DE102024205546B3 - Method and system for compensating lateral guidance disturbances of a vehicle - Google Patents

Abstract

The invention relates to a method for compensating a disturbance variable (S) which exerts a force on a vehicle (F) running transversely to the direction of travel and thereby causes a transverse offset error (d) of the transversely controlled vehicle (F), the method comprising the following steps:
- determining a lateral deviation error (d) of the vehicle (F) with respect to a planned trajectory (T) (S10);
- estimating a disturbance (S) causing the lateral deviation error (d) of the vehicle (S11);
- Determining a compensation value (K) based on the disturbance value (S) (S12);
- applying an activation function to the compensation quantity (K), thereby obtaining a modified compensation quantity (K _mod ) (S13);
- changing a controller control variable (δ _{regier_soll} ) provided by a transverse controller (4) of the vehicle (F) based on the modified compensation variable (K _mod ), thereby forming a control variable (δ _soll ) (S14); and
- Supplying the manipulated variable (δ _soll ) to the steering control device (LR) of the electric steering (3) of the vehicle (F) for lateral guidance of the vehicle (F) (S15).

Description

The invention relates to a method and a system for compensating disturbances that cause a lateral deviation of the vehicle from a planned trajectory, hereinafter also referred to as lateral deviation errors.

Lane-guided driver assistance systems use recorded environmental and road data to control vehicles laterally and longitudinally along a selected path (also called a trajectory). In most cases, an electric power steering (EPS) system is used as the actuator for lateral guidance, which receives appropriate control signals from a control unit of the advanced driver assistance system (ADAS).

A key quality characteristic of lateral control is the vehicle's lateral deviation from the planned trajectory. Ideally, the vehicle travels exactly along the planned trajectory without any lateral deviation error. In reality, however, the system is subject to a multitude of disturbances that must be compensated for by the lateral control system. Since this can only be achieved with finite dynamics and accuracy, the lateral control error is not zero, but rather has a transient or permanent non-zero value, depending on the characteristics of the disturbance and the lateral control system's design.

A typical disturbance affecting a lateral-controlled vehicle is the transverse inclination of the road surface. This causes a force oriented perpendicular to the direction of travel, acting in the direction of an increase in the lateral deviation error along the lateral gradient of the road surface.

Lateral controllers can be classified, for example, according to their steady-state control accuracy. The simplest controller without steady-state accuracy, suitable for controlling a vehicle along a trajectory, is a PD controller, which reads the lateral offset error as a controlled variable, processes it accordingly, and outputs a steering angle as a setpoint to the underlying steering angle control system.

The simplest lateral controller with the potential for steady-state accuracy is a PID controller. Although the integral component generates a zero lateral deviation error in the ideal steady-state condition, a transient lateral deviation error still occurs, which depends on the temporal progression of the road's bank angle. In practice, especially at high driving speeds, the dynamic change in the bank angle is so high that the transient increase in the lateral deviation error cannot be neglected. The transient control deviation when using a PID controller can be up to 0.5 m, depending on the controller design. Depending on the parameterization, it can typically take 2 to 10 seconds for the control deviation to be compensated.

With both the PD controller and the PID controller as a lateral controller, there is a risk that the vehicle will cross the lane markings and thus the risk that the vehicle will collide with oncoming vehicles.

To avoid this, measures have already been taken to compensate for disturbances resulting from the transverse inclination of the roadway.

The problem with existing systems is that they tend to overcompensate, undercompensate, or even miscompensate for the lateral tilt error at times. This can result in the vehicle's lateral control behavior being even worse than without the use of lateral tilt compensation.

The DE 10 2018 104 473 A1 describes an automatic driving system with an electric power steering device and a control device that performs automatic steering control to control the electric power steering device to cause the vehicle to travel along a target travel route. The control device is configured to determine whether the automatic driving system is malfunctioning based on a tracking state of the target travel route by the vehicle. The control device is further configured to increase, in the event that it is determined that the automatic driving system is malfunctioning, a degree to which a driver's steering intervention is reflected in the control amount of the electric power steering device. The determination of whether the automatic driving system is malfunctioning is made based on a magnitude of a lateral deviation of the vehicle with respect to the target travel route, a magnitude of a steady-state deviation extracted from the lateral deviation, and/or a magnitude of a transient deviation extracted from the lateral deviation. In the event that it is determined that the automatic driving system is not operating well and, in addition, that a magnitude of a steady-state deviation, which is included in a lateral deviation of the vehicle with respect to the target travel route, is larger than a predetermined threshold, an assistance force to assist a driver's steering operation is increased. In the event that it is determined that the automatic driving system is not operating well and, in addition, that a magnitude of a temporary deviation, which is included in a lateral deviation of the vehicle with respect to the destination route is greater than a specified threshold, the control gain of the automatic steering control is reduced.

DE 10 2010 029 245 A1 describes a method for crosswind compensation in vehicles, in which a crosswind currently acting on the vehicle is calculated from sensor-determined vehicle state variables and, as a compensation measure, an intervention in an active steering system of the vehicle is carried out to influence the driving dynamics state, wherein a lateral acceleration is determined, and wherein the crosswind compensating measure is only carried out if the signs of a crosswind force and the lateral acceleration match.

The DE 10 2005 049 071 A1 describes a device for keeping a vehicle in its lane under different driving conditions, such as crosswinds or inclined road surfaces, comprising a reference model which receives geometric data relating to the position of the vehicle in the lane and data relating to the lane course from a lane recognition system and calculates therefrom a target value for a control, wherein an I controller controls a transverse deviation of the vehicle.

It is an object of the invention to provide a method for compensating disturbances acting transversely to the direction of travel which cause transverse offset errors of the vehicle, which method offers a high quality and accuracy in compensating disturbances acting on the vehicle's transverse guidance.

