Method for distinguishing between real obstacles and apparent obstacles in a driver assistance system for motor vehicles

Abstract

A method for distinguishing between real obstacles and apparent obstacles in a driver assistance system for motor vehicles having a positioning system (18) for determining the vehicle's own location and a sensor (12) for detecting objects in the vehicle's surroundings, in which method location information for objects detected as apparent obstacles is stored in a database (22, 24), and the driver assistance system, when it detects a stationary object at a specific location, queries the database (22, 24) to determine whether an apparent obstacle is stored for this location, characterized in that after the database query, a verification algorithm is carried out with the aim of verifying the detected object as a real obstacle based on the data from the sensor (12) and / or the data from additional sensors (14), that the driver assistance system initiates the storage of the location information of a stationary object detected by the sensor as an apparent obstacle,if the location of this object was not yet saved and the verification as a real obstacle failed.

Description

The invention relates to a method for distinguishing between real obstacles and false obstacles in a driver assistance system for motor vehicles, which comprise a locating system for determining the own location and a sensor for detecting objects in the environment of the vehicle, in which method location information for objects detected as false obstacles is stored in a database and the driver assistance system, when it detects a stationary object at a specific location, queries in the database whether a false obstacle is stored for this location.Electronic driver assistance systems are known which assist the driver in driving the vehicle on the basis of information which is provided in particular by a radar sensor, for example by outputting a warning in the event of an acute risk of an accident or automatically triggering an intervention in the form of an emergency braking operation.In order that traffic safety is increased by such driver assistance systems, it is important that incorrect interventions, such as incorrect warnings or unnecessary braking interventions-and thus unexpected for the following traffic-are avoided as far as possible. This presupposes that the system is able, on the basis of the available information, to reliably distinguish between real obstacles which require a warning or an intervention and false obstacles which the radar sensor recognizes, but which do not represent a real obstacle. Examples of such false obstacles are, for example, gulli lids or channel lids made of metal, cow gates and the like, which are located by the radar sensor because of their high reflectivity for radar waves, but do not represent any actual obstacles since they can be traveled over by the vehicle without any problem. Due to their limited angular resolution capability, the usual radar sensors are not capable of reliably estimating the dimensions, in particular the height, and the exact position of the located object in such a way that these objects can be distinguished from real obstacles such as stationary vehicles and the like.By comparing the relative speed of a located object with the speed of the own vehicle, the driver assistance system is able to distinguish between objects that are absolutely (relative to the roadway) and moving objects. Moving objects on the roadway whose relative speed is negative (i.e., which approach one another) can generally be assumed to be genuine obstacles. In contrast, in the case of stationary radar targets, the distinction is difficult. Although evaluation algorithms have been developed that allow additional plausibility checking of the obstacles, these algorithms do not operate reliably in all situations.It is also known to use the data of additional sensors for plausibility checking or verifying obstacles, for example the data of a video camera and an associated image processing system. However, these more complex evaluation methods require considerable computing power or computing time.A method according to the preamble of claim 1 is known from JP 2010-72 947 A. The use of databases for classifying stationary objects is known from DE 103 35 898 A1 and US 2010 / 0 152 967 A1.The object of the invention is to provide a method which allows simpler, faster and more reliable differentiation between real obstacles and false obstacles. This object is achieved in that after the database query a verification algorithm is carried out with the aim of verifying the detected object as a genuine obstacle on the basis of the data of the sensor and / or the data of additional sensors, in that the driver assistance system initiates the storage of the location information of a stationary object detected by the sensor as a false obstacle if the location of this object has not yet been stored and the verification as a genuine obstacle has failed.The invention makes use of the fact that most motor vehicles equipped with a driver assistance system today also have a positioning system, for example a GPS navigation system, which allows the determination of the own location of the vehicle. With the aid of this locating system, it is also possible to determine the locations of the objects detected by the sensor at the current time so precisely that they can be matched to the location information stored in the database. If the query in the database reveals that a false obstacle is already stored for the relevant location, the currently located stationary object can be reliably qualified as a false obstacle and a miswarranted or misinteraction can be avoided. The request in the database is carried out before further processing steps are carried out for a closer plausibility check or verification of the detected stationary radar target. If the query reveals that the object is a false obstacle, additional processing steps that are costly in terms of computing can then be saved. If, however, the result of the query is that no false obstacle is yet stored for the relevant location, an attempt is made to verify the probable obstacle as a genuine obstacle using known methods. If it is found that this is a false obstacle, the database is updated by the located object being included in the database as a false obstacle.Advantageous refinements and refinements of the invention are specified in the dependent claims.It is also possible to search the database virtually continuously during the trip for false obstacles which are stored along the current travel route, so that the object, when it is located by the sensor, can be qualified immediately as a false obstacle. In this case, the method can also be used to detect any omission of the (radar) sensor and / or to check the accuracy of the positioning system.The database in which the false obstacles are stored may be located on board the vehicle equipped with the driver assistance system. In this case, new false obstacles can be stored whenever the vehicle first travels on the route on which the false obstacle is located. The sensor will then locate the false obstacle and either based on a failed verification or at the latest when the putative obstacle is then traveled over by the vehicle, the driver assistance system can recognize that it is a false obstacle. In this way, the driver assistance system thus "learns" the false obstacles present on any routes, so that these will no longer lead to false warnings or false interventions in the future. A false obstacle can of course also be stored if the driver assistance system has actually triggered a false warning or a false intervention and this has been corrected by active intervention of the driver.In another embodiment, the database is not located on board the vehicle, but on a server which communicates with the driver assistance system in the vehicle through a wireless communication network (mobile radio with Internet access, WLAN or the like). In this case, the location data of false obstacles