void 11 is created for the lane edges, and a repulsive void 12 is created for the lane center (for example, by road marking detection, guardrail detection, and road edge structure detection). Similarly, vehicles 10a-10d can be represented. In this case, the vehicles recognized by the sensors are represented, for example, as a short transition history. For example, an attractive void 13 is represented by vehicle 10c, or its travel path, in that void 13 is created based on the transition history of vehicle 10c. These measurements or the identified voids are likewise inserted into the list of previous measurements (object list) and their positions are corrected according to the defined rules.

[0031] It is to be noted that errors in lane progression estimation, for example due to observation of paths travelled by other vehicles 10a-10d, can be reasonably corrected by the ego vehicle 1 or by the vehicles implementing their respective trajectory planning, taking into account predefined rules. Examples of rules include: "guardrails are parallel to the lanes", "lanes have at least approximately constant width", "vehicles run parallel to the lanes", "guardrails are centred at the measuring points", "guardrails have no bends or branches" and similar rules. This automatically results in a path or trajectory whose progression is parallel to the guardrails ("emergent behaviour"). Furthermore, the measurements can be weighted in a configurable way, for example with an αβ filter.

[0032] An increase in the measurement accuracy is achieved in particular by the fact that the matching rules converge all measurements (each corresponding to a void) of the same point in a feature space, which then consists of the position and orientation of the ego vehicle 1. This principle can surprisingly be extended to other traffic participants or vehicles as well, for example "vehicles running parallel to the lane" (i.e. the same rules as for static objects, except that now the modeled positions of the vehicles change), "vehicles do not collide with each other", "vehicles in the same lane are aligned with each other" ("platooning"), "vehicles do not collide with guardrails" and similar rules.

[0033] Alternatively, it is also possible to choose the space of the trajectory's clothoid parameters as feature space. In this case, the individual voids are a number of individual measurements over time. Here, these voids may be fixed, for example, in the vertical direction and can only move laterally and with respect to their curves according to a rule. It is practical to erase these voids, in this case, when the ego vehicle 1 has passed through them. This allows, among other things, to save memory and calculation time. In addition, several voids representing the same object in the real space (for example, the voids form a compact cluster with a predefined variance) can be merged to save memory and calculation time.

[0034] In the following, we show how the components of the OSB (Object-Selection-Boid) group are integrated into a self-organizing system with emergent behavior, where the concept of OSB refers to one group member in the OSB group, and how the system or OSB group is applied to situation recognition for the ACC function. In this case, after every update cycle of the sensors, a new surrounding model is generated, which is the time when the sensors perform measurements and the surrounding model is generated. The surrounding model contains a list of all traffic participants (TP = Traffic Participant) captured by the sensors for this update cycle. This list therefore contains information such as routes and the positions of the traffic participants that are used for the self-organizing system or OSB group. Once every update cycle, the OSB group is transmitted about the traffic participants and routes in this list, and each OSBs assigns the traffic participants the corresponding roles and saves them in the list. That is, after every update cycle of the surrounding model, the OSB fleet is simulated once for a certain number of simulation steps, traveling through the entire route with the traffic participants.

[0035] The simulation can be divided into three essential method steps: initialization, use of the behavioral norms, and simulation according to a configurable behavior model. In the first simulation step, the OSBs are initialized, i.e., the OSBs are placed in the ego coordinate system of the surrounding model. The current number of selected traffic participants, the direction angles θ and steering angles β of the OSBs are set to zero. The speeds v of all OSBs are calculated based on the initial speed v init The OSB representing the ego lane, i.e., the OSB traveling in front of the ego vehicle, is set on the ego vehicle, and the remaining OSBs are set at intervals d lat,initThe x-position in the ego coordinate system is initially zero for all objects. A further variation features that the orientation angle θ, steering angle β and / or x- and y-position of the OSBs are initialized from sensor data or navigation map data. This step is performed anew for each new simulation or each new update cycle of the surrounding model.

[0036] In a second step, the behavior of the OSBs is determined using the rules of conduct or attraction-repulsion rules as well as the navigation (particularly the navigation module). For this, first the traffic participants and routes in the current surrounding model as well as the neighboring OSBs must be transformed from the ego coordinate system to the OSB coordinate system. In parallel, it is calculated which traffic participants and routes are in the field of view of each OSB. Then, using the transformed coordinate system, the respective rules of conduct are adapted, e.g. to follow one / vehicle and / or to maintain a distance between one / vehicle and / or multiple traffic participants and / or to maintain the initial state of the initialization step (particularly to maintain the distance between formations and groups).

[0037] The code of conduct then gives a speed change Δv or a steering angle change Δβ. The navigation module then adds these changes to the current speed v and current steering angle β of each OSB. A prioritization of the changes can be applied. It is possible to prioritize and only use certain steering angle changes if a route or a traffic participant is in the field of view of each OSB. If there is no route in the field of view, only the formation change can be used. In parallel, the navigation module can check whether there is a discrepancy between the formation change and the changes of the individual traffic participants. If the discrepancy exceeds a configurable limit, the priority of the traffic participant's following for each OSB is ignored for n simulation steps and the OSB is dispersed by the formation change and pushed back or moved into the formation. For all OSBs, if a route or traffic participant is within its field of view, each navigation module of the OSB is given a global steering angle to a global target, which can be selected as the steering angle for the OSB until the route is found. Other global targets can be, for example, the navigation map, other sensor data, and / or V2X communication data. Each OSB then determines its own behavioral norm, i.e. it only considers its neighbors or traffic participants and routes within its field of view. The movement is determined decentrally by its local neighborhood, which is a necessary feature for a self-organizing system. An example of the resulting navigation module functionality for each OSB is shown in Figure 4.