This object is achieved by a method having the features of independent patent claim 1. Preferred embodiments are the subject of the subclaims. A system for lateral guidance of a vehicle is the subject of independent patent claim 15.

• Zunächst wird ein Querablagefehler ermittelt. Der Querablagefehler gibt die laterale Abweichung des Fahrzeugs von einer Trajektorie an, die durch das Fahrerassistenzsystem bereitgestellt wird.

According to a first aspect, a method for compensating for a disturbance that exerts a force transverse to the direction of travel on a vehicle and thereby causes a transverse offset error of the vehicle is disclosed. The method comprises the following steps:

• First, a lateral deviation error is determined. The lateral deviation error indicates the lateral deviation of the vehicle from a trajectory provided by the driver assistance system.

Subsequently, a disturbance causing the vehicle's lateral deviation is estimated. This disturbance could, for example, be a lateral force caused by a road bank and/or crosswind.

Based on the estimated disturbance, a compensation variable is determined. The compensation variable can have a dimension identical to the dimension of the manipulated variable supplied to a steering control system to compensate for the lateral deviation error. For example, the compensation variable can have the dimension of an angle.

An activation function is applied to the compensation variable, resulting in a modified compensation variable. The activation function is designed to change the amount of the compensation variable depending on the magnitude of the cross-deviation error.

Based on the modified compensation variable, a controller manipulated variable provided by a transverse controller of the vehicle is modified, thereby forming a manipulated variable. The manipulated variable is preferably obtained by adding the modified compensation variable and the controller manipulated variable.

The control variable is fed to the steering control device of the vehicle's electric steering system in order to effect lateral guidance of the vehicle.

The technical advantage of the method is that the disturbance variables can be estimated using lower-quality sensors without significantly degrading the lateral control performance. Furthermore, the method eliminates the need for complex processes for detecting and compensating sensor errors. This allows for more reliable lateral control of a vehicle supported by a driver assistance system.

According to one exemplary embodiment, the activation function has an activation factor dependent on the determined lateral deviation error. Preferably, the activation factor is also dependent on the sign of the compensation variable, for example, such that the activation factor is greater than zero, and thus the compensation is effective, only if the signs of the compensation variable and the lateral deviation error are the same. This allows the effect of the disturbance compensation to be adjusted depending on the lateral deviation error, in particular such that the activation factor is smaller for lateral deviation errors of small magnitude than for lateral deviation errors of larger magnitude. This can prevent undesired oscillation of the vehicle's lateral control.

Preferably, the activation function for absolute values of the cross-departure error Above a cross-rotation error threshold, the activation factor has an activation factor of 1, i.e., the compensation variable is activated without reduction in this value range of the cross-rotation error. For absolute values of the cross-rotation error below the cross-rotation error threshold, the compensation variable is attenuated. Preferably, the activation function has a continuously decreasing function curve in this range, such that the activation factor provided by the activation function drops from the value 1 at the cross-rotation error threshold to a minimum value, preferably to zero or essentially zero. This achieves an increasing attenuation of the activation factor from the cross-rotation error threshold to a cross-rotation error of zero.

According to one embodiment, the activation function has a first functional range associated with a first value range of the determined cross-rotation error. The first value range of the determined cross-rotation error is preferably a range in which the magnitude of the cross-rotation error is greater than the magnitude of the cross-rotation error threshold. Provided that the compensation variable is suitable for reducing the cross-rotation error in terms of its sign, i.e., its physical direction of action, the first functional range is configured to leave the compensation variable unchanged, so that the modified compensation variable is equal to or substantially equal to the compensation variable. This allows for direct application of the compensation variable in the case of high cross-rotation errors, so that, with precise determination of the disturbance variable, complete compensation of the disturbance causing the cross-rotation error is achieved.

According to one embodiment, the activation function has a third functional range assigned to a third value range of the determined cross-slip error, wherein the third functional range is designed to set the compensation variable to zero, so that if a determined cross-slip error in the third value range is equal to or substantially equal to zero, a modified compensation variable is output. The third functional range of the activation function is particularly useful when the cross-slip error and the disturbance variable or the compensation variable contradict each other, i.e., the compensation variable does not act to reduce the cross-slip error, but rather to amplify it. Thus, incorrect compensation, which would lead to an amplification of the cross-slip error, is avoided by means of the third functional range of the activation function.

According to one embodiment, the activation function has a second functional range located between the first and third functional ranges. The second functional range is designed to form a smooth transition between the values of the activation function in the first and third functional ranges. Thus, the second functional range provides an activation factor that mitigates the disturbance compensation, so that the disturbance compensation is reduced in the case of small lateral offset errors. This can prevent oscillation of the lateral control due to an abrupt compensation application.

According to one embodiment, the first functional range of the activation function is assigned to magnitude-based cross-offset errors greater than a first cross-offset error threshold. Thus, for magnitude-based high cross-offset errors, the compensation variable is applied to a maximum by means of this first functional range.

According to one embodiment, the second functional range of the activation function is assigned to cross-offset errors between a first and a second cross-offset error threshold. Thus, for cross-offset errors that are smaller in magnitude than the first cross-offset error threshold, the second functional range is applied, in which a damped application of the compensation variable occurs.

According to one embodiment, the first lateral offset error threshold is adjusted depending on the driving situation, the driving speed, and/or the width of the roadway on which the vehicle is traveling. This allows the disturbance compensation and thus the lateral control of the vehicle to be modified depending on the current situation and conditions in order to achieve the best possible lateral control of the vehicle for the respective situation.

According to one embodiment, the second lateral deviation error threshold is 0 or, depending on the sign of the curve curvature of a curve along which the vehicle is moving, is not equal to 0 but smaller in magnitude than the first lateral deviation error threshold. By selecting the second lateral deviation error threshold, a desired lateral target deviation from the planned trajectory can be achieved, for example, such that the vehicle is not kept exactly in the center of the lane in a curve, but is offset toward the inside of the curve. This is perceived by the human occupant as a more natural driving behavior.

