Method for creating a localization map, robot unit, computer program product and localization system

Abstract

A method for creating a localization map (10) of a robot unit (2) in an environment (13) is proposed, comprising at least the following steps: a. Sensory acquisition of location-related environmental information (9) by means of the robot unit (2); b. Storage of the acquired environmental information (9) in the localization map (10) in the robot unit (2); c. Determining a communication possibility (11) to another robot unit (2) and transmitting environmental information (9); d. Verification of received environmental information (9) based on at least one predefined criterion; and e. Updating the localization map (10) with the received environmental information (9) if at least one criterion is met. Robot units (2), computer programs, and localization systems are further specified.

Description

The present invention relates to methods for creating a localization map, a robot unit (in particular an autonomous vehicle), a computer program product and a localization system.It is known that, in particular for transport purposes, vehicles navigate independently through a predetermined environment, in particular on the basis of a navigation map which comprises the paths to be driven, and in the process access a map system, referred to here as a localization map with the information about the environment.For example, there is localization software for reliably determining the position and orientation of a wide variety of vehicle types in changing environments. This is made possible by a powerful algorithm which automatically detects the environment with the aid of a sensor on the vehicle, locates itself in a digital map of the environment (map) created from the environment data, and constantly renews this location in the map of the environment (map update).Mobile robots and / or forklifts (vehicles) sense their environment through sensors. Common examples of such sensors are laser scanners, ultrasonic sensors, cameras and / or interfaces for reading out QR codes.The map serves the vehicle for orientation. By matching the immediate environment with the map, the vehicle can determine its position. Flexible adaptation to changing environments is made possible by automatic updating of the map. The localization software automatically detects changes in the environment and continuously updates the map. The map is managed centrally on a server. Vehicles transmit their determined map deviations to the server. Newly recorded maps are stored here, as well as updates of the maps are calculated and distributed further to the vehicles. For this purpose, a WLAN connection (e.g. Wifi) between the vehicle and the server is necessary.The maps are currently managed on a central server. This requires a regular network connection covering as much surface as possible. For this purpose, a WLAN connection is generally established and operated. The vehicles need this connection to obtain a current map and / or upload changes to the map. A high bandwidth is necessary in order to be able to load all the generated changes of the map onto the server in real time and to be able to distribute them to all vehicles. If vehicles travel almost exclusively and / or for a longer time in areas without and / or with only a poor WLAN connection, then no changes to the map for these vehicles and no changes to the map in this area are possible.The installation and secure operation of a program on a server is complicated and complicated.One field of application is thus intralogistics, i.e. the logistic material and goods flows within an operating site. The intralogistics for goods transport, in which material and goods flows take place outside an operating site, is to be delimited.It is therefore an object of the invention to at least partially solve the problems described at the beginning and in particular to provide a method with which a map update can take place promptly and without a high bandwidth being required. It is also necessary to provide a method in which vehicles can receive and transmit changes to a map even in areas without and / or with only a poor WLAN connection.To achieve this object, a method for creating a localization map of a robot unit in an environment contributes, comprising the steps of a. sensorially capturing location-related environmental information by means of the robot unit; b. storing the captured environmental information in the localization map in the robot unit; c. establishing a communication possibility to a further robot unit and transmitting environmental information; d. verifying obtained environmental information on the basis of at least one predefined criterion; and e. updating the localization map with the obtained environmental information if this at least one criterion is fulfilled.This enables, in particular, a distribution mechanism for local map changes among the robot units. The method allows changes to the map to communicate decentrally between the robot units and then to be managed internally.Steps a. to e. can be carried out at least once in a predefined sequence, but this is not absolutely necessary. It is possible in particular for the steps to be at least partially superimposed in terms of time or to be carried out simultaneously. It is also possible for the steps to be carried out with different numbers of times and / or with a different repetition rate within the scope of the method.A robot unit relates in particular to a mobile robot and / or a mobile vehicle.A sensor detection in step a. can comprise in particular a scanning and / or filming by the robot unit. A sensory detection of the environmental information can comprise in particular the determination of a position, orientation and / or type of obstacles. The environmental information can be acquired in absolute terms and / or relative to other environmental conditions. A position and the type of obstacles can provide, as environmental information, indications as to whether it is a fixed obstacle, such as a wall, for example, or a movable obstacle, such as a human being or a means of transport, for example.The information about the position and the type of obstacles can also provide information about how sensitive the obstacle is. Various scenarios can be set up according to which the robot unit has to behave in emergency situations in which a collision can no longer be prevented. Thus, a collision with a human would be to be evaluated differently than a collision with a solid wall.Step a. is carried out in particular during a movement and / or during a navigated movement of a robot unit. In other words, the robot unit acquires environmental information along the predetermined or traveled path.According to step b., this acquired environmental information can be stored in the robot unit's own localization map. The storage can be effected in such a way that the stored environmental information is used or taken into account for the navigation of the robot unit directly or optionally after verification.It is possible that the data on the environment information is stored together with additional information. Additional information can be selected in particular from the following group: time of the detection, type of means / sensor for detection, height of a means / sensor for detection above the ground, orientation of a means / sensor for detection.The time of the detection can provide information about how an environment behaves at one time in comparison to another time. An obstacle may have been present at a location at a first time. At a subsequent second point in time, however, this obstacle has been moved and is no longer located at the previously detected location. The information about the point in time of the detection can provide information about whether the obstacle has been moved to a location or moved away from a location.The manner of using the sensing may allow for more accurate sensing of environmental information. Means for capturing environmental information can comprise, in particular, laser scanners, ultrasonic sensors, cameras and / or QR codes. Each of these means provides different data which must be put into a context in order to be able to be used correctly. It is possible to calculate possible inaccuracies in a detection of the environmental