Control system and control method for sampling-based planning of possible trajectories for motor vehicles

DE502019012460D1Active Publication Date: 2024-11-21ZF AUTOMOTIVE GERMANY GMBH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
DE502019012460
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Priority Date
2018-10-31
Filing Date
2019-10-30
Publication Date
2024-11-21
Estimated Expiration
2039-10-30

AI Technical Summary

Technical Problem

Current control systems for semi-autonomous and autonomous vehicles face challenges in determining safe and comfortable trajectories in dynamically changing traffic scenarios, particularly due to the complexity of recognizing and responding to lane changes, speed variations, and collisions, which leads to increased resource consumption and reduced real-time performance.

Method used

A control system that utilizes environmental sensors to detect lanes, road boundaries, other vehicles, and objects, determining lateral and longitudinal positions, speeds, and stopping points for a motor vehicle, employing spline-based interpolation to calculate trajectories, and incorporating adaptive discretization and collision checks to optimize driving maneuvers.

Benefits of technology

This approach enables efficient and accurate determination of optimal trajectories, enhancing driving safety and comfort by improving the assessment of the current driving situation and accounting for static and dynamic collisions, while minimizing resource usage and ensuring real-time performance.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader
Need to check novelty before this filing date? Find Prior Art

Description

Background of the invention

[0001] This document describes a control system and a control method for determining a trajectory that a vehicle should follow to execute a driving maneuver. The control system and the control method are based in particular on environmental sensors in the vehicle and support a driver or an autonomously driving vehicle. In semi-autonomous and autonomously controlled vehicles, this helps to increase the safety and driving comfort of the vehicle's occupants. State of the art

[0002] The ever-increasing interest in the diverse functions of self-driving vehicles is giving rise to an increased need for reliable and fast control systems and algorithms for motion planning (trajectory planning) of at least partially autonomous vehicles. A major challenge is not only the growing number of road users, but also the cognitive tasks that road users must master in highly complex traffic scenarios. In the architecture of control systems for partially autonomous or autonomous vehicles, decision-making and motion planning, in addition to the recognition and control of specific traffic situations, significantly influence the overall performance of the respective vehicle.

[0003] The increasingly sophisticated control systems and algorithms for trajectory planning of semi-autonomous vehicles in recent years demonstrate the complexity of automated driving. The challenge regarding motion planning typically lies in determining a comfortable, executable, and collision-free trajectory within a limited time frame, thereby ensuring real-time performance of the semi-autonomous vehicle systems in a dynamically changing environment.

[0004] In the past, various approaches to trajectory planning have been developed. These include, for example, motion planning algorithms or sampling-based trajectory planning techniques. The latter typically uses certain simplifications of the motion planning problem to meet specific requirements, such as limited computation time. The (simplification) goal is to cover a continuous space with a finite number of samples.

[0005] In today's motor vehicles, advanced driver assistance systems (ADAS) offer a variety of monitoring and alerting functions to make driving safer. These systems monitor the vehicle's surroundings based on data obtained from one or more sensors on the vehicle, with regard to the driver's current driving progress.

[0006] For example, known driver assistance systems determine whether the vehicle is within a lane and whether the driver is unintentionally drifting to one side of the lane or about to leave it. These driver assistance systems use the acquired environmental data to generate an "image" of the road, and in particular the lane. Objects such as curbs, lane markings, directional arrows, etc. are detected and tracked while driving.

[0007] So-called "blind spot monitors" are also part of today's driver assistance systems. These use radar, lidar, video, or similar technology to determine whether another vehicle, a road user, or an object is located to the side and / or behind the vehicle, so that changing lanes or turning the vehicle could lead to a collision.

[0008] Furthermore, in so-called ACC (Adaptive Cruise Control) systems, the driver's own vehicle's automatic cruise control is adjusted to the speed of a vehicle ahead. A specific distance from the vehicle ahead is always maintained. To this end, such systems determine the direction of travel and / or the speed of the vehicle ahead in order to prevent the driver's own vehicle from crossing the path of the vehicle ahead in such a way that a critical situation arises. This applies, on the one hand, to lane changes or turning maneuvers, and, on the other hand, to rear-end collisions.

[0009] In human-driven vehicles, driver assistance systems typically provide a warning function to warn the driver of a critical situation or a corresponding maneuver, or to suggest a suitable maneuver for the vehicle. Driver assistance systems can also be used in autonomously controlled vehicles to provide the autonomous control system with the relevant environmental data.

[0010] For example, DE 10 2016 205 442 A1 discloses a method for optimizing a vehicle's path planning. Further technological background can be found in DE 10 2015 217 891 A, DE 10 2016 211 129 A1, and T. Sartini et al.: "A resolution-adaptive strategy for probabilistic motion planning," World Automation Congress, 2002, Proceedings of the 5th Biannual June 9-13, 2002, January 1, 2002, pages 591-596. Underlying problem

[0011] In road traffic, situations can arise that require a driver or a (semi-)autonomous driver assistance system of a vehicle to perform a driving maneuver. For example, a curved lane may require a corresponding maneuver by the vehicle.

[0012] However, the current situation in which one's own vehicle finds itself is not constant, but is constantly changing in reality. For example, other road users may intentionally or inadvertently change lanes or speed due to an accident. Furthermore, the current driving situation of one's own vehicle changes due to one's own driving behavior and / or the changing lane course. Responding appropriately and in a timely manner to such changes in the current situation presents a major challenge for both conventional driver assistance systems and human drivers. The dynamic nature of many traffic situations and ever-increasing traffic volumes are also reflected in the increasing resource consumption of conventional driver assistance systems.

[0013] The objective is therefore to provide a control system and a control method for a motor vehicle that increase the driving safety and comfort of the vehicle depending on the current traffic situation in which the vehicle is located. Furthermore, resources for the control system should be saved by efficiently calculating trajectories for subsequent driving maneuver decisions for the vehicle. Proposed solution

[0014] This object is achieved by a control system having the features of claim 1 and a control method having the features of claim 12.

[0015] Preferred embodiments will become apparent from subclaims 2 to 11 and 13 and the following description.

[0016] One aspect relates to a control system configured and intended for use in a user's own motor vehicle. This control system detects lanes, lane boundaries, lane markings, other motor vehicles, and / or other various objects based on environmental data obtained from at least one environmental sensor assigned to the user's own motor vehicle. The at least one environmental sensor is configured to provide an electronic controller of the control system with environmental data representing the area in front of, to the side of, and / or behind the user's own motor vehicle. The control system is configured and intended at least to determine information regarding a current driving situation of the user's own motor vehicle and at least one other motor vehicle based on the provided environmental data.Furthermore, the control system is at least configured and intended to determine a plurality of lateral positions based on the information regarding the current driving situation of the own motor vehicle and / or the other motor vehicle and to determine a plurality of longitudinal positions and / or speeds based on the information regarding the current driving situation of the own motor vehicle and / or the other motor vehicle.Furthermore, the control system is configured and intended to determine, from the plurality of lateral positions and the plurality of longitudinal positions and / or speeds, stopping points for a trajectory for the own motor vehicle, which the own motor vehicle is to follow when executing a driving maneuver, and to determine the trajectory for the own motor vehicle by means of a spline-based interpolation between the determined stopping points and corresponding to the lateral positions at the stopping points.

[0017] When determining lanes, lane boundaries, lane markings, other motor vehicles and / or other various objects, the control system can, for example, be set up and determined to determine a lane in which the vehicle's own motor vehicle is located and another lane in which another motor vehicle is located in the area in front of, to the side of and / or behind the vehicle's own motor vehicle, and to determine the other motor vehicle in the area in front of, to the side of and / or behind the vehicle's own motor vehicle as another motor vehicle to be detected. The other lane can correspond to the vehicle's own lane or be adjacent to the vehicle's own lane. This can ensure that other motor vehicles traveling in the same lane as the vehicle's own motor vehicle as well as in a lane adjacent to the vehicle's own motor vehicle can be determined as the other motor vehicle to be detected.

