Environment assessment system in a vehicle with sensor means for detecting objects in the environment of the vehicle

Abstract

Environment assessment system in a vehicle with sensor means for detecting objects in the environment of the vehicle and with evaluation means for assessing objects in the environment of the vehicle with regard to the relevance of the objects as target objects for a warning, control or regulation system in the vehicle, wherein in a first step each detected object (i; F1, F2) is evaluated independently of the other detected objects (F2; F1) with regard to its lateral relevance (rel y,i ) and separately regarding its longitudinal relevance (rel x,i ) for the own vehicle (FE) is evaluated, that in a second step for each recorded object (i) an overall relevance (rel i ) from its lateral relevance (rel y,i ) and its longitudinal relevance (rel x) is calculated and that in a third step the detected object with the maximum overall relevance is selected as the only target object for the own vehicle (FE), characterized in that a physical quantity (TTC_NO) for determining the longitudinal relevance (rel x,i ) of a first detected object (F2) depending on a defined physical quantity (x pos1 , v x,1 ) of a second detected object (F1) is evaluated, which is the longitudinally closest detected object (F1) relative to the own vehicle (FE), if the first detected object (F2) and the second detected object (F1) are not identical.

Description

The invention relates to an environmental evaluation system in a vehicle having sensor means for detecting objects in the environment of the vehicle according to the preamble of patent claims 1 and 3.These are in particular speed control systems which actuate the drive and / or brake systems of a vehicle in such a way that either a setpoint speed or a setpoint distance from the vehicle (target object) driving in front is set. In the case of mass-produced vehicles, such speed control systems are also known as ACC (Adaptive Cruise Control) systems.The known longitudinal control systems have hitherto been designed in such a way that automatic longitudinal acceleration to the setpoint speed is carried out when a following travel is ended. A following travel with a predetermined target distance is ended either when the preceding vehicle (target object) changes lanes or when the own vehicle changes lanes. Up to now, reliable detection of a lane change is not possible even using information from a navigation system, since firstly the road networks stored in the navigation system have error tolerances and secondly the sensor used up to now for determining the position of the own vehicle (GPS) is still too inaccurate.The correct evaluation of whether or not a preceding vehicle (target object) is present on the own lane cannot be made with sufficient probability even with the information of sensors for distance measurement (radar, ultrasonic, infrared, and image processing), because the detection range (detection angle) of the sensors is relatively small.Adaptive cruise controls (ACC) as a development of cruise control functions are today state of the art in the field of longitudinal guidance assistance for motor vehicles. Further developments today have the integration of congestion functions with the aim of being able to adjust the distance from the front vehicle even in slow-flowing traffic and congestion situations to a standstill. From the standstill, an automatic or driver assisted, controlled starting is implemented (Stop&Go), depending on the service life. In addition to comprehensive environment detection, these functions require in particular the suitable interpretation of the measured environment data with the aim of evaluating objects with respect to their relevance and selecting a target object for control processes.The laid-open specification DE 10 2007 029 483 A1 describes a distance control device for motor vehicles, having a locating system for locating vehicles driving in front on the own lane and on secondary lanes. In addition, the laid-open specification discloses a controller for regulating the speed of the host vehicle as a function of the distance from a vehicle traveling ahead on the host lane. An evaluation module for evaluating a cut-in probability for vehicles located on a side lane is also disclosed.The document DE 199 54 536 A1 describes a method for evaluating objects in the course of a vehicle, in which the distance and / or the speed of the target objects is determined with a sensor. In the case of the presence of more than one target object, only the target object is to be included as a new object in the evaluation, instead of the current target object, which is located in a range which is limited by at least one parameter defined by the position relative to the vehicle. In addition, other parameters may be included as conditions for the selection of a new target object.It is an object of the invention to ensure a predictive and unique selection of a target object also from a plurality of detected objects.This object is achieved by the features of claims 1 and 3. Advantageous further developments of the invention are the subject matter of the dependent claims.According to the invention, in the case of a surroundings evaluation system in a vehicle having sensor means for detecting objects in the surroundings of the vehicle and having evaluation means for evaluating objects in the surroundings of the vehicle with regard