Method, system and computer program product for determining the pose of a mobile unit

Description

[0001] The invention relates to a method, system and computer program product for determining the pose of a mobile unit.

[0002] Localization methods, which can be used in robotics or autonomous driving, for example, serve to precisely determine the pose or position of a vehicle. Landmarks, such as traffic signs or traffic lights, are detected in an image captured by a camera, classified, and then assigned to a known landmark with predetermined position information. From this assignment, the pose or position of the vehicle can then be determined, for example.

[0003] The accuracy of such a method depends sensitively on the number of detected landmarks and the unambiguous assignment of the detected landmarks to the known landmarks. In particular, in areas where few or no landmarks can be detected, or where an unambiguous assignment cannot be made, such a method cannot be implemented or the accuracy of the method is insufficient.

[0004] EP 3 444 693 A1 discloses a method and system for guiding an autonomous vehicle to extract a drivable road area. The method comprises detecting the road area in front of the autonomous vehicle using a plurality of sensors. A plurality of images of the road in front of the autonomous vehicle are then captured with a camera. The detected road area and the plurality of images are mapped and compared (see also US2018 / 131924 A1).

[0005] The object of the present invention is therefore to avoid the disadvantages indicated and to propose a reliable and accurate method, system and computer program product for determining the pose of a mobile unit.

[0006] According to the invention, the object is achieved by the features mentioned in claims 1, 13, and 15. Advantageous variants result from the features mentioned in the subclaims.

[0007] The invention relates to a method for determining the pose of a mobile unit having at least one sensor device configured to capture images of the surroundings of the mobile unit, and a map in which at least one drivable area is marked. The map is stored in an electronic storage unit. In the method, at least one image of the surroundings is captured by means of the at least one sensor device, and at least one collision-free area is calculated on the basis of the at least one captured image of the surroundings by means of an electronic evaluation and control unit. The pose of the mobile unit is determined by means of a comparison of the at least one collision-free area with the at least one drivable area marked on the map by means of the electronic evaluation and control unit.

[0008] The proposed method allows the pose of a mobile unit to be determined particularly reliably and flexibly. The method can be used particularly advantageously in environments of the mobile unit in which few or no landmarks can be detected or for which few or no landmarks are known or marked on a map, for example in small side streets or in remote and little-traveled areas. In particular, the method can be used when the assignment of detected to known landmarks cannot be clearly or reliably determined in a landmark-based localization method. In such challenging situations, a collision-free area can still be reliably and robustly calculated based on the at least one acquired environmental image and the comparison can be carried out with the at least one drivable area marked on the map.

[0009] The mobile unit is a land vehicle, e.g. a motor vehicle or a robot, or a watercraft, e.g. a boat or a ship. The pose of the mobile unit can at least comprise position information. Position information can be formed, for example, using values ​​for two or three coordinates. The pose can also comprise orientation angle information. Orientation angle information can be formed, for example, using values ​​for two or three orientation angles. Preferably, the pose of the mobile unit comprises both position information and orientation angle information. For example, a pose can be represented by a vector with six components. Three components can correspond to coordinates, e.g. geographical coordinates or GPS coordinates (GPS = Global Positioning System). Three further components can correspond to orientation angles.

[0010] The at least one marked drivable area can correspond at least partially to a partial area of ​​a road surface or a water surface. The map preferably has map position information for the at least one marked drivable area. A map position information can be formed with values ​​for two or three coordinates in a map coordinate system. For example, the at least one marked drivable area can be determined by a two-dimensional or three-dimensional set of points or pixels. Preferably, the at least one marked drivable area corresponds to a two-dimensional area or set of points in a two-dimensional or three-dimensional map coordinate system. The map coordinate system can be a global reference system, which can be formed, for example, with geographical coordinates or GPS coordinates.

[0011] The map is a semantic map. Alternatively or additionally, the map includes at least one known landmark with map position information. The at least one landmark can be, for example, a traffic sign, a traffic light, or a fire hydrant. The map position information of a landmark can include values ​​for two or three coordinates of the respective landmark in the map coordinate system.

