Method and device for carrying out an evasive maneuver

Abstract

Method for carrying out an evasive maneuver of a motor vehicle with the following steps: - Recording current status data (z) of an object in the vicinity of the motor vehicle with which the motor vehicle is on a collision course, - Determining at least one optimal trajectory section for the evasive maneuver of the motor vehicle, wherein the optimal trajectory section is determined by means of a non-linear program and the optimal trajectory section is determined by means of the non-linear program by determining at least one optimal control variable (x'(c1,c2,..cx)) depending on a predetermined cost functional and / or in compliance with predetermined restrictions (K&R), and - influencing a steering system (L) of the motor vehicle depending on only a first part of the determined optimal control variable (x'(c1,c2,..cx)) from the determined optimal trajectory section.

Description

[0001] The invention relates to methods and a device for carrying out an evasive maneuver of a motor vehicle.

[0002] A variety of driver assistance systems for collision avoidance through steering intervention are currently known. For example, EP 1926646 B1 discloses a method and a device for collision avoidance by determining an evasive trajectory for the evasive maneuver and influencing the steering system depending on the determined evasive trajectory. The evasive trajectory is specified by a sigmoid, the shape of which is determined by at least one speed-dependent parameter or a parameter dependent on the desired maneuver width.

[0003] The disadvantage here is that the resulting maneuver does not correspond to an intuitive evasive maneuver, as a human would drive one, and the full potential of driving physics is not exploited. Furthermore, this method is not suitable for cyclically replanning the evasive trajectory, but is limited to a one-time planning and maintaining the originally determined trajectory throughout the maneuver. This new environmental information cannot be taken into account, particularly in dynamic traffic situations (e.g., a sudden change in course of the detected hazard).

[0004] From DE10 2007 013 303 A1 a method for calculating a collision-avoiding trajectory for a vehicle is known.

[0005] The object of the invention is to provide a method and a device for carrying out an evasive maneuver that are improved compared to the prior art.

[0006] This object is achieved by a method according to claim 1 and a device according to claim 6. Advantageous further developments emerge from the dependent claims.

[0007] The basic idea of ​​the invention is to use a numerical calculation method to determine the avoidance trajectory, which also allows dynamic traffic situations to be taken into account. To this end, the method according to the invention is based on so-called "(nonlinear) model predictive control," in which input signals are cyclically specified for a given route model or dynamic model, so that the resulting avoidance trajectory behaves optimally with respect to a given criterion, at least for a given finite time horizon.

[0008] According to the invention, the method for carrying out an evasive maneuver of a motor vehicle comprises at least the following steps: - Recording current status data (at least position) of an object in the vicinity of the motor vehicle with which the motor vehicle is on a collision course, - Determining at least one optimal trajectory section for the evasive maneuver of the motor vehicle, wherein the optimal trajectory section is determined by means of a non-linear program and - Influencing a steering system of the motor vehicle depending on the determined optimal trajectory section.

[0009] Similarly, the device for carrying out an evasive maneuver of a motor vehicle - an environment detection device with which at least one object in the environment of the motor vehicle can be detected, - an evaluation device with which the current status data of the object can be determined, - a decision-making device for making a decision regarding the need for an evasive maneuver, - a calculation unit with which at least one optimal trajectory section for the evasive maneuver of the motor vehicle is determined in the event of a necessary evasive maneuver, wherein the optimal trajectory section is determined by means of a non-linear program and - a control device with which a steering actuator of a steering system of the motor vehicle can be controlled as a function of the determined optimal trajectory section.

[0010] The nonlinear program comprises a discrete-time or continuous-time dynamic vehicle model, which is used to determine a trajectory segment. Ideally, a simple vehicle model serves as the vehicle model.

[0011] Using a nonlinear optimizer (nonlinear program), the trajectory section is optimized depending on a given cost functional and, if necessary, subject to additional constraints.

[0012] The cost functional can consider various optimization criteria such as comfort and / or low steering effort (steering effort, energy expenditure) and / or low risk for the driver of the evasive vehicle or for all road users (=probabilistic consideration of the vehicle movement of other vehicles and the road surface condition). Ideally, the cost functional is composed of a weighted integral of longitudinal and lateral acceleration and / or their time derivatives (lateral and longitudinal pressure) and / or predefined approach hazard thermics.

[0013] The restrictions are strict conditions or specifications that must be strictly observed when determining the optimal trajectory section, as deviations from these prohibitions make it impossible to reach or maintain the trajectory section. In particular, these include, for example, a maximum possible steering angle and / or a maximum steering angle rate or steering torque and / or physical limits and / or driving dynamics characteristics and / or a mandatory safety distance from other road users and / or static obstacles and / or lane markings and / or final position limits that must be observed and / or the alignment of the vehicle after the evasive maneuver has been completed.

[0014] Advantageously, the non-linear program is used to determine the optimal trajectory section by determining at least one optimal manipulated variable taking into account the specified cost functional and / or the restrictions, and then the steering system is influenced depending on the determined manipulated variable.

[0015] The optimal trajectory section or the optimal control variable can be determined once at the beginning of the evasive maneuver or as a cyclical replanning process depending on the current driving situation. To continuously incorporate new framework conditions, ideally the steering system is only influenced based on an initial portion of the determined optimal trajectory section or the determined control variable, i.e., only an initial portion of the control variable or the input signal of the actuator is applied to the track. The optimization process then begins again. A time-consuming optimization process can also be performed offline, i.e., outside the vehicle, if necessary, and the "result" stored as a table in the vehicle.

