EP3526624A1

EP3526624A1 - Automated open space identification by means of difference analysis for vehicles

Info

Publication number: EP3526624A1
Application number: EP17825393.6A
Authority: EP
Inventors: Torsten Baier; Christian KLIER; Navid Nourani-Vatani
Original assignee: Siemens Mobility GmbH
Current assignee: Siemens Mobility GmbH
Priority date: 2016-12-06
Filing date: 2017-12-04
Publication date: 2019-08-21
Also published as: WO2018104191A1; DE102016224212A1

Abstract

A method for detecting a collision obstacle (KH) is described. In the method, sensor data (SD) are captured from an area (Bu) surrounding a vehicle (11) in a location-dependent manner with the aid of a sensor unit (12). A grid-based occupancy map (GBK) is then generated on the basis of the captured sensor data (SD). A grid-based reference occupancy map (GRBK) corresponding to the grid-based occupancy map (GBK) is also determined on the basis of the location. A check is then carried out in order to determine whether a collision obstacle (KH) occurs. This checking step is carried out on the basis of a comparison of the current grid-based occupancy map (GBK) with the determined grid-based reference occupancy map (GRBK). An automated anti-collision system (50) is also described.

Description

description

Automated free space detection using differential analysis for vehicles

The invention relates to a method for detecting a collision obstacle. In addition, the invention relates to a automa ^¬ tisiertes anti-collision system.

Driver assistance systems and Fahrzeugautomatisierungstechnolo ^¬ technologies for road vehicles and rail vehicles require in ^¬ mation relating to the environment in which they move to capture the current traffic situation and pretieren to international. For this purpose, with the help of sensor systems a

Illustration of the environment and a generation of editable data made. On the basis of an evaluation of the data, objects are perceived and an attempt is made to determine their relevance to their own locomotion on the basis of various assessment methods and classifications. By means of specialized sensors or evaluation we perform an explicit recognition of objects. These objects are then placed in Rela ^¬ tion for their own position or locomotion and evaluated for their relevance. In order to enable such object detection, the situation must be described in advance for the relevant sensor system. This can be done eg by mathematical models and classifiers. A wide ^¬ rer approach that even without a specific CLASSIFICA ^¬ tion of detected objects can get along is called a grid-based assignment card, in English as "Occupancy grid map", can be realized. This all detected by the sensor objects are a surrounding grid Occupied grid cells are then to be evaluated with regard to their relevance to their own movement laws.

To capture the environment of a vehicle mainly LiDAR systems, radar systems and / or cameras are used. LiDAR and radar can be used to determine distances to detected objects in the vicinity of the vehicle. There are two reasons that the most can not serve as the basis for an object recognition using the Senso ^¬ ren captured raw data directly: The sensor data are mostly affected by noise, so that the position of de- tektierten object can not be determined accurately; and usually the raw data are so extensive that the required computing capacities are too expensive or could not be accommodated in a vehicle. For this reason, as part of the "Occupancy grid map" -

Method an area to be traveled, for example, a Be ^¬ zirk, a state or the entire earth's surface, divided into small grid squares or grid areas, for example, with a size of 0.1 m * 0.1 m, and thus a grid-based occupancy map created. Each sensor event, which can be assigned to one of the mentioned screen areas, leads to an increase of the respective screen area to ^¬ ordered count. Total is thus obtained for each Ras ^¬ terfläche a count, which provides information about whether and how often each raster area using the

Sensor was detected and whether this grid area is occupied by an object.

Since the sensor is typically impaired by the effects of noise, it makes sense, an occupancy of a raster surface is not in a single hit, so a count value equal to 1, but only when it exceeds the count value of a predetermined threshold, for example the value 10 provide festzu ^¬ , Depending on the count values, different probability values for an occupancy of a grid area with an object can also be defined.

US 2003/0 004 644 A1 describes methods for identifying stationary objects.

In DE 10 2011 086 433 AI a maneuvering method for a vehicle is described, in which a surrounding environment representing surrounding data are stored so that they a restart of the system are available. Upon a restart, the environment of the vehicle is detected again and carried out a comparison of the recorded environment data with the stored environment data.