[0018] In addition, lateral movements (of the longitudinal axes) of the driver's own motor vehicle and / or the other motor vehicle in the direction of a lane boundary of a lane in which the driver's own motor vehicle and / or the other motor vehicle are located can be determined. In particular, the control system can determine another motor vehicle as the other motor vehicle that is located in front of and / or behind and / or to the side of and / or to the side of the driver's own motor vehicle and is closest to the driver's own motor vehicle in the current driving situation (at the current time).

[0019] The control system can determine the current lane and the other lane, for example, based on the provided environmental features, such as lane boundaries and / or lane markings. The control system can also recognize these environmental features based on the environmental data obtained by the at least one environmental sensor. Furthermore, the control system can also receive the environmental features from an external system, such as a GPS system.

[0020] The other motor vehicle can be detected over a predetermined period of time or continuously by means of the at least one environmental sensor of the own motor vehicle, and the corresponding data can be provided to the control system of the own motor vehicle. The control system can be configured and determined to determine, over a predetermined period of time or continuously, a change in a distance between a longitudinal axis of the other motor vehicle and a center line, at least one lane boundary, or at least one lane marking of the associated lane in which the other motor vehicle is located. The center line and the lane boundary of the associated lane can be a virtual current center line or current lane boundary of the associated lane determined by the control system.Similarly, the control system may detect, during the predetermined period of time or continuously, a change in the distance between a longitudinal axis of the other motor vehicle and a virtual or real lane marking or lane boundary on which the own motor vehicle is located.

[0021] The control system can be set up and intended to determine a distance between the own motor vehicle and another motor vehicle located in front of, to the side of and / or behind the own motor vehicle, as well as a speed difference and / or an acceleration difference between the own motor vehicle and the other motor vehicle, in order to determine the current traffic situation, wherein the own motor vehicle and the other motor vehicle can be in the same lane or in adjacent lanes.

[0022] The determined distances, speed differences and / or acceleration differences between the own motor vehicle and the other motor vehicle can be combined and related to each other as desired by the control system.

[0023] It is understood that the control system can also be configured and intended to continuously detect the other motor vehicles participating in traffic using the at least one environmental sensor over a predetermined period of time or continuously in order to determine the current traffic situation. Thus, the control system can also continuously determine changes in the previously described relative distances, speed differences, and / or acceleration differences over a predetermined period of time, specifically for all other motor vehicles or other objects that are in the immediate vicinity of the vehicle in a current traffic situation.

[0024] In certain embodiments, the control system may further be configured and determined to determine, based on the provided environmental data, a basic driving maneuver that the vehicle is to perform, and to determine at least one of the plurality of lateral positions based on the determined basic maneuver and / or to assign a value to at least one of the lateral positions that characterizes a basic maneuver class, wherein the basic maneuver class is contained in a set of basic maneuver classes that is composed of at least basic maneuvers for changing lanes left, changing lanes right, and keeping lanes.

[0025] Alternatively or additionally, the set of basic maneuver classes may include the additional basic maneuvers lane keeping left and / or lane keeping right.

[0026] The control system may further be configured and determined to determine the plurality of longitudinal positions based on a longitudinal acceleration of the own motor vehicle and / or the other motor vehicle.

[0027] This could, for example, be the maximum possible acceleration of one's own vehicle based on the engine power of one's own vehicle. Alternatively or additionally, it could be maximum acceleration values ​​of one's own and / or the other vehicle derived from safety and comfort considerations, which could be based on the current driving situation or specified in another suitable manner.

[0028] In certain embodiments, the control system may further be configured and determined to determine the plurality of longitudinal positions by means of adaptive discretization, wherein an acceleration of the own motor vehicle and / or the other motor vehicle is used as a control parameter for the adaptive discretization.

[0029] This may, for example, be one of the longitudinal and / or lateral accelerations described above or corresponding maximum values.

[0030] Furthermore, this may in particular be the acceleration in the direction of travel of the own motor vehicle and / or in the direction of travel of the other motor vehicle at the respective time of determination of the current driving situation of the own motor vehicle and / or the other motor vehicle.

[0031] By means of adaptive discretization, a non-linear image of the linearly sampled longitudinal positions and / or velocities of the own motor vehicle and / or the other motor vehicle can be created.

[0032] In addition, with the help of adaptive discretization, varying longitudinal accelerations of one's own vehicle and / or the other vehicle can be taken into account when determining the trajectory for one's own vehicle, particularly when evaluating a collision scenario with a motorway and / or expressway.

[0033] The control system may further be configured and intended to carry out the determination of the trajectory for the own motor vehicle taking into account a static collision check, wherein the static collision check is based on the plurality of lateral positions of the own motor vehicle and / or the other motor vehicle.

[0034] The control system may further be configured and intended to carry out the determination of the trajectory for the own motor vehicle taking into account a dynamic collision check, wherein the dynamic collision check is based on the plurality of lateral positions and / or on the plurality of longitudinal positions and / or speeds of the own motor vehicle and / or the other motor vehicle.

[0035] The control system can further be designed and configured to carry out the dynamic collision test taking into account the maximum acceleration of the vehicle itself and / or the other vehicle.

[0036] The control system may further be configured and intended to perform the static collision check and / or the dynamic collision check using the separation set.

[0037] The control system may further be configured and intended to check the trajectory for the own motor vehicle by means of an objective function based on a cost function.

[0038] A target state can be assigned to the objective function, and deviation(s) of the objective function from the target state can be used as objectives of the objective function. Alternatively or additionally, further objectives of the objective function in the form of the lateral and / or longitudinal acceleration of the vehicle and / or in the form of the lateral and / or longitudinal jerk of the vehicle can be used as objectives of the objective function.

[0039] In this way, a comprehensive holistic objective function can be determined, the individual terms of which model the dynamic collision avoidance, the static collision avoidance, the feasibility and the driving comfort for the respective current traffic situation of the own motor vehicle 12 in order to select the trajectory for the own motor vehicle for each current driving situation and for each maneuver class.

[0040] According to the invention, the control system is configured and intended to determine the plurality of lateral positions and / or the plurality of longitudinal positions and / or speeds for the own motor vehicle and / or for the other motor vehicle in curvilinear coordinates with respect to a reference lane provided to the control system of the own motor vehicle by the environmental data, and to check the trajectory for the own motor vehicle in orthogonal coordinates.

[0041] Interpolation can also be performed in orthogonal coordinates.

[0042] This allows trajectory planning to be carried out independently of the road geometry and can be used for journeys on straight and curved road sections.

[0043] In contrast to evaluating candidate trajectories in curvilinear coordinates, this eliminates the need for a time-consuming point-by-point transformation of trajectories determined for obstacles (e.g., the other vehicle). Furthermore, transforming the stopping points, including interpolation and evaluation, into orthogonal coordinates (vehicle coordinates) eliminates the need to transform each individual trajectory point. For accuracy reasons, the resolution of the stopping points can be selected according to the road curvature, thus minimizing any deviation from a reference lane (reference trajectory) caused by the interpolation.

[0044] The control system may further be configured and determined to determine the trajectory using an objective function based on a cost function.

[0045] The control system can further be configured and intended to perform a static collision check when determining the trajectory for the host motor vehicle. This static collision check is based on the plurality of initial sampling values.

[0046] The control system can further be configured and intended to perform a dynamic collision check when determining the trajectory for the host motor vehicle. The dynamic collision check is based on the plurality of second sampling values.

[0047] The control system may further be designed and configured to carry out the dynamic collision test taking into account the maximum acceleration of the vehicle itself.

