METHOD FOR SELF-LOCATION IN A NAVIGATION ENVIRONMENT AND SELF-LOCATION VEHICLE
Patent Information
- Application Number
- DE502022002023
- Authority / Receiving Office
- DE · DE
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-12-13
- Publication Date
- 2024-11-14
- Estimated Expiration
- 2042-12-13
AI Technical Summary
Indoor navigation without satellite signals faces challenges due to the lack of infrastructure for existing localization methods, which are often complex and require high-quality sensors and algorithms, and existing optical marker systems do not support scalable and flexible self-localization in dynamic environments.
A method using groups of at least three optical markers with limited code values, where each marker encodes a single digit or number, allowing for a combination code that clearly identifies positions, enabling self-localization with a simple camera and reducing the need for complex infrastructure or algorithms, and allowing for scalable navigation in one, two, or three dimensions.
This approach provides a cost-effective and robust self-localization method that is suitable for indoor navigation, enabling precise positioning of vehicles and robots without requiring complex systems like SLAM or GPS, and allows for easy scaling and integration with higher-level systems for data analysis and route optimization.
Description
[0001] The invention relates to a method for self-localization in a navigation environment and a self-localizing vehicle according to the preamble of claims 1 and 12, respectively.
[0002] A prerequisite for successful navigation is reliable knowledge of one's own position through self-localization. Satellite-based methods (GPS, Global Positioning System) are available for this outdoors. For indoor use without satellite signal reception, alternative methods are available that rely on a large number of stationary transmitters using various technologies such as ultrasound, radar, Wi-Fi, or mobile communications. However, the infrastructure designed for this is generally not available. Image-based methods such as SLAM (Simultaneous Localization and Mapping) also exist, which require very high-quality sensors and algorithms, especially if the error rate is to be kept low. Navigation, especially indoors, is therefore a very complex task that requires a great deal of effort.The difficulties become even greater when adaptation to new situations and scenarios is required.
[0003] Self-localization can traditionally be supported by artificial optical markers. For example, it is known to record the positions of artificial markers on a map. Furthermore, a marker can communicate its own position using optical codes. Complementary methods exist that determine the pose of a camera capturing the marker. Fiducial markers, which provide sufficient features to reconstruct the pose, are particularly suitable for this purpose. In the simplest case, the features are the four corners of the marker; however, additional robustness, including the possibility of error detection and correction, can be achieved through other internal features of the marker.
[0004] A well-known vision system based on fiducial markers is AprilTag, see https: / / april.eecs.umich.edu / software / apriltag.html, including a bibliography. It includes a library implemented in C with functions for determining, among other things, the 3D position, orientation, and identity of the markers. An alternative is ArUco, with a tutorial at https: / / docs.opencv.org / 4.x / d5 / dae / tutorial_aruco_de-tection.html. VuMarks offers another class of special artificial markers, see https: / / library.vuforia.com / objects / vumarks.
[0005] US 2021 / 311205 A1 and US 10 223 666 B1 disclose a navigation system with self-describing fiducial markers that optically communicate navigation-supporting information, including the marker's own position. It is also possible to access a database based on the self-description and thus obtain additional information. According to one embodiment, a self-describing fiducial marker can take the form of an array of LEDs.
[0006] US 10 223 666 B1 presents a method for managing a warehouse in which autonomous vehicles move. They orient themselves using fiducial markers recorded on a reference map. The value range of the fiducial markers is small, so that ambiguities regarding the identity of a detected fiducial maker initially arise. This ambiguity is resolved by additional properties, such as different colors, orientations, or sizes of the fiducial markers. DE 10 2019 202 733 B4 (D2) discloses a control unit for navigating an autonomous vehicle. It follows a trail of markers, in particular LEDs, which can be assigned to g groups with identical optical properties. By detecting N consecutive markers, a position is inferred from the sequence of optical properties.