DE 10 2011 100 820 A1 describes a method for detecting an obstacle in the surroundings of a vehicle during the stance phase of the vehicle. US 2011/0 040 481 A1 describes a collision warning system for a vehicle which compares GPS coordinates of possible collision objects with position data of the vehicle. The approaches described all share the problem that objects in the environment must be explicitly recognized. However, this is never completely reliable mög ^¬ Lich. In addition, increases in computational effort for the conventional methods.

There is therefore the problem, a method and entspre ^¬ sponding device for obstacle detection to develop, requiring little effort and are reliable. This object is achieved by a method for detecting a collision obstacle according to claim 1 and by an automated anti-collision system according to claim 12. In the method according to the invention for detecting a collision obstacle, sensor data from a surrounding area of a vehicle are initially detected with the aid of a sensor unit, depending on location. The surrounding area may include, for example, an area located in front of the vehicle. This area located in front of the vehicle is intended to include the Be ^¬ reaching the direction of travel in any case. In addition, the surrounding area can also include sections located on the right and left of the direction of travel and on the side of the vehicle. rich ones. Furthermore, the surrounding area can also have areas located on the left and right behind the vehicle and also an area directly behind the vehicle. For example, a moving collision object behind the vehicle may start to overtake or be in collision with a turning maneuver of the vehicle.

As a vehicle can, for example, a stretch-guided vehicle, in particular a road vehicle or a rail vehicle ver ^¬ applies to. The sensor unit for sensor-based detection of the surroundings of the vehicle can be arranged, for example, on the respective vehicle. However, the sensor data can also be determined with the aid of stationary trackside arranged sensors which are integrated in a trackside surveil ^¬ monitoring system that can communicate with the vehicle.

Based on the acquired sensor data, a grid-based occupancy map is generated. The grid-based occupancy map includes information about which grid area is occupied by an object and which not. As the grating-based, Be ^¬ legungskarte only occupied and unoccupied grid areas has ^¬, the data amount of grid-based allocation map is quite small. Therefore, such a grid-based occupancy ^¬ card can be quickly created and also involved in a real-time process.

Furthermore, one of the grid-based allocation map is Corresponding reference grating-based allocation map determined in depen ^¬ dependence on the location. That is, a grid-based reference occupancy map is searched whose associated reference sensor data is assigned to the same locations where the currently acquired sensor data was acquired. Because only in this case, the data affect the same location area or the same recording area and can be the two records also compare. In the simplest case it is sufficient if the reference sensor data for the grid-based reflection Reference occupancy map were recorded at the same point of a route as the currently recorded sensor data. In this example it is assumed in a vehicle-side arrangement of the sensors that the vehicle or the sensors of the vehicle at the actual recording of the sensor data are ge ^¬ nauso oriented, as in the recording of reference sensor data. In an improved embodiment, deviations of the orientation of the vehicle or of the sensors, if vehicle-side sensors are present, are taken into account in the comparison of the position of the sensor data with the position of the reference sensor data.

The grid-based reference assignment card may be he ^¬ witness, for example, during a test drive along a predetermined route, wherein the environment is abge ^¬ samples with the aid of sensors and the grid-based reference assignment map ( "occupancy grid map") with the described count values is created Since, so that visibility of the grid area is occupied by an object and which are not. the grating-based allocation map so obtained may be used as a grating-based Refe ^¬ rence assignment card, is carried out with at a RETRY ^¬ th traveling the route a comparison. comprises grid-based reference assignment map only occupied and unoccupied grid spaces, the data amount of Re ^¬ conference assignment map is quite small. Therefore, such Re ^¬ conference assignment map can be quickly created and also involved in a real-time process.

In addition, it is checked whether a collision obstacle occurs.

That is, the lattice-based occupancy map successively created on re-driving the route, or their individual rasters, are compared with the corresponding raster areas of the reference occupancy map. If free grid areas in the reference occupancy map are suddenly occupied in the currently created occupancy map, this may indicate the occurrence of an obstacle. A simple sub ^¬ traction of two areas of the map can be used to detect changes. Since during the test drive the route is drive, it is also known which grid areas are located on the trajectory of the vehicle. Is an occupied grid ^¬ area on the route, this can be regarded as a threat to a colli ^¬ sion.