[0012] The at least one sensor device can comprise a camera, preferably a monocamera. Calibration is particularly preferably performed with the at least one sensor device. By means of the calibration, at least one coordinate, for example, a height relative to the local ground plane of the mobile unit, can be assigned to a point or pixel of the at least one environmental image.

[0013] The local ground plane of the mobile unit may correspond to at least a portion of the drivable surface of the mobile unit. For example, the local ground plane may be determined by the contact points of the wheels of the mobile unit on the drivable surface of the mobile unit.

[0014] The calibration can be performed using a camera coordinate system, wherein the optical axis of the camera as at least one sensor device can correspond to a coordinate axis of the camera coordinate system. Furthermore, a height of the at least one sensor device, the mobile unit, or the optical axis relative to the local ground plane of the mobile unit can be predetermined and / or taken into account during calibration.

[0015] The at least one collision-free area can be a two-dimensional or three-dimensional set of points or pixels. For example, the at least one collision-free area can correspond to a two-dimensional or three-dimensional set of points in the camera coordinate system. The at least one collision-free area can be calculated as a sub-area of ​​the surface captured with the at least one environmental image and drivable by the mobile unit, for example the road surface or the water surface. In particular, the at least one collision-free area can form the area of ​​the drivable surface of the mobile unit that can be reached from the mobile unit in the direction of travel. The at least one collision-free area can in particular be free of obstacles, such as other mobile units, sidewalks or fire hydrants.

[0016] The at least one collision-free region can be extracted from the at least one environmental image as a mask, wherein a bit value can be assigned to each pixel of the at least one environmental image. For example, a pixel with the bit value 1 can be assigned to the collision-free region, while a pixel with the bit value 0 can be assigned to the complement of the collision-free region. The mask defined in this way can be represented by a binary matrix, wherein the rows and columns of the binary matrix can correspond to the row and column arrangement of the pixels of the at least one environmental image. Preferably, the at least one collision-free region is calculated based on the at least one environmental image using semantic segmentation. Particularly preferably, the at least one collision-free region is calculated using a machine learning method, in particular using a trained neural network.

[0017] The method may include an initialization step. For example, a position or pose of the mobile unit may be initially estimated using at least one further sensor device. The at least one further sensor device may, for example, comprise a satellite positioning system for acquiring GPS data, at least one speed sensor for acquiring odometry data, and / or a LIDAR system (LIDAR = light detection and ranging). For example, the map may be retrieved or obtained based on the initially estimated pose or position of the mobile unit via a mobile communications network, via an external server, and / or from an electronic storage unit of the mobile unit.

[0018] The method may also include a geometric projection. For example, the calculated at least one collision-free area can be projected onto the local ground plane of the mobile unit, preferably based on the calibration. Preferably, the at least one drivable area marked in the map is also projected onto the local ground plane of the mobile unit, for example based on the initially estimated pose or position of the mobile unit or based on a hypothesis for the pose of the mobile unit. Such a geometric projection allows the comparison of the at least one calculated collision-free area with the at least one drivable area marked in the map to be carried out particularly quickly and efficiently.

[0019] Particularly advantageously, the proposed method can also be combined with a known localization method, in particular a landmark-based localization method. For example, at least one hypothesis for the pose of the mobile unit can first be determined or calculated using a known localization method, in particular a landmark-based localization method.

[0020] A plausibility check can be performed when comparing the at least one calculated collision-free area with the at least one drivable area marked on the map. Using the plausibility check, the plausibility of the at least one hypothesis can be verified or determined, or a hypothesis, preferably a plausible hypothesis, can be selected from several hypotheses and taken into account when determining the pose of the mobile unit. Alternatively or additionally, the plausibility of the initially estimated pose of the mobile unit can also be determined using the plausibility check.

[0021] Preferably, the localization method is a landmark-based localization method. The landmark-based localization method can comprise a detection and / or a classification of at least one landmark in the at least one environmental image. For example, detection can be performed using a trained neural network.

[0022] Alternatively or additionally, classification can be performed using a trained neural network.

