Method for automatically locating an ego vehicle using at least one registration pattern for one or more environment sensors

Abstract

The invention relates to a method for the automatic localization of an ego-vehicle by means of at least one registration pattern (M A1 , M A2 , M E , M U ) consisting of a multitude of individual elements (1), for one or more environmental sensing sensors comprising the following steps: - Recording (S1) an environment of the ego vehicle with at least one environment detection sensor of the ego vehicle; - Evaluating (S2) the recording from at least one environmental detection sensor; - Creating (S3) an environment map based on the sensor recordings; - Determine (S4) at least one registration pattern (M A1 , M A2 , M E , M U , M X ) and an arrangement of at least some of the multitude of individual elements (1) in the vicinity of the Ego vehicle; - Determining (S5) various localization information based on the arrangement of the individual elements (1); - Locating (S6) the Ego vehicle in the created environment map based on the determined localization information.

Description

[0001] The invention relates to a method for the automatic localization of an ego vehicle by means of at least one registration pattern consisting of a plurality of individual elements for one or more environment detection sensors.

[0002] The state of the art for localization is the simultaneous localization and mapping (SLAM) method, in which an autonomous vehicle must simultaneously create a map of its surroundings and estimate its spatial position within this map. It is therefore used to detect obstacles and thus supports autonomous navigation. If the absolute position of the autonomous vehicle is also known, e.g. from additional GPS sensors, a map can be created. The autonomous vehicle measures the relative position of possible obstacles and, using the known position, can then determine the absolute position of the obstacles, which are then entered into the map. Solving SLAM requires solving the data association problem, i.e. it must be determined which (environmental) features correspond. This problem is particularly difficult because features cannot be extracted with absolute certainty.

[0003] Other state-of-the-art methods use visual markers (e.g., "April Days"). However, these are only detectable by optical sensors. Influences such as occlusion, smoke, or sensor contamination make it difficult to perceive the visual markers or other visual features between the optical sensors. Therefore, redundancy is required for a fail-safe implementation.

[0004] It is therefore an object of the present invention to provide a method by means of which a reliable localization of an ego vehicle is possible independently of the sensors used and without the use of GPS.

[0005] This object is solved by the subject matter of independent claim 1. Further advantageous embodiments and embodiments are the subject matter of the subclaims.

[0006] Initial considerations were to enable navigation of an autonomous vehicle in buildings, such as a parking garage, or tunnels without or at least with limited GPS reception.

[0007] In parking garages, underground car parks, or tunnels, the navigation of an autonomous vehicle without GPS reception, with rapidly changing lighting conditions, and significant occlusion, is very limited or very inaccurate. Therefore, it is usually unclear on which floor and exactly where the autonomous vehicle is located.

[0008] Even with good GPS reception, it should be noted that regular GPS can easily have an inaccuracy of 10 meters.

[0009] Orientation based on or solely on radar detections is not possible with visual markers.

[0010] According to the invention, a method for automatically locating an ego vehicle by means of at least one registration pattern consisting of a plurality of individual elements for one or more environment detection sensors is therefore provided, comprising the following steps: - Recording an environment of the ego vehicle with at least one environment detection sensor of the ego vehicle; - Evaluating the recording of at least one environment detection sensor; - Creating an environment map based on the sensor recordings; - Determining at least one registration pattern and an arrangement of at least some of the plurality of individual elements in the environment of the ego vehicle; - Determining various localization information based on the arrangement of the individual elements; - Localize the ego vehicle based on the determined localization information in the created environment map.

[0011] The at least one environment detection sensor is a camera, a radar, a lidar or a combination of several of these sensors. To create the environment map, all detections of the sensor can be entered into the map. The three-dimensional positions of the individual elements can be known a priori, whereby each element is uniquely identifiable via the three-dimensional distances to the other elements and whereby the distances are known from the a priori known three-dimensional positions of the elements. Thus, each of the elements is uniquely identifiable via its n-nearest neighbors. Through this identification of the individual elements, the arrangement of the individual elements can be recognized or calculated and the registration pattern can be uniquely recognized. Corresponding information for each registration pattern can be stored in the ego vehicle.Each registration pattern is advantageously designed to have a unique arrangement and thus provide unambiguous localization information. By uniquely identifying the individual elements, the registration pattern does not need to be completely captured by an environment detection sensor or multiple sensors. The arrangement can transfer the information, for example, similar to Braille or star formations, where the arrangement of different points / celestial bodies also conveys specific information.

[0012] The ego vehicle is preferably an autonomous or at least partially autonomous vehicle.

[0013] In a preferred embodiment, the localization information comprises the absolute position of the pattern and / or information about the surroundings of the pattern. The absolute position of the pattern can be in world coordinates. With knowledge of the absolute position, the position of the ego vehicle in the surrounding map can be determined precisely and accurately in relation to the registration pattern. Furthermore, with knowledge of the absolute position of the pattern, it would be possible to use the registration pattern in addition to the localization for the calibration of the environmental sensors used. Information about the surroundings can include, for example, whether the ego vehicle is located in a parking garage or an underground car park, and also on which floor the ego vehicle is located.

