DE102020107899A1 - Device for correcting deviations in localization information of a planning level and an execution level - Google Patents

Abstract

A device (300) is disclosed for correcting deviations in localization information (115, 215) of a planning level (100) and an execution level (200). The planning level (100) is designed to repeatedly determine first localization information (115) and, based on this, to create planning data (125) for controlling an object, and the execution level (200) is designed to repeatedly supply second localization information (215) determine. The device (300) comprises a receiving module (310) which is designed to continuously receive the first localization information (115) and the second localization information (215) and to continuously receive the planning data (125), an evaluation module (320), which is designed to determine a deviation for first localization information (115) and simultaneous second localization information (215), and a correction module (330) which is designed to correct the planning data (125) or using at least one determined deviation for at least one of the second location information (215).

Description

The present invention relates to a device for correcting deviations in localization information of a planning level and an execution level, a system with separate planning and execution of a control, an object controlled by such a system, a method for correcting deviations in localization information of a planning level and a Execution level, and in particular a method for correcting different localization data in machine-controlled systems.

BACKGROUND

Systems that plan a target trajectory under machine control and then execute a corresponding movement - such as moving robots, autonomous vehicles or robotized aircraft (e.g. drones) - can have different organizational architectures. For a rough classification, a first class of such systems can be found in which planning and execution are coupled with one another or integrated into one another, for example in such a way that execution-related deviations of the movement from the target trajectory are taken into account when planning a next target trajectory . In a second class of such systems, however, planning and execution are distributed on different levels or on different devices. The levels are connected via an interface via which the planning only transfers a description of the planned behavior (for example the target trajectory in the form of the temporal sequence of positions and alignments to be implemented) to the executing level.

The organizational architecture of the second class of systems offers advantages or freedoms, particularly in the design phase, which otherwise cannot be achieved. This includes, for example, the possibility of making changes within the planning or execution without having to carry out an often necessary check of all functions of the system afterwards. Rather, such a test can be limited to the respective level of the changes and thus tasks in the production and maintenance of the system can be better distributed.

At the same time, however, the separation creates new problems that need to be solved for the architecture to be used successfully. Such a problem occurs in particular when there is more than one service for localization, that is to say for determining actual values of positions and orientations. This can be desirable or useful for reasons of redundancy, for example. As a result of the double localization, inconsistencies or systematic errors can occur between the two positions and alignments obtained, since both services lead to different deviations from true values depending on the sensor technology used - for example due to different measuring devices, but also due to different environmental conditions. Depending on how the localization information obtained is used, such an inconsistency can lead to serious problems. If the localization information from the two services differs, which is generally assumed in operation, this leads, for example, to a control deviation, which in turn leads to the specification of undesired control commands for the actuators. Such behavior is undesirable and must be prevented during operation. Deviations from the rules may only arise from an actual difference between the planned target localizations and the actual actual localizations.

There is therefore a need to compensate for deviating localizations within the framework of a modular implementation of planning and execution, by means of which incorrect execution specifications can be avoided.

BRIEF DESCRIPTION OF THE INVENTION

Such an object is at least partially achieved by an apparatus according to claim ₁ , a system according to claim 6, an object according to claim 9 and a method according to claim 12. The dependent claims relate to advantageous developments of the subjects of the independent claims.

The present invention relates to a device for correcting deviations in localization information of a planning level and an execution level. The planning level is designed to repeatedly determine first localization information and, based on this, to create planning data for controlling an object. The execution level is designed to repeatedly determine second localization information. The device comprises a receiving module which is designed to continuously receive the first and the second localization information and to continuously receive the planning data. The device further comprises an evaluation module which is designed to determine a discrepancy for simultaneous first and second localization information. The device also detects a correction module which is designed to carry out a correction for the planning data or for at least one of the second localization information items using at least one determined deviation. The correction module can also use the Correct planning data as well as the localization information, for example by transferring all localizations into planning data and second localization information in a further reference system.

The term localization information is intended to stand for information about a state, in particular a position and / or orientation at a specific point in time, of an object or object to be controlled, so that a sequence of chronologically successive localization information enables the object to be controlled. The above-mentioned deviations, in particular between simultaneous localization information from the planning and execution level, arise, for example, from the use of different sensor devices or from different processing errors in the context of the creation of localization information in the planning and execution level.

The terms planning level or execution level can be understood here in each case to mean devices made up of one or more devices and their connections, the use of which can mostly be assigned to planning a control or an execution of a control.

In exemplary embodiments, the localization information includes in particular poses from a position and an orientation of the object to be controlled as well as a time stamp which determines the point in time at which the pose was determined. Such localization information is repeated, for example at a specific time frequency or also according to conditions resulting from movements of the object or object, in or for the planning level and in or for the execution level.

In exemplary embodiments, the planning level is particularly responsible for creating desired trajectories, which it outputs as planning data, for example in the form of a sequence of desired future localization information. The execution device processes planning data as well as its own localization information into control commands for an actuator system that causes the object or object to move.

The present device now determines a correction for planning data or at least for localizations of the execution level which compensates for deviations between the localizations from the planning and execution level. The receiving module of the device can in particular have a memory in which localization information and / or planning data are held, for example, for a certain period of time. The evaluation module can determine simultaneous pairs from the first and second localization information and determine deviations between two localization information. Simultaneousness is to be understood in the context of a certain tolerance, which can be caused in particular by the measurement accuracy and the frequency of the creation of localization information. The correction is determined and applied in the correction module of the device, it being possible for a correction that has been determined once to be applied multiple times - for example until it is determined again. The modules of the device - in particular the receiving module, the evaluation module and the correction module - do not have to be structurally present as separate units, but can be combined, as non-independent components of other parts of the device or functionally inseparable.

Optionally, the correction module is designed to redetermine the correction at regular time intervals and / or at the request of the receiving module according to predetermined criteria, for example after receiving planning data or localization information.

In exemplary embodiments, planning data are created, in particular, at a lower frequency than localization information in the execution level. As already mentioned, the correction module can then, in particular when correcting second localization information, redefine a correction when planning data is received, and then continuously change the second localization information with the same correction.

Optionally, the correction module is also designed to redetermine the correction according to a predefined condition on two second localization items of information offset in time.

Such a condition can include, for example, the condition that two temporally successive localization functions have a deviation which lies beyond a threshold value condition and may be due to a sudden correction of a measurement error or a reinitialization of sensors to determine the localization information.

Optionally, the evaluation module is also further designed to determine and store a degree of dispersion such as a standard deviation or a variance from localization information, the condition just mentioned being applied to the two temporally offset second ones Localization information takes this degree of dispersion into account. For example, the condition can include that a piece of localization information lies wholly or partially outside a range of the degree of dispersion that applies to unrestricted functionality.

Optionally, the device further comprises a warning module which is designed to output an electronic warning message for the deviation or for the correction when a predefined condition is fulfilled.

Exemplary embodiments also relate to a system for planning and controlling a movement of an object. The system comprises a planning level which is designed to repeatedly determine first localization information and to create planning data for a controller, an execution level which is designed to repeatedly determine second localization information and to execute a control based on the planning data, and a device to correct deviations between the localization information of the type described above.

The planning level and / or the execution level are optionally designed to cause the correction module to redetermine a correction after a predefined condition of localization information.

In exemplary embodiments, localