Method, control device, computer program and machine-readable storage medium for determining features for positioning and / or mapping

Abstract

A method for determining features in the environment of at least one mobile unit for performing localization and / or mapping by a control device is disclosed. The method comprises receiving sensor measurement data of the environment, converting the received sensor measurement data into a cost function using an orientation algorithm, creating a cost map based on the cost function, creating a convergence map based on the orientation algorithm, determining and storing at least one feature from the cost map and / or the convergence map, and providing the at least one feature for optimizing localization and / or mapping. Furthermore, a control device, a computer program, and a machine-readable storage medium are disclosed.

Description

Technical Field

[0001] The invention relates to a method for ascertaining features in the environment of at least one mobile unit for carrying out localization and / or mapping by a control device. The invention also relates to a control device, a computer program, and a machine-readable storage medium. Background Art

[0002] Vehicles with automated driving capabilities are becoming increasingly important in road traffic. These vehicles have the potential to prevent congestion and accidents caused by human error. Successful implementation of automated driving requires robust localization and accurate environmental mapping.

[0003] To create a map for implementing automated driving functions, features need to be extracted from the environment and stored in a digital map. Unique or prominent features can be particularly useful for subsequent localization within the digital map. Examples of such unique features include traffic signs, unique buildings, or masts. Furthermore, there are multiple periodically repeating features. Repeating features can be formed, for example, by lane markings or guideposts. However, the extraction or use of repeating features can lead to ambiguity in mapping or localization. Summary of the Invention

[0004] The object underlying the present invention can be seen as providing a method for extracting additional qualitative and quantitative features for use in mapping and / or localization.

[0005] This object is achieved by means of the corresponding subject matter of the present invention. Advantageous embodiments of the present invention are each the subject matter of preferred developments.

[0006] According to one aspect of the present invention, a method is provided for determining features in the environment of at least one mobile unit for performing localization and / or mapping by a control device. In one step, sensor measurement data of the environment is received. Alternatively or additionally, processed map data may also be received.

[0007] An orientation algorithm is created based on the received sensor measurement data. A cost function is part of the orientation algorithm and is specified. A cost map is then created based on the cost function. The cost map can be configured, for example, as a discrete cost function. The cost map can be created based on sensor measurement data acquired by the mobile unit's environmental sensor system.

[0008] A convergence diagram is then created based on the orientation algorithm. Unlike a cost diagram, a convergence diagram shows a large number of trajectories that the orientation algorithm follows under different initial conditions. The initial conditions, or starting conditions, of the orientation algorithm depend on different positions and orientations of the mobile unit. Thus, while a cost diagram describes the behavior of the cost function, a convergence diagram describes the behavior of the orientation algorithm.

[0009] In another step, at least one feature is determined from the cost map and / or the convergence map and stored. The at least one feature is provided for optimizing positioning and / or mapping. The at least one feature can be the occurrence of one or more minima in the cost map and / or the convergence map.

[0010] An important aspect of map creation is the ability to automatically extract multi-valued and / or single-valued areas or features and to use this additional information during map creation or localization. The use of cost maps and convergence maps allows the determination of additional functionality necessary to expand the features used in digital maps.

[0011] The cost map and / or convergence map can preferably be created from sensor measurement data from a video sensor or camera sensor, a stereo camera sensor, a 3D camera sensor, a 360° camera system, a LIDAR sensor, a radar sensor, an ultrasonic sensor, or the like. Furthermore, sensor measurement data from different sources can be combined or fused to create a combined cost map and / or convergence map.

[0012] An orientation algorithm based on an iterative method or a step-by-step method can be used. Such an orientation algorithm can be configured, for example, as a so-called iterative closest point algorithm or a step-by-step algorithm.

[0013] In particular, cost maps and convergence maps can be considered as compressed representations of the actual sensor measurement data. This compressed representation can be achieved, for example, by the number of minimum values within the map, which can be used as an additional feature of the map and can be used by the positioning unit during operation of the mobile unit. These features can be used, for example, to limit the positioning unit's determination of the mobile unit's position within the digital map to areas in which the convergence map has a single-valued minimum. In areas where the minimum values have a periodic nature, the positioning unit can use targeted strategies. For example, the positioning unit can focus its calculations on areas of the cost map with lower cost values, which are visible in the cost map and / or the convergence map.

[0014] According to another aspect of the present invention, a control device is provided, configured to implement the method. The control device may be, for example, a control device located in a mobile unit or an external control device. For example, the control device may be connected to a control device of the mobile unit for implementing automated driving functions, or may be integrated into such a control device. An external control device may, for example, be a cloud-based server unit external to the vehicle.