[0048] The control system may further be configured and determined to perform the static collision check and / or the dynamic collision check using the separation set.

[0049] A further aspect of the invention relates to a control method according to claim 12.

[0050] Yet another aspect relates to a motor vehicle comprising a control system as described above.

[0051] Compared to conventional driver assistance systems, the solution presented here improves the accurate assessment and recognition of the current driving situation of the driver's own vehicle and other vehicles. Furthermore, a real-time, two-stage planning and evaluation approach for trajectory candidates is provided, enabling the precise and efficient determination of the best possible (optimal) trajectory for a future driving maneuver of the driver's own vehicle.

[0052] This optimal trajectory can therefore be determined as an appropriate response to the current traffic situation in which the vehicle is located. The environmental data obtained by at least one environmental sensor constantly changes according to the real traffic and driving situation and can be updated cyclically.

[0053] Using the best possible trajectory for a future driving maneuver of the driver's own vehicle increases the driving comfort and safety of the driver's own vehicle, as not only are static and dynamic collision tests performed, but the driver's own vehicle's driving dynamics are also taken into account when determining the trajectory. The control system can also ensure the dynamic feasibility of the best possible trajectory, which further increases driving safety. Furthermore, in other embodiments of the control system, the driver's own vehicle's driving behavior when following the trajectory can be optimized by taking the driver's own vehicle's driving dynamics into account. Driving dynamics here include, for example, the longitudinal and lateral acceleration of the driver's own vehicle.

[0054] It will be apparent to those skilled in the art that the aspects and features described above can be combined as desired in a control system and / or a control method. Although some of the features described above were described with reference to a control system, it is understood that these features may also apply to a control method. Likewise, the features described above with reference to a control method may apply correspondingly to a control system. Short description of the drawing

[0055] Further objects, features, advantages, and possible applications will become apparent from the following description of non-limiting embodiments with reference to the accompanying drawings. All described and / or illustrated features, individually or in any combination, illustrate the subject matter disclosed herein. The dimensions and proportions of the components shown in the figures are not to scale. Identical or equivalent components are provided with the same reference numerals. Figure 1 schematically shows a motor vehicle according to certain embodiments, comprising a control system and at least one environment sensor. Figure 2 schematically shows the architecture for a two-stage approach for planning and evaluating candidate trajectories according to certain embodiments. Figure 3 schematically shows lateral reference paths (reference trajectories) for five different lateral maneuver classes according to certain embodiments. Figure 4 schematically shows a nonlinear mapping from linear to adaptive discretization using different exemplary acceleration values ​​as control parameters according to certain embodiments. Figure 5 schematically shows a mapping of the relative costs of adaptive discretization compared to linear discretization for a different number of longitudinal accelerations according to certain embodiments.Figure 6 shows schematically the transformation from curvilinear coordinates. K in vehicle coordinates Faccording to certain embodiments. Figure 7 schematically shows a set of determined candidate trajectories according to certain embodiments for a straight (upper image) and a curved (lower image) road section. Figure 8 schematically shows an optimal trajectory determined by the control system from the set of determined candidate trajectories according to certain embodiments for a straight (upper image) and a curved (lower image) road section. Figure 9 schematically illustrates the separation set applied according to certain embodiments. Figure 10 schematically shows the mean runtime for dynamic collision checking according to certain embodiments. Figure 11 shows a schematic representation for dynamic collision checking using the separation set according to certain embodiments.Figure 12 shows a schematic representation of the two-stage trajectory planning and evaluation approach according to certain embodiments, with the first stage of sampling-based determination of the candidate trajectories (left) and the second stage of collision testing (right). Figure 13 schematically shows (from top to bottom) the first five representations of a highway merging scenario according to certain embodiments using the two-stage trajectory planning and evaluation approach. Figure 14 schematically shows (from top to bottom) the second five representations of the highway merging scenario according to certain embodiments using the two-stage trajectory planning and evaluation approach. Figure 15 schematically shows the speed of the motor vehicle during the collision described in the . Figures 13 and 14 shown collision scenarios onto a motorway. Detailed description of the drawings

[0056] In the following disclosure, certain aspects are described primarily with reference to the control system. However, these aspects are naturally also valid within the scope of the disclosed control method, which can be executed, for example, by a central control unit (ECU) of a motor vehicle. This can be done by performing suitable write and read accesses to a memory assigned to the motor vehicle. The control method can be implemented within the motor vehicle in both hardware and software, or a combination of hardware and software. This also includes digital signal processors, application-specific integrated circuits, field-programmable gate arrays, and other suitable switching and computing components.

[0057] Figure 1schematically shows a motor vehicle 12 (own motor vehicle) that includes a control system 10. The control system 10 is coupled to at least one environmental sensor 14, 16, 18 located on the motor vehicle 12 in order to receive environmental data from the at least one sensor 14, 16, 18. The control system 10 can include an electronic control unit (ECU; not shown in the figure). For example, the present control system 10, with the aid of the ECU and / or other electronic control systems, can be configured and intended at least to determine a trajectory for the motor vehicle 12 and / or to determine a trajectory for another motor vehicle that is located in the environment of the own motor vehicle 12 in a current driving situation. For example, the ECU receives signals from the environmental sensors 14, 16, 18, processes these signals and the associated environmental data, and generates control and / or output signals.

[0058] In the Figure 1 Three environmental sensors 14, 16, 18 are shown, which send corresponding signals to the control system 10 or the electronic control unit ECU. In particular, at least one environmental sensor 14 is arranged on the motor vehicle 12, directed forward in the direction of travel of the motor vehicle 12, which detects an area 22 in front of the motor vehicle 12. This at least one environmental sensor 14 can be arranged, for example, in the area of ​​a front bumper, a front lamp, and / or a front radiator grille of the motor vehicle 12. As a result, the environmental sensor 14 detects an area 22 directly in front of the motor vehicle 12.

[0059] At least one additional or alternative environment sensor 16, also directed forward in the direction of travel of the motor vehicle 12, is shown in the region of a windshield of the motor vehicle 12. For example, this environment sensor 16 can be arranged between an interior rearview mirror of the motor vehicle 12 and its windshield. Such an environment sensor 16 detects an area 24 in front of the motor vehicle 12. Depending on the shape of the motor vehicle 12, an area 24 directly in front of the motor vehicle 12 cannot be detected due to the front section (or its geometry) of the motor vehicle 12.

[0060] Furthermore, at least one environmental sensor 18 can be arranged on the side and / or at the rear of the motor vehicle 12. This optional environmental sensor 18 detects an area 26 that lies to the side and / or behind the motor vehicle 12 in the direction of travel of the motor vehicle 12. For example, the data or signals from this at least one environmental sensor 18 can be used to verify information detected by the other environmental sensors 14, 16 and / or to determine a curvature of a lane traveled by the motor vehicle 12.

[0061] The at least one environment sensor 14, 16, 18 can be implemented in any way and can include a front camera, a rear camera, a side camera, a radar sensor, a lidar sensor, an ultrasonic sensor, and / or an inertial sensor. For example, the environment sensor 14 can be implemented in the form of a front camera, a radar, lidar, or ultrasonic sensor. A front camera is particularly suitable for the higher-positioned environment sensor 16, while the environment sensor 18 arranged at the rear of the motor vehicle 12 can be implemented in the form of a rear camera, a radar, lidar, or ultrasonic sensor.

[0062] The electronic control unit ECU processes the environmental data obtained from the environmental sensor(s) 14, 16, 18 located on the motor vehicle 12 in order to obtain information regarding the static environment (immobile environmental objects such as road markings) and the dynamic environment (moving environmental objects such as other motor vehicles or road users) of the motor vehicle 12.