[0016] In contrast to the simple determination of geometric curves, which are parameterized, for example, based on the distance traveled, this method enables the calculation of a time-parameterized maneuver trajectory or at least an optimal trajectory section. This allows for predictive responses not only to static but also to dynamic traffic situations.

[0017] The invention will be explained in more detail with reference to the following exemplary embodiment. Fig. 1 a simplified representation of a driver assistance system in a motor vehicle for carrying out an evasive maneuver, and Fig. 2 a simplified block diagram showing a non-linear program for determining an optimal control variable for an optimal trajectory section.

[0018] The Fig. Figure 1 shows a highly simplified representation of a driver assistance system in a motor vehicle for carrying out an evasive maneuver, wherein the individual control, evaluation, and calculation components can be implemented as software modules. Objects or object data o in the surroundings of the motor vehicle are continuously detected by means of an environment detection device UE, which can comprise various sensors such as ultrasound or imaging sensors. From the object data o, the current status data (absolute or position data relative to the vehicle) are determined in an evaluation unit AE, and a decision (y / n) regarding the necessity of an evasive maneuver is made in a directly downstream decision device EE from the determined status data z (and the known vehicle data f).

[0019] If the necessity of an evasive maneuver is determined (j), at least one optimal trajectory section, in particular the optimal manipulated variables x' required for the optimal trajectory section for the evasive maneuver of the motor vehicle, is determined in a downstream calculation unit BE depending on the determined relevant state data z. In order to avoid having to carry out time-consuming optimization online, the determination of an optimal trajectory or optimal manipulated variables can be carried out in advance, i.e. offline, using state data provided in advance and stored in a corresponding table. The result or the table is then stored in the vehicle so that the corresponding manipulated variables can be read from the table depending on the current situation and are thus quickly available.

[0020] The optimal manipulated variable x' consists of a predetermined number of constant individual values ​​c1, c2, ... cx, where the number is determined from a predetermined prediction horizon (time interval) and a predetermined time period, for which a constant value is assumed within the prediction horizon.

[0021] The calculation unit BE then outputs only a first part of the determined optimal manipulated variable x'(c1) for reaching the optimal trajectory section to a control unit SE, which then outputs a control signal l to a steering actuator L of a steering system of the motor vehicle.

[0022] Based on Fig. 2, the nonlinear model used to determine an optimal trajectory section is explained in detail using the example of a model predictive control. For the sake of simplicity, the model predictive control can be divided into three blocks: vehicle model F, cost functional and constraints K&R, and optimizer O. In the vehicle model F block, the vehicle model, which can be designed as a simple discrete-time, dynamic vehicle model, is stored in the form of difference number equations (DGL), with the help of which the optimal driving trajectory is calculated depending on the "finitely parameterized" input variable curve x'(c1, c2, ...cx) to be determined. In order to be able to determine an optimal trajectory section for the current situation, the current state data u (see Fig. 1) was used.

[0023] Taking into account a given cost functional and given restrictions K&R, an input variable curve x' that is optimal with regard to the given restrictions and criteria is determined in the optimizer O by a numerical method, ie a control or optimal control variable is determined that is cost-optimal with regard to the given restrictions and criteria.

Claims

[1] Method for carrying out an evasive maneuver of a motor vehicle, comprising the following steps: - Recording current status data (z) of an object in the vicinity of the motor vehicle with which the motor vehicle is on a collision course, - Determining at least one optimal trajectory section for the evasive maneuver of the motor vehicle, wherein the optimal trajectory section is determined by means of a non-linear program and the optimal trajectory section is determined by means of the non-linear program by determining at least one optimal control variable (x'(c1,c2,..cx)) depending on a predetermined cost functional and / or in compliance with predetermined restrictions (K&R), and - influencing a steering system (L) of the motor vehicle depending on only a first part of the determined optimal control variable (x'(c1,c2,..cx)) from the determined optimal trajectory section. [2] Method according to claim 1, characterized by that a trajectory section is determined by means of the non-linear program as a function of a given time-discrete or time-continuous dynamic vehicle model (F). [3] Method according to claim 1 or 2, characterized by that the cost functional is specified from suitably accumulated longitudinal and lateral acceleration and / or their time derivatives and / or specified approach hazard terms or steering movements. [4] Method according to one of the preceding claims, characterized by that the determination of the optimal trajectory section occurs once or cyclically. [5] Method according to one of the preceding claims, characterized by that the optimal trajectory section is determined from the data of a table (T) stored in the vehicle, whereby the data are previously determined externally to the vehicle by means of the non-linear program. [6] Device for carrying out an evasive maneuver of a motor vehicle, comprising: - an environment detection device (UE) with which at least one object in the environment of the motor vehicle can be detected, - an evaluation device (AE) with which current status data (z) of the object can be determined, - a decision device (EE) with which a decision is made regarding (y / n) the need for an evasive manoeuvre, - a calculation unit (BE) with which at least one optimal trajectory section for the evasive maneuver of the motor vehicle is determined in the event of a necessary evasive maneuver, wherein the optimal trajectory section is determined by means of a non-linear program and the optimal trajectory section is determined by means of the non-linear program by determining at least one optimal manipulated variable (x'(c1,c2,..cx)) depending on a predetermined cost functional and / or in compliance with predetermined restrictions (K&R), and - a control device (SE) with which a steering actuator (L) of a steering system of the motor vehicle can be controlled as a function of only a first part of the determined optimal control variable (x'(c1,c2,..cx)) from the determined optimal trajectory section.

Images