[0015] Furthermore, according to one aspect of the present invention, a computer program is provided, comprising instructions which, when executed by a computer or a control device, cause the computer or control device to implement the method according to the present invention. According to another aspect of the present invention, a machine-readable storage medium is provided, on which the computer program according to the present invention is stored.

[0016] According to the German Federal Institute for Transport Research (Bundesanstalt Für Straßenwesen, BASt) standards, the mobile unit can be operated in an assisted, partially automated, highly automated, and / or fully automated manner, or in other words, without a driver. The mobile unit can be embodied, in particular, as a vehicle, such as a passenger car, bus, commercial vehicle, truck, or the like. Furthermore, the mobile unit can be configured as a robot, drone, helicopter, airplane, ship, shuttle, self-driving taxi, or the like.

[0017] According to one embodiment, the number of minimum values is extracted from the cost map and / or the convergence map as at least one feature. The cost map and / or the convergence map can thus be considered as a compressed representation of the sensor measurement data. A cost function is a function defined by the user or the application, which is based on the position of the mobile unit and the received sensor measurement data. In the context of mapping, the cost function defines how accurately the sensor measurement data of different runs agree with one another at a given position of the mobile unit. This can be used, for example, in the orientation of the sensor measurement data of different runs. In the context of positioning, the cost function defines how accurately the sensor measurement data agree with an existing map at a given position of the mobile unit.

[0018] The convergence diagram relates to the underlying properties of the orientation algorithm used, which is based on an iterative or step-by-step method, and defines trajectories in the so-called pose space of the orientation algorithm, which start with different initial conditions. The initial conditions can also include, for example, the position of the mobile unit.

[0019] According to another embodiment, periodically occurring features are determined by a plurality of detected minima in the cost map and / or convergence map, and uniquely occurring features are determined by a single minimum in the cost map and / or convergence map. This measure allows the number of determined minima to provide an indication of whether the extracted feature is periodic or recurring, or unique or uniquely occurring. Features indicated by the minima in the cost map can, for example, repeat in time or space.

[0020] According to another embodiment, the cost function is configured to create a two-dimensional or three-dimensional cost map. By this measure, different degrees of explicitness can be used depending on the available computing performance and memory. The cost function obtained by the orientation algorithm can thus be determined in different spatial directions and stored in a memory.

[0021] According to another embodiment, the clarity of the extraction cost map The cost function can be used to determine the shape and / or form of the cost function and to process the sensor measurement data. In particular, the sharpness or slope of the cost function can be determined within a minimum or multiple minimum regions. The shape of the minima shown in the cost map can be used, for example, to limit preprocessing steps in mapping or localization, such as clustering or simulating environmental influences.

[0022] According to another embodiment, the differences between the minimum values determined in the cost map are determined. In particular, the determined differences in the shape and value of the different local minima can be extracted and stored as further features. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The preferred embodiment of the present invention will be further described below based on a highly simplified schematic diagram. Here, it is shown:

[0024] Figure 1 A schematic flow chart is shown for illustrating a method according to one embodiment.

[0025] Figure 2 An exemplary diagram showing a cost map with one minimum,

[0026] Figure 3 An exemplary diagram showing a convergence graph with one minimum,

[0027] Figure 4 a schematic diagram showing the environment of a mobile unit to illustrate the repetitive features,

[0028] Figure 5 shows a schematic convergence diagram reflecting the repetitive features,

[0029] Figure 6 Another schematic convergence diagram reflecting the repetitive characteristics is shown,

[0030] Figure 7 Schematic diagram showing sensor measurement data of a radar sensor, the sensor measurement data being used to create Figure 6 The convergence diagram used in

[0031] Figure 8 Another schematic convergence diagram reflecting the only occurring features is shown,

[0032] Figure 9 Schematic diagram showing sensor measurement data of a radar sensor, the sensor measurement data being used to create Figure 8 The convergence diagram used in . DETAILED DESCRIPTION

[0033] exist Figure 1 A schematic flow chart for explaining a method 1 according to one embodiment is shown in FIG. The method 1 is for determining features in an environment U of at least one mobile unit 2 for the purpose of performing localization and / or mapping by means of a control device 4. The mobile unit 2 has a control device 4, which is designed as an autonomous vehicle and is Figure 4 Shown in.

[0034] In a first step 10 of method 1 , sensor measurement data of the environment U are received. The sensor measurement data can be ascertained, for example, by an environment sensor system 6 and received and evaluated by control device 4 . Alternatively, already existing map data can be retrieved.