[0023] The at least one hypothesis for the pose of the mobile unit can be determined based on the association of at least one landmark detected in the at least one captured environmental image with at least one known landmark marked on the map. For this purpose, the localization method can comprise, for example, a particle filter or a Kalman filter, preferably a multi-hypothesis Kalman filter.

[0024] It may happen that a landmark-based localization method cannot be used to unambiguously assign multiple detected landmarks to multiple known landmarks. In this case, multiple hypotheses can be determined. Each hypothesis can correspond to a specific assignment of the multiple detected landmarks to the multiple known landmarks. For each assignment or hypothesis, a hypothesis for the pose of the mobile unit can then be determined using the landmark-based localization method. In particular, multiple hypotheses for the pose of the mobile unit can be determined in this way.

[0025] If multiple hypotheses are determined using the localization method, the plausibility check can advantageously be performed for all of these hypotheses. If multiple plausible hypotheses are then determined, the comparison or plausibility check can also include weighting of the multiple plausible hypotheses.

[0026] The plausibility check may include determining the sub-area or subset of the at least one collision-free area that can be clearly assigned to the at least one passable area marked on the map based on the at least one hypothesis. For the comparison or plausibility check, the at least one collision-free area and the at least one passable area can be transformed into a common coordinate system based on the at least one hypothesis and a corresponding coordinate transformation.

[0027] Preferably, the at least one collision-free area and the at least one drivable area are projected onto a surface in a common coordinate system based on the at least one hypothesis and / or the calibration. The surface can, for example, correspond to the local ground plane of the mobile unit. The surface can also correspond to the area defined in the map by the marked drivable area. The surface can also correspond to a surface or hypersurface defined by the at least one marked drivable area in the map coordinate system or in the camera coordinate system.

[0028] A hypothesis can be considered plausible if the ratio of the size of a sub-area of ​​the collision-free area that can be clearly assigned to the marked drivable area based on at least one hypothesis and its complement exceeds a critical value. A weighting of the plausible hypotheses can also be determined based on the ratio thus determined between the size of the identified sub-area and its complement.

[0029] For example, an intersection of the at least one collision-free area projected onto a common surface and the at least one marked drivable area can be determined. The number of valid pixels can correspond to the number of pixels contained in the intersection (pixels that can be clearly assigned to the at least one marked drivable area). The number of invalid pixels can correspond to the number of pixels contained in the at least one collision-free area but do not belong to the intersection (pixels that cannot be clearly assigned to the at least one marked drivable area).

[0030] The plausibility check can then involve comparing the ratio of the number of valid pixels to the number of invalid pixels with a predefined critical value. If the predefined critical value is exceeded, at least one hypothesis can be classified as plausible. A weighting of several plausible hypotheses can then be determined based on the ratio of the number of valid pixels to the number of invalid pixels, with the highest weight corresponding to the largest ratio of the number of valid pixels to the number of invalid pixels.

[0031] During the comparison, the pose of the mobile unit or a new estimate of the pose of the mobile unit can be determined based on the plausibility check. For example, the pose of the mobile unit determined by means of the comparison can correspond to a plausible hypothesis. The pose of the mobile unit determined by means of the comparison can also correspond to the initially estimated pose of the mobile unit. For example, if no plausible hypothesis can be determined by means of the plausibility check, the pose of the mobile unit determined by means of the comparison can correspond to the initially estimated pose of the mobile unit. If several plausible hypotheses are determined by means of the plausibility check and the comparison includes a weighting of the several plausible hypotheses, the pose of the mobile unit determined by means of the comparison can also correspond to the plausible hypothesis with the greatest weight.

[0032] In addition to or as an alternative to the localization method, several hypotheses for the pose of the mobile unit can also be determined as initial values ​​or initial vectors in the form of a regular or random grid. The grid can, for example, be determined based on the initially estimated position or pose of the mobile unit or contain this as an initial value or as an initial vector. The position information and / or the orientation angle information of the hypotheses determined in this way can correspond as a point cloud in a coordinate system, for example, to a two-dimensional or three-dimensional Bravais grid, preferably a square or cubic grid. Alternatively, the several hypotheses can also be randomly generated as initial vectors using a random generator.

[0033] In addition or alternatively to the plausibility check, an update of the pose or position of the mobile unit can take place.