to the relevance of the objects as target objects for a warning control or regulating system in the vehicle, in a first step each detected object is evaluated independently of the other detected objects with regard to its lateral relevance and separately therefrom with regard to its longitudinal relevance for the host vehicle. In a second step, an overall relevance is calculated for each detected object from its lateral relevance and its longitudinal relevance. In a third step, the detected object having the maximum overall relevance is selected as the single target object for the own vehicle.Alternatively or additionally, according to the invention, in the case of an environmental evaluation system in a vehicle having sensor means for detecting objects in the environment of the vehicle and having evaluation means for evaluating objects in the environment of the vehicle with respect to the relevance of the objects as target objects for a warning control or regulating system in the vehicle, in a first step each detected object is evaluated independently of the other detected objects at least with respect to its longitudinal relevance for the host vehicle. The longitudinal relevance of each detected object is determined in each case on the basis of a plurality of defined physical variables. A physical variable for ascertaining the longitudinal relevance of a first detected object is evaluated as a function of a defined physical variable of a second detected object, which, with respect to the own vehicle, is the detected object closest longitudinally if the first detected object and the second detected object are not identical.The lateral relevance and the longitudinal relevance are preferably each determined on the basis of a plurality of defined physical variables of the respectively detected objects.Preferably, for determining the longitudinal relevance of an object, the time gap between the own vehicle and this object and / or the longitudinal position sequence, which results from the longitudinal position of the object relative to all other detected objects, and / or the collision time, after which a collision would occur if the relative speed between the own vehicle and the object remained the same, are defined as physical variables.To determine the lateral relevance of an object, the lateral position to the own vehicle and / or the lateral speed to the own vehicle are preferably defined as physical variable.In a development of the invention, the influence of the defined physical variables is weighted differently at least for ascertaining the longitudinal relevance.In a preferred exemplary embodiment, the relevance values are defined in a range between 0 and 1, wherein the overall relevance of an object is calculated by multiplying its lateral relevance by its longitudinal relevance.The environment evaluation system according to the invention is preferably used in a longitudinal control system for distance control and / or collision avoidance, wherein drive and / or brake actuators for maintaining a predefined distance or a predefined speed from the target object selected in the front region by the environment evaluation system according to the invention are actuated in a fundamentally known manner by the longitudinal control system. For example, regulation can be made to a predefined distance from a selected target object with penetrating longitudinal relevance or to a predefined speed in order to avoid impermissible passing processes (right-hand passing prohibition on German freeways) in the case of ACC with penetrating lateral relevance.The term longitudinal relevance is to be understood in particular as the degree of approach of an object in the longitudinal direction. The term lateral relevance is to be understood in particular as the degree of the approach of an object in the transverse direction, i.e. the probability of an object changing to its own lane.A particularly advantageous application of the invention is in relation to objects in the front area of the own vehicle. However, applications relating to objects in the lateral and return range, for example in the context of lane change assistants, are also possible. Alternatively or additionally to the application in control and regulating systems with active driving or braking interventions, an acoustic, visual or haptic warning of a selected "target object" can also be output.For a detailed explanation of the invention in connection with the description of particularly relevant exemplary embodiments, reference is made to the drawing. It shows FIG. 1 shows an object scenario with high longitudinal relevance of a "front-front vehicle", FIG. 2 shows an object scenario with increasing lateral relevance of a vehicle (object) changing to the own lane FIG. 3 shows a transformation of possibly curved course profiles to a rectilinear course profile, which is preferably carried out before a relevance determination, in order to simplify the determination of the lateral and longitudinal relevance, FIG. 4 shows a schematic functional block diagram for the main components of the environment evaluation system according to the invention, FIG. 5 shows a schematic illustration of a particularly advantageous calculation of the longitudinal relevance, and FIG. 6 shows a schematic illustration of a particularly advantageous calculation of the lateral relevance.The prior art relevance score generally provides to select the immediately closest