[0007] Various geocoding methods are known for naming locations, including postal addresses, but these are unsuitable for indoor navigation or even navigation outside of built-up areas. A special type of geocoding is called What3Words and is described, for example, in WO 2014 / 170646 A1. Here, three words from a fixed dictionary are assigned to each location with a resolution of approximately 3 meters. For example, the European Patent Office is located in Munich at / / / umfragen.melodram.baum. There is no obvious traceability of the three words assigned to a location, nor are there any simple neighborhood relationships. What3Numbers at https: / / github.com / hfu / what3num-bers uses a similar principle with three numbers instead of three words. Mapcode uses two groups of letters and numbers, possibly with a country code; here the address of the European Patent Office is Deu C57.MB.
[0008] However, such codes are purely abstract and only accessible via computer systems. There is no way to use sensors on-site to read the What3words code and thereby determine one's own location. Only the reverse is possible: first determine the position, for example, using GPS, and then query the corresponding code. However, this does not support self-localization, but remains solely based on GPS. Furthermore, there is no way to integrate a dedicated indoor area into such a system with self-selected subdivisions.
[0009] The as yet unpublished European patent application with the file number 22193749.3 describes a camera device that reliably determines its pose relative to an optical marker, in particular a fiducial marker, by using two camera channels. This allows the camera device to locate itself relative to the marker, but this does not contribute to solving the problem of determining the position of the marker itself.
[0010] It is therefore an object of the invention to further improve self-localization by optical markers.
[0011] This object is achieved by a method for self-localization in a navigation environment and a self-localizing vehicle according to claim 1 and 12, respectively. The navigation environment can be arbitrary, whereby the method is particularly suitable for indoor use where no satellite-based localization method is available. Optical markers are attached at a plurality of positions in the navigation environment. A code reader detects a marker within its field of vision and reads the code content of the marker in order to determine the position of the marker. Any device known for this purpose can be used as a code reader, for example a camera. Effectively, only the code reader localizes itself directly, but thus also its carrier, preferably an autonomous vehicle or a mobile robot (AMR, autonomous mobile robot).
[0012] The invention is based on the basic idea of combining optical markers into groups of at least three markers. Each individual marker has only a limited value range, meaning there are fewer, and preferably significantly fewer, distinguishable individual markers than positions with markers. For example, a single marker encodes only one digit, one letter, or only a two-digit number, with numerical code contents being preferred, while hundreds, thousands, or even many more positions can be distinguished. Only the code contents of the at least three markers in a group together unambiguously identify a position. The code contents for this purpose are preferably combined to form a combination code. The groups are preferably each formed by exactly three markers as an optimum between a small number of markers and a sufficient number of uniquely coding groups.A descriptive name for this idea, given the state of the art discussed in the introduction, is What3markers.
[0013] The process is computer-implemented, running, for example, in the code reader, a computing unit connected via a network or cloud, or, for example, a vehicle control system. Distributed implementation is also possible. Only the optical detection of the marker needs to be performed locally; any further processing can be outsourced. However, it is advantageous to at least perform the decoding locally, i.e., in the code reader or at least the code reader's carrier, so that the information to be exchanged externally remains limited.
[0014] The invention has the advantage of enabling a particularly cost-effective and at the same time very robust form of self-localization. Since the codes only have a small value range, for example just one digit, a very simple camera is sufficient to read the code content. The images from this camera can also be used for other tasks, such as obstacle detection. The principle of three markers in a group nevertheless makes a large navigation environment with high spatial resolution accessible. The density of the markers even allows the size of the navigation environment and spatial resolution to be scaled very easily. Self-localization is possible in one dimension, for example along a rail, in two dimensions, and even in three dimensions in a building with multiple floors or high shelves.Complex technical solutions such as indoor GPS or related systems such as ultrasonic, radar, Wi-Fi, or cellular navigation are unnecessary, as are complex algorithms for localization and mapping (such as SLAM). The concept enables a kind of swarm intelligence for multiple vehicles in the same navigation environment. Connection to higher-level systems, such as warehouse management, risk analysis, or data analysis and optimization of routes and other processes, is easily possible.
[0015] The navigation environment is preferably divided into a plurality of cells, in particular in a grid, each cell being assigned a group of markers. A desired position resolution can be specified via the cells. The cells are preferably regular and of equal size, as is particularly the case with a grid. Alternatively, cells can also be individual or irregular, or positions with a respective group of markers can be arranged arbitrarily. It is conceivable to leave at least one cell in a grid without a group of markers, especially when the attachment of markers is not possible in some places. As explained below, however, a group of markers can provide information about the environment, even about a cell without its own group of markers.