[0063] The electronic control unit processes the environmental data obtained from the environmental sensor(s) 14, 16, 18 located on the motor vehicle 12 in order to detect a lane traveled by the motor vehicle 12 with a first and a second lateral lane boundary in front of the motor vehicle 12. In addition, the electronic control unit ECU processes the environmental data obtained from the environmental sensor(s) 14, 16, 18 located on the motor vehicle 12 in order to detect a lane occupied by another object (which is adjacent to the lane traveled by the vehicle's own vehicle, where adjacent means that one or more other lanes can also lie between the adjacent lanes) as well as its lateral lane boundaries in front of, to the side of, and / or behind the motor vehicle 12.The other object here is another motor vehicle moving along the lane adjacent to that of the driver's own motor vehicle, or any other possible obstacle in the lane in front of that other motor vehicle.

[0064] For this purpose, the environmental sensors 14, 16, 18 provide the electronic control unit (ECU) with the environmental data representing the area in front of, to the side of, and / or behind the vehicle. For this purpose, the control system 10 is connected to the vehicle via at least one data channel or bus (in Fig. 1 (shown in dashed lines) is connected to at least one environmental sensor 14, 16, 18. The data channel or bus can be implemented via cable or wirelessly.

[0065] Alternatively or additionally, the control system 10 or its electronic control unit (ECU) can also receive data from one or more other assistance systems 20 or another control system 20 of the motor vehicle 12, which indicate the lanes traveled by the vehicle 12, the other vehicle, and other vehicles, including their lateral lane markings, or can be derived therefrom. Thus, data and information already determined by other systems can be used by the control system 10.

[0066] Furthermore, the control system 10 or its electronic control unit (ECU) determines a driving situation using the environmental sensors, i.e., based on the environmental data acquired using the at least one environmental sensor 14, 16, 18. Alternatively or additionally, an existing assistance system 20 or an electronic control unit 20 can also provide data and / or information that define a driving situation or from which a driving situation can be quickly derived. Depending on the determined driving situation, at least one possible trajectory is then determined, which the driver's own motor vehicle 12 should follow during the further course of the journey.

[0067] The driver assistance system 20 or the electronic control system 20 can further be configured and intended to partially (autonomously) control the motor vehicle. In this case, the control system 10 is configured and intended to output data to the driver assistance system 20 or the electronic control system 20 for autonomous driving. In particular, the control system 10 (or its ECU) can output data to the component 20 indicating a course of the specific trajectory that the vehicle 12 is to follow in the further course (depending on the current traffic situation). The data can also be transmitted via a data channel or bus, either wired or wirelessly.

[0068] The information obtained from the environmental data and provided to the control system 10 includes, for example, positions and / or speeds and / or accelerations of the host motor vehicle 12 and / or the other motor vehicle in lateral, longitudinal, or a combination of lateral and longitudinal directions. Furthermore, this information includes, for example, relative speeds and / or relative accelerations between the host motor vehicle 12 and the other motor vehicle in the respective directions.

[0069] An overview of the approach presented here, based on real-time sampling of the environmental data provided to the control system 10, to plan one or more trajectories for the own motor vehicle 12 and to determine the best possible (optimal) trajectory of the planned trajectories for the further course of the own motor vehicle, which in particular underlies the control system 10 and the control method presented here, is provided by Figure 2 . This shows that control system 10 initially generates trajectory candidates. These trajectory candidates are based on the lateral maneuver classes of left lane change (SWL), left lane keeping (SHL), center lane keeping (SHM), right lane keeping (SHR), and right lane change (SWR). However, the present disclosure is not limited to these; alternatively, more or fewer, or other maneuver classes, may be defined.

[0070] The control system / control method then selects the best possible trajectory candidates from the trajectory candidates based on the maneuver classes using an objective function. The objective function incorporates target states relating to the dynamic and static environment of the host motor vehicle 12 in the current driving situation, as well as the driving comfort and feasibility of the best possible trajectory candidates. Within the scope of this disclosure, a trajectory is considered unfeasible, for example, if the driving characteristics of the host motor vehicle 12 are unable to generate the necessary acceleration.

[0071] Static and dynamic collision checks are then performed to determine the optimal trajectory for the host vehicle 12 from among the best possible trajectory candidates. The host vehicle 12 then follows this optimal trajectory throughout the current traffic situation.

[0072] The proposed sampling strategy further enables an independent longitudinal (in-direction-of-travel) operating strategy. Desired operating characteristics at a higher level are represented by target states. Nevertheless, the decision regarding the optimal trajectory is made at the trajectory planning level using a two-level candidate selection strategy. The first level represents the objective function described above, which delivers the best possible trajectories from the maneuver classes. The second level is characterized by the static and dynamic collision checks.

[0073] However, before the sampling strategy is described in detail, the mathematical principles used in trajectory planning are described below.

[0074] The control system and the control method presented in this disclosure are suitable for generating example states (or sample states) for stopping points (also support points or nodes) xl in relation to corresponding temporal states tl. In other words, at least one stopping point is assigned to each temporal state tl. Thus, at least lateral and longitudinal position values ​​of the own motor vehicle and / or the other motor vehicle are included in the determination of the stopping points. In the context of the following disclosure and the figures, an earth coordinate system, a motor vehicle coordinate system, and a curvilinear coordinate system follow the notations E, F or KThe Earth coordinate system and the vehicle coordinate system follow an orthogonal (e.g., Cartesian) framework. The corresponding letters, placed in superscript before the respective values, indicate the corresponding coordinate system.

[0075] In order to be able to interpolate between sampled states, a spline is determined which represents a movement of the own motor vehicle 12 in F< x ( t ) and F< y ( t ) describes the interpolation x ( t ) is defined as follows. x t = { s l t t l ≤ t ≤ t l + 1 0 anderenfalls .

[0076] For n - 1 spline segments and provided that t ∈ { t l , t l +1}, the spline is s defined as follows. s l t = c υ , l t − t l ν + c ν , l t − t l ν − 1 + … + c 1 , l t − t l + c 0 , l .

[0077] The properties of the spline require that it passes through the breakpoints s l ( t l ) = x l A further requirement is that the spline is continuous between the respective spline intervals, i.e. s l ( t l +1 ) = s l +1 ( t l +1 ) is sufficient.

[0078] With corresponding start and end conditions for x l = [ F< x l , F< y l ] T< and the corresponding derivatives, the spline coefficients c 0 ... v,l be determined. The l = 1 ... n Support points of each trajectory determined in this way (underlined in the following formula) are: B = x _ 1 , x _ ˙ 1 , … , x _ 1 N , t _ 1 , … , x _ η , … , x _ η N , T _ p

[0079] For the above relationship, N = (v - 1) / 2 also applies. To determine the optimal time for the transition from one interval of the spline to the next interval of the spline, time instances t l used in sampling. The continuous spline-based formulation also allows the selection of a resolution with respect to time Δ T independent of the support points l and independent of a fixed forecast scope TP . For this reason, interpolation n = TP / Δ T + 1 the trajectory points x k = [ F< xk , F< yk ] T<. These trajectory points are then summarized in the trajectory (which represents a candidate trajectory) as follows. x ego = x 1 , t 1 , x 2 , t 2 , … , x k , t k , … , x n , T P T .

[0080] Constant time intervals are t ∈ [ tk, tk + 1] for k = 1... n is used.

[0081] The two-stage approach for trajectory planning and evaluation presented here is superior to conventional approaches, among other things, because the spline-based formulation and generation of the trajectory described above is combined with curvilinear coordinate transformation. Lateral and longitudinal positions of the determined stopping points are sampled in curvilinear coordinates and later transformed into vehicle coordinates. Lane boundaries and / or lane markings of a currently occupied lane, for example, serve as references for the transformation into vehicle coordinates. The transformed stopping points are combined with the spline-based interpolation described above. This results in a set of candidate trajectories, which are then evaluated in vehicle coordinates.