information with the knowledge about the means of detection and / or to combine different means of detection in order to reduce inaccuracies.Information about a height and / or an orientation of the means for detection makes it possible to take distortions into account, for example, during image analysis. Distortions can arise, for example, if the means for detecting can only partially recognize environmental information or in edge regions. Distortions can also arise if environmental information has been detected too quickly for the recording. Accurate detection is then not possible.It is possible to guide a plurality of versions of the localization map-for example based on different sensor systems, different heights and / or alignments of the sensors and / or different reflectors-in different planes. In particular, robot units with a special sensor system, for example comprising cameras and / or GPS, can guide special map planes which do not have to be guided and updated by all robot units.It is possible to distinguish moving objects and static objects from each other and define classes for moving objects and classes for static objects. Man or other moving robot units would thus be recognizable and could be located directly with a filter function. This could make it possible to create a localization map that excludes moving objects and only concentrates on (at least temporarily) stationary or stationary objects.In step c., a communication possibility is established if a data-exchanging connection between the robot units is possible. A possible connection is checked or evaluated, in particular, via an availability and a quality of a data transmission possibility. In a possibly defined short range, a possible connection can be given or can be evaluated as given. In this short range, communication between the robot units can be realized simply, largely protected / trouble-free and with a low energy expenditure. In particular, communication by means of radio, in particular by means of ad-hoc Wifi, Bluetooth, mobile radio (in particular LTE, 5G, 6G) and / or via a mesh may be successful.It is possible that a communication process is initiated as soon as a robot unit recognizes the presence of a further robot unit in the short range. It is possible for the robot units to first prepare together the transmission of environmental information, for example via the activation of suitable communication modules and / or channels. It is possible in particular for the robot units to decide on the basis of the preparation communication which robot unit transmits which data or environmental information. It is possible that only one robot unit sends environmental information and the other robot unit receives environmental information only.In step c., (also) mutual communication can take place via the sensory detection of environmental information. There is a mutual communication when the robot units both transmit and receive environmental information within the scope of a communication.In step c., a communication can additionally take place via the sensory detection of the environmental information. At least one of the above-mentioned additional information items can be communicated.In step c., an exchange of environmental information can take place in such a way that a relay takes place only in the case of environmental information which another robot unit does not yet have. Thus, environmental information already available to a robot unit does not need to be transmitted.According to step d., the robot unit has at least one criterion on the basis of which the robot unit can decide whether or not the environmental information obtained is to become part of the localization map to be used. It is possible that several criteria are used for this verification. In particular, the above-mentioned likewise obtained additional information can be taken into account. In particular, a history and / or compatibility of the environmental information can be compared and evaluated.All the environmental information is transmitted regularly or always. This contains a complete history of the ascertained environmental information. Each environment information is stored as a packet. In step d., therefore, in particular the transmission of all packets takes place.Whether a robot unit already has environmental information can (likewise) be detected on the basis of predefined criteria. The criteria can in particular comprise the time and the location of the detection of the environmental information. For example, two robot units may have environment information from the same location. The robot unit that has passed the location later thus has more recent information. This information is transmitted to the robot unit that has passed the location earlier. For conflicting environment information related to the same location, older environment information may be overwritten with new environment information. Such an overwriting uses information about the time of the detection of environmental information and thus determines which information is more current and thus more relevant. Older information that is no longer relevant is overwritten.In particular, starting from the result of step d., the updating of the entire localization map or only a part of the localization map can be carried out in the context of step e. By transmitting all packets of environmental information, it is possible to exchange only the packets that have changed in the history of a location map. By such a transmission of all environmental information, which also contains the history of the detection, intermediate states can also be detected, in particular, which characterize a transition from one state to another state. Thus, in a first state, for example (morning), a bearing could be filled and in another state (evening), the bearing could be empty. All intermediate states while the bearing is emptying over the day could thus also be detected.It is advantageous here that the packets of environmental information can be successively incorporated into a localization map. A consistent state of the localization map can thus be ensured and / or validated. It is thus possible, for example, to track and understand the stay of a box which is initially moved to a specified location and is then removed or moved. For each transmitted environmental information, it can be decided whether the localization map becomes more up-to-date and whether it is still consistent. In particular, the chronological sequence can be taken into account in this case, so that adaptations of the localization map take place correctly.In particular, it can be established within the scope of step c. that only those items of environmental information are transmitted which are relevant and verified (that is to say suitable for use with the own localization map). This results in a lower requirement for a transmission bandwidth.In particular, this method creates a map update and / or a (local) map update (automatically) by the robot unit.This specifies card management with the aid of decentralized version management. Incremental map updates are commidated. The sum of all commits applied in the correct sequence to one another or to an initial map yields a current overall map.At least one of the following possibilities can be realized: exchange of relevant commits in a wireless near-field encounter of the mobile robot units, management of the map in different semantic levels on the basis of sensor types (camera, GPS, etc.), management of the map in z-direction levels (heights); assignment of metadata to commits (date, robot unit, further attributes); qualification of the map data by metadata.To achieve this object, a robot unit contributes, having at least one sensor for detecting environmental information and a drive for moving the robot unit. Furthermore, a computing unit is provided which comprises at least one navigation unit and a memory and which is configured to regulate the drive. In addition, a communication unit and means are provided which are configured to execute a method for creating a localization map of a robot unit in the manner proposed here.With regard to the configuration of the robot unit, reference can be made in its entirety to the above explanations and / or the explanations in connection with the figures.A sensor for