The test is thus based on a comparison of the grid-based occupancy map based on the acquired location-dependent sensor data with the determined grid-based reference occupancy map. In the comparison, differences between the grid-based reference occupancy map and the current grid-based occupancy map are interpreted as indicative of a potential collision obstacle.

Advantageously, the recorded sensor data need not necessarily be interpreted. It is therefore necessary by the sensor no explicit Whether ^¬ jekterkennung. It is sufficient to compare the sensor data with reference data in order to obtain clues for an existing collision obstacle. The Sen ^¬ sordaten and the reference data is transformed for comparison in grid-based assignment cards. Alternatively or additionally, the sensor data and the reference data can also be further evaluated and processed, for example, interpreted in ^¬ and include semantic additional information. Changes in the environment may be detected by determining differences between the sensor data and the reference sensor data. Thus, the required computational effort for the evaluation of the acquired sensor data is reduced. The automated anti-collision system of the invention for a vehicle includes a sensor unit for location-dependent He ^¬ take sensor data from an area surrounding a vehicle. Furthermore, automatic ^¬ catalyzed anti-collision system of the invention comprises an occupancy map generation unit for generating a grating-based allocation map based on the acquired sensor data. Part of the automated anti-collision system according to the invention is also a reference occupancy map determination unit for determining one of grid-based occupancy map corresponding lattice-based reference occupancy map depending on location. The dung OF INVENTION ^¬ modern anti-collision system further comprises a checking unit for checking whether an obstacle collision occurs, WO at the test is carried out on the basis of a comparison of gitterbasier ^¬ th assignment map at the determined lattice-based reference assignment map. The automatic ^¬ catalyzed anti-collision system according to the invention divides obstacle the advantages of the method for detection of a collision.

Some essential components of the automated anti-collision system according to the invention can be in the form of software components ^¬. This concerns in particular the reference sensor data set acquisition unit and the Prüfein ^¬ ness.

In principle, however, these components can also be partly realized, in particular in the case of particularly fast calculations, in the form of software-supported hardware, for example FPGAs or the like.

A largely software implementation has the advantage of being already used up control system of vehicles that take eg automated assistance systems to ^¬, nachge by a software update ^¬ can be upgraded easily to work in the inventive manner. In this respect, the object is also achieved by a computer program product which can be loaded directly into a memory of a control device, with program code sections in order to carry out all steps of the method according to the invention when the program is executed in the control device. Such a computer program product can next to the computer program ^¬ any additional ingredients such as egg ^¬ ne documentation and / or additional components and hardware components, such as hardware keys (dongles, etc.) to use the software.

A computer-readable medium, for example a memory stick, a hard disk or another portable or permanently installed data carrier can serve for transport to the control device and / or for storage on or in the control device, on which program sections of the computer program which can be read and executed by a computer unit are stored. The computer unit may be for example a purpose or more cooperating micro ^¬ processors or the like. A wireless transmission of the computer program is possible. The dependent claims and the following description each contain particularly advantageous embodiments and further developments of the invention. The claims of claim category may also be further formed in accordance with depen ^¬ Gigen claims of a different category of claim and the description parts in particular. In addition, in the context of the invention, the various features of different embodiments and claims can also be combined to new Ausfüh ^¬ tion examples. In a variant of the method according to the invention, at least one of the following sensor types is used to acquire the sensor data:

- radar,

- LiDAR,

- a laser scanner,

- a camera,

- a camera system.

Radar can be used, for example, in poor visibility conditions to get an idea of the surroundings of a vehicle. Laser-based sensor systems are suitable for a particularly accurate ^{measurement of} an environment of a vehicle. Cameras allow one high-resolution scanning of the environment with a relatively low technical effort.

The geographical position of the vehicle is particularly preferred ermit ^¬ telt wherein the position-dependent detection of the sensor data. The knowledge of the geographical position of the vehicle already allows a basic local assignment of the detected sensor data to an absolute position. As already mentioned, the sensors can be aligned differently when taking the reference sensor data, and in the case of a vehicle-side sensor arrangement, the vehicle can also be oriented differently. For additional clarification, a relative position of objects assigned to the sensor data to the vehicle can therefore additionally be determined. The relative position can be given eg by an orientation of the sensors as well as the distance between sensor and object.