According to one embodiment, the magnitude of the estimated disturbance or the compensation variable is limited. This allows very high magnitude disturbance estimates to be excluded from the disturbance compensation.

According to one embodiment, the limitation is based on the statistical distribution of the expected values of the disturbance variable or the statistical distribution of the expected values of the compensation variable. For example, statistically very rare values of the disturbance variable or the compensation variable can be excluded. Due to the rare occurrence of these values, this only affects track guidance in very rare cases, but prevents the undesired application of high compensation variables that may have arisen due to an incorrect disturbance estimate.

According to one embodiment, the modified compensation variable is low-pass filtered. The controller manipulated variable is changed based on the low-pass filtered modified compensation variable. This allows for a frequency-dependent application of the compensation variable, which offers advantages for the stability of the cross-control behavior at low values of the first cross-shift error threshold.

According to one embodiment, the estimated disturbance variable is a disturbance variable dependent on the road's cross slope or a crosswind. An estimation device is provided that, based on measured values supplied by the vehicle, provides output information regarding the magnitude of the road's cross slope or the crosswind.

According to one embodiment, the controller manipulated variable specifies a target steering angle, a target torque, or a tie rod force, which is fed to a steering control system of the vehicle as an input variable. This allows the compensation method to be applied to different interface types of the electric steering system.

• - Ermitteln eines Querablagefehlers des Fahrzeugs in Bezug auf eine geplante Trajektorie;
• - Schätzen einer Störgröße, die den Querablagefehler des Fahrzeugs verursacht;
• - Ermitteln einer Kompensationsgröße basierend auf der Störgröße;
• - Anwenden einer Aktivierungsfunktion auf die Kompensationsgröße, wodurch eine modifizierte Kompensationsgröße erhalten wird;
• - Verändern der Reglerstellgröße basierend auf der modifizierten Kompensationsgröße, wodurch eine Stellgröße gebildet wird; und
• - Zuführen der Stellgröße zur Lenkregeleinrichtung der elektrischen Lenkung des Fahrzeugs zur Querführung des Fahrzeugs.

According to a further aspect, a system for lateral guidance of a vehicle along a planned trajectory is disclosed. The system comprises a control unit with a lateral controller, which provides a controller control variable. Furthermore, the system comprises an interface by means of which a control variable is supplied to an electric steering system. The control unit is configured to perform the following steps:

• - Determining a lateral deviation error of the vehicle with respect to a planned trajectory;
• - Estimating a disturbance that causes the vehicle's lateral deviation error;
• - Determining a compensation value based on the disturbance;
• - applying an activation function to the compensation variable, thereby obtaining a modified compensation variable;
• - Changing the controller manipulated variable based on the modified compensation variable, thereby forming a manipulated variable; and
• - Supplying the control variable to the steering control device of the vehicle's electric steering system for lateral guidance of the vehicle.

The terms “approximately”, “essentially” or “about” mean, in the sense of the invention, deviations from the exact value by +/- 10%, preferably by +/- 5% and/or deviations in the form of changes that are insignificant for the function.

Further developments, advantages, and possible applications of the invention will become apparent from the following description of exemplary embodiments and from the figures. All described and/or illustrated features, individually or in any combination, are fundamentally part of the invention, regardless of their summary in the claims or their reference back to them. The content of the claims is also incorporated into the description.

• 1 beispielhaft eine schematische Darstellung eines Fahrzeugs mit einer elektrischen Lenkung und einem Fahrassistenzsystem zur lateralen Spurführung;
• 2 beispielhaft eine schematische Darstellung eines in Bezug auf eine geplante Trajektorie bestehenden Querablagefehlers, der sich aufgrund einer Fahrbahnquerneigung einstellt;
• 3 beispielhaft ein schematisches Funktionsdiagramm eines Fahrzeugs, das eine Steuereinheit zur Querregelung und eine elektrische Lenkung mit einem Lenkwinkelregler aufweist;
• 4 beispielhaft eine für positive Querablagefehler wirkende Aktivierungsfunktion, die abhängig von der Größe des Querablagefehlers d einen Aktivierungsfaktor a bereitstellt;
• 5 beispielhaft eine für negative Querablagefehler wirkende Aktivierungsfunktion, die abhängig von der Größe des Querablagefehlers d einen Aktivierungsfaktor a bereitstellt; und
• 6 beispielhaft ein Blockdiagramm, das die Schritte eines Verfahrens zur Kompensation einer Störgröße, die einen Querablagefehler hervorruft, veranschaulicht.

The invention is explained in more detail below with reference to exemplary embodiments and the figures. They show:

• 1 an example of a schematic representation of a vehicle with electric steering and a driver assistance system for lateral lane guidance;
• 2 by way of example, a schematic representation of a lateral deviation error with respect to a planned trajectory, which occurs due to a transverse inclination of the road surface;
• 3 by way of example, a schematic functional diagram of a vehicle having a control unit for lateral control and an electric steering system with a steering angle controller;
• 4 by way of example, an activation function acting for positive cross-deviation errors, which provides an activation factor a depending on the size of the cross-deviation error d;
• 5 by way of example, an activation function acting for negative cross-deviation errors, which provides an activation factor a depending on the size of the cross-deviation error d; and
• 6 by way of example, a block diagram illustrating the steps of a method for compensating for a disturbance that causes a cross-deviation error.

1 schematically shows a vehicle F having a system 1 for lateral guidance of the vehicle along a planned trajectory. The vehicle F has, in particular, an electric steering system 3 with a steering control device LR. The steering control device LR is coupled to a control unit 2, which provides it with a manipulated variable δ _soll . The manipulated variable δ _soll can be, for example, the desired steering angle, the desired steering torque, or a desired tie rod force. The control unit 2 has means for compensating for a disturbance variable S that influences the lateral guidance of the vehicle F on the trajectory T.