vehicle or obstacle (object) traveling in the own lane in front of the own vehicle as a relevant target object. Here, the algorithm is generally limited to evaluating, after discrete lane assignment of surrounding objects to the own lane, the next surrounding object with the highest overall relevance assigned longitudinally to the own lane. This procedure leads to a strong focusing of the following travel on the direct front vehicle without far-sighted forward view. Particularly in dynamic situations which occur either longitudinally (e.g. a strong braking of the front vehicle and not yet a deceleration of the direct front vehicle-see scenario according to FIG. 1 ) and / or laterally (e.g. a vehicle cutting in from the adjacent lane-see scenario according to FIG. 2 ) in the front space, a predictive driver would include all surrounding objects visible to him in his longitudinal guidance behavior.The driver is in this case able to fuse all laterally and / or longitudinally dynamic transition situations of object movements to form an overall relevance. For its overall evaluation, it perceives lateral as well as longitudinal decision variables for this purpose. A lane assignment to the future course to be driven, which is initially carried out, and the subsequent selection of the next vehicle in the longitudinal direction generally do not reflect the forward-looking environmental interpretation of the driver and thus causes many longitudinal guidance interventions, which the driver evaluates as synthetic and less intuitive. If the system does not drive approximately "how the driver himself would drive", this leads to a reduced acceptance of the longitudinal guidance assistance system and thus also to a reduction in comfort by means of permanently required system monitoring by the driver.According to the invention, therefore, a system for predictive object evaluation and environment interpretation is proposed, which evaluates all objects (here, for example) in the front space separately at any point in time with respect to lateral and longitudinal relevance for a selection to the control object by individual relevance measurement variables. In order to enable the strict separation after lateral and longitudinal relevance assessment and in this case to achieve disruptive dependencies on curved travel course profiles, the transformation of curved course profiles to a rectilinear course profile is preferably carried out first on the basis of course prediction (see FIG. 3 ). All physically relevant measured variables of surrounding objects, i.e. positions, speeds and accelerations, are to be transformed. After the transformation, lateral and longitudinal relevance variables are determined on the basis of physical models and separately fused to form a lateral and longitudinal overall relevance. Lateral relevance variables are always completely decoupled from longitudinal relevance variables in their calculation. After ascertaining the lateral and longitudinal overall relevance, an overall relevance measure for all objects in the front space is ascertained by a further fusion algorithm. The surrounding object that is the most highly evaluated with regard to the overall relevance becomes the regulating object, regardless of whether it is the currently closest surrounding object in the own lane at the respective point in time.A schematic representation of the overall approach illustrates FIG. 4. input signals E 1, such as object data, the predicted course of the own vehicle and vehicle data, lead to a first function block B 1, in which the aforementioned course transformation takes place. The course-transformed object data and vehicle data enter the function blocks B 2 and B 3 as input signals E 2 and E 3, respectively. In function block B 2, the lateral relevance calculation takes place and in function block B 3, the longitudinal relevance calculation takes place. The output signals of the function block B 2 are the total lateral relevance values for all detected objects which enter the function block B 4 as input signals E 4. The output signals of the function block B 3 are the longitudinal total relevance values for all detected objects which are input as input signals E 5 to the function block B 4. In function block B 4, the overall relevance values are ascertained for all objects which enter as input signals E 6 into function block B 5, in which the target object selection, in particular in the form of a maximum selection, takes place. The output signal E7 of the function block B5 is the selected object which is taken into account by the control, regulating or warning system which is respectively processed further.The course transformation which is carried out in function block B 1 is explained in more detail with reference to FIG. 3 :For the course transformation, the predicted curvature κ for the course course course of the own vehicle (FE) to be driven in the future is provided as an input variable from a course prediction. This input variable is based on an approximation of the course profile to a circular path, i.e. the vehicle moves at a time t approximately on a circular path with a constant radius R. Furthermore, all object data from a surrounding sensor system are made available for the transformation in an arbitrary vehicle coordinate system to the course transformation.According to FIG. 3, from the origin of the predicted circular path, the lot is dropped to the predicted course by