[0016] Preferably, a pose of the code reader is determined with respect to a detected marker in order to determine an offset from the position designated by the group of markers. The code reader thus localizes itself relative to the group of markers in up to six degrees of freedom of position and rotation. A respective optical marker preferably has at least three reference features of known positions within the optical marker. In particular, this is a fiducial marker (fiducial marker, frame marker). The three reference features enable pose determination in the manner of a tripod by defining a 2D geometry in space, with four or more reference features preferably being provided for redundant and overall more robust pose determination. Fiducial markers are known per se, and compliance is additionally referred to.By determining the pose, the position of the group of markers is known from their code content, as is the relative position of the code reader with respect to the marker, and thus the exact position of the code reader and thus its carrier. The mounting position of the code reader with respect to the carrier must also be taken into account. A camera with distance determination, i.e., a 3D camera, is particularly suitable for determining the pose of the code reader with respect to a detected marker. Any known 3D method is suitable for this, for example, stereoscopy based on human vision, triangulation methods for a known projected illumination pattern, or image sensors with time-of-flight (ToF) measurement in their pixels.
[0017] The pose is preferably determined multiple times for several markers, especially all markers in a group. Each marker is suitable for determining the relative pose of the code reader to the marker. Each group thus offers multiple possibilities for pose determination. An average pose can then be calculated.
[0018] The code contents of at least three markers in a group are preferably combined into a tuple of at least three. The code contents thus form a combination code, in the preferred optimal case with three markers per group, a triplet or triplet. This can be written as CodeContent1.CodeContent2.CodeContent3, in the preferred case of digits or two-digit numbers corresponding to c1.c2.c3, with c1...c3 as a digit or two-digit number. It should be noted that these are coordinates at most in the virtual parameter space of the possible unique identifiers, which preferably have nothing to do with world coordinates. A small difference between two combination codes cannot then be used to infer spatial proximity. The triplets are unique, but preferably without an intuitively comprehensible order. Particularly preferably, the order is irrelevant, so there is no difference between c1.c2.c3 and c2.c3.c1, and neither when attaching nor reading the markers, the order within the group needs to be taken into account. In the combinatorics of combination codes, this corresponds to drawing without replacement and would be completely unthinkable for conventional world coordinates. Alternatively, it is conceivable that the order is important after all. In this case, a symmetry breaker can be attached and recognized for each group—an additional optical feature that specifies a reading direction.
[0019] The code content of at least one marker in a group preferably contains additional information that goes beyond the unambiguous designation of the position with the other markers in the group. There are then, so to speak, two code families in the navigation environment: the inventive groups of at least three markers and additional optical markers that can be attached to the group. The additional information is not needed for the self-localization explained so far; it can, for example, relate to other navigation instructions or information for an application other than navigation. Furthermore, it is conceivable for a marker to fulfill a dual function in both code families. A different appearance, such as a different color, different size, or the like, is conceivable, whereby the marker of the second code family can also function as a symmetry breaker.However, the additional information can simply be found in the appearance of at least one marker, which is particularly larger to facilitate pose determination.
[0020] The markers are preferably displayed, at least partially, on a variable display. This can be a tablet, electronic paper, or the like. A variable display allows switching between visual markers, for example, in the event of an emergency such as a fire alarm. Alternatively, the markers, once applied, are fixed, typically in the form of stickers.
[0021] The code contents of at least three markers in a group are preferably used to query additional information from a database. The information content of a group is sufficient to determine a unique position in the navigation environment. However, there is often a need to learn more about the group or position. This could only be encoded in the markers to a very limited extent and would also counteract the idea of simple markers. However, using the unique group or the combination code or the triplet of code information, a database query can be implemented, which removes the limitation on the amount of information for practical purposes. The database is preferably located externally, especially in a cloud.At least parts of the database can be stored in the code reader or its carrier, for example, a vehicle, whereby this can also be built up gradually or temporarily as a buffer of previous queries and results. This creates a certain degree of independence, at least in the immediate vicinity or on frequently traveled routes. However, for an advantageous embodiment with a dynamic database, which will be described later, a regular comparison should then be performed.