[0082] In Figure 3An exemplary driving situation of the own motor vehicle 12 is shown. The own motor vehicle 12 is moving in the middle lane 30 of a three-lane road / carriageway (for example, motorway) 32. There are no obstacles such as other motor vehicles on the roadway 32. In addition, Figure 3 lateral (lateral) reference paths (reference trajectories) 28a to 28e are shown, which characterize the maneuvers described above: lane change left (reference trajectory 28a), lane keeping left (reference trajectory 28b), lane keeping center (reference trajectory 28c), lane keeping right (reference trajectory 28d) and lane change right (reference trajectory 28d).

[0083] The manner in which the lateral position of the host motor vehicle 12 is sampled is primarily based on the structure of the current environment of the host motor vehicle 12. The sampled values ​​are determined from the environmental data. The set of lateral (side) maneuver classes used is based on the higher-level planning of operating characteristics of the host motor vehicle 12 (for example, by the electronic controller 20 or any other suitable electronic controller of the host motor vehicle 12). The maneuver classes mentioned, lane change left (SWL), lane keep left (SHL), lane keep center (SHM), lane keep right (SHR), and lane change right (SWR), represent selected maneuver classes, but the present disclosure is not limited thereto. For example, the control system 10 (as well as the control method) can alternatively operate with three selected maneuver classes (e.g., lane change left, lane keep, and lane change right).

[0084] However, the larger number of maneuver classes included results in greater variability (in contrast to systems that classify maneuvers only according to SWL, SWR, and lane keeping, for example) in the maneuvers of motor vehicle 12, particularly with regard to the lane 30 it is currently occupying. Thus, even complex, compound maneuvers can be classified (for example, using SHM and SHR). One such maneuver would be the approach of the vehicle 12 to another lane (located to the left or right of the lane currently occupied) in order to close a gap there, for example. The use of curvilinear coordinates K This significantly simplifies the scanning of the lateral position, since each of the five maneuver classes is directly associated with a fixed value for K< ytIn other words, the lateral scanning of the (static) vehicle environment at each scanning time provides a value that is at least based on the determined maneuver class or assigned to this maneuver class.

[0085] Next, the scanning in the direction of travel (longitudinal scanning) is carried out with reference to Figure 4 described there. A nonlinear mapping of a linear discretization to an adaptive discretization is shown there. The adaptive discretization is used for longitudinal sampling, i.e., the determination of longitudinal positions (or position values) and / or speeds (or speed values) of the host motor vehicle 12 and / or the other motor vehicle.

[0086] Due to the limited computing resources in semi-autonomous (autonomous) vehicles, it is necessary to generate a finite yet meaningful set of possible trajectories (trajectory candidates). For this reason, in addition to the state-based sampling strategy for lateral sampling, an action-based sampling strategy is used within the present disclosure to determine longitudinal parameters such as the position K< xt and / or the speed K< vt The longitudinal acceleration (in the direction of travel of the vehicle) K< ax ( t ) of the own vehicle and / or the other vehicle is used as a control parameter. A second-order model is used to generate longitudinal samples for the position K< xt and the speed K< vt with respect to the curvilinear coordinate system K(in the following relationship again superscripted in front of the respective value) according to the following relationship. κ x ˙ t κ υ t = 0 1 0 0 κ x t κ υ t + 0 1 κ a x t .

[0087] Then a finite set of control inputs K< ax ∈ for the adaptive discretization. Assuming that the best possible control input for a currently executed planning step (planning cycle) of trajectory planning is close in time to that of the last planning step, a higher resolution is required in the range of the actual longitudinal acceleration. Nevertheless, minimum and maximum acceleration values ​​must be considered.

[0088] In the resulting adaptive (situation-adapted) discretization strategy, as in Figure 4 shown, therefore two parables f 1 and f 2 is used to obtain the linearly sampled values z(which are determined from the environmental data) online (i.e. in real time or essentially in real time) and depending on the actual acceleration value z act of the vehicle 12 and / or the other vehicle. The necessary coefficients are determined from the following conditions: f 1 z min = z min f 1 z ‾ = z akt f 1 ′ z ‾ = 0 f 2 z max = z max f 2 z ‾ = z akt f 2 ′ z ‾ = 0 , where z as z = ( zmax - z min ) / 2. The adaptive (adjusted) sampled value f ( z ) is then determined as follows. f z = { f 1 z z < z max − z ‾ f 2 z z ≥ z max − z ‾ .

[0089] The result of the adaptive discretization for current (for the respective valid sampling situation) acceleration values ​​of the own motor vehicle 12 and / or the other motor vehicle a act = -2 and a act = 0 is in Figure 4The closely spaced horizontal grid lines in the figure show the desired higher resolution in the range of the current value for the adaptive discretization with respect to a act = -2.

[0090] The advantage of adaptive discretization is evident when evaluating a merging scenario onto a motorway and / or expressway. This scenario contains varying longitudinal accelerations that are necessary to successfully handle dynamic everyday traffic. For comparison, Figure 4 a linearly discretized reference trajectory with quasi-continuous values ​​is shown, over which the adaptive discretization 4 is plotted. This approximates the best possible solution with respect to the objective function.

[0091] The relative costs ΔΦ are calculated from the difference of the open-loop results between a reference solution Φ opt and the suboptimal (next best) solution Φ of the approach with the corresponding longitudinal sampling strategy. Four variants were tested, each of which serves to compare linear and adaptive discretization with respect to the number of sampled longitudinal accelerations ζ = 5 and ζ = 10. The results are shown in Figure 5 This illustrates the advantage of adaptive sampling due to its lower relative costs compared to linear sampling.

[0092] If one compares the adaptive sampling strategies with each other (the two right parts of the Figure 5 at ζ ad = 5 and ζ ad = 10), the occurrence of a reduced number of outliers at ζ ad= 10 can be explained by the fact that the last best possible acceleration value determined in the current planning step (in the last planning step before the current one) is used for the discretization.

[0093] It should be noted that the reference solution for the cost function Φ opt also represents only an approximation of the best possible (optimal) solution, enabling relative costs less than zero. Changes in the dynamic environment are slower (i.e., they require a relatively longer time) than the computation time for a planning cycle, so the control system or control method adapts smoothly and finds the suboptimal acceleration depending on the current traffic situation.

[0094] The two-stage planning and evaluation approach as part of the control system 10 is designed for simultaneous planning of longitudinal and lateral movements, since the trajectory to be planned for the host vehicle 12 should follow any curve and simultaneously take into account the traffic situation directly in front of, to the side of, and behind the host vehicle 12. Therefore, lateral and longitudinal sampling states are transformed and combined to generate the trajectory candidates.

[0095] The transformation strategy used is described below with reference to Figure 6 In order to transform the curvilinear coordinates into vehicle coordinates, the control system 10 or its electronic control system uses a reference lane marking detected by the environment sensors (see, for example, the lane marking 32 in Figure 3) is selected. The reference lane marking 32 is defined as a cubic polynomial f ref To calculate the arc length, the finite integral is approximated and the trapezoidal rule is applied. In this way, a look-up table is created for equally spaced queries within the range of the detected lane marking F< 11 ∈ [ xs , xe ] This results in a mapping of K< x to the initial solution of the transformation F< x 0 . The initial value for the lateral part is given by F< y 0 = ( f F< x 0 ) is given. With the orthogonal slope m ⊥ = − 1 / f ref ′ x <mprescripts / > <none / > F The transformed position in x-direction and y-direction of the vehicle coordinates is: F x = x 0 <none / > <mprescripts / > <none / > F + y <mprescripts / > <none / > K ∗ 1 / 1 + m ⊥ 2 F y = F y 0 + K y ∗ m ⊥ 1 + m ⊥ 2

[0096] Figure 6 illustrates how a transformation from the curvilinear coordinate system K in vehicle coordinates FThe evaluation of the look-up table for K< x = L returns the position F< x 0 (position in the x-direction in curvilinear coordinates). The arc length is then calculated by numerical integration of the reference polynomial. This gives the position F< y 0 (position in y-direction in curvilinear coordinates). Taking into account the normal distance K< y = N , the desired coordinates in the vehicle coordinate system are obtained F from trigonometric considerations.

