DE102007023000B4 - Method and device for controlling a vehicle - Google Patents

Abstract

Method for controlling a vehicle by means of driver assistance systems which act directly on the vehicle via actuating elements of the vehicle drive system and vehicle braking system, and which signals measured and determined by the driver assistance systems are changed or further signals are generated, whereby influences due to the vehicle system arise when the signals are changed and / or generated the driving behavior of the vehicle are represented in the changed and / or generated signals and the changed and / or generated signals are fed into the driver assistance system, characterized in that a vehicle acceleration is predicted by means of a predictor by dividing the current vehicle acceleration a (n) by the Change in a vehicle engine torque and a vehicle braking torque is determined and a deceleration signal is generated between the current vehicle acceleration a (n) and the temporally preceding vehicle acceleration a (n-1) and is multiplied by a vehicle-specific variable K and added to the current vehicle acceleration a (n).

Description

The present invention relates to the field of vehicle follow-up control and, in particular, to a method and a device for controlling a vehicle by means of a driver assistance system.

Driver assistance systems are known which support the driver in the longitudinal guidance of the vehicle (acceleration and deceleration) and / or in lateral guidance (lane keeping, steering). The functions of such guidance systems range from simple cruise control to a desired speed that the driver can select, through adaptive cruise control (ACC), which also takes into account the distances to vehicles in front, to fully autonomous guidance of the vehicle. Further examples of functions of such a driver assistance system are the automatic generation of collision warnings or the automatic initiation of emergency braking or evasive maneuvers to avoid or mitigate the effects of collisions. The traffic situation is recorded with the help of sensors attached to the vehicle, the signals of which are fed to a control unit as input variables. The input variables relate on the one hand to the movement variables of the guided vehicle itself, for example its driving speed, acceleration, yaw rate and the like, and on the other hand to information about the traffic environment, in particular location data from vehicles in front and other obstacles and possibly information about the course of the road, the road surface and the like.

One or more distance sensors are typically provided for the acquisition of the location data, for example a radar sensor for measuring the distances and relative speeds of radar targets, in the case of an angularly resolving radar sensor also for measuring the azimuth angles of the radar targets, or lidar sensors or camera systems, in particular stereo camera systems with electronic image processing. On the basis of the input variables supplied by these sensors, the control unit calculates manipulated variables that act directly on the vehicle via control elements of the drive system and, if necessary, of the brake system. An example of an ACC system of this type is described in SAE Paper No. 96 10 10 “Adaptive Cruise Control, System Aspects and Development Trends”, Winner et al., 1996.

The calculation of the manipulated variables by the control unit is dependent on a number of parameters, some of which can also be changed dynamically during normal operation, depending on the traffic situation.

For example, with the radar-based distance control, several objects are usually located by the radar system at the same time. An object list is created in which the individual objects are represented by their distance, relative speed and angle data. In the periodically repeated radar measurements, a so-called tracking procedure is used to identify the objects detected in the current measurement with the objects detected in previous measurements, and the movements of the individual objects are tracked.

The input signals transmitted from the sensors to the control unit are more or less noisy in practice and must therefore be processed with the aid of suitable filters. Each of these filter procedures is influenced by one or more parameters which in particular determine the temporal resolution of the filter, for example integration times, decay rates or the selection of frequency ranges in the frequency spectrum of the signal. The filters must be parameterized in such a way that, on the one hand, sufficient noise and interference signal suppression is achieved, but, on the other hand, changes in the input variables are passed on so quickly that the management system can react in good time.

Often predictive controllers are used in the control unit, which extrapolate the movements of the own vehicle and movements of the located objects into the future and thus predict the traffic situation at a point in time in the future. The manipulated variables are then calculated in such a way that an optimal adaptation to the predicted traffic situation is achieved within a certain optimization time interval. Longer forecast periods and optimization time intervals lead to a "predictive" behavior of the guidance system and thus to a high level of driving comfort. However, they have the disadvantage that the probability of incorrect forecasts increases and, under certain circumstances, sudden changes are not adequately responded to.

Furthermore, when predicting the future development of an input variable, for example the distance to an object, it must be decided whether a linear extrapolation should be carried out, based on the assumption that the relative speed remains constant, or a quadratic extrapolation assuming a constant acceleration, or else an even higher order extrapolation. If necessary, in the event of sudden changes in state, for example in the event of an abrupt braking maneuver of the vehicle driving ahead, the prediction must also be plausible Assumptions are made about how long this condition will last.