Finally, a location associated with the sensor data is determined on the basis of the determined geographical position of the vehicle and the determined relative position.

In the most favorable case, a kind of three-dimensional image recording takes place during the detection of the sensor data, so that individual pixels can be assigned relative positions to the sensor unit. To be able to perform a comparison with a reference image, knowing the position of the vehicle and the orientation angle of the Sen ^¬ sensors is sufficient already. A knowledge of the position and, where appropriate, ^¬ orientation of the vehicle, the sensors are rigidly or trackside angeord ^¬ net, so is sufficient. By means of said additional information, the choice of a suitable reference sensor data set or a suitable grid-based reference occupancy map can be facilitated. Optionally, sensor data recorded from different directions can also be correspondingly corrected or converted in order to be better comparable with a reference sensor data set or a grid-based reference occupancy map. In a preferred embodiment of the method according to the invention, a database is searched for the determination of the grid-based reference assignment card, which includes a multi ^¬ number of grid-based reference assignment cards. A database can be designed, for example, as a vehicle-side non-volatile data memory. It may also be designed as a stationary data storage station that communicates with one or more automated anti-collision systems. From the database, the grid-based reference occupancy map is taken, which comprises the positions assigned to the currently detected sensor data. Advantageously, the removed from the database reference Bele ^¬ supply map of the grid-based assignment map generated from the current sensor data largely corresponds to the case that has not changed at the same position since the recording of the sensor data for the grid-based reference assignment map the scenario. Differences between the grid-based occupancy map based on the currently recorded sensor data and the lattice-based reference occupancy map thus provide indications of an occurrence of possible collision obstacles.

In a preferred variant of the inventive method for the case that the positions of currently detected sensor data are not precisely known, the proper grid based reference assignment map on the basis of a Ver ^¬ equalization of several grid-based reference assignment cards with the currently acquired sensor data determined grid-based occupancy map. In this variant, a choice of a suitable grid-based reference occupancy map is then made based on a similarity between the current grid-based occupancy map and the respective grid-based occupancy maps. Consequently, an appropriate grid-based Refe ^¬ ence allocation map can be selected from the database even when only imprecisely known spatial dependence of the currently recorded sensor data. In one embodiment of the method according to the invention, when checking whether a collision obstacle occurs, one or more collision obstacle candidates are determined at positions at which there is a difference between the currently generated grid-based occupancy map and the determined grid-based reference occupancy map. That is, it ^¬ the determined based on the sensor data and the grid-based allocation map produced therefrom spatial positions or Kartenpo ^¬ sitions in which differences between the aktu- eil grid-based allocation map generated and the grid-based reference assignment map available. The Posi ^¬ these functions assigned to map areas are classified as a collision obstacle candidates and subjected to detailed analysis. Advantageously, in this variant, only a few occupancy card areas must be examined more closely, so that the computational effort for the evaluation of the sensor data or the grid-based occupancy maps based thereon is greatly reduced. The more detailed analysis can then take place, for example, on the basis of high-resolution data, such as image data, in the identified candidate areas.

In a particularly effective variant of the method according to the invention, the grid-based reference occupancy map is determined by location-dependent detection of sensor data along an obstacle-free route. Advantageously, in this variant, an assignment of the reference sensor data to the positions at which the reference sensor data were recorded.

Thus recorded late sensor data can be retrieved targeted which include the positio ^¬ nen the later recorded sensor data appropriate grid-based reference assignment cards, without a comparison must be made for the selection of a suitable grid-based reference assignment card. Furthermore, the recording of the sensor data for the grid-based reference occupancy ^¬ map takes place in the absence of collision obstacles, so that Un- Differences in lattice-based occupancy maps based on later recorded sensor data can be categorized as potential collision obstacles to lattice-based occupancy maps.

The aufgenom ^¬ menen for the grating-based reference assignment map sensor data can be obtained with a vehicle along the obstacle-free route when performing a test run, wherein the sensor data is collected depending on the location of an area surrounding the vehicle using a vehicle-mounted sensor unit. This variant can be used particularly advantageously if the surrounding area is also detected by vehicle-side sensors during later runs in regular operation. Thus, the reference sensor data with the sensor data recorded later or their assigned occupancy maps are easily comparable due to the same recording perspective.