capturing environmental information can comprise in particular a laser scanner, an ultrasonic sensor, a camera and / or a QR code reader.A drive of a robot unit can comprise, in particular, electrical drives, such as accumulators, with which the robot unit can be moved, adjusted and / or moved. The drive or its operation offers further possibilities for capturing sensor data. Thus, the speed can be used to assist in acquisition of sensor data in that (image) data relating to environmental information can be reduced.A computing unit makes it possible to process sensor data. The computing unit comprises at least one navigation unit and a memory. The drive can be controlled by the computing unit on the basis of data from the navigation unit and the memory. The computing unit can control the drive in such a way that it follows a route specified by the navigation unit. The route may be modified with environmental information that may be stored in memory.The computing unit is furthermore connected to a communication unit and to means with which the method can be carried out. The computing unit may verify obtained environmental information from other robot units and update a location map in the memory with the obtained environmental information. The computing unit can thus store an updated localization map in the memory, with which it is possible to improve control of the drive.To achieve this object, a computer program product is included, comprising instructions that cause a robot unit to execute the method proposed here.With regard to the configuration of the computer program product, reference can be made in its entirety to the above explanations and / or the explanations in connection with the figures.To achieve this object, a localization system contributes, comprising a plurality of robot units that move autonomously in an environment by means of a locally stored localization map, wherein the robot units have sensors for detecting environmental information and, during the movement, detect environmental information by sensors and exchange it with one another, wherein the robot units are configured to carry out the method.With regard to the configuration of the localization system, reference can be made in its entirety to the above explanations and / or the explanations in connection with the figures.The robot units that move autonomously in an environment by means of a locally stored localization map are in particular so-called "automated guided vehicle" (AGV).The locally stored localization map of the localization system can be used exclusively as a target.(i.) itself detected by sensors and / or(ii.) Environmental information acquired directly by sensors by other robot units and communicated directly and / or indirectlyThe updates can be updated.It is possible to communicate data directly detected by other robot units by sensor directly and / or indirectly as environmental information. Thus, the data acquired by a robot unit by sensor can be passed on directly as environmental information to a further robot unit when communication takes place between two robot units. It is also possible to pass data acquired by a robot unit directly as environmental information to a third robot unit. Thus, a third robot unit can obtain data from a first robot unit via a further robot unit which communicates first with the first robot unit and then with the third robot unit.In other words, the invention can also be described as follows: A decentralized graph-based distribution mechanism for local updates of the localization map enables a localization map to be maintained and updated in a so-called "peer-to-peer" communication. Individual robot units can exchange each other in a decentralized manner, without a server, in pairs, align their maps with each other and update the localization maps on the basis of the shared knowledge. Updates of localization maps can thus be introduced and fetched by advancing and retracting robot units in areas without a central network connection.Thus, remote locator map management is proposed that can store, apply, and combine updates to the locator map. This corresponds to a kind of Git version management for localization maps. The version manager may store small-scale updates of the location map (e.g., a single incoming container as an obstacle) as a commit. The sum of all commits applied in the correct order and / or on an initial one yields the current localization map.If two robot units meet, they exchange only the relevant commits with one another, i.e. only those which the other robot unit does not yet have and which it can install in its graph. Version management allows this to be done in small data packets, which are also applicable individually if the connection is only short. Also, those packets that are relevant can be quickly identified over the commit graph.The entire history of the map can be carried in the version management graph, so that more recent changes can be distinguished from older changes and can be installed successively in time. The bandwidth required for the transmission is comparatively low. Only incremental changes are transmitted.A central server may be omitted, which saves administration effort and may make the system robust against failures of the central software component. The robot units can thus continue to operate even in the event of failure of an individual robot unit. It is also possible for robot units to be operated independently of an existing infrastructure and thus also to be able to operate in areas without a central radio connection and to exchange updates of the localization map there.The localization map created by the robot units, as well as the updates of the localization map, can be stored in a commit graph similar to a Git version management. Each robot unit forms a decentral node. It participates in managing all locator maps by each node or robot unit. All robot units check new data at the nodes and can thus participate in the generation of a new localization map and check it.Conflicts in location map updates can be avoided by maintaining a "Nice value" (criterion) at which vehicles can assess the quality of their location maps mutually.The invention and the technical field are explained in more detail below with reference to three figures. Identical components are identified by the same reference numerals. The representations are schematic and are not provided to illustrate size relationships. The explanations given with reference to individual details of a figure can be extracted and freely combined with facts from other figures or the above description, unless something else necessarily arises for a person skilled in the art or such a combination is explicitly prohibited. They show schematically: FIG. 1 : robot unit in a side view FIG. 2 : shows an embodiment of the method on the basis of a history in a graph-based representation form, and FIG. 3 : a localization map with three robot units.FIG. 1 shows a robot unit 2 in a side view. The robot unit 2 has a drive 7 and a sensor 1. The arithmetic unit 3, a navigation unit 4 and a (data) memory 5, configured for regulating the drive 7, and a communication unit 6 are also schematically shown.FIG. 2 illustrates the method based on a history in a graph-based representation form similar to Git version management. The movements 8 of three robot units 2 can be seen. The direction of movement 12 indicates the time profile of the events. The robot units 2 first use their own stored localization map 10 (not shown, see FIG. 3 ). When a communication possibility 11 is established, the robot units 2 transmit environmental information 9 to one another. The environmental information 9 obtained (not shown, see FIG. 3 ) is verified on the basis of at least one criterion and the localization map 10 is updated if this at least one criterion is fulfilled.FIG. 2 also illustrates how environmental information 9 can be passed on to robot units 2 if there is no direct communication possibility 11. Environment information 9 can be passed indirectly from a first robot unit 2 to a second robot unit 2 when the environment information 9 is passed on via a third robot unit 2.FIG. 3 illustrates a localization map 10 with three robot units 2. The robot units 2 move in an environment 13 and acquire environment information 9 during the movement.