In general, the determination of the parameters that determine the behavior of the various control functions of the control unit requires an assessment of the traffic situation. So far, either the measured or derived kinetic state variables of the own vehicle and the detected objects have been used as situation-specific variables for this assessment, e.g. B. the speed of your own vehicle, the distance and the relative speed of the vehicle in front, etc., or simple variables that are derived from these state variables, z. B. the "time to collision" (TTC), ie the calculated time until the impact.

For the reasons mentioned, existing ACC regulations aim to set specified acceleration and deceleration requirements with the lowest possible control deviations. For this purpose, a permanent actual / target comparison of the correspondingly measured or calculated values is carried out. This means that system changes, such as increasing or "adding" vehicle engine torque or building up vehicle brake pressure, would have to be included in the actual / target comparison immediately at the corresponding signal path in order to be able to design the vehicle follow-up control accordingly.

Since the vehicle is a complex system made up of various inertias, elasticities, etc., delays arise in the mapping of the system change in the measured or measured and determined signals or calculated signal courses. This fact creates system-related control deviations which are initially corrected incorrectly and subsequently lead to overshoot processes which have to be compensated for by a corresponding system reaction. These system reactions generate an oscillating, uncomfortable regulation, which also, named as an example, increases the load change of the EBS.

From the DE 10 2004 058 359 A1 a method for driving motion control is known in which setpoints are fed into a controlled system and real measured values are taken from the controlled system to determine updated setpoint values. For this purpose, the controlled system is modeled, with modeled measured values being determined from the setpoint values in the modeled controlled system. The modeled measured values are then compared with the real measured values, the modeled controlled system (9) being adjusted in the event of a deviation and the parameters describing the adjustment being taken into account when determining the setpoint values.

Furthermore describes the DE 10 2004 051 759 A1 a method and a corresponding control system for integrating a vehicle model into a driving system device. The model parameters can be modified while driving depending on the incoming sensor data. The model updated in real time is used to control the vehicle components by first checking the effect of setting values on the vehicle model and correcting them if necessary.

The object of the invention is to reduce the aforementioned system-related influences.

The object is achieved by the features of claim 1 and claim 7. Advantageous configurations and developments of the invention are specified in the subclaims.

Inertia is understood to mean all "system-related" situations of all influencing factors that generate a delayed vehicle reaction and thus a time delay in the signal.

In addition to the “classic inertia”, elasticities caused by the tires, the chassis, the clearance and other system delays are advantageously recorded and evaluated according to the invention. This variable, which takes into account the inertia and elasticity, represents a predictor which, in dynamic situations, advantageously predetermines the vehicle reaction in order to prevent overshoot processes or to reduce them to a minimum.

Driver assistance systems can be, for example, parking assistance (sensor arrays for obstacle and distance detection), brake assist (BAS), cruise control, adaptive cruise control or adaptive cruise control (ACC), distance warning, turning assist, traffic jam assist, lane detection system, lane keeping assist / lane assist (lateral guidance support, lane departure warning (LDW )), Lane keeping support, lane change assistance, lane change support, Intelligent Speed Adaption (ISA), adaptive cornering lights, tire pressure monitoring system, driver status recognition, traffic sign recognition, platooning, automatic emergency braking (ANB), on - and dimming assistant for the headlights, night vision system (Night Vision).

In the following, embodiments of the invention are based on the 1 explained.

The engine torque request ( 10 ) as well as brake pressure requirements ( 20th ) are used, as these requirements evaluate the current system deviations, but without taking into account the tracking of the variables to be influenced. The variables to be influenced, such as the vehicle engine torque and / or vehicle brake pressure, are determined via the control unit of the driver assistance system ( 30th ) requested. Only after a certain time ( 8th ) the requested value is set. The evaluation of the target / actual deviation is only one influencing factor, however, in order to transfer the current vehicle reaction to a vehicle reaction that is just beginning to appear. Another element is the torsion of the tire sidewall ( 40 ). In other words, when a force or a moment is transmitted from the motor or the brake to the road surface, the tire is first twisted in order to then enable a force build-up between the tire and the road. Powertrain and chassis also create a deceleration ( 8th ). Based on the evaluation of the torsion of the tire ( 40 ) the influence is determined, i.e. if a requirement ( 50 ) to vehicle engine ( 10 ) or vehicle brake ( 20th ) is set, after a certain time ( 8th ) the set state in the vehicle deceleration ( 70 ) reflect. However, since this represents a primary control variable of an ACC or cruise control system, this variable is modified by a predictor a (pred) in such a way that the overall control is suppressed in under- or overshoots.