the position of a surrounding object. The result is the angle φ according to the equation:Using equation (1.1), the length of the arc segment described by the angle φ and the radius R=1 / κ is calculated as follows:The transformed longitudinal position is described by x pos. The lateral position y pos is calculated as:In addition, the speeds v' x, v' y and the accelerations a' x, a' y become as shown in Eq. 1.4) and Gl. (1.5):The longitudinal relevance calculation which is carried out in function block B 3 is explained in more detail with reference to FIG. 5 :For the longitudinal relevance calculation according to FIG. 5, physical model variables for calculating a longitudinal relevance for the selection of a rule object are mapped via a mapping function to a relevance value of 0 to 1. These relevance values are summed up in a weighted manner and divided by the sum of the weights (w TG,... w TTC_NO). The following longitudinal relevance variables are calculated:The time gap TG is a distance measure normalized to the own speed v ego of the own vehicle FE ("ego vehicle") from a (foreign) object i according to Eq. (2.1):The longitudinal position order PF x,i, which is derived from the longitudinal position of a (foreign.) This results in objects relative to all other (foreign) objects located ahead. The associated relevance measure is calculated for a total object number of N (foreign) objects as:The collision time (time-to-collision) TTC occurs after a collision occurs with constant relative speed v rel between the own vehicle (FE) and (foreign) object i. The TTC is calculated accordingly as:In order to be able to optionally also control a front vehicle with high longitudinal dynamics in the front space if this has a high deceleration, but if the direct front vehicle does not yet show this reaction, a predictive collision time TTC_NO between the longitudinally closest (foreign) object to the own vehicle and all other longitudinally further (foreign) objects is calculated. A predictive driver would likewise already use the slower front vehicle as a basis for his actions at an early stage. For an additional relevance measure, the TTC_NO is calculated as the input variable as the collision time between the next longitudinal object with position x pos,NO and speed v x,NO and all other longitudinally further (foreign) objects i with position x pos,i and speed v x,i.The lateral relevance calculation which is carried out in function block B 2 is explained in more detail with reference to FIG. 6 :For the lateral relevance calculation according to FIG. 6, a relevance measure rel Ypos is formed via the input variables track width SB, signals for expanding or removing the lateral track assignment on secondary tracks NS and lateral position y pos,i of a (foreign) object i via an mapping function. In addition, the relevance measure rel vy is calculated from the lateral speed v y,i of a (foreign) object i by a further mapping function. The relevance measure rel vy has the particular property that a value range from -1 to 1 is permissible for this measure. Both relevance measures are summed and mapped via a limiter to a normalized lateral overall relevance rel y with a value range from 0 to 1.The mapping function for calculating the relevance measure rel Ypos is, according to FIG. 6, for example, embodied as a trapezoid. A stationary track assignment is carried out via this trapezoid on the basis of the lateral position y pos,i. The boundaries Y2r and Y2l vary depending on an estimated and measured track width, respectively. In the case of an identified lane change to the adjacent lane or further criteria, the trapezoid is dynamically expanded laterally in the direction of the adjacent lane and correspondingly dynamically relieved upon movement to the adjacent lane. Further criteria in this case are the expansion laterally in the direction of a fixed edge construction (guardrail) directly next to the current lane and / or a direct adjacent lane which was identified as a lane for oncoming vehicles. In addition, it is pointed out that the lane width SB can be widened in two lanes by means of the signal NS when the host vehicle FE changes lanes.The form of the mapping function for calculating the relevance measure rel vy is illustrated according to the latest diagram in FIG. 6. Starting from a certain minimum speed v1r, v1l, a relevance in the range from -1 to 1 is formed. A positive relevance indicates that a foreign object i cuts into the current lane. In the case of negative relevance, a foreign object i scissors off from the current lane. This measure allows lateral forward-viewing of when foreign objects are about to enter or leave their own lane.Overall relevance calculation according to function block B 4 in FIG. 4 : The lateral and longitudinal relevance measure rel x, rel y or rel x,i, rel y,i is combined by multiplication for each foreign object i to form an overall relevance rel i: Target Object Selection according to Function Block B 5 in FIG. 4 :After the calculation of the total relevance rel i for all foreign objects i in the front space of the own vehicle, the target object is selected for the longitudinal control by maximum decision.A concrete example regarding a front-front vehicle will be illustrated with reference to FIG. 1. A specific example taking into account a merging vehicle is illustrated with reference to FIG. 2. The object F 2( ti- 1) is in each case illustrated in dashed lines