[0022] The database provides at least one of the following pieces of group information for a position designated by a group of at least three markers: a detailed description of the position or its surroundings, an offset of the group of at least three markers from the position designated by it, and at least one navigation rule. This is called group information because it is assigned to a group and thus to a position. Group information can be used to describe the position in more detail, for example that it is building X or department Y. It can provide a description of the surroundings, such as information about obstacles or hazards in certain directions, or even a local map with neighboring cells and groups of markers there or their combination codes. One possible piece of group information indicates an offset of the coded position from the actual position of the group.This allows markers to be placed in a clearly visible or accessible position within or even outside their cell, especially deviating from a regular grid. A navigation rule provides instructions for the next movements or actions, and the next paragraph provides examples. Group information can also contain metadata such as timestamps, administrative data, or events, such as the age of the marker, a history of when and by which device this group was previously recorded, or the age and origin of navigation rules, or the time of their last modification.
[0023] The navigation rule preferably includes at least one of the following instructions: a direction, a distance, a speed specification, a request to pick up or put down an object, or a request to recharge a battery. It can therefore be a navigation rule in the narrower sense that specifies certain movements, such as move forward, turn left, turn right, turn around, or travel a certain distance, or that limits movements, for example, with regard to a speed or acceleration limit. However, other actions related to navigation are also conceivable, such as the instruction to follow a specific object, wait a certain time or for a certain event, pick up or put down something, or recharge the battery at a charging station.
[0024] At least one piece of group information in the database is preferably changed dynamically, in particular to react to a current task, to people, or to vehicles in the navigation environment. This enables flexible, situation-specific navigation. Through the indirection of the dynamic database, which is accessed via the group of markers or their triplet of code information or combination code, variable information can be effectively assigned to fixed optical markers. This allows a task to be presented differently or, for example, to react to vehicles or people in the navigation environment, in particular to adapt the routes depending on the situation.
[0025] A self-locating vehicle according to the invention, in particular an autonomous vehicle or a mobile robot, moves in a navigation environment in which positions are designated by groups of at least three markers. The vehicle and the navigation environment provided with the markers thus form a common system. A control and execution unit, for example the vehicle controller, is designed for self-localization and navigation by reading the codes of a group and determining the position uniquely designated thereby, in particular by an embodiment of the method according to the invention. The functionality of the control and execution unit can be distributed across the code reader, the vehicle, and connected computing units, in particular a cloud.
[0026] The invention will be explained in more detail below with regard to further features and advantages, using exemplary embodiments and with reference to the accompanying drawings. The figures of the drawing show: Fig. 1 is a schematic representation of a vehicle as an exemplary carrier of a code reader for detecting a group of optical markers, from whose code contents a position is determined; Fig. 2 is a schematic representation illustrating the division of a navigation environment into cells with respective groups of optical markers; Fig. 3 is a schematic representation of the communication connections of a vehicle with a code reader according to Figure 1and an associated database; Fig. 4 shows an exemplary representation of a group of optical markers with an additional symmetry breaker; Fig. 5 shows an exemplary representation of a group of optical markers from two code families; and Fig. 6 shows an exemplary representation of an alternative group of optical markers from two code families.
[0027] Figure 1shows a schematic representation of a vehicle 10 with a code reader 12 for detecting a group 14 of optical markers 16a-c, from whose code contents a position of the code reader 12 and, taking into account its mounting pose, also of the vehicle 10 is determined. The vehicle 10 is, for example, an autonomous vehicle (AGV, automated guided vehicle) or a mobile robot (AMR, autonomous mobile robot) and represents any carrier of the code reader 10, whereby this carrier can in principle also be a person. The group 14 of optical markers 16a-c enables self-localization of the vehicle 10.
[0028] Self-localization, i.e., knowledge of one's own position in a navigation environment 18, then enables navigation of the vehicle 10. Examples of the indoor navigation environment 18 include logistics centers, factories, administrative buildings, airport buildings, hospitals, exhibitions, or even vertical farming. The application is initially not limited to outdoor areas either, but is primarily aimed at limited areas not accessible by conventional geocoding such as postal addresses, such as ports, airports, loading zones of logistics centers, agricultural areas, or vineyards.