[0097] After the positions have been transformed into vehicle coordinates, stopping points (also called nodes) are determined by subsampling, and a set of trajectory candidates is calculated using the spline interpolation described above. Such a resulting set of trajectory candidates (Reference number 36) is in the context of an exemplary traffic scenario (also driving situation or traffic situation) in Figure 7 In each case, the own motor vehicle 12 can be seen following another motor vehicle 28 on the right lane 34 of a two-lane roadway 32'. The upper illustration of the Figure 7 shows a scenario with the other motor vehicle 28 as a (static or dynamic) obstacle for a straight road section. In the lower figure of the Figure 7 A corresponding scenario for a winding road section is described.

[0098] In other words, the Figure 7a plurality (also set) of combined candidate trajectories 36 for the own motor vehicle 12, i.e. candidate trajectories composed or determined from lateral and longitudinal samples (positions and / or speeds). The individual trajectories that make up the set of candidate trajectories 36 are each shown as grey lines and are connected by the stopping points indicated on the lines in dot form.

[0099] An important aspect of sampling-based trajectory planning is the precise and rapid determination of candidate trajectories 36. Therefore, the two-stage planning and evaluation approach mentioned above is used in this disclosure. This combines the aspects of comfortable driving for humans and precise collision avoidance. This is done against the background that the candidate trajectories 36 (and especially the optimal trajectory for the current traffic situation of the vehicle 12 ultimately determined from the candidate trajectories) should, on the one hand, reflect human driving behavior, but, on the other hand, safe and precise maneuvering must also be ensured.Therefore, the candidate trajectories 36 of each maneuver class are evaluated (checked) using a comprehensive objective function that models physically inspired safety distances for dynamic collision avoidance and also takes into account the static collision avoidance, feasibility (executability), and comfort goals for the respective current traffic situation of the vehicle 12. Thus, the optimal (best possible) trajectory is selected for each current driving situation and for each maneuver class.

[0100] For safety reasons, a follow-up test is conducted for each selected optimal trajectory with an exact geometric representation of the vehicles involved. If the follow-up test fails, a backup strategy not described in detail here can be used.

[0101] In order to determine the (global) optimal trajectory from the trajectory candidates 36, the best possible trajectories of the individual maneuver classes are compared. This comparison is performed using the described cost function Φ, which is why the optimal trajectory simultaneously represents the best maneuver class for the current driving situation of the vehicle 12. The following formula represents the comparison for determining the optimal trajectory. x ego ∗ .

[0102] Here, Φ denotes the cost of the objective function. The expressions bd and bs are defined as follows: b d = { 1 bei dynamischer Kollision 0 anderenfalls . b s = { 1 bei statischer Kollision 0 anderenfalls .

[0103] In addition, the expressions ω d and ω s represent weighting parameters for exact collision control. Figure 8 shows results of the calculation of the global optimal trajectory x ego ∗ among the trajectory candidates 36 for which already Figure 7presented road sections or driving situations. The optimal trajectory x ego ∗ with stronger and darker points compared to the other trajectories of the trajectory candidates 36.

[0104] Next, the calculation of the objective function is described, which is used to evaluate (determine) the best possible trajectory x ego ∗ for the vehicle's own motor vehicle 12 is defined as a comprehensive objective function. Therefore, individual terms of the objective function take into account the dynamic collision avoidance using a dynamic (environmental) field, the static collision avoidance using a static (environmental) field, the feasibility / executability of the best possible trajectory, and the driving comfort of the vehicle's own motor vehicle 12 when pursuing the optimal trajectory. x ego ∗ . Both in the upper picture of the Figure 8 , where the optimal trajectory x ego ∗ for a straight road section, as well as in the lower image of the Figure 8 , where the best possible trajectory x ego ∗ for a curved road section, the evaluation leads to a lane change maneuver to the left, so that the own vehicle 12, if it follows this determined optimal trajectory x ego ∗ in the further course of the current traffic situation, the other motor vehicle 28 can overtake, for example because it is traveling slower than the own motor vehicle 12 and / or has to carry out a braking maneuver that forces the own motor vehicle 12 to perform an evasive maneuver.

[0105] The cost function included in the objective function is defined as follows: Φ = e T Ω e .

[0106] In doing so, Ω a weighting matrix Ω ∈ ℝ d err × d err The expression e denotes an error vector which is determined as follows: e = o 1 , o 2 , … , o z , χ h 1 , χ h 2 , … , χ h n − 1 T

[0107] The expressionso i represent the objectives of the objective function. Inequality conditions are defined by χ( h k ) = max{[0, hk ]} is taken into account. In this way, the conditions imposed on the current driving situation of the vehicle 12 by the static and dynamic environment are taken into account. In addition, conditions relating to the vehicle dynamics of the vehicle 12, such as its maximum acceleration (which can be defined, for example, by Kamm's circle) and non-holonomic vehicle properties are recorded. The deviation from the desired target state is also formulated as the target of the objective function. This target state is defined by a higher-level planning module and is not described in detail here. In order to achieve the objectives related to driving comfort ( oi ), the lateral and longitudinal acceleration as well as the jerk (the derivative of the acceleration with respect to time) in the lateral and longitudinal direction of the own motor vehicle 12 when following the optimal trajectory x ego ∗ minimized.

[0108] Next, the principles of collision testing used here are explained with reference to the Figure 9described. In order to ensure collision avoidance for the candidate trajectories 36 for the own motor vehicle 12, an exact geometric model of the own motor vehicle 12 and the other motor vehicle 28 (i.e., in the general case, models of the potentially colliding motor vehicles) is important in the present case. Within the scope of this disclosure, bounding boxes are used to represent the shape of the own motor vehicle 12 and the other motor vehicle 28 and / or other motor vehicles involved in the current traffic situation. This is determined, for example, from Figure 9 The rectangular boxes are particularly suitable for depicting convex shapes, such as those frequently used in vehicle bodies, with high accuracy.

[0109] Within the context of dynamic collision testing, a high model accuracy also leads to greater computational effort for the control system performing the collision test (in this case, control system 10). To prevent this, for example, hierarchical pruning of possible collisions can be performed, which, however, in turn leads to a reduction in the number of exact collision tests. In contrast, multiple interference tests require the testing of each individual candidate trajectory point by point against all obstacle trajectories, i.e., trajectories that the host motor vehicle 12 determines for, for example, the other motor vehicle 28 based on its current driving behavior and the current driving situation of the host motor vehicle 12 and / or the other motor vehicle 28.This point-by-point check corresponds to the worst-case scenario in terms of the computing resources of the executing control system, but for safety reasons, its calculation must always be considered within a feasible timeframe. To adequately evaluate the dynamic environment, the collision check is performed in space and time. The separation theorem (Eidelheit's theorem) is used to represent temporal overlaps between two rectangles.

[0110] The Figure 9 clarifies the separation theorem. It can be seen that the own motor vehicle 12 does not collide with the other motor vehicle 28 (both are shown as rectangles) if a line exists onto which both rectangles (i.e. objects) can be projected and the objects do not overlap on this line. This is in the Figure 9This is the case in the example shown. Within the framework of the dynamic collision check, all trajectory candidates 36 of the own motor vehicle 12 are compared with trajectory data for the other motor vehicle 28 determined by the own motor vehicle 12 for each current time tk checked for overlaps.