can also be available in the database, which location data have been recognized and reported by other vehicles, so that substantially more complete information about the false obstacles is available for all vehicles involved and can already be used by an individual vehicle when this vehicle first travels the relevant route.A combination of the two variants described above is particularly advantageous, i.e. a database on a fixed server, which communicates with local databases on board the vehicles involved. The less extensive database on board the vehicle can then be updated from time to time depending on the current location of this vehicle, for example whenever a suitable data connection to the server exists.The method is not limited to radar sensors in the actual sense, but can also be applied analogously to lidar systems, for example.An exemplary embodiment is explained in more detail below with reference to the drawings.The following are shown: FIG. 1 shows a block diagram of a driver assistance system for motor vehicles; and FIG. 2 is a flow chart for explaining the operation of the driver assistance system in detecting obstacles.The driver assistance system shown in FIG. 1 has an electronic control unit 10 having a processor or a plurality of processors, which receives and evaluates data from a radar sensor 12 and / or a video camera 14. Radar sensor 12 is, for example, an FMCW radar installed in the front of the vehicle, which is used to measure the distances and relative speeds of vehicles traveling ahead and other obstacles on the roadway. The control device 10 also receives information about the intrinsic speed of the vehicle equipped with the driver assistance system from a vehicle speed sensor, not shown. If the relative speed of an object located by radar sensor 12 corresponds in magnitude to the own speed, it can thus be determined in the control unit that the located object is a stationary object, for example a traffic sign or guardrail post at the edge of the roadway or also an object such as a channel cover or a stationary vehicle on the roadway.In addition to an ACC function (Adaptive Cruise Control), in which the speed of the host vehicle is controlled as a function of the measured distance from the vehicle traveling ahead, the driver assistance system described here has a further assistance function, which consists in outputting a warning message to the driver or actively initiating emergency braking if a collision with an obstacle located by radar sensor 12 is imminent. For this purpose, the assistance system also has an output unit 16 which is capable of outputting a warning to the driver via a human / machine interface with a display and / or a loudspeaker and which optionally also permits an active intervention in the braking system of the vehicle.The video camera 14 is installed in the vehicle so as to monitor the front of the vehicle. The image information supplied by the video camera is evaluated in the control device 10 by image processing software and can be used, for example, in the context of a lane keeping assistance function. In addition, the data allows the video camera 14 to verify the data supplied by the radar sensor 12 if the latter has located a presumed obstacle. If, for example, the radar sensor 12 reports a stationary object on the roadway and indicates the approximate location of this object on the basis of the distance measurement and its angular resolution capability, the object present at this location can be qualified in more detail by evaluating the video image, and it is possible in particular to distinguish whether it is a genuine obstacle such as a parked vehicle or else a false obstacle such as a channel cover, which does trigger a radar echo but can be traveled over by the own vehicle without problems.This control device 10 also communicates with a GPS-based positioning system (navigation system) 18 and an interface 20 to a mobile data network and with a local database 22. The control device 10 then causes this location information to be stored in the database 22. If the vehicle then next travels the same route and the radar sensor 12 relocates the false obstacle, the complicated verification based on the data from the video camera 14 can be omitted and it is determined, merely based on the entry in the database 22, that a radar target is located at this location which is not a genuine obstacle. If a network connection exists via the interface 20, then the entry stored in the local database 22 is additionally transmitted to an external database 24, which is stored on a fixed server. In this way, the information that a radar target that is not a true obstacle is to be expected at the relevant location is made available not only for the host vehicle but also for other vehicles.FIG. 2 illustrates a typical method sequence on the basis of a flow chart.In step S 1, control unit 10 determines, on the basis of the data supplied by radar sensor 12, whether a stationary target has been detected in the driving path (i.e., in the predicted roadway). As long as this is not the case, step S 1 is repeated periodically at short time intervals. If a stationary target is detected, a query is made in the databases 22 and 24 in step S 2 as to whether a false obstacle is stored for the location at which the vehicle is currently located. Alternatively, instead of the location of the own vehicle, the location of the false obstacle itself may also be stored and queried, which may be determined in more detail on the basis of the distance and direction data of the radar sensor.If the database query reveals that a false obstacle is already stored for this location, the target currently located by the radar sensor is identified with the false obstacle in step S 3 and a return to step S 1 takes place.If the radar target could not be identified as a false obstacle in step S 3, a verification algorithm is carried out in step S 4 by the control device, in which a test is carried out to verify the recognized radar target, which now has to be viewed as a potential obstacle, as a genuine obstacle on the basis of supplementary information from the radar sensor 12 and / or the video camera 14. If this verification is successful and if it is also the result of the distance and speed data for this obstacle measured by the radar sensor that a collision is to be feared, a warning is output to the driver via the output unit 16. Depending on urgency, emergency braking can also be triggered directly by intervention in the brake system.If the presumed obstacle cannot be verified in step S4, step S5 is skipped.In both cases, a check is then made in step S 6 as to whether the obstacle has been falsed. If the verification in step S 4 was not successful, the false verification in step S 6 may consist, for example, in the presumed obstacle being traveled over by the own vehicle. It is thus established that it has not dealt with a genuine obstacle. Likewise, the obstacle is falified if a warning has been issued to the driver in step S 5, but the driver has ignored this warning and then the obstacle is traveled over. Furthermore, the obstacle is falified in step S 6 if emergency braking has been actively triggered in step S 5, but the driver has actively stopped this braking process OR falified if a camera recognizes that no relevant target object is present (e.g. gullicap).If the obstacle has been falified in step S 6, an entry is made in the databases 22 and 24 in step S 7. Depending on the specific embodiment, either the location coordinates of the located false obstacle are stored or the location coordinates that the host vehicle had at the point in time at which the radar target was initially located by the radar sensor.In a modified embodiment, in addition to the location data, further information about the obstacle can also be stored. For example, by evaluating the image supplied by the video camera, the type of false obstacle can be specified in more detail, so that the relevant obstacle class can then also be stored in the databases in addition to the location of the obstacle.