and is intended to illustrate only one scenario before the current (ti) scenario which differs from the current scenario and as a result of which a target object change takes place.In principle, the following applies to both examples: the host vehicle FE contains an environment evaluation system according to the invention having sensor means which currently detect the objects F 1 and F 2 in its environment. The environment evaluation system has evaluation means (see function blocks B 1 to B 5 according to FIG. 4 ) for evaluating the objects F 1 and F 2 with regard to the relevance of the individual objects i (F 1: i=1 and F 2=i=2). In a first step, according to the function blocks B 2 and B 3, each detected object F 1 and F 2 is evaluated independently of the respective other detected object with respect to its lateral relevance rel y,1 and rel y,2 and separately therefrom with respect to its longitudinal relevance rel x,1 and rel x,2 for the own vehicle FE. In a second step, for each detected object F 1 and F 2, a total relevance rel 1 and rel 2 is calculated from its lateral relevance and its longitudinal relevance (function block B 4 according to FIG. 4 ). In a third step, the detected object having the maximum overall relevance is selected as the single target object for the own vehicle FE.In the example according to FIG. 1, the objects F 1 and F 2 are both located on their own lane, so that the same (highest) lateral relevance rel y,1 and rel y,2 of 1 results therefrom for both objects. In this example, the preferred determination of the longitudinal relevance rel x,1 and rel x,2 is to be discussed in more detail in particular:The longitudinal relevances rel x,1 and rel x,2 are determined on the basis of the defined physical variables already illustrated in connection with FIG. 5. In the specific example, this means the following:1) for object F1, i.e. i=1: wherein PF x,1= 1 and N=2Here, the first detected object F 1 is identical to the second closest object F 1, therefore no TTC_NO value is present. By means of the exemplary embodiment according to FIG. 5, the weighting w TTC_NO can be set to zero in this case in order to disable the entire TTC_NO path for the relevance consideration.2) for the object F2(ti), i.e. i=2: wherein PF x,2= 2 and N=2This is because x pos,NO= x pos,1 and v x,i= v x,2 and v x,NO= v x,1 and v x,2< v x,NOAt low speed v x,2 the result of TTC_NO is assumed to be a time value that is less than the defined time limit T 1 (see FIG. 5 ). Therefore, the highest relevance value of 1 results for the variable TTC_NO. If the remaining variables result in lower relevance variables, the object F 2 is selected as the target object.In the example according to FIG. 2, object F 1 is located in the own lane and object F 2 is to change from the adjacent lane to the own lane at current point in time ti. If object F 2 is recognized as arriving in the own lane by the lateral relevance calculation, both objects F 1 and F 2 are again given the highest lateral relevance. Due to the closer longitudinal position x pos,2, the high position sequence PF and the higher TTC of the object F 2 compared to the object F 1, the target object changes from object F 1 to object F 2 at the time ti.In this example according to FIG. 2, in particular the increase in the lateral relevance rel y,2 of the object F 2 is also illustrated in more detail. The lateral relevance rel y,2 is determined on the basis of the defined physical variables already illustrated in connection with FIG. 6. In the specific example, this means that the following: the lateral position y pos,2 and the lateral speed v y,2 change at the time tito values which indicate that the object F2, which was on the adjacent lane at the time ti-1 or behind the own vehicle FE and thus had neither longitudinal nor lateral relevance, is now located on the own lane.In contrast to conventional methods of object assessment for longitudinal guidance assistance systems, this approach is not limited to the aim of selecting the closest foreign object as target object in a driving path to be driven in the future. Rather, all foreign objects in the front space are generally included in the evaluation. Furthermore, a strict decoupling of the evaluation of lateral and longitudinal relevance variables is created in this case in order to achieve preview with respect to longitudinal situations (addressed by the regulation on front-end vehicles) or with respect to lateral situation (addressed by the consideration of cut-in / cut-out processes on the basis of laterally dynamic measurement variables). The object assessment presented achieves the claim of simulating the look-ahead typical of the experienced driver by measures of the environmental interpretation. In many cases, this makes it possible to access a foreign object that becomes relevant in short in the regulation more early, regardless of whether another foreign object is currently closest in the lane of the ego vehicle at this point in time. In particular in urban traffic situations-excellent due to higher dynamics and higher complexity-this form of preview provides a significantly higher customer acceptance. The driver is forced to have fewer longitudinal control interventions and experiences brake and acceleration interventions resembling his own driving style in kinesthetic fashion. This approach of interpreting the environment therefore represents a distinct value of added value for the customer.