In one embodiment of the method according to the invention the advertising to different points in time he holds sensor data from the surrounding area of the vehicle with the aid of a sensor unit ^¬. Furthermore, a grating-based occupancy ^¬ map based on the acquired sensor data for the different points in time is created in each case. In addition, a grid-based assignment to map corresponding grid-based Refe ^¬ ence assignment card is determined or selected from a database forth ^¬. For example, it is possible to use the knowledge of a position of the vehicle taken at different times. Finally, it is checked whether a collision obstacle occurs. The test is performed on the basis of the equalization Ver ^¬ determined for different time points gitterba ^¬ overbased occupancy cards with the determined reference grid-based allocation map. Information about the dynamics of an environment scenario of a vehicle are advantageously included in this Vari ^¬ ante so that an assessment be ^¬ züglich can be specified to a collision of a moving vehicle with a possibly even moving obstacle. In a variant of the method, a ge ^¬ underestimated path as a function of time is determined when examining whether a collision obstacle occurs. That is, it is ermit ^¬ telt future trajectory of the vehicle. The future Trajektiorie can, for example using a route based on a road or ei ^¬ nem track history, and based on a current direction and a current steering angle are determined. Then, a position of a collision obstacle candidate is compared with the estimated driving route. In this variant, the dynamic behavior of the vehicle is included in the ^{estimation of} a collision. In this variant, determination of collision obstacles can advantageously be carried out more accurately than without consideration of the future travel path of the vehicle.

Preferably, in the test of whether a collision obstacle occurs, a time-dependent trajectory of a collision obstacle candidate is determined and it becomes the estimated

Track as a function of time compared with the determined Tra ^¬ jectorie the collision obstacle candidate.

In this variant, the dynamic behavior of the collision obstacle candidates is also included in the determination of collision obstacles, so that a more accurate determination, in particular of moving collision obstacles can be made. In one embodiment of the method according to the invention of a collision obstacle candidates when considering whether a collision obstacle occurs, whether ^¬ jekteigenschaften determined. Advantageously, on the basis of the object properties, an evaluation of potential collision obstacles with regard to their dangerousness can take place. This review can be used to ANPAS ^¬ sung a response to a particular obstacle to ^¬. In the comparison method used, the risk assessment is based on the difference information between the setpoint and actual sensor signal or sensor data and reference sensor data or grid-based occupancy maps generated therefrom. In order to determine a rating or quality, in one embodiment of the method according to the invention a suitable combination of different parameters determined by comparison is used: eg the temporal development of the difference between the occupancy maps, a geometric value (location, extent ...), the own Trajectory, the trajectory of the supposed obstacle, location information, etc. These values are used to determine a confidence interval. The determined quality or the determined confidence ^{interval are} then used to determine whether a difference flows directly into a hazard assessment or is subjected to a further classification / assessment step.

In a specific embodiment of the method according to the invention, a position of the vehicle is determined, for example by self-localization of the vehicle. For this, e.g. at least one of the following technologies:

- a global civil satellite navigation system,

- real-time kinematics,

- mobile radio-based procedures,

- local transmit / receive systems,

- Sensor-based technologies.

Real-time kinematics-based methods also operate using satellite signals from global civil satellite navigation systems. This satellite signals are received simultaneously. On the basis of phase-based correction methods, greater accuracies are achieved than with simple satellite navigation.

Mobile networks are usually divided into cells. Mobi le ^¬ transmitting / receiving devices can be a cell in which they happen to be, assign and locate that. Local transmission / reception systems include, for example, RFID systems, beacon systems and others on short-range com munication ^¬ based devices.

Sensor-based technologies include imaging systems or systems equipped with additional emission units, such as LiDAR or radar. These technologies can also be used in combination to achieve increased redundancy in Selbstlokali ^¬ tion and to make the collision warning even more robust. The invention will be explained in more detail below with reference to the beige ^¬ added figures using ^exemplary embodiments. Show it:

1 is a schematic representation of a path on which a vehicle moves with a collision warning system according to an embodiment of the invention,

2 shows a flow chart illustrating the acquisition of reference sensor data,

3 shows a flowchart illustrating a method of detecting a collision obstacle according to a ^¬ execution example of the invention,

FIG 4 is a flow diagram showing a comparison step of the process illustrated in FIG 3 the method in detail, FIG 5 is a schematic representation of a Kollisionswarnsys ^¬ tems according to an embodiment of the invention.