2 shows, by way of example, a situation in which the vehicle F is guided along a planned trajectory T, but has a lateral deviation error d compared to this trajectory T due to a disturbance variable S. In the exemplary embodiment shown, the disturbance variable S is a roadway cross slope with the cross slope angle α, which results from a different height h of the roadway along the lane width b. Due to the roadway cross slope to the right, a force acting transversely to the direction of travel and directed to the right is exerted on the vehicle F, which leads to a lateral deviation of the vehicle F to the right from the trajectory T.

3 shows the system 1 for lateral guidance of the vehicle F and the means for compensating the lateral deviation error d in more detail.

System 1 comprises a control unit 2 that creates a target lane definition or receives such a target lane definition from a higher-level control unit. The target lane definition defines the trajectory T along which vehicle F is to be moved.

Subsequently, the lateral deviation error d of the vehicle F is calculated based on a comparison of the actual position of the vehicle F, i.e. the actually driven trajectory and the target trajectory T.

This cross-shift error d is fed to a cross-shift controller 4. The cross-shift controller 4 is configured to provide a controller manipulated variable δ _{regier_soll} based on the cross-shift error d. The controller manipulated variable δ _{regler_soll} is a manipulated variable output by the cross-shift controller 4, which is fed to a summation point.

In addition, the control unit 2 has a compensation branch 10 by means of which the disturbance variable is compensated.

The control unit 2 has an estimation unit 5 for estimating the disturbance variable S or an interface via which the disturbance variable S is received from an external estimation unit. The disturbance variable S is, for example, a disturbance variable caused by the transverse slope of the roadway on which the vehicle F is moving. In an alternative embodiment, the disturbance variable S is caused by the crosswind acting on the vehicle.

The disturbance variable S is then preferably fed to a transformation device 6. The transformation device 6 is configured to transform the disturbance variable S into a compensation variable K, wherein the compensation variable K _has the unit of the manipulated variable δ setpoint that the steering control device expects as an input variable (i.e., for example, a steering angle, torque, or tie rod force). After creating an identical unit of the controller manipulated variable δ _setpoint of the transverse controller 4 and the output variable provided by the compensation branch 10, this can be directly applied to the controller manipulated variable δ setpoint _controller to compensate for the disturbance variable S and thus compensate for the disturbance.

The transformation device 6 is preferably designed to take into account the purely static or the static and dynamic transfer behavior between the manipulated variable and the point of application of the disturbance.

The compensation variable K provided by the transformation device 6 is preferably fed to a limiting device 7. The limiting device 7 limits the amount of the compensation variable K, resulting in a limited compensation variable K _lim .

In the event that the disturbance variable S relates to the transverse slope of the roadway, i.e. the disturbance variable S specifies a transverse slope angle α of the roadway, the compensation variable K can, for example, be limited in terms of its magnitude in such a way that its values correspond to a maximum transverse slope angle α _max , i.e. the limited compensation variable K _lim assumes values that correlate with the value range [-α _max ; α _max ] of the transverse slope angle α. The limitation can also be based on the statistical distribution of the expected magnitude of the disturbance variable S, for example in such a way that the outermost ends of the statistical distribution of the disturbance variable S are disregarded. An example of this would be the limitation in such a way that the limited compensation variable K _lim corresponds to 95% of the sta covers all statistically occurring cases of transverse inclination angles α of the roadway.

The limited compensation value K _lim , or in the case of no limitation, the compensation value K itself, is fed to a modifier 8. This modifier 8 also receives the cross-deviation error d.

The modifier 8 is designed to apply an activation function A to the limited compensation variable K _lim or the compensation variable K. The activation function A is used to check the plausibility of the limited compensation variable K _lim or the compensation variable K with respect to the determined cross-slip error d. As part of the plausibility check, it is checked in particular whether the estimated disturbance variable S, based on which the limited compensation variable K _lim or the compensation variable K is calculated, can be the cause of the determined cross-slip error d according to its sign. 2 shown road cross slope to the right (road is lower at the right edge of the road than at the left) leads to a force acting to the right, which causes a lateral deviation error d of the vehicle F to the right. In the opposite case, a road cross slope to the left (road is lower at the left edge of the road than at the right) leads to a force acting to the left, which causes a lateral deviation error d of the vehicle F to the left. However, if the disturbance variable S and the lateral deviation error d are directed differently, i.e. the disturbance variable S indicates a road cross slope to the left, but there is a lateral deviation error d to the right (or vice versa), this signals an incorrect estimation or overcompensation. In this case, the activation factor a of the activation function A is set to zero, so that the compensation branch 10 of the control unit 2 is deactivated and thus no longer any compensation of the controller manipulated variable δ _{controller_setpoint} of the lateral controller 4.

On the other hand, the activation function A implements a modification of the compensation variable K that depends on the magnitude of the cross-slip error d or - if the compensation variable K is limited - a modification of the limited compensation variable K _lim . In particular, the activation function A provides an activation factor a that depends on the magnitude of the cross-slip error d and is preferably multiplied by the compensation variable K or the limited compensation variable K _lim . The activation factor a has a maximum value at a high magnitude of the cross-slip error d and preferably decreases continuously to zero from a first cross-slip error threshold value d ₀ . In this way, overcompensation or incorrect compensation by the compensation branch 10 can be prevented by means of the activation function A.

The manipulated variable δ _soll is formed by summing the controller manipulated variable δ _{regler_soll} of the transverse controller 4 and the modified compensation variable K _mod of the compensation branch 10. This manipulated variable δ _soll is fed to the electric steering system 3, in particular to the steering control device LR of the electric steering system 3. Depending on the design of the electric steering system 3, the manipulated variable δ _soll can be, for example, the desired steering angle, the desired steering torque, or a desired tie rod force. Thus, the system 1 can advantageously implement the lateral guidance of the vehicle F.