[0035] In a further step 12 , an orientation algorithm is provided and a cost function is created by the orientation algorithm from the received sensor measurement data.

[0036] Then, a cost map 14 is created based on the orientation algorithm and the cost function. In a further step, a convergence map 16 is created based on the orientation algorithm.

[0037] In a further step, at least one feature is extracted from the cost map 14 and / or the convergence map 16 and the at least one feature is stored 18 .

[0038] The at least one feature is then provided for optimized positioning and / or mapping 20 .

[0039] Figure 2 An exemplary illustration of a cost map 14 is shown with a local minimum 8. The minimum 8 has such a low value that the mobile unit 2 can travel in the region of the minimum 8 without the risk of collision.

[0040] exist Figure 3 An exemplary illustration of a convergence diagram 16 with a minimum value 8 is shown in FIG. Figure 2The cost graph 14 shown in FIG. 1 shows the orientation algorithm. Point 22 shows the starting point for the iteration point. Line or trajectory 24 corresponds to the direction in which the corresponding point 22 converges. By using the convergence graph 16, the performance of the analysis algorithm can be graphically illustrated.

[0041] Figure 4 A schematic diagram of the environment U of the mobile unit 2 is shown to illustrate a repetitive feature 26. The repetitive feature 26 may be, for example, a lane marking or a guardrail fixture.

[0042] The mobile unit 2 is designed as a vehicle, for example, and has a control device 4. The control device 4 is connected in a data-transmitting manner to an environment sensor system 6. The control device 4 can thereby receive sensor measurement data from the environment sensor system 6.

[0043] Surroundings sensor system 6 can, for example, include a camera sensor, a radar sensor, a lidar sensor, an ultrasonic sensor, or the like, and provide ascertained sensor measurement data to control device 4 in analog or digital form.

[0044] Corresponding to the repeated feature 24, Figure 5 A schematic convergence diagram 16 is shown in FIG, which reflects a recurring feature 26. Here, the number of minima 8 gives information about the uniqueness of the feature 26. A unique feature 25 leads to a single minimum 8. Periodically occurring features 26 lead to multiple local minima 8. This relationship is shown in FIG. Figures 6 to 9 Instructions.

[0045] exist Figure 7 and Figure 9 The sensor measurement data of the radar sensor are shown in FIG. Figure 7 Sensor measurement data for a highway section with a large number of guide posts configured as a repeating feature 26 is shown. Figure 6 A corresponding convergence diagram 16 with a plurality of minima 8 is shown in FIG.

[0046] Figure 9 The sensor measurement data of the radar sensor at the highway exit are shown. The highway exit represents a unique or only occurring feature 25. The resulting convergence diagram 16 is Figure 8 , and has a single minimum 8. Based on the number of minima 8, conclusions can be drawn about the uniqueness of the respective feature 25, 26.

Claims

1. A method for ascertaining features (25, 26) in an environment (U) of at least one mobile unit (2) for the purpose of performing localization and / or mapping by means of a control device (4), wherein: receiving (10) sensor measurement data of said environment (U), converting (12) the received sensor measurement data into a cost function by means of an orientation algorithm and creating a cost map (14) based on the cost function, Creating a convergence graph (16) based on the orientation algorithm, determining (18) at least one feature (25, 26) from the cost map (14) and / or the convergence map (16) and storing the at least one feature (25, 26), providing (20) said at least one feature (25, 26) for optimized positioning and / or mapping, wherein the number of minima (8) is extracted from the cost map (14) and / or the convergence map (16) as at least one feature (25, 26), In this case, periodically occurring features (26) are determined by multiple detected minimum values (8) in the cost diagram (14) and / or convergence diagram (16), and singly occurring features (25) are determined by a single minimum value (8) in the cost diagram and / or convergence diagram.

2. The method according to claim 1, wherein The cost function is used to create a two-dimensional or three-dimensional cost map (14).

3. The method according to claim 1 or 2, wherein: The sharpness and / or shape of the minimum (8) in the cost map (14) is determined and used for processing the sensor measurement data.

4. The method according to claim 1 or 2, wherein: The difference between the minimum values (8) found in the cost map (14) is determined.

5. A control device (4) configured to carry out the method according to any one of claims 1 to 4.

6. A computer program comprising instructions which, when executed by a computer or a control device (4), cause the computer or the control device to carry out the method according to any one of claims 1 to 4. 7 . A machine-readable storage medium having stored thereon the computer program according to claim 6 .

Images