Claims

Method for creating a localization map (10) of a robot unit (2) in an environment (13), comprising the steps of a. sensorially acquiring location-related environmental information (9) by means of the robot unit (2); b. storing the acquired environmental information (9) in the localization map (10) in the robot unit (2); c. establishing a communication possibility (11) to a further robot unit (2) and transmitting environmental information (9); d. verifying acquired environmental information (9) on the basis of at least one predefined criterion; and e. updating the localization map (10) with the acquired environmental information (9) if this at least one criterion is fulfilled.Method according to the preceding claim, wherein step a. is carried out during a movement (8) of a robot unit (2).Method according to one of the preceding claims, wherein step b. leads to an update of the localization map (10).Method according to one of the preceding claims, wherein in step c. a communication possibility (11) is determined if a connection between the robot units (2) is possible.Method according to one of the preceding claims, wherein in step c. a mutual communication of environmental information (9) takes place.Method according to one of the preceding claims, wherein in step c. additionally a communication takes place via the sensory detection of the environmental information (9).The method of any preceding claim, wherein step e. performs the updating of only a portion of the locator map (10).Robot unit (2) having at least one sensor (1) for detecting environmental information (9), a drive (7) for moving the robot unit (2), a computing unit (3) comprising at least one navigation unit (4) and a memory (5) configured to regulate the drive (7), and a communication unit (6) and means configured to carry out a method according to the preceding claims.A computer program product comprising instructions that cause a robotic unit (2) according to claim 8 to perform the method according to any one of claims 1 to 7.Localization system, comprising a plurality of robot units (2) moving autonomously in an environment (13) by means of a locally stored localization map (10), wherein the robot units (2) have sensors (1) for acquiring environment information (9) and acquire environment information (9) sensorially during the movement and exchange it with one another, wherein the robot units (2) are configured to carry out a method according to one of claims 1 to 7.Localization system according to claim 10, wherein the locally stored localization map (10) is updated exclusively with (i.) itself sensorially detected and / or with (ii.) environment information (9) directly detected by other robot units (2) and directly communicated.

Images