The compensation procedure is described in greater detail below, based on the example of a disturbance caused by a transverse inclination of the road and a compensation that counteracts the transverse inclination of the road. A transverse inclination of the road causes a force oriented perpendicular to the vehicle's direction of travel F, which acts in the direction of an increase in the lateral deviation error along the lateral gradient of the road. To prevent the increase in the lateral deviation error and, ideally, to completely compensate for it, a compensation variable K _mod in the form of a compensation steering angle δ _comp is required to be applied to the controller control variable δ _{regier_soll} .

To determine this compensation steering angle δ _comp as the compensation variable K _mod , one can take advantage of the fact that, from the perspective of the lateral controller 4, it is irrelevant whether a lateral acceleration results from a road curvature or a road bank. The steady-state relationship between lateral acceleration and the resulting steering angle can therefore be derived from the equation of the steady-state single-track model: $${\delta }_{komp}=a_q\cdot \left(\frac{l+EG\cdot v^2}{v^2}\right)$$

aq

Querbeschleunigung aufgrund der Fahrbahnquerneigung;

I

Radstand des Fahrzeugs;

EG

Eigenlenkgradient des Fahrzeugs;

v

Geschwindigkeit des Fahrzeugs;

This includes:

aq

Lateral acceleration due to the road slope;

I

Wheelbase of the vehicle;

EC

Self-steering gradient of the vehicle;

v

speed of the vehicle;

The lateral acceleration a caused by the road surface slope with the slope angle α is given by: $$a_q=g\cdot \text{sin }\alpha$$

g

Gravitationskonstante;

α

Querneigungswinkel;

This includes:

G

gravitational constant;

α

bank angle;

To compensate for the transverse inclination of the road, as previously described in connection with 2 As described above, a compensation steering angle δ _comp is determined by the compensation branch 10 of the control unit 2 and added to the output variable δ _{regier_soll} of the transverse controller 4. This results in a reference variable δ _soll related to the steering angle according to the following formula: $${\delta }_{soll}={\delta }_{regler\_soll}+{\delta }_{komp}$$

By applying the compensating steering angle δ _comp, the task of compensating for the road's cross slope is relieved of the lateral controller 4, provided the influence of the cross slope is ideally compensated statically and dynamically. According to Equation (2), this requires that the information on the lateral acceleration a _q due to the road's cross slope and, furthermore, the parameters of the single-track model according to Equation (1) are available with high accuracy.

v

Fahrgeschwindigkeit;

r

Fahrzeug-Gierrate;

ay

Querbeschleunigung des Fahrzeugs;

However, this is often not the case, since the self-steering gradient EG in particular cannot be assumed to be constant over the vehicle's lifetime. For example, a change in load or the fitting of different tires leads to a change in the self-steering behavior. Furthermore, the lateral acceleration a _q resulting from the road's cross inclination is not directly available as a measured value, but must be estimated using existing vehicle inertial sensors. A common procedure essentially consists of calculating the lateral acceleration a _q resulting from the road's cross inclination using the relationship $$a_q=v\cdot r-a_y$$ to be calculated. Where:

v

driving speed;

r

vehicle yaw rate;

ay

Lateral acceleration of the vehicle;

This relationship is used in the literature, for example, in conjunction with (extended) Kalman filters to ensure the best possible estimation even in the presence of signal interference and noise components from the sensors used. However, this requires that the noise and signal interference can be assumed to be temporally uncorrelated and normally distributed. However, this is not the case for a variety of sensor errors. These include, for example, offset and linearity errors, which are also subject to temperature and long-term drift. While functions for linearizing and offset compensating the sensor values are used during sensor signal processing, especially for errors with high drift behavior, such as a temperature-dependent sensor offset, the compensation can only correct these with a time delay. This leaves uncompensated error components in the yaw rate and lateral acceleration signals, which, due to error propagation, result in an inaccurate estimate of the lateral pitch acceleration a _q .

Another factor that influences the estimation of the road's cross slope is a possible roll angle of the vehicle F due to asymmetrical loading or centrifugal force when cornering, especially with a high center of gravity. Here, too, there is a risk of over- or undercompensation of the actual road's cross slope.

Furthermore, the estimation of the transverse inclination is always only possible with a phase shift compared to the actually existing transverse inclination due to the necessary signal filtering of the vehicle dynamic measured variables used and the signal latency times during the synthesis and transmission of the data on the communication bus of the vehicle F. The dynamic delay of the estimation thus also leads to incorrect compensation in the event of rapid changes in the physical road transient cross inclination.

A rapid change in the road's cross slope also occurs, for example, when vehicle F travels at high speed at an acute angle along road ruts. Each time the vehicle crosses one of the shoulders of the ruts, the cross slope acceleration changes immediately, which then only affects the compensation signal after a time delay. The result is lateral oscillation of the vehicle, since the required compensation is almost never activated at exactly the right time and, for example, is still present after the shoulder of the rut has already been overcome. The result is further driving over the rut until the compensation signal has dissipated again.

If the compensation steering is too high transiently or permanently due to sensor or model errors or the roll influence or the delayed provision of the estimated lateral inclination If the angle δ _{comp_setpoint} is determined, overcompensation occurs when the vehicle is activated. This causes the vehicle to cross the center of the road transiently (e.g., with a lateral PID controller) or permanently (e.g., with a lateral PD controller) in the direction of the rising road gradient. Undercompensation occurs when the vehicle is only partially compensated for the transverse gradient, resulting in a lateral deviation in the direction of the falling transverse gradient.

In addition to the aforementioned over- and undercompensation, the case of miscompensation must also be considered. This is characterized by the sign of the compensating steering angle being permanently or transiently opposite to the desired physical effect. This further increases the lateral offset of the vehicle relative to the trajectory, reducing the accuracy and quality of the tracking despite the active compensation function.