In Figure 1, a railway line section 10 is shown on which a locomotive 11 moves. The locomotive 11 comprises an anti-collision system designed as a collision warning system 50 with a sensor unit 12, with which it scans a surrounding area Bu lying in front of it, and a collision warning device 50a serving as an evaluation device. The sensor data SD determined during the sampling are stored in the Collision warning system 50 evaluated, as described in connection with FIG 4. In the scenario shown in FIG. 1, an object 0 is located at the edge of the environmental region Bu scanned by the sensor unit 12. Reaches the object 0 in the sampled area surrounding Bu, it is detected by the collision warning system 50 and it is output a warning to the Fah ^¬ rer the locomotive 11 that an obstacle on the route of the locomotive 11 is located. FIG. 2 shows a flowchart 200 with which a

Test drive, in which a section without existing col ^¬ lisionshindernisse is traversed, is illustrated. In this test drive is a locomotive 11 (see FIG 1) on a track section 10 and departs a predetermined route. In the step 2.1 a Posi tion ^¬ P of the locomotive is first determined on the track. The position P may for example be read with reference to markings on the side of the track, or are determined by means of another Selbstlokalisie ^¬ approximation method. In the step 2. II sensor data SD are then detected at an ascertained position P from an environmental region B 0 of the locomotive 11 situated in front of the locomotive 11 with the aid of a sensor unit 12. An ^¬ closing in the step 2. III generated based on the acquired data SD as well as the determined position P of the locomotive 11, a grating-based reference assignment map GRBK and stored in a database DB. In step 2 IV, it is determined whether the engine 11 has reached the end of the EoT to he ^¬ comprehensive route section 10 and at the end point of the test ^¬ ride. If the end EoT is not yet reached, which is marked with "n" in FIG

Etc returned Step 2.1, and again a position P of the saw ^¬ leaders locomotive 11 determines .. If the end EOT of the route section is reached, which is shown in FIG 2 with "j" marked in ^¬ distinguished, it proceeds to step 2.V and Test drive finished.

FIG. 3 shows a flow chart with which a method for detecting a collision obstacle in accordance with an embodiment of the invention is shown. embodiment of the invention is illustrated. In step 3.1, a position P of the locomotive is first determined on the track. Subsequently, in the step 3. II at the determined position P sensor data SD from a located in front of the locomotive 11 surrounding area Bu of

Locomotive 11 detected by means of a sensor unit 12. After ^¬ following, in step 3. III based on the sensor ^¬ data SD a current grid-based occupancy card GBK he ^¬ averages. Further, in the step 3 is IV searches a database DB to a respective grating-based reference Bele ^¬ supply card GRBK. In step 3.V, it is then determined by comparing the current grid-based occupancy map GBK with the grid-based reference occupancy map GRBK whether one or more collision obstacles KH are present. In the event that no collision obstacles KH were determined, which is characterized in FIG 3, with "n", the flow returns to step 3.1 and the process wei ^¬ tergeführt. In contrast, if one or more collision obstacles KH determined, which in Figure 3 with the " j is characterized ", it proceeds to step 3. VI and are output to the type and position of the collision obstacle KH Informa ^¬ functions, and output a warning W.

FIG. 4 shows a flowchart 400, with which the comparison step 3.V is illustrated in greater detail. In a substep 3.Va first Kollisionshindernis- candidates K-KH be determined on the basis of comparison between the aktuel ^¬ len grid-based allocation map GBK and a gitterba ^¬ overbased reference assignment map GRBK over several time points. Subsequently, in sub-step 3.Vb trajectories T (K-KH) of the collision obstacle candidates K-KH are determined. Furthermore, in the sub-step 3.Vc, a travel path of the locomotive 10 is determined. Finally, in sub-step 3.Vd, a comparison is made between the trajectories T (K-KH) of the collision obstacle candidates K-KH and the determined travel path FW. In the event that at least one of the trajectories T (K-KH) of one of the collision obstacle candidates K-KH meets with the determined travel path FW, which is shown in FIG "J" is marked, is transferred to the step 3. VI and there are collision obstacles KH and a corre sponding W warning ^¬ issued.However, if no collision obstacle KH is determined, which is marked in Figure 4 with the "n", then as also shown in FIG 3, returned to the step 3.1.