[0029] The code reader 12 can be designed as a simple camera, as shown, with an image sensor (not shown separately), which is preferably designed as a matrix-shaped recording chip and records a rectangular pixel image, for example, a CCD or CMOS sensor. The code content of a marker 16a-c is then read using known image processing methods. The invention is not limited to any specific code reader 12, as long as the markers 16a-c used are read. Image recording can occur as a secondary function of any image recording device, such as an on-board camera for obstacle detection or a tablet or mobile phone carried on board.
[0030] A control and evaluation unit 20 is connected to the code reader 12. In Figure 1, this is a vehicle control system. In general, however, any computing units carried on board or connected via a preferably wireless network connection are possible, which are based on a wide variety of hardware, for example one or more digital components such as microprocessors, ASICs (Application Specific Integrated Circuit), FPGAs (Field Programmable Gate Array), GPUs (Graphics Processing Unit) or hybrid forms thereof, wherein the digital components can be at least partially integrated into the code reader 12. External computing units include, in particular, an edge device or, as shown, a cloud 22. The control and evaluation functionality can be distributed as desired among these elements, which will be explained later with reference to Figure 3 preferred division of tasks are explained.
[0031] AprilTags are used as markers 16a-c, for example, although other optical codings are also conceivable. However, fiducial markers are preferred because they simplify the determination of a relative pose from code reader 12 to markers 16a-c. Such markers are also read much more robustly. The introduction provides further examples of optical codes and fiducial markers. The code content of each marker 16a-c is a digit or number; in the example shown, these are the numbers 04, 02, and 07. The value range of digits or even two-digit numbers is far too small to mark a significantly large navigation environment 18 with usable spatial resolution using unique individual markers 16a-c. Large numbers, in turn, make code reading more difficult, making it impossible to work with simple code readers 12, longer ranges, poor visibility conditions, and the like.
[0032] Therefore, markers 16a-c are grouped into groups 14. These are preferably groups of three or triplets, but the number k=3 is not initially fixed. Group 14 together exponentially expands the effective value range in which positions can be uniquely designated. The individual code contents of markers 16a-c are combined into a combination code or triplet, in the example of Figure 1 as of February 4, 2007, in any syntax. This can be referred to as "What3markers," in reference to the prior art cited in the introduction.
[0033] Depending on the embodiment, the order of the markers 16a-c and their code contents may or may not play a role, with the latter facilitating detection since no orientations or reading directions need to be taken into account. For a value range n, such as n=10 for a digit, for the individual markers 16a-c and a number k, such as k=3, of markers 16a-c per group, the following results when observing the order: n< and without observing the order ( n k ) unique combination codes corresponding to a drawing with / without replacement from the combinatorics. For the AprilTag family "36h11", n=587 applies, so that from three corresponding markers 16a-c, even in the somewhat less favorable case with regard to combinatorics, 201,229,470 unique combination codes can be formed without observing the sequence. This is sufficient for an area of approximately 201 km² with one group of 14 per m². Should the number of available combination codes still be insufficient, it is possible to assign higher-level codes to navigation environments 18, such as "X" floors in "Y" buildings in "Z" countries. This also has the advantage that large amounts of data can be summarized and evaluated across locations.
[0034] The combination code uniquely designates a position, but in a format unusable for navigation. This is because the distribution of the combination codes across the navigation environment 18 is preferably not based on the geometric arrangement at all, but rather on internal criteria, a history of repeatedly added groups 14, or even the specific requirement to keep spatially adjacent groups 14 as distinct as possible to avoid confusion. In embodiments in which the order of the markers 16a-c is disregarded, it is immediately apparent that there can be no meaningful geometric relationship between the combination code and the position in terms of spatial coordinates.
[0035] To obtain processable position information, the control and evaluation unit 20 preferably queries a database 24 with the combination code for the associated unique position. The database 24 is preferably implemented in a cloud 22, but can alternatively also be carried on the vehicle 10. The database 24 also removes the limitation on the information content of optical markers, because any additional group information can be made available in addition to the combination code. Group information means that this information is assigned to the group and thus to the position. This can include, in particular, navigation rules. The group 14 then functions as a type of traffic controller.