In all three cases of faulty compensation, it is then necessary for the cross controller 4 to correct the undercompensated, overcompensated or incorrectly compensated component, resulting in a transient deviation (when using cross controllers with integration of the lateral control deviation such as PID controllers) or a permanent cross offset error (in the case of controllers without integration of the lateral deviation such as PD controllers) related to the planned trajectory.

In summary, especially in cases of overcompensation or miscompensation, the lateral control behavior of the vehicle may even be worse than without the use of lateral tilt compensation.

In the present disclosure, it is proposed to calculate a compensation steering angle δ _comp under the assumption of ideal sensor data. For example, the equations Eq. 1 and Eq. 2 can be used as a basic calculation rule. As a plausibility check, the limiting device 7 limits the compensation variable K to an interval that corresponds to a defined value range of the bank angle α of [- α ₀ , + α ₀ ]. This results in a limitation of the compensation steering angle δ _comp . Using the equations Eq. 1 and Eq. 2, one obtains $${\delta }_{komp}=min\left(\left(g\cdot \text{sin}\left({\alpha }_0\right)\right),\text{abs}\left(a_q\right)\right)\cdot sgn\left(a_q\right)\cdot \left(\frac{l+EG\cdot v^2}{v^2}\right)$$

aq

geschätzte Querbeschleunigung aufgrund der Fahrbahnquerneigung;

g

Gravitationskonstante;

α0

limitierter Querneigungswinkel der Fahrbahn;

I

Radstand des Fahrzeugs;

EG

Eigenlenkgradient des Fahrzeugs;

v

Geschwindigkeit des Fahrzeugs;

This includes:

aq

estimated lateral acceleration due to the road bank;

G

gravitational constant;

α0

limited cross slope angle of the roadway;

I

Wheelbase of the vehicle;

EC

Self-steering gradient of the vehicle;

v

speed of the vehicle;

The maximum possible amount of the compensation steering angle δ _comp is defined by $${\delta }_{komp,max}=g\cdot \left|\text{sin}\left({\alpha }_0\right)\right|\cdot \left(\frac{l+EG\cdot v^2}{v^2}\right)$$

The angle α ₀ in Eq. 5 and Eq. 6 can correspond to the typical maximum cross slope angle of the roadways in the respective country in which the vehicle is operated. This limitation serves to prevent incorrectly determined compensating steering angles δ _comp from only being activated up to a maximum expected height. When selecting the angle α ₀ , the statistical distribution of the cross slope angles α can also be used as a basis, as already described. The value of the angle α ₀ can be selected such that only a subset of all cross slopes is covered, for example only 95% of the statistically occurring cross slope angles α. By omitting, for example, 5% of the statistically occurring cross slope angles α, the range of the compensating steering angle δ _comp can be specifically restricted, with an acceptable loss of lateral control quality related to the distance covered over the vehicle's lifetime.

Subsequently, the limited compensation variable K _lim , in particular the limited compensation steering angle, is subjected to a classification in such a way that the signs of the limited compensation variable K _lim and the current lateral deviation error d are related to each other.

The counting arrow of the lateral deviation error d is defined such that a vehicle F located to the right of the target lane, as seen in the direction of travel, causes a positive lateral deviation error d. A positive control variable δ _soll, for example, causes the vehicle F to yaw counterclockwise, i.e., in the mathematically positive direction of rotation.

If the signs of the limited compensation variable K _lim and the offset error d differ, this is classified as incorrect compensation. As a result, the limited compensation variable K _lim is not taken into account when calculating the manipulated variable δ _soll, or the comp compensation steering angle δ _comp provided by the compensation branch is 0: $${\delta }_{komp}=0f\ddot{\text{u}}rsgn\left(K_{lim}\right)\ne sgn\left(d\right)$$

When classified as faulty compensation, the control quality of the cross-control is only as good as if there was no disturbance compensation, but the cross-control is not impaired by faulty compensation.

If, however, the signs of K _lim and d are identical, a disturbance feedforward is applied, ie the controller manipulated variable δ _{regler_soll} is subjected to the compensation steering angle δ _komp according to Eq. 3.

In this case, the cases of undercompensation, exact compensation, or overcompensation can occur. Which of these cases is present is determined by the reaction to the activation of the compensation steering angle. To account for these different cases, the compensation steering angle is activated depending on the resulting lateral deviation error d, for example, according to the regulation $$K_{mod}=K_{lim}\cdot min\left(\mathrm{1,}\left(\frac{d}{d_0}\cdot sgn\left(d\right)\right)\right)$$ if: sgn(K _lim ) = sgn(d);

Kmod

modifizierte Kompensationsgröße;

Klim

limitierte Kompensationsgröße;

d

Querablagefehler;

d0

erster Querablagefehlerschwellwert.

This includes:

Kmod

modified compensation quantity;

Climate

limited compensation amount;

d

Cross-laying error;

d0

first cross-departure error threshold.

The sgn function, also called the sign function or signum function, produces a value of -1 for a negative argument and + 1 for a positive argument.

The functional part $min\left(\mathrm{1,}\left(\frac{d}{d_0}\cdot sgn\left(d\right)\right)\right)$ can also be referred to as an activation function A, which, depending on the cross-slip error d and a first cross-slip error threshold d ₀ , which forms an activation threshold, outputs an activation factor a of 1 (if the amount of the cross-slip error d is greater than the amount of the first cross-slip error threshold d ₀ ), or an activation factor a with the value $\frac{d}{d_0}$ (if the magnitude of the cross-slip error d is smaller than the magnitude of the first cross-slip error threshold d ₀ ).

4 shows the activation function A for positive values of the offset error d and the compensation quantity and 5 the activation function A for negative values of the offset error d and the compensation quantity.