Claims

Environment evaluation system in a vehicle having sensor means for detecting objects in the environment of the vehicle and having evaluation means for evaluating objects in the environment of the vehicle with respect to the relevance of the objects as target objects for a warning control or regulating system in the vehicle, wherein in a first step each detected object (i; F1, F2) is evaluated independently of the other detected objects (F2; F1) with respect to its lateral relevance (rel y,i) and separately therefrom with respect to its longitudinal relevance (rel x,i) for the own vehicle (FE), in a second step, for each detected object (i), a total relevance (rel i) is calculated from its lateral relevance (rel y,i) and its longitudinal relevance (rel x) and in a third step, the detected object having the maximum total relevance is selected as the single target object for the own vehicle (FE), characterized in that a physical variable (TTC_NO) is evaluated for determining the longitudinal relevance (rel x,i) of a first detected object (F2) as a function of a defined physical variable (x pos1, v x,1) of a second detected object (F1), the detected object (F1) closest longitudinally to the own vehicle (FE) when the first detected object (F2) and the second detected object (F1) are not identical.Environment evaluation system according to Patent Claim 1, characterized in that the lateral relevance (rel y,i) and the longitudinal relevance (rel x,i) are in each case determined on the basis of a plurality of defined physical variables of the respectively detected objects (i).Environment evaluation system in a vehicle having sensor means for detecting objects in the environment of the vehicle and having evaluation means for evaluating objects in the environment of the vehicle with regard to the relevance of the objects as target objects for a warning control or regulating system in the vehicle, wherein in a first step each detected object (F1; F2) is independent of the other detected objects (F2; F1) is evaluated with respect to its longitudinal relevance (rel x,i) for the host vehicle (FE), characterized in that the longitudinal relevance (rel x,i) of each detected object is determined in each case on the basis of a plurality of defined physical variables, and in that a physical variable (TTC_NO) for determining the longitudinal relevance (rel x,i) of a first detected object (F2) is evaluated as a function of a defined physical variable (x pos,1, v x,1) of a second detected object (F1), which is the longitudinally closest detected object (F1) with respect to the host vehicle (FE), if the first detected object (F2) and the second detected object (F1) are not identical.Environment evaluation system according to one of the preceding patent claims, characterized in that, in order to determine the longitudinal relevance (rel x,i) of an object, the time gap (TG) between the own vehicle and this object is defined as a physical variable.Environment evaluation system according to one of the preceding patent claims, characterized in that, in order to determine the longitudinal relevance (rel x,i) of an object (F i) the longitudinal position sequence (PF x,i), which results from the longitudinal position (x pos,i) of the object relative to all other detected objects with respect to the own vehicle (FE), is defined as a physical variable.Environment evaluation system according to one of the preceding patent claims, characterized in that, in order to determine the longitudinal relevance (rel x,i) of an object (F i) the collision time (TCC) after which a collision would occur if the relative speed between the host vehicle and the object remained the same is defined as the physical variable.Environment evaluation system according to one of Patent Claims 2, 4, 5 or 6, characterized in that, in order to determine the lateral relevance (rel y,i) of an object (F i) the lateral position (y pos, i) to the own vehicle is defined as a physical variable.Environment evaluation system according to one of Patent Claims 2, 4, 5, 6 or 7, characterized in that, in order to determine the lateral relevance (rel y,i) of an object (F i) the lateral speed (v pos, i) to the own vehicle is defined as a physical variable.Environment evaluation system according to one of the preceding patent claims, characterized in that the influence of the defined physical variables is weighted differently at least for ascertaining the longitudinal relevance (rel x,i).Environment evaluation system according to one of Patent Claims 2, 7 or 8, characterized in that the relevance values are established in a range between 0 and 1, and in that the overall relevance (rel i) of an object (F i) is calculated by multiplying its lateral relevance (rel y,i) by its longitudinal relevance (rel x,i).Environment evaluation system according to one of the preceding patent claims, characterized in that the relevance values (rel x,i, rel y,i) are determined only after a transformation of curved course profiles to a rectilinear course profile carried out on the basis of course prediction.Environment evaluation system according to one of the preceding patent claims, characterized in that it is used in a longitudinal control system for distance control and / or collision avoidance, drive and / or brake actuators being controlled by the longitudinal control system in order to maintain a predefined distance or a predefined speed from the target object selected in each case by the environment evaluation system according to the invention.

Images