[0111] The runtime of the dynamic collision check is in Figure 10 shown. In the present exemplary embodiment, an average runtime for the precise dynamic collision check using bounding boxes is assumed, which was determined for 10,000 test runs with trajectories consisting of 31 trajectory points. It is clearly evident that the average runtime depends on the number of obstacles (e.g., other motor vehicles) in the vicinity of the vehicle 12. Even for a larger number of obstacles (e.g., seven), an average runtime of approximately 100 µs results.

[0112] Figure 11 shows an example of exact dynamic collision checking in the space-time domain using the separation theorem. The own motor vehicle 12 and the other motor vehicle 28 are modeled as rectangles. Each individual rectangle 12 shown (the rectangles in the Figure 11 whose position in the x-direction changes over time) for each position of the own motor vehicle 12 in the space-time domain. The stacked rectangles 28 each represent a position of the other motor vehicle 28 in the space-time domain. The optimal trajectory x ego ∗ is compared with one or more trajectories determined for the other motor vehicle 28 (the control system 10 of the own motor vehicle 12 is set up and intended for this purpose) for each point in time tk checked for overlaps. A detected overlap indicates a possible collision.

[0113] A static collision check is also performed, for example, to detect an impending collision of the host vehicle 12 with a road marking. This is achieved by calculating the shortest distance between the bounding box of the host vehicle 12 and the corresponding lane marking.

[0114] The two-stage or two-level trajectory planning and evaluation approach or its two planning stages are in Figure 12 shown again for clarity. The left image, marked with a one in a circle, shows the first planning phase. There, the 36 trajectory candidates are determined from the objective function (see also the Figure 2 and the corresponding description). The second planning step concerns the collision check (static and dynamic) just described and is shown on the right side of the Figure 12 shown.

[0115] With reference to the Figures 13 and14 The functions of the control system 10 and the two-stage trajectory planning and evaluation approach are now illustrated using a realistic traffic scenario. Figures 13 and 14 Each consists of five images, which are to be viewed in order from top to bottom. The traffic scenario is the entry onto a motorway, where several other vehicles (all vehicles that are in the Figures 13 and 14 not referenced with the number 12), such as the other motor vehicle 28. Between each of the five images in the Figures 10 and 11 are 2 seconds.

[0116] The Figures 13 and 14The vehicle 12 shown in the specific example described is equipped with radar sensors on its front, rear, and sides, and additionally has a camera that monitors the area in front of the vehicle 12. In this way, static and dynamic environmental information is obtained from the environmental data provided to the control system 10. The reference lane for the transformation of curvilinear coordinates K into orthogonal vehicle coordinates Fis selected in the current driving situation by the control system 10 of the driver's own motor vehicle 12 based on the camera data. In the current driving situation or sampling situation, the lane marking selected as the reference lane marking is that lane marking from which the camera of the driver's own motor vehicle 12 can capture the largest area compared to other existing lane markings. By further incorporating map data, which is made available to the control system 10, for example, by a navigation system present in the driver's own motor vehicle 12 or by any other suitable means, wirelessly or wired, the accuracy and thus the quality of the results is increased.

[0117] The control system 10, for example, operates with a model predictive control (receding horizon control), since the environmental data used for trajectory planning and collision avoidance are updated cyclically.

[0118] For the example, which is not to be considered restrictive for the functionality of the control system 10, an average runtime of 40 ms is used for a single-core processor with a clock speed of 3.30 GHz and 6 MB cache for a set of 100 trajectories. For the spline-based interpolation to generate the candidate trajectories 36, an order of v = 5 is specified, and for the forecast scope TP three seconds are selected.

[0119] Trajectories for the basic maneuver lane keeping center (SM) are defined as a spline with n = 3 support points with a time in between t l =2 = 1.5 sFor the trajectories of the other maneuver classes, the number of support points is n = 2.

[0120] As part of the Figures 13 and 14 During the illustrated rear-end maneuver onto the motorway, the vehicle marked ID:7 in the figure is initially selected by the control system 10 as the other motor vehicle 28. This selection is based on the relevance of all other motor vehicles for the current driving situation of the own motor vehicle 12, which is determined by the control system 10. The own motor vehicle 12 follows the other motor vehicle 28 in the situation shown (first image from the top in Figure 13) in the same lane 34 and is currently traveling at a higher speed than the other motor vehicle 28. Lane 34 is a curved entrance lane to the motorway. The driver's own motor vehicle 12 and the other motor vehicle 28 then change from lane 34 to the right lane of the motorway.

[0121] At this point in time, the driver or the driver assistance system 20 of the own motor vehicle 12 wishes to increase the longitudinal speed of the own motor vehicle 12 to 90 km / h. An overtaking maneuver performed by the own motor vehicle 12 is not possible at the beginning of the driving scenario, even if the other motor vehicle 28 is currently traveling at a slower speed, for example, 70 km / h. Trajectories that entail a lane change to the left would most likely lead to collisions with the other motor vehicle 36 that is overtaking itself. Later on (starting with the fourth image from the top of the Figure 13) an overtaking maneuver of the own motor vehicle 12 is prevented by the other motor vehicle 38. Therefore, the control system 10 will select a trajectory for lane keeping (SHM) and the own motor vehicle 12 will be braked, for example, by the driver assistance system 20, in order to maintain a constant distance to the other motor vehicle 28 driving ahead (see Figures 2 to 5 from the top of the Figure 13 ).

[0122] In the top picture of the Figure 14 It can be seen that the control system 10 considers an overtaking maneuver for the current situation. The control system 10 therefore selects the optimal trajectory x ego ∗ a trajectory for lane keeping left (SHL, see also reference path 28b in Figure 3) from the trajectory candidates 36. The own motor vehicle 12 thus performs a driving maneuver in which it approaches the lane to the left of its current lane in order to initiate the overtaking maneuver.

[0123] In the second picture from the top the Figure 14 It can then be seen that a trajectory for center lane keeping is again selected from the trajectory candidates 36 by the control system 12. The overtaking maneuver was thus aborted (for example, by the control system 10 and / or by the driver assistance system 20) and the vehicle 12 is again essentially in the center of the lane it is currently traveling in.

[0124] The reason for this is shown in the middle picture of the Figure 14recognizable. A motor vehicle 40 traveling at a higher speed than the driver's own motor vehicle 12 is approaching the driver's own motor vehicle 12 from behind in the overtaking lane. This, in turn, poses the risk of a collision, which is why an overtaking maneuver by the driver's own motor vehicle 12 is not possible.

[0125] In the fourth picture from the top in Figure 14 It can be seen that the faster moving motor vehicle 40 has now a relatively large distance in the longitudinal direction ( E x in the Figures 13 and 14 ) to the driver's own motor vehicle 12, so that an overtaking maneuver by the driver's own motor vehicle 12 is now possible. Even in the overtaking lane, no other faster-moving motor vehicle is approaching from behind at this time.

[0126] Since the own motor vehicle 12 is to be accelerated to an exemplary speed of 90 km / h, but the other motor vehicle 28 is traveling slower, the other motor vehicle 28 no longer represents the most relevant road user for trajectory planning as soon as an overtaking maneuver is possible. The control system 10 therefore selects the other motor vehicle 40 as the reference (road user) for trajectory planning and calculates an optimal trajectory as described above. x ego ∗ from the maneuver class lane change left (see the third and fourth pictures from the top in Figure 14 ). The own motor vehicle 12 then follows the other motor vehicle 40 in the overtaking lane based on an optimal trajectory x ego ∗ from the maneuver class lane keeping center and overtakes the slower moving other motor vehicle 28.