As can be seen from the 4 and 5 As can be seen, the activation function A is selected such that if the amount of the lateral deviation error d is greater than the first lateral deviation error threshold value d ₀ , the modified compensation variable K _mod or the compensation steering angle δ _komp is initially applied unchanged and thus acts directly on the controlled system, provided that the signs of the compensation variable and the lateral deviation error are the same. In other words, the modified compensation variable K _mod is equal to the limited compensation variable K _lim . As a result, the disturbance variable S is compensated by the compensation branch 10, and the unmodified application of the compensation steering angle δ _komp initially reduces the amount of the lateral deviation error.

If the cross-shift error d is reduced only to a residual value greater than d ₀ , this is considered undercompensation. The control quality of the cross-shift control is thus already better than without the application of disturbance compensation, since the cross-shift controller 4 is at least partially supported. In the case of a cross-shift controller 4 with an integral component, the remaining control error is reduced to 0. The compensation branch 10 acts as a feedforward control, which reduces the control error more quickly than without the application of disturbance compensation.

If the modified compensation variable K _mod or the compensation steering angle δ _comp is calculated exactly without any signal and model errors, an ideal and fastest possible disturbance compensation is achieved until the first lateral deviation error threshold d ₀ is reached.

If the lateral deviation error d is further reduced from d ₀ , the activation factor a decreases continuously according to Equation (8) (or according to the activation function A), and the compensation steering angle δ _comp is thereby reduced to zero. For lateral controllers 4 with PD characteristics, for example, the steady-state value for d will be in the interval [sgn(K _lim )* d ₀ ... 0]; for controllers with an integral component, a lateral deviation error of zero is ultimately achieved in the steady state. In the case of exact compensation, therefore, a more dynamic disturbance compensation is achieved overall than without the application of the compensation method.

In case of a strong overcompensation, the cross deviation error d will change its sign and as a consequence the condition $$sgn\left(K_{lim}\right)\ne sgn\left(d\right)$$ according to equation (7), after which a compensation steering angle δ _comp = 0 is applied. With less severe overcompensation or with sufficient damping of the lateral controller 4, a sign reversal of the lateral deviation error d and thus a lateral overshoot can also be avoided, since the lateral controller 4 has sufficient time to counteract the overcompensation or the damping of the lateral control 4 leads to an aperiodic approach to a lateral deviation error d = 0 despite overcompensation.

The purpose of using an activation function A with a first cross-slip error threshold value d ₀ is to avoid damping or an increased tendency of the cross-controller 4 to oscillate in the area of the falling or rising edges of the activation function A. This is because the smaller the first cross-slip error threshold value d ₀ is selected, the more the compensation is reduced depending on the cross-slip error d. In the limiting case d ₀ → 0, a change in the sign of the cross-slip error d results in a hard change between the application of the full compensation steering angle δ _comp and the value 0. In the event of overcompensation, this structure switch can excite the cross-control to a limit cycle in the form of a continuous oscillation with the result of unacceptable control quality. By multiplicatively masking out the compensation using the activation function A with its finite gradient in the transitions (cf. 4 and 5 ), however, this effect can be effectively avoided. The choice of the first cross-shift error threshold d ₀ can be limited by the amplitude reserve of the cross-control loop and must be defined accordingly.

The first lateral deviation error threshold d ₀ can be dynamically adjusted depending on the driving situation, the route, the road width b and/or the speed of the vehicle F. On country roads, for example, the requirements for the accuracy of lane guidance are generally higher than on motorways due to the narrower road width, so that, for example, in the speed range below 110 km/h or on roads classified as country roads, the first lateral deviation error threshold d ₀ is selected to be smaller than at higher speeds or on roads classified as motorways.

Preferably, the modified compensation variable K _mod or the compensation steering angle δ _comp can be low-pass filtered before activation, for example, using a PT1 filter. Since this reduces the amplitude reserve of the lateral control to a lesser extent, a higher gradient of the activation function A and thus a smaller first lateral deviation error threshold d ₀ can be selected. In contrast, low-pass filtering leads to a delayed activation of the disturbance variable compensation; in individual cases, an adequate compromise can be found through simulation or road testing. Filter time constants as low as 0.1 s typically allow a reduction of the first lateral deviation error threshold d ₀ .

It should be noted that the activation function A according to the 4 and 5 represent only an exemplary curve. Deviating from the linear curve for values between 0 and the first cross-shift error threshold d ₀ , other function curves are also conceivable, for example, with a continuously differentiable curve with corresponding flattening at the edges. A key feature of alternative activation functions A is that they reduce the application of disturbance compensation to smaller values, preferably continuously down to zero, starting at a certain cross-shift error threshold d ₀ .

Furthermore, it should be noted that the cross-slip error d, to which the minimum activation factor a is assigned, is also referred to here as the second cross-slip error threshold d ₁ (see 4 and 5 ), does not necessarily have to be zero, but can also be shifted in a positive or negative direction. For example, when cornering, depending on the sign of the road curvature, the minimum activation function value can be selected so that the vehicle tends to drive closer to the inside of the curve, which drivers generally perceive as more comfortable than driving near the outer lane markings.

6 shows a block diagram illustrating the process steps of a method for compensating a disturbance variable that exerts a force transverse to the direction of travel on a vehicle and thereby causes a transverse offset error of the transversely controlled vehicle.

First, a lateral deviation error of the vehicle with respect to a planned trajectory is determined (S10).

Subsequently, a disturbance variable causing the vehicle's lateral deviation error is estimated (S11).

Based on the estimated disturbance, a compensation value is determined (S12).

Subsequently, an activation function is applied to the compensation variable, resulting in a modified compensation variable (S13).

Based on the modified compensation variable, a controller control variable provided by a transverse controller of the vehicle is changed, thereby forming a control variable (S14).

This control variable is fed to the steering control device of the vehicle's electric steering system for lateral guidance of the vehicle (S15).

The invention has been described above using exemplary embodiments. It is understood that numerous changes and modifications are possible without departing from the scope of protection defined by the patent claims.