[0036] The markers 16a-c are preferably permanently arranged in the navigation environment 18. Suitable for this purpose are stickers that are printed accordingly. Alternatively, electronic displays are conceivable, such as tablets or e-book readers, with which the displayed markers 16a-c can be dynamically changed. For example, during maintenance or in an emergency, different markers 16a-c or completely different information can be displayed, which can also be used for diagnosis or the initial setup of the system. Electronic displays are primarily aimed at people. In a functioning system with an automatic sequence, changes are preferably made via the database 24, i.e. the markers 16a-c and thus the combination codes remain the same, but the group information stored for them can change.
[0037] Figure 2shows a schematic diagram illustrating the subdivision of a navigation environment 18 into cells 26a-b with respective groups 14 of optical markers 16a-c. Scalable subdivisions into regular grids or irregular cells 26a-b are possible, as shown by two alternative sizes with large cells 26a on the one hand or small cells 26b on the other. It is conceivable that an automatic distribution of cells 26a-b is performed, which determines an optimal grid constant or optimally fits an arrangement of cells 26a-b into a navigation environment 18.
[0038] The small cells 26b of the Figure 2can alternatively be understood differently as the result of a finer subdivision of the navigation environment with more groups 14 of optical markers 16a-c. In some embodiments, it is possible for the code reader 12 to determine its pose relative to a marker 16a-c in up to six degrees of freedom, taking advantage of where a marker 16a-c was detected within the field of view 28 of the code reader 12. This is preferably achieved using fiducials. AprilTags already incorporates a corresponding mechanism for self-localization relative to the marker. This then enables more precise self-localization corresponding to the small cells 26b, even though a group 14 of markers 16a-c is only attached in a respective large cell 26a. Since a group 14 has multiple markers 16a-c, the pose can even be determined multiple times with a systematic offset corresponding to the arrangement of the markers 16a-c within its group 14.Pose determination can thus become even more robust by averaging or comparing individual results.
[0039] It is not mandatory for each cell 26a-b to be provided with a group 14 of markers 16a-c. This is sometimes not possible because columns, shelves, doors, and the like are in the way. The vehicle 10 can then determine its position based on the pose relative to a group 14 of markers 16a-c of a neighboring cell 26a-b. Another possibility is for the group 14 of a cell 26a-b to be mounted outside its cell 26a-b and to communicate the offset as group information.
[0040] Figure 3 shows a schematic representation of the communication connections of a vehicle 10 according to Figure 1 , whose substructure is only shown in outline, and the associated database 24 preferably in a cloud. As already Figure 1As explained, the code reader 12 uses appropriate image processing to read the markers 16a-c of the group and combines their code contents into a unique combination code 30. The combination code 30 is used to query the database 24, and the unique position is returned. Further group information can be communicated, for example, navigation rules. The query via the combination code can optionally be secured with encryption or a checksum such as a CRC.
[0041] In the database 24, each combination code 30 forms an entry that links the combination code 30 to the associated unique position and possibly group information. This may include metadata such as timestamps for read markers 16a-c and combination codes 30 derived therefrom, timestamps for group information or changes thereto, or which code reader 12 used which image to query it. The mere position can be supplemented with further details of the environment of the respective cell 26a-b, for example, where exactly is the group 14 located in its cell 26a-b, in which building, which facility, which country is the cell 26a-b located, what is the environment like, be it with regard to existing objects, obstacles, hazards, which cells 26a-b are located in the neighborhood, in particular in the form of a local, area of interest of a global map, for example in the form of the eight surrounding cells 26a-b.
[0042] Navigation rules constitute a special group of information that tells the vehicle 10 how it should move or what movement options are available. These can be directional instructions such as straight ahead, left, right, or back; distance information regarding a destination or the next movement step; or specifications for speeds or accelerations. Other examples include requests to wait a certain time, to go to a specific position, to follow a specific object, to avoid an object, to choose or not choose certain routes, and the like. Navigation rules in a somewhat broader sense trigger an action, such as picking up or dropping off objects or charging a battery of the vehicle 10.