Claims

Method for distinguishing between real obstacles and false obstacles in a driver assistance system for motor vehicles, which comprise a locating system (18) for determining the own location and a sensor (12) for detecting objects in the environment of the vehicle, in which method location information for objects detected as false obstacles is stored in a database (22, 24), and the driver assistance system, when it detects a stationary object at a specific location, queries in the database (22, 24) whether a false obstacle is stored for this location, characterized in that after the database query a verification algorithm is carried out with the aim of verifying the detected object as a real obstacle on the basis of the data of the sensor (12) and / or the data of additional sensors (14), the driver assistance system causes the storage of the location indication of a stationary object detected by the sensor as a false obstacle if the location of this object has not yet been stored and the verification as a genuine obstacle has failed.Method according to Claim 1, in which the location information of the false obstacles is stored in a local database (22) on board the own vehicle.Method according to Claim 1 or 2, in which the location indication of the detected stationary object is stored as a false obstacle if the host vehicle passes over this object and the location of this object has not yet been stored.Method according to one of the preceding claims, in which the location indication of the detected stationary object is stored as a false obstacle if the driver assistance system had initiated an intervention in the braking system of the vehicle and the driver breaks off this intervention and the location of this object was not yet stored.Method according to one of the preceding claims, in which the location information of the false obstacles is stored in an external database (24) with which the driver assistance system communicates via an interface (20) of a wireless data network.

Images