List of reference symbols

1

system

2

Control unit

3

Electric steering

4

Aileron

5

Estimation unit

6

Transformation facility

7

Limiting device

8

modifier

10

Compensation branch

α

bank angle

A

Activation function

A1

first functional area

A2

second functional area

A3

third functional area

a

Activation factor

b

Lane width

d

Cross-placement error

d0

first cross-deviation error threshold

d1

second cross-deviation error threshold

δtarget

Control variable

δcontroller_setpoint

Controller control variable

δcomp

Compensation steering angle

F

vehicle

h

Road height

K

Compensation value

Climate

limited compensation size

Kmod

modified compensation variable

S

Disturbance

T

Trajectory

Claims (15)

Method for compensating for a disturbance variable (S) that exerts a force on a vehicle (F) that runs transversely to the direction of travel and thereby causes a transverse deviation error (d) of the transversely controlled vehicle (F), the method comprising the following steps: - determining a transverse deviation error (d) of the vehicle (F) with respect to a planned trajectory (T) (S10); - estimating a disturbance variable (S) that causes the transverse deviation error (d) of the vehicle (S11); - determining a compensation variable (K) based on the disturbance variable (S) (S12); - applying an activation function to the compensation variable (K), thereby obtaining a modified compensation variable (K _mod ) (S13), the activation function being designed to change the magnitude of the compensation variable (K) depending on the magnitude of the transverse deviation error (d); - changing a controller manipulated variable (δ _{regier_soll} ) provided by a lateral controller (4) of the vehicle (F) based on the modified compensation variable (K _mod ), whereby a manipulated variable (δ _soll ) is formed (S14); and - supplying the manipulated variable (δ _soll ) to the steering control device (LR) of the electric steering system (3) of the vehicle (F) for lateral guidance of the vehicle (F) (S15). Procedure according to Claim 1 , characterized in that the activation function (A) has an activation factor (a) dependent on the determined transverse offset error (d). Procedure according to Claim 1 or 2 , characterized in that the activation function (A) has a first functional range (A1) which is assigned to a first value range of the determined transverse offset error (d), wherein the first functional range (A1) is designed to leave the compensation variable (K) unchanged, so that the modified compensation variable (K _mod ) is equal to or substantially equal to the compensation variable (K). Method according to one of the preceding claims, characterized in that the Activation function (A) has a third functional area (A3) which is assigned to a third value range of the determined transverse offset error (d), wherein the third functional area (A3) is designed to set the compensation variable (K) to zero, so that in the case of a determined transverse offset error (d) in the third value range, a modified compensation variable (K _mod ) equal to or substantially equal to zero is output. Method according to one of the preceding claims, characterized in that the activation function (A) has a second functional area (A2) which is arranged between the first and third functional areas (A1, A3), and in that the second functional area (A2) is designed to form a continuous transition between the values of the activation function (A) in the first and third functional areas (A1, A3). Method according to one of the Claims 3 until 5 , characterized in that the first functional range (A1) of the activation function (A) is assigned to magnitude-related transverse offset errors (d) greater than a first transverse offset error threshold value (d ₀ ). Procedure according to Claim 5 or 6 , characterized in that the second functional range (A2) of the activation function (A) is assigned to cross-deviation errors (d) between a first and a second cross-deviation error threshold value (d ₀ , d ₁ ). Procedure according to Claim 6 or 7 , characterized in that the first transverse offset error threshold value (d ₀ ) is adapted depending on the driving situation, the driving speed and/or the width of the roadway on which the vehicle (F) is moving. Procedure according to Claim 7 or 8 , characterized in that the second transverse deviation error threshold value (d ₁ ) is 0 or, depending on the sign of the curve curvature of a curve along which the vehicle (F) is moved, is not equal to 0, but is smaller in magnitude than the first transverse deviation error threshold value (d ₀ ). Method according to one of the preceding claims, characterized in that the estimated disturbance variable (S) or the compensation variable (K) is limited in terms of amount. Procedure according to Claim 10 , characterized in that the limitation is based on the statistical distribution of the expected values of the disturbance variable (S) or the statistical distribution of the expected values of the compensation variable (K). Method according to one of the preceding claims, characterized in that the modified compensation variable (K _mod ) is low-pass filtered and that the controller manipulated variable (δ _{regler_soll} ) is changed based on the low-pass filtered modified compensation variable. Method according to one of the preceding claims, characterized in that the estimated disturbance variable (S) is a disturbance variable dependent on the roadway transverse inclination or a disturbance variable dependent on the crosswind. Method according to one of the preceding claims, characterized in that the controller manipulated variable (δ _{regier_soll} ) indicates a desired steering angle, a desired torque or a tie rod force which is supplied to a steering control device (LR) of the vehicle (F) as an input variable. System for the lateral guidance of a vehicle (F) along a planned trajectory (T), comprising a control unit (2) with a lateral controller (4), by means of which a controller manipulated variable (δ _{regler_soll} ) is provided, and an interface via which the manipulated variable (δ _soll ) is supplied to an electric steering system (3), wherein the control unit (2) is designed to carry out the following steps: - determining a lateral deviation error (d) of the vehicle (F) with respect to a planned trajectory (T); - estimating a disturbance variable (S) which causes the lateral deviation error (d) of the vehicle (F); - determining a compensation variable (K) based on the disturbance variable (S); - applying an activation function (A) to the compensation variable (K), thereby obtaining a modified compensation variable (K _mod ), wherein the activation function is designed to change the amount of the compensation variable (K) depending on the amount of the lateral deviation error (d); - changing the controller manipulated variable (δ _{regier_soll} ) based on the modified compensation variable (K _mod ), thereby forming a manipulated variable (δ _soll ); and - supplying the manipulated variable (δ _soll ) to the steering control device (LR) of the electric steering system (3) of the vehicle (F) for lateral guidance of the vehicle (F).