$$\text{a }\left(\text{präd}\right)=\left({\Delta }_{\text{Ver}}*\text{K}\right)+\text{a}$$

mit

• a: Fahrzeugverzögerung

A further advantageous embodiment of the method according to the invention is the consideration of the current change in the vehicle deceleration, thus the vehicle acceleration. If the engine and / or braking torque is changed, after a certain time delay ( 8th ) an initial vehicle reaction. This is, at least for a certain time, usually a linear requirement, the deceleration signal or the deceleration sensor signal is modified by adding a vehicle-specific factor. This factor contains the current change in the deceleration signal Δ _Ver , as well as a vehicle-specific variable that describes the deceleration behavior (K) of the vehicle. $${\Delta }_{\text{Ver}}=\text{a}\left(\text{n}\right)-\text{a}\left(\text{n-1}\right)$$

$$\text{a}\left(\text{präd}\right)=\left({\Delta }_{\text{Ver}}*\text{K}\right)+\text{a}$$

With

• a: vehicle deceleration

The vehicle-specific factor K can either be determined in advance as a starting value for each vehicle and stored in the control unit ( 60 ) or it is learned individually for each vehicle by means of a learning phase and a learning algorithm and stored in the control unit ( 60 ) processed.

As a function of the rate of change of the deceleration due to, for example, a rapid change in torque, the signal of the predictor is modified in such a way that Δ _{Ver is} only partially included and thus highly dynamic situations are taken into account.

In a further advantageous embodiment, it is contemplated according to the invention to modify the predictor with regard to vehicle speed and / or pressure range in which the vehicle is operated.

In a further advantageous embodiment of the predictor, a prediction of the next, thus expected future vehicle acceleration takes place on the basis of the sum torque from the vehicle equation. The prediction of the acceleration a (n) to be expected is calculated from the sum of the instantaneous vehicle engine torques TM (n-1) and vehicle braking torques TB (n-1).

The generated signal a (pred) is used to replace the original deceleration signal in the driver assistance system, such as an ACC system and / or a cruise control, and thus to use it as a control variable. A situation-dependent combination of the known delay signal and a (präd) is also considered and advantageous.

For this purpose, it is provided that the inertia and elasticity can be assessed and weighted by means of an artificial neural network. The weighted signals are fed to the predictor and the signal a (pred) is generated for the ACC system, for example. It should be mentioned here that in addition to neural networks, other data mining methods, such as cluster analysis, main component analysis, factor analysis and classification methods, etc. are being considered.

Although the present invention has been described above with reference to preferred exemplary embodiments, it is not restricted thereto, but rather can be modified in many ways.

Claims (7)

Method for controlling a vehicle by means of driver assistance systems, which act directly on the vehicle via actuating elements of the vehicle drive system and vehicle braking system, and which signals measured and determined by the driver assistance systems are changed or further signals are generated, with the change and / or generation of the signals Vehicle system-related influences on the driving behavior of the vehicle are represented in the changed and / or generated signals and the changed and / or generated signals are fed into the driver assistance system, characterized in that a vehicle acceleration is predicted by means of a predictor by the current vehicle acceleration a (n ) is determined by changing a vehicle engine torque and a vehicle braking torque and a deceleration signal is generated between the current vehicle acceleration a (n) and the previous vehicle acceleration a (n-1) and multiplied by a vehicle-specific variable K and the current vehicle acceleration a (n) is added. Procedure according to Claim 1 , characterized in that the vehicle-specific inertia and vehicle-specific elasticity when changing and / or generating the signals are evaluated as influences caused by the vehicle system. Procedure according to Claim 1 , characterized in that the system-related properties of the tires and the system-related properties of the chassis are evaluated to determine the elasticities. Method according to one of the preceding claims, characterized in that air gaps and system delays of the vehicle are determined and evaluated in order to determine the elasticities. Method according to one of the preceding claims, characterized in that the vehicle-specific variable K is applied as a starting value in a control unit. Method according to one of the preceding claims, characterized in that the vehicle-specific variable K is learned by means of a learning phase and a learning algorithm. Device (60) containing technical devices for performing one of the methods according to one of the preceding claims.

Images