[0034] To update the pose or position of the mobile unit, a cost function can be minimized when comparing the at least one calculated collision-free area with the at least one drivable area marked on the map. The cost function can specify, for at least one hypothesis, a characteristic distance between the at least one calculated collision-free area and the at least one drivable area marked on the map in a common coordinate system. The minimum of the cost function can be determined using an optimization method, for example, using the least squares method, using Monte Carlo simulation, using linearization in the correction step of a Kalman filter, or using the simulated annealing method.The pose of the mobile unit determined by the comparison can correspond to the minimum of the cost function. The at least one characteristic distance can be determined based on the shortest distance between a respective three-dimensional point attributable to the at least one collision-free area and the area defined in the map by the at least one marked drivable area. For example, the at least one characteristic distance can correspond to an average value or the sum of all shortest distances thus determined.

[0035] Preferably, the minimization of the cost function is used to determine an update of a position or pose of the mobile unit.

[0036] In a preferred embodiment, the cost function is minimized for one or more hypotheses determined using a localization method. Thus, the cost function minimization or the optimization method can also be combined with a landmark-based localization method.

[0037] It would also be possible to carry out the minimization of the cost function for several ad hoc hypotheses determined as initial values ​​or initial vectors. Minimizing the cost function or the optimization method can then be used advantageously, particularly in environments of the mobile unit in which no landmark can be detected or for which no known landmarks are marked on the map and therefore a landmark-based localization method is not feasible. It can also be provided that an initially estimated pose of the mobile unit is determined by means of the further sensor device, at least one hypothesis is determined by means of a localization method and / or at least one further hypothesis is determined by means of the minimization of a cost function. A plausibility check can also be carried out for the initially estimated pose, the at least one hypothesis and / or the at least one further hypothesis.By comparing the at least one calculated collision-free area with the at least one drivable area marked in the map, the pose of the mobile unit can be determined such that it corresponds to the initially estimated pose, the at least one hypothesis, or the at least one further hypothesis.

[0038] The invention also relates to a system comprising a mobile unit with at least one sensor device, an electronic storage unit, and an electronic evaluation and control unit. The at least one sensor device is configured to capture at least one image of the surroundings of the mobile unit. A map having at least one drivable area marked on the map is stored in the electronic storage unit. The electronic evaluation and control unit is configured to calculate at least one collision-free area based on at least one image of the surroundings captured by the at least one sensor device. The electronic evaluation and control unit is also configured to determine a pose of the mobile unit by comparing the at least one collision-free area with the at least one drivable area marked on the map.The map is a semantic map and / or includes at least one known landmark with map position information. Preferably, the at least one sensor device comprises a calibrated mono-camera. Particularly preferably, the electronic evaluation and control unit is configured to calculate the at least one collision-free area based on an environmental image sequence captured by the calibrated mono-camera.

[0039] It would also be possible for the at least one sensor device to have a 3D sensor, for example a LiDAR sensor.

[0040] The at least one sensor device may also comprise a combination of one or more 3D sensors and / or one or more 2D sensors, such as multiple cameras.

[0041] The electronic storage unit can be part of the electronic evaluation and control unit or electronically connected to an external server. The electronic storage unit can be a read-only memory or a working memory.

[0042] The electronic evaluation and control unit may comprise a CPU (central processing unit), a GPU (graphical processing unit), and / or a computing unit. The system, the mobile unit, and / or the electronic evaluation and control unit may also comprise at least one further sensor device and / or a mobile communication system.

[0043] The at least one further sensor device can be configured to initially estimate a pose or position of the mobile unit. For example, the at least one further sensor device can comprise a satellite positioning system, preferably a GPS (Global Positioning System), or can be configured as part of a satellite positioning system. The at least one further sensor device can also comprise at least one speed sensor for estimating odometry data.

[0044] The mobile communication system can be a WLAN-enabled (WLAN = wireless local area network) and / or a mobile radio-enabled communication system. For example, the mobile communication system can be configured to retrieve or capture the card from an electronic storage unit via an external server.