[0038] After the speed and steering angle of the OSBs are changed based on the behavioral norm in the second step, the OSBs are simulated in the third step, which is the simulation step. In that time, all OSBs are simulated and updated. Then, the second and third steps are repeated for a configurable number of simulation steps until the OSBs move through the entire field of view of the ego vehicle, the simulation of the OSBs for one update cycle is completed, and the correct roles are assigned, or multi-object selection is achieved. To improve this selection / selection, previous roles for previous cycles can be saved. If a traffic participant is not assigned a role in the latest update cycle of the OSBs, the role from the previous cycle is inherited. This assignment is maintained for three cycles, unless a new role assignment is made in the meantime. If the assignment is not possible after several cycles, the traffic participant is no longer assigned a role. In addition, using the previous roles allows hysteresis to be realized. The most frequent role in three cycles, i.e. the most recent cycle and the two previous cycles, is always selected, thereby avoiding switching between two roles in a short period of time.

[0039] Furthermore, as a behavior model it is possible to assume, for example, a bicycle model or a single lane model or a half-vehicle model, which includes, for example, three state values: the direction angle θ, the x- and y-positions for the ego vehicle. The speed v and the steering angle β can be used as input values ​​that can influence the model and thereby control each OSB of the group. For the OSBs, a direct change of the steering angle can be assumed in order to simplify the model and the control of the OSBs. This simplification is possible because the OSBs are only required for the multi-object selection and are not given the actual trajectory that the ego vehicle should follow. In addition, it goes without saying that it is also possible to assume other behavior models known in the prior art. [Explanation of symbols]

[0040] 1 EGO vehicle 2. Control Means 3 Steering system 4 Engine / Motor 5 Brakes 6. Camera 7 Lidar Sensor 8 Radar Sensor 9a-9d Ultrasonic sensors 10a Vehicle 10b Vehicle 10c Vehicle 10d Vehicle 11 Void (road markings, lane edges) 12 Void (Road marking, center of lane) 13 Void (Vehicle 10c Movement)

Claims

1. an ego vehicle (1) including a control means (2) and at least one sensor for capturing surroundings and objects; A trajectory plan is performed based on the captured object to steer the ego vehicle (1); Voids (10, 11, 12) are created for the objects, by which attraction and repulsion rules are defined, and The trajectory planning is carried out based on the void (10, 11, 12) A method for assisted or automatic vehicle steering of an ego vehicle (1), comprising: A method characterized by the following method steps: Initialization, i.e., the void (10, 11, 12) is brought into the coordinate system; using attraction and repulsion rules; and A step of simulating according to a defined behavioral model.

2. 2. The method of claim 1, wherein the attraction and repulsion rules are determined by defining objects that are adjacent to each other and arranged in parallel as attracting voids, while objects that are arranged in parallel to each other but with a large gap between them as repulsive voids.

3. 3. A method according to claim 2, characterized in that the static objects are defined as repulsive voids (11, 12) and the mobile objects as attractive voids (13).

4. 4. A method according to claim 2 or 3, characterized in that the movable object is observed over time, a transition history is created and, on the basis of said transition history, attractive voids (13) are defined.

5. 10. A method according to any one of the preceding claims, characterized in that the captured objects and / or voids are stored in an object list.

6. 10. The method according to claim 9, wherein a feature space is defined based on the position and direction of movement of the ego vehicle (1), and the matching rules for all voids (13) converge to a point in the feature space.

7. 7. The method of claim 6, wherein the feature space is defined based on clothoid parameters of the trajectory plan.

8. 8. The method according to claim 6 or 7, characterized in that the feature space is extended to other traffic participants.

9. 10. The parking assistance system according to claim 9, further comprising at least one camera sensor (6) and / or a lidar sensor (7), a radar sensor (8) or an ultrasonic sensor (9a-9d) for detecting the surroundings.

10. 10. A method according to any one of the preceding claims, characterized in that the attraction and repulsion rules are expressed as a combination of simple behavioural norms.

11. 10. A method according to any one of the preceding claims, characterized in that the attraction and repulsion rules are realized geometrically, by feedback control or logically.

12. 10. A method according to any one of the preceding claims, characterized in that the attraction and repulsion rules are realised on the basis of priorities, weights and / or averaging.

13. A driver assist system for an ego vehicle (1) for assisted or automatic vehicle steering of the ego vehicle (1) comprising a control means (2) and at least one sensor for capturing surroundings and objects, characterized in that the control means (2) performs trajectory planning based on a method according to any one of the preceding claims.

Citation Information

Patent Citations

  • Vehicle control system, vehicle control method and vehicle control program

    JP2018037100A

  • Vehicle control device

    JP2019119259A

  • Swarm autopilot

    US20170068243A1

  • Method for determining an automation level that can be used for at least partially automated vehicle movement

    DE102015205135A1