In addition, in the event that a collision obstacle has been detected, automated countermeasures, such as e.g. the initiation of a braking maneuver be performed.

In FIG 5 an automated collision warning system 50 is shown ge ^¬ Mäss an embodiment of the invention. The au ^¬ tomatisierte collision warning system 50 may be, for example, part of an au- tomatisierten driver assistance system of a vehicle 11, as shown in Fig. 1 The automated collision ^¬ warning system 50 includes a sensor unit 12 and a Kollisi ^¬ onswarneinrichtung 50a. The sensor unit 12, in this exemplary embodiment a camera system, records sensor data SD from the surroundings of the vehicle and sends the sensor data SD to the collision warning device 50a. The camera system captures an area of the vehicle in front of the vehicle. The collision warning device 50 a comprises a sensor data input interface 51, which receives the sensor data SD from the sensor unit 12. The sensor data SD comprising image data in the ^exemplary embodiment is transmitted via the sensor data input interface 51 to an occupancy-card generation unit 54a. The Bele ^¬ supply map generating unit generates a grid-based loading legungskarte GBK based on the acquired sensor data SD.

The assignment card determination unit 54a receives, via a position determining unit 52 position information P ^¬ züglich be a current position of the vehicle. Further, the occupancy map determination unit 54a receives also, since ^¬ th SA with respect to a orientation of the camera via a sensor SA orientation data acquisition unit 53, by means of which a spatial dependence of the sensor data SD or gitterba ^¬ overbased occupancy map GBK is determined.

The grid-based occupancy map GBK is transmitted to a reference occupancy map determination unit 54b. Is then Da ^¬ tenbank DB retrieved a grid-based reference assignment map GRBK and together with the grid-based allocation map GBK transmitted from the reference assignment map obtaining unit 54b from a to a checking unit 55 based on the grid-based allocation map GBK. The check unit 55 checks on the basis of the received data GBK, GRBK whether a collision obstacle KH occurs, and performs a comparison of the grid-based occupancy map GBK with the determined grid-based reference occupancy map GRBK. In the event that a collision obstacle KH has been determined, the corresponding information KH is output via an output interface 56. In addition, the result KH can also be used to initiate automated countermeasures, such as performing a braking maneuver.

It is finally pointed out once again that this han ^¬ punched in the above-described methods and devices by only preferred embodiments of the invention and that the invention can be varied by the skilled person without departing from the scope of the invention in so far as determined by the claims is. Thus, the method and the device have been explained primarily in connection with the use of rail vehicles. However, said method and collision warning system described are not limited to railroad applications but may also be used in connection with road vehicles or other environments such as off-road off-road or other vehicles such as off-road vehicles or trams. For the sake of completeness, it is also pointed out that the use of indefinite articles does not exclude "a" or "one", that the characteristics in question can also be present multiple times. Similarly, the term "entity" does not exclude that This consists of several components, which may possibly be distributed spatially ^¬ .

Claims

claims

1. A method for detecting a collision obstacle (KH), comprising the steps:

location-dependent detection of sensor data (SD) from a surrounding area (Bu) of a vehicle (11) by means of a sensor unit (12),

Generating a grid-based occupancy map (GBK) on the basis of the acquired sensor data (SD),

Determining a lattice-based occupancy map (GRBK) corresponding to the lattice-based occupancy map (GBK) depending on the location,

Checking for a collision obstacle (KH) based on a comparison of the Grid Based Occupancy Map (GBK) with the determined grid based occupancy map

(GRBK).

2. The method of claim 1, wherein the location-dependent Erfas ^¬ sen of the sensor data (SD) comprises the steps of:

- Determining the geographical position (P) of the vehicle

(11),

Determining a relative position of objects assigned to the sensor data (SD) to the vehicle (11),

- Determining a sensor data (SD) associated location (P) based on the determined geographical position of the vehicle (11) and the determined relative position.