[0014] Particularly preferably, the arrangement of the individual elements is analyzed using pattern recognition. Pattern recognition can, for example, be implemented as a trained neural network.

[0015] In another particularly preferred embodiment, the pattern recognition is trained using a variety of different arrangements and corresponding information. If a neural network is used for pattern recognition, the neural network can be trained accordingly using the different arrangements.

[0016] Furthermore, the individual elements are particularly preferably designed such that they are equally detectable by different sensor types. The individual elements thus have sub-elements with different properties. Thus, one sub-element can be designed for detection with an optical sensor, in particular a camera, and another sub-element can be designed for detection with a radar sensor. Accordingly, it is irrelevant which sensors the ego vehicle is equipped with or which sensor detects the registration pattern, since the detectability of all sensors is equally ensured by the design.

[0017] In a further preferred embodiment, a correction and / or error calculation of a GPS position of the ego vehicle is performed based on the localization of the ego vehicle. The GPS position can be compared with the localization of the ego vehicle. This is particularly ensured if the absolute position of the registration pattern is known via the localization information. In this way, the ego vehicle can also be clearly localized, allowing a comparison with the current GPS position to be performed.

[0018] Further advantageous embodiments are shown in the drawings, which show: Fig. 1: a schematic flow diagram of a method according to an embodiment of the invention; Fig. 2: a schematic representation of an application of a registration pattern according to an embodiment of the invention; Fig. 3: a schematic representation of an application of a registration pattern according to a further embodiment of the invention;

[0019] Fig. 1 shows a schematic flow diagram of a method according to one embodiment of the invention. In step S1, an environment of the ego vehicle is recorded using at least one environment detection sensor of the ego vehicle. In step S2, the recording of the at least one environment detection sensor is evaluated. In step S3, an environment map is created based on the sensor recordings. In step S4, at least one registration pattern and an arrangement of at least some of the plurality of individual elements in the environment of the ego vehicle are determined. In step S5, various localization information items are determined based on the arrangement of the individual elements. In a further step S6, the ego vehicle is localized in the created environment map based on the determined localization information.

[0020] Fig. 2 shows a schematic representation of the application of a registration pattern according to an embodiment of the invention. In this illustration, for example, a registration pattern Mu is applied to an entrance of a parking garage P. The ego vehicle enters the parking garage P in the direction of travel F and detects the pattern Mu. Via this pattern Mu, for example, the localization information can be transmitted, so that the autonomous vehicle receives the information that it is entering a parking garage P with a certain number of floors E. On the floors E of the parking garage P, for example, further patterns M E which provide the vehicle with information on which floor E the vehicle is currently located. The registration patterns M E and Mu each consist of a multitude of individual elements 1.

[0021] The Fig.Figure 3 shows a schematic representation of the application of a registration pattern according to a further embodiment of the invention. Here, two registration patterns M A1 and M A2 located next to a motorway A in the vicinity of a rest area. If an ego vehicle detects one of the patterns M A1 and M A2 It knows the absolute position of the pattern and can also locate itself precisely and accurately. This allows for correction and error calculation of the GPS positioning by comparing the GPS position of the GPS system with the position of the ego vehicle based on the absolute position of the registration pattern M. A1 or M A2 is compared.

[0022] The invention is not limited to the above-mentioned models M E , M U , M A1 , M A2These are merely illustrative examples. A variety of different registration patterns M X be installed at different locations, which provide an autonomous ego vehicle with corresponding information about the respective location. List of reference symbols 1 single element A motorway E Floor F Direction of travel M A1 , M A2 , M E , M U , M X Registration template P Parking Garage S1-S6 process steps

Claims

[1] Method for the automatic localization of an ego vehicle by means of at least one registration pattern (M A1 , M A2 , M E , M U ) consisting of a plurality of individual elements (1), for one or more environment detection sensors comprising the following steps: - recording (S1) an environment of the ego vehicle with at least one environment detection sensor of the ego vehicle; - evaluating (S2) the recording of the at least one environment detection sensor; - Creating (S3) an environment map based on the sensor recordings; - Determining (S4) at least one registration pattern (M A1 , M A2 , M E , M U , M X ) and an arrangement of at least some of the plurality of individual elements (1) in the environment of the ego vehicle; - determining (S5) various localization information based on the arrangement of the individual elements (1); - Localizing (S6) the ego vehicle in the created environment map based on the determined localization information. [2] Method according to claim 1, characterized by that the localization information contains the absolute position of the pattern (M A1 , M A2 , M E , M U , M X ) and / or information about the environment of the pattern (M A1 , M A2 , M E , M U , M X ) include. [3] Method according to claim 1, characterized by that the arrangement of the individual elements (1) is analyzed by means of pattern recognition. [4] Method according to claim 3, characterized by that pattern recognition is trained with a variety of different arrangements and corresponding information. [5] Method according to claim 1, characterized bythat the individual elements (1) are designed in such a way that they can be detected equally by different types of sensors. [6] Method according to claim 1, characterized by that based on the localization of the ego vehicle, a correction and / or an error calculation of a GPS position of the ego vehicle is carried out.

Images