[0127] The speeds that the own motor vehicle 12 in the individual situations (pictures) of the Figures 13 and 14 are in Figure 15 The Figures 13 and 14 The overtaking maneuver shown starts at a time t = 8 s the Figure 15 and ends at a time t = 26 s the Figure 15 As mentioned, there are gaps between the individual images in the Figures 13 and 14 two seconds each. As can be seen Figure 15 As can be seen, the vehicle can only be parked shortly before reaching t = 26 s accelerate to reach the desired 90 km / h (preset, for example, by the driver assistance system 10). This characterizes the speed in the last two images (from the top) of the Figure 14 illustrated overtaking maneuver carried out by the own motor vehicle 12.

[0128] The present disclosure relates to a control system (10) for sampling-based planning of candidate trajectories. Suitable trajectory candidates 36 are generated and evaluated in real time. Lateral sampling is based on lateral basic maneuver classes. This allows the lane-discrete structure to be utilized for semi-autonomous driving. In contrast to this lateral sampling of lateral values ​​in the state space, longitudinal samples (i.e., samples in the direction of travel) are generated from the (dynamic) action space, using the longitudinal acceleration of the host vehicle 12 as a control parameter. Adaptive discretization increases sampling efficiency.

[0129] The spline-based formulation of the trajectories and their application in curvilinear coordinates offers the advantage that the control system 10 can operate (essentially) in real time and can be used for both curved and straight road sections.

[0130] The two-stage approach for planning and evaluating candidate trajectories 36 serves to avoid collisions with the trajectories of other road users who, in a current driving situation, are in the immediate vicinity of the driver's own motor vehicle 12. These trajectories are also determined by the control system 10. The required driving comfort and the feasibility of a possible optimal trajectory are also taken into account in the evaluation.

[0131] The lateral and longitudinal driving characteristics of the host vehicle 12 are not predetermined by a higher control level, as this merely provides a target state. The presented sampling and evaluation strategy therefore inherently enables, in addition to trajectory planning for lateral maneuvers such as lane changes, the execution of other driving maneuvers such as following a leading vehicle or merging into a lane as the number of lanes decreases.

[0132] It is understood that the exemplary embodiments explained above are not exhaustive and do not limit the subject matter disclosed herein. In particular, it will be apparent to those skilled in the art that they may combine the features of the various embodiments and / or omit various features of the embodiments without deviating from the subject matter disclosed herein.

Claims

1. A control system (10) which is set up and designed for use in a subject motor vehicle (12), based on environment data obtained from at least one environment sensor (14, 16, 18) associated with the vehicle, to detect lanes, road boundaries, road markings, other motor vehicles and / or objects in a region (22, 24, 26) in front of, laterally adjacent to and / or behind the subject motor vehicle, wherein the at least one environment sensor is set up and designed to provide an electronic controller of the control system (10) with environment data reflecting the region in front of, laterally adjacent to and / or behind the subject motor vehicle, and wherein the control system is at least set up and designed to, - determine information relating to a current driving situation of the subject motor vehicle (12) and at least one other motor vehicle on the basis of the environment data provided, - determine a plurality of lateral positions based on the information relating to the current driving situation of the subject motor vehicle (12) and / or the other motor vehicle, - determine a number of longitudinal positions and / or speeds based on the information relating to the current driving situation of the subject motor vehicle (12) and / or the other motor vehicle, and - determine, from the plurality of lateral positions and the number of longitudinal positions and / or speeds, stopping points for a trajectory for the subject motor vehicle (12), which the subject motor vehicle (12) is to follow to carry out a driving maneuver, characterized in that the control system (10) is further set up and designed to - determine the trajectory for the subject motor vehicle (12) by means of a spline-based interpolation between the determined stopping points and corresponding to the lateral positions at the stopping points, and - determine the plurality of lateral positions and / or the plurality of longitudinal positions and / or speeds for the subject motor vehicle (12) and / or for the other motor vehicle in curvilinear coordinates with respect to a reference lane provided to the control system (10) of the subject motor vehicle (12) by the environment data, and to check the trajectory for the subject motor vehicle in orthogonal coordinates.

2. The control system (10) according to claim 1, which is further set up and designed to: - determine a basic driving maneuver based on the provided environment data, which the subject motor vehicle (12) is to carry out, and - determine at least a plurality of lateral positions based on the determined basic maneuver and / or to assign at least one of the lateral positions a value that denotes a basic maneuver class, wherein the basic maneuver class is contained in a set of basic maneuver classes composed at least of basic maneuvers for a lane change to the left, a lane change to the right and lane keeping.

3. The control system (10) according to any one of the preceding claims, which is further set up and designed to - determine the plurality of longitudinal positions based on a longitudinal acceleration of the subject motor vehicle (12) and / or the other motor vehicle.

4. The control system (10) according to any one of the preceding claims, which is further set up and designed to determine the plurality of longitudinal positions by means of adaptive discretization, wherein an acceleration of the subject motor vehicle (12) and / or the other motor vehicle is used as a control parameter for the adaptive discretization.

5. The control system (10) according to any one of the preceding claims, which is further set up and designed to carry out the determination of the trajectory for the subject motor vehicle (12) taking into account a static collision check, wherein the static collision check is based on the number of lateral positions of the subject motor vehicle and / or the other motor vehicle.

6. The control system (10) according to any one of the preceding claims, which is further set up and designed to determine the trajectory for the subject motor vehicle (12) taking into account a dynamic collision check, wherein the dynamic collision check is based on the plurality of lateral positions and / or on the plurality of longitudinal positions and / or speeds.

7. The control system (10) according to claim 6, which is further set up and designed to carry out the dynamic collision check taking into account the maximum acceleration of the subject motor vehicle (12) and / or the other motor vehicle.

8. The control system (10) according to any one of claims 6 to 7, which is further set up and designed to perform the static collision check and / or the dynamic collision check using the separation theorem.

9. The control system (10) according to any one of the preceding claims, which is further set up and designed to check the trajectory for the subject motor vehicle (12) by means of a target function based on a cost function.

10. The control system (10) according to claim 9, wherein a target state is assigned to the target function, and wherein deviation(s) of the target function from the target state are used as targets of the target function.

11. The control system (10) according to claim 10, wherein further targets of the target function in the form of the lateral and / or longitudinal acceleration of the subject motor vehicle (12) and / or in the form of the lateral and / or longitudinal jolt of the subject motor vehicle are used as target states of the target function.

12. A control method, which detects, in a subject motor vehicle (12) and based on environment data from at least one environment sensor (14, 16, 18) associated with the subject motor vehicle (12), lanes, road boundaries, road markings, other motor vehicles and / or objects in a region in front of, laterally adjacent to and / or behind the subject motor vehicle (12), wherein the control method is carried out in particular by means of a control system (10) according to any one of the preceding claims, and wherein the control method includes at least the following steps: - determining information relating to a current driving situation of the subject motor vehicle (12) and of at least one other motor vehicle based on the environment data provided, - determining a plurality of lateral positions based on information regarding the current driving situation of the subject motor vehicle (12) and / or of the other motor vehicle, - determining a plurality of longitudinal positions and / or speeds based on information relating to the current driving situation of the subject motor vehicle (12) and / or of the other motor vehicle, and - determining stopping points for a trajectory for the subject motor vehicle, which is to be followed by the subject motor vehicle when performing a driving maneuver, characterized by - determining the trajectory for the subject motor vehicle (12) by means of a spline-based interpolation between the determined stopping points and according to the lateral positions at the stopping points, - determining the plurality of lateral positions and / or the plurality of longitudinal positions and / or speeds for the subject motor vehicle (12) and / or for the other motor vehicle in curvilinear coordinates with respect to a reference lane provided to the control system (10) of the subject motor vehicle (12) by the environment data, and - checking the trajectory for the subject motor vehicle in orthogonal coordinates.

13. A motor vehicle comprising a control system according to any one of claims 1 to 11.