[0045] The invention also provides a computer program product. The computer program product comprises a computer program (or a sequence of instructions). The computer program comprises software means for carrying out a method as described above or for controlling a system as described above when the computer program is executed in a computing unit.

[0046] Preferably, the computer program product can be loaded directly into an internal electronic memory or storage unit of the computing unit or is already stored therein and typically comprises portions of a program code for carrying out the described method when the computer program product runs or is executed on the computing unit. The program code or portions of the program code can be formulated in a scripting language or a compiler language, such as C, C++, or Python. The computer program product can be stored on a machine-readable carrier, preferably a digital storage medium.

[0047] Embodiments of the invention are illustrated in the drawings and are described below with reference to Figures 1 to 2 explained: Showing: Fig. 1 a schematic representation of a captured environmental image with a calculated collision-free area and Fig. 2a schematic representation of a map with a marked drivable area.

[0048] The following are recurring features in the Figures 1 and 2 each provided with identical reference symbols.

[0049] Figure 1 shows a schematic representation of an environmental image PIC of a vehicle as a mobile unit 1 (not shown), captured by a monocamera. Based on the environmental image PIC, a collision-free area 3 is calculated using a machine learning method and marked in the environmental image PIC. The collision-free area 3 corresponds to the part of the road surface 5 that can be driven over in the direction of travel. Figure 1The collision-free area 3 borders on a vehicle 1.1 driving ahead of the mobile unit 1 and on a vehicle 1.2 traveling towards the mobile unit 1. Furthermore, a traffic sign is detected as a landmark 4.1 using a trained neural network based on the surrounding image PIC.

[0050] Figure 2 shows a schematic representation of a semantic map MAP with a marked drivable area 2 and a known landmark 4.2. The map MAP also includes map position information for the known landmark 4.2.

[0051] The pose of the mobile unit 1 is determined by comparing the at least one collision-free area 3 with the at least one drivable area 2 marked in the map MAP. For this purpose, a position of the mobile unit 1 is initially estimated using a GPS system. Based on the initially estimated position of the mobile unit 1, the landmark 4.1 detected in the environmental image PIC is assigned to the known landmark 4.2 marked in the map MAP using a known landmark-based localization method. Based on the assignment and the map position information of the known landmark 4.2, a hypothesis for the pose of the mobile unit 1 is determined using the known landmark-based localization method.

[0052] Based on the hypothesis for the pose of mobile unit 1 and a predetermined calibration of the mono camera, the Figure 1The collision-free area 3 marked and calculated in the surrounding image PIC is projected onto the drivable area 2 marked in the map MAP. In Figure 2 the projection of the collision-free area 3 marked and calculated in the surrounding image PIC onto the drivable area 2 marked in the map MAP corresponds to the sub-areas 3.1, 3.2.

[0053] During the comparison, a plausibility check of the hypothesis for the pose of mobile unit 1 is then carried out. During the plausibility check, the sub-area 3.1 of the calculated collision-free area 3, 3.1, 3.2 is determined which can be clearly assigned to the drivable area 2 marked in the map MAP. The number of valid pixels then corresponds to the number of pixels contained in sub-area 3.1. The number of invalid pixels corresponds to the number of pixels contained in sub-area 3.2. Sub-area 3.2 is the complement of sub-area 3.1. The higher the number of valid pixels, the more plausible the hypothesis for the pose of mobile unit 1. In particular, the ratio of the number of valid pixels to the number of invalid pixels is compared with a predetermined critical value. Plausibility exists if the ratio thus determined is greater than the predetermined critical value. For the Figure 2In the example shown, the specified critical value is 4. The ratio of the number of valid pixels to the number of invalid pixels is 5. Thus, the hypothesis determined using the landmark-based localization method is plausible. The pose of mobile unit 1 determined by comparing the calculated collision-free area 3 and the drivable area 2 marked in the map MAP then corresponds to the plausible hypothesis for the pose of mobile unit 1.

[0054] In a further embodiment, several landmarks 4.1 are detected in the environmental image PIC. The map MAP also includes several known landmarks 4.2, each with map position information. Based on the initially estimated position of the mobile unit 1, several hypotheses are determined using the known landmark-based localization method. Each hypothesis assigns a known landmark 4.2 to each of the detected landmarks 4.1. For each assignment hypothesis, a hypothesis for the pose of the mobile unit 1 is then determined based on the map position information of the respectively assigned known landmarks 4.2.