[0043] The markers 16a-c themselves are preferably fixed. An embodiment with electronic displays was briefly discussed above, but even this limits the variability, especially since the amount of information that can be displayed and captured by the code reader 12 is preferably already largely exhausted by the combination code. The system as a whole can nevertheless become highly dynamic by updating the database 24. There is no practically relevant limit to the group information stored in the database 24. The fixed combination code thus provides access to an adaptable, yet still clearly position-related, information source.
[0044] This enables situational navigation, not only in the sense that specific group information is provided at specific positions, but also by providing new group information at each specific position. The adaptation of navigation rules is Figure 3 summarized as a common external influence on the database 24 in block 32. In fact, there may be various reasons which, contrary to the representation, may also be based in the cloud 22, the database 24, or the vehicle 10.
[0045] Particularly advantageous is the ability to modify the navigation rules in a dynamic database 24. There can be many reasons for this, such as optimizing routes, fulfilling new tasks, safety measures, or maintenance work. Adapting routes appears particularly relevant, for example, depending on the number of vehicles 10 or people in the navigation environment 18, traffic jams, restricted areas, or other obstacles. If it is known that groups 14 of markers 16a-c are no longer legible, it may be advisable to avoid the corresponding cells 26a-b if possible, since self-localization will fail there, thus making navigation at least significantly more difficult.
[0046] Higher-level systems 34 in particular influence the navigation rules. Examples include fleet management, dynamic risk assessment (behavior-driven risk assessment), or maintenance information. Fleet management is responsible for ensuring that the system functions efficiently as a whole and, accordingly, dynamically specifies navigation rules for individual positions. It can collect the positions of the vehicles 10 and thus coordinate swarm intelligence. The risk assessment depends, for example, on the number of people and their current locations and tasks. Legal requirements, which may also be country-specific, can also play a role. Environmental information was already discussed above as an example of group information. This, too, does not have to be static, because objects in the navigation environment 18 change.Even virtual objects are conceivable, especially walls, that are communicated via group information. This allows sections of the building to be temporarily inaccessible, for example, when people are currently working there or performing maintenance. It is advisable to create a backup copy of the database 24 and keep a record of the changes, especially secured via a blockchain.
[0047] A publish-and-subscribe mechanism is preferably used as the underlying software architecture (a German term for this is not established; it can be roughly translated as "publish and subscribe"). Each arrow pointing to the database 24 can be interpreted in this sense. The combination code is published. An algorithm subscribes to this information, stores it, and determines associated position or group information, particularly navigation rules, from the database 24. This information is then published in reverse and subscribed to by the vehicle 10, which thus always receives up-to-date group information and navigation rules. The adaptation of navigation rules at block 32 can also be based on publish-and-subscribe.
[0048] An advantageous implementation utilizes cloud-native technologies such as K8s, K3s, or Minikube. Connecting to a cloud 22 requires at least an indirect connection of the code reader 12 to the internet or an intranet with the cloud 22. Such cloud-native systems already provide for changes to a database 24 from multiple sources, monitoring such changes, and reliable forwarding to the relevant subscribers. It is also possible to configure which information is forwarded at which times. Despite cloud technologies, part of the database 24 can be stored on the vehicle 10 itself to reduce data traffic and corresponding latencies.There may also be navigation rules that should not be changed due to safety regulations, legal requirements or company policy, which is either taken into account in the database 24 or is ensured by the vehicle 10 knowing such navigation rules and, if necessary, ignoring or overwriting specifications of the database 24.
[0049] Figure 4 shows an exemplary representation of a group of 14 optical markers 16a-c with an additional symmetry breaker 36. This refers to an optical element with which the group 14 is given a desired orientation or reading direction. This is particularly interesting when the order of the individual code information in the combination code is relevant. The specific graphic design of the symmetry breaker 36 is purely exemplary; any design elements, including company logos and the like, can be used.
[0050] Figure 5 shows an exemplary representation of a group 14 of optical markers 16a-c from two code families. This means that at least one of the optical markers 16a has an additional function that is not limited to the contribution to the combination code. The code content of the at least one optical marker 16a may be any additional information, which in particular replaces a naturally only static group information, without having to access the database 24. In the example of Figure 5The at least one optical marker 16a is simply larger to facilitate the accuracy and robustness of the pose determination discussed above. This effect can also be achieved through arrangements and other graphic elements, including the possibility of using a different coding, e.g., no April day. Furthermore, the at least one optical marker 16a advantageously also functions as a symmetry breaker.