The method of claim 1 or 2, wherein determining the grid-based reference occupancy map (GBK) comprises the steps of:

- searching a database (DB) which comprises a plurality of location-dependent grid-based reference occupancy maps (GRBK),

- Remove the grid-based reference occupancy card

(GRBK) from the database (DB), its assigned position

(P) corresponds to the position (P) of the grid-based occupancy map (GBK) generated by the currently acquired sensor data (SD).

4. The method of claim 3, wherein for the case that the position (P) of the currently detected sensor data (SD) and / or the position (P) of the grid-based reference occupancy maps (GRBK) of the database (DB) are not known exactly , the correct reference grating-based allocation map (GRBK) with the sensor data currently recorded (SD) assigned based on a comparison of several grid-based reference occupancy ^¬ cards (GRBK) grid-based allocation map (GBK) is determined.

5. Method according to one of the preceding claims, wherein when checking whether a collision obstacle (KH) occurs, one or more collision obstacle candidates (K-KH) are determined at positions (P) at which a difference between the actual generated lattice-based occupancy map (GBK) and the determined grid-based reference occupancy map (GRBK) is present.

6. The method according to any preceding claim, wherein the grating-based reference assignment map (GRBK) by ortsab ^¬ hängiges detecting sensor data (SD) along an obstacle-free route (10) is determined.

7. The method according to claim 6, comprising the steps:

Performing a test drive with a vehicle (11) along the obstacle-free route (10),

Location-dependent detection of the sensor data (SD) from a surrounding area (Bu) of the vehicle (11) with the aid of a vehicle-side sensor unit (12),

Create a grid-based reference occupancy map

(GRBK) based on the collected sensor data (SD).

8. The method according to any one of the preceding claims, wherein - at different times sensor data (SD) from the surrounding area (Bu) of the vehicle (11) by means of a Sen ^¬ soreinheit (12) are detected, in each case a grid-based occupancy map (GBK) is created on the basis of the acquired sensor data (SD),

in each case one of the grid-based occupancy map (GBK) ent ^¬ speaking grid-based reference occupancy map (GRBK) is determined,

- The check of whether a collision obstacle (KH) occurs, based on a comparison of he ^¬ averaged time ^¬ averaged grid-based occupancy maps (GBK) with the determined grid-based reference occupancy maps (GRBK) takes place.

9. The method of claim 5, wherein the checking whether a collision obstacle (KH) occurs comprises:

the determination of an estimated travel path (FW) as a function of time,

a comparison of a position of a collision obstacle candidate (K-KH) with the estimated travel path (FW).

10. The method of claim 9, wherein examining whether a collision obstruction (KH) occurs comprises:

- Determining a time-dependent trajectory ((K-KH)) of a collision obstacle candidate (K-KH) and

- comparing the estimated travel path (FW) as a function of time with the determined trajectory ((K-KH)) of the collision obstacle candidate (K-KH).

11. The method according to any one of claims 5 to 10, wherein the checking whether a collision obstacle (KH) occurs, the Ermit ^¬ stuffs of object properties of a candidate Kollisionshindernis- (K-KH) comprises.

12. Automated anti-collision system (50), comprising:

- A sensor unit (12) for location-dependent detection of

Sensor data (SD) from a surrounding area (Bu) of a vehicle (11),

an occupancy map generating unit (54a) for generating a grid-based occupancy map (GBK) on the basis of the acquired sensor data (SD), a reference occupancy map determination unit (54b) for determining a lattice-based occupancy map (GBK) corresponding lattice-based occupancy map (GRBK) depending on the location,

- A test unit (55) for checking whether a collision obstacle (KH) occurs, based on a comparison of gitterba ^¬ sierten assignment card (GBK) with the determined gitterba ^¬ based reference occupancy card (GRBK).

13. Computer program product with a computer program wel ^¬ Ches directly into a memory device of a control device of a vehicle (11) is loadable, cut with Programmab ^¬ to perform all the steps of a method according to one of claims 1 to 11, when the computer program executed in the control device becomes.

14. The computer-readable medium on which-loadable by a computer unit and executable portions are ge ^¬ stores to carry out all steps of a method according to any one of claims 1 to 11 when the Programmab ^¬ sections are executed by the computer unit.