[0055] For each of the hypotheses determined in this way, the calculated collision-free area 3 is projected onto the drivable area 2 marked on the map MAP, and the ratio of the number of valid pixels to the number of invalid pixels is determined. Based on this ratio, a weighting of the respective hypotheses is then determined, with the hypothesis with the largest ratio being assigned the greatest weight. The pose of mobile unit 1 determined by the comparison then corresponds to the hypothesis for the pose of mobile unit 1 with the greatest weight.

[0056] In a further embodiment, no landmark-based localization method is performed. Instead, 10,000 hypotheses for the pose of the mobile unit 1 are randomly determined using a random number generator. During the comparison, a cost function is then minimized, with the cost function specifying a characteristic distance between the calculated collision-free area 3 and the drivable area 2 marked in the map MAP in a common coordinate system for each of the randomly determined hypotheses.To determine the characteristic distance, a coordinate transformation of the calculated collision-free area 3 from the camera coordinate system to the map coordinate system is performed based on a respective hypothesis, and the shortest distances between a respective three-dimensional point of the at least one collision-free area 3 and the area defined in the map MAP by the marked drivable area 2 are determined. The sum of the shortest distances thus determined then corresponds to the characteristic distance.

[0057] The cost function is minimized using linearization in the correction step of a Kalman filter. The pose of mobile unit 1 determined from the comparison then corresponds to the minimum of the cost function and is used to determine an update of the pose of mobile unit 1.

[0058] The methods described in the exemplary embodiments are carried out using a system comprising a mobile unit 1 with a monocamera as a sensor device, an electronic storage unit, and an electronic evaluation and control unit. The system also has a GPS system. The monocamera is configured to capture images of the surroundings of the mobile unit 1. The map MAP with the marked drivable area 2 is stored in the electronic storage unit. The electronic evaluation and control unit is configured to calculate the collision-free area 3 based on a sequence of environmental images captured by the mono camera. The electronic evaluation and control unit is also configured to determine the pose of the mobile unit 1 based on the comparison of the collision-free area 3 with the drivable area 2 marked in the map MAP.

[0059] Only features of the various embodiments disclosed in the exemplary embodiments can be combined with one another and claimed individually.

Claims

1. A method for determining the pose of a mobile unit (1), which is formed as a land vehicle or a watercraft, with at least one sensor device, which is configured for capturing environmental images (PIC) of the mobile unit (1), and a map (MAP), in which at least one drivable area (2) is marked, wherein the map (MAP) is stored in an electronic storage unit and the map (MAP) is a semantic map and / or includes at least one known landmark (4.2) with map position indications, in which at least one environmental image (PIC) is captured by means of the at least one sensor device and at least one collision-free area (3, 3.1, 3.2) is calculated by means of an electronic evaluation and control unit based on the at least one captured environmental image (PIC), and the pose of the mobile unit (1) is determined by means of the electronic evaluation and control unit by a comparison of the at least one calculated collision-free area (3, 3.1, 3.2) to the at least on drivable area (2) marked in the map (MAP).

2. The method according to claim 1, characterized in that the pose of the mobile unit (1) includes position indications and / or orientation angle indications and / or the at least one marked drivable area (2) corresponds to a two-dimensional point set or a three-dimensional point set in a map coordinate system.

3. The method according to any one of the preceding claims, characterized in that the at least one collision-free area (3, 3.1, 3.2) corresponds to a two-dimensional point set or a three-dimensional point set in a camera coordinate system and / or the at least one collision-free area (3, 3.1, 3.2) corresponds to a surface drivable without collision starting from the mobile unit (1) in direction of travel and / or the at least collision-free area (3, 3.1, 3.2) is calculated based on at least two sequentially captured environmental images (PIC) and / or the at least one collision-free area (3, 3.1, 3.2) is calculated based on a monocular environmental image sequence and / or the at least one collision-free area (3, 3.1, 3.2) is calculated by means of semantic segmentation and / or the at least one collision-free area (3, 3.1, 3.2) is calculated by means of a machine learning method, in particular by means of a trained neural network.