[0051] Figure 6 shows an exemplary representation of an alternative arrangement of optical markers 16a-d from two code families. The overall arrangement of the markers 16a-d forms a group 14a, now understood somewhat differently. In this case, the optical marker 16d of the second code family is not part of the actual group 14 and therefore does not contribute to the combination code, while the at least one marker 16a of the Figure 5 also has a dual function as part of the combination code.
Claims
1. A method for self-localization in a navigation environment (18), in which a respective fiducial marker (16a-c) is attached at a plurality of positions, each of which comprises at least three reference features of known positions within the fiducial marker that enable a pose determination in the manner of the tripod, wherein a fiducial marker (16a-c) is detected with an optical code reader (12) and the code content thereof is read, characterized in that the position of the fiducial marker (16a-c) is determined from the code content by attaching the fiducial markers (16a-c) in groups (14) of at least three, wherein a value range of the code content of the fiducial markers (16a-c) is smaller than the number of positions having a fiducial marker (16a-c), and the code contents of the at least three fiducial markers (16a-c) of a group (14) are combined to form a combination code and thus together unambiguously designate the position of the group (14).
2. The method according to claim 1, wherein the navigation environment (18) is subdivided into a plurality of cells (26a-b), in particular in a grid, and the cells (26a-b) are each assigned a group (14) of fiducial markers (16a-c).
3. The method according to claim 1 or 2, wherein a pose of the code reader (12) is determined with respect to a detected fiducial marker (16a-c) to determine an offset from the position designated by the group (14) of fiducial markers (16a-c).
4. The method according to claim 3, wherein the pose is determined a plurality of times for a plurality of fiducial markers (16a-c), in particular all fiducial markers (16a-c), of a group (14).
5. The method according to any of the preceding claims, wherein the code contents of the at least three fiducial markers (16a-c) of a group (14) are combined to form an at least three-tuple, in particular without taking into account the sequence.
6. The method according to any of the preceding claims, wherein the code content of at least one fiducial marker (16a, 16d) of a group (14) contains further information which goes beyond the unique designation of the position with the other fiducial markers (16a-c) of the group (14).
7. The method according to any of the preceding claims, wherein the fiducial markers (16a-c) are at least partially displayed on a variable display.
8. The method according to any of the preceding claims, wherein the code contents of the at least three fiducial markers (16a-c) of a group (14) are used to query additional information from a database (24), in particular a database (24) in a cloud (22).
9. The method according to claim 8, wherein the database (24) provides at least one of the following group information for a position designated by a group (14) of at least three fiducial markers (16a-c): a more detailed description of the position or its surroundings, an offset of the group (14) of at least three fiducial markers (16a-c) from the position designated thereby, at least one navigation rule.
10. The method according to claim 9, wherein the navigation rule comprises at least one of the following instructions: a direction indication, a distance indication, a speed indication, a request to pick up or put down an object, a request to recharge a battery.
11. The method according to any of claims 8 to 10, wherein at least one group information of the database (24) is dynamically modified, in particular to react to a current task, to persons or to vehicles in the navigation environment (18).
12. A system comprising a self-locating vehicle (10) in a navigation environment (18) and a plurality of fiducial markers (16a-c) each comprising at least three reference features of known positions within the fiducial marker that enable a pose determination in the manner of the tripod and that are attached at a plurality of positions in the navigation environment, the vehicle (10) comprising an optical code reader (12) for detecting the fiducial markers (16a-c) and for reading their code content, as well as a control and evaluation unit (20) configured to carry out self-localization of the vehicle (10), characterized in that the fiducial markers (16a-c) are attached in groups (14) of at least three fiducial markers (16a-c), wherein a value range of the code content of the fiducial markers (16a-c) is smaller than the number of positions having a fiducial marker (16a-c) and the code contents of the at least three fiducial markers (16a-c) of a group (14) together uniquely designate the position of the group (14), and in that the control and evaluation unit (20) is configured to combine the code contents of the at least three fiducial markers (16a-c) of a group to form a combination code in order to determine their position.