4. The method according to any one of the preceding claims, characterized in that the calculated at least one collision-free area (3, 3.1, 3.2) and / or the at least one drivable area (2) marked in the map (MAP) are projected to the local ground plane of the mobile unit (1).

5. The method according to any one of the preceding claims, characterized in that at least one hypothesis for the pose of the mobile unit (1) is calculated by means of a localization method and a plausibility check of the at least one calculated hypothesis is performed in the comparison.

6. The method according to claim 5, characterized in that the localization method is a landmark-based localization method, and / or the at least one hypothesis for the pose of the mobile unit (1) is determined based on the association of at least one landmark (4.1) detected in the captured environmental image (PIC) with at least one known landmark (4.2) marked in the map (MAP), and / or multiple hypotheses for the pose of the mobile unit (1) are determined by means of a landmark-based localization method.

7. The method according to claim 5 or 6, characterized in that a determination of that partial area (3.1) of the at least one collision-free area (3, 3.1, 3.2) is performed in the plausibility check, which can be uniquely associated with the at least one drivable area (2) marked in the map (MAP) based on the at least one hypothesis, and / or the at least one hypothesis is plausible if the ratio of a size of a partial area (3.1) of the at least one collision-free area (3, 3.1, 3.2) capable of being associated with the at least one marked drivable area (2) based on the at least one hypothesis and the complement (3.2) thereof exceeds a critical value.

8. The method according to any one of claims 5 to 7, characterized in that the pose of the mobile unit (1) determined by the comparison corresponds to a plausible hypothesis, and / or a pose of the mobile unit (1) is initially estimated by means of a further sensor device and the pose of the mobile unit (1) determined by the comparison corresponds to the initially estimated pose of the mobile unit (1) or to a plausible hypothesis, and / or multiple plausible hypotheses are determined by means of the plausibility check and the comparison includes weighting of the multiple plausible hypotheses and the pose of the mobile unit determined by the comparison corresponds to the plausible hypothesis with the greatest weight.

9. The method according to any one of the preceding claims, characterized in that at least one hypothesis for the pose of the mobile unit (1) is determined and a minimization of a cost function is performed in the comparison, wherein the cost function for the at least one hypothesis indicates at least one characteristic distance between the at least one calculated collision-free area (3, 3.1, 3.2) and the drivable area (2) marked in the map (MAP) in a common coordinate system, and the pose of the mobile unit (1) corresponds to the minimum of the cost function.

10. The method according to claim 9, characterized in that the at least one characteristic distance is determined based on the shortest distances between a respective three-dimensional point of the at least one collision-free area (3, 3.1, 3.2) and the surface defined in the map (MAP) by the marked at least one drivable area (2).

11. The method according to claim 9 or 10, characterized in that an update of the pose of the mobile unit (1) is performed by means of the minimization of a cost function.

12. A system including a mobile unit (1), which is formed as a land vehicle or a watercraft, with at least one sensor device, which is configured to capture at least one environmental image (PIC) of the mobile unit (1), an electronic storage unit, in which a map (MAP) with at least one marked drivable area (2) is stored, wherein the map (MAP) is a semantic map and / or includes at least one known landmark (4.2) with map position indications, and an electronic evaluation and control unit, which is configured to calculate at least one collision-free area (3, 3.1, 3.2) based on at least one environmental image (PIC) captured by means of the at least one sensor device, and to determine a pose of the mobile unit (1) based on a comparison of the at least one collision-free area (3, 3.1, 3.2) to the at least one drivable area (2) marked in the map (MAP).

13. The system according to claim 12, characterized in that the system and / or the mobile unit (1) and / or the electronic evaluation and control unit comprise at least one further sensor device, wherein the at least one further sensor device includes a satellite tracking system and / or a mobile communication system.

14. A computer program product, which comprises a computer program, which comprises software means for performing a method according to any one of claims 1 to 11 or for controlling a system according to any one of claims 12 or 13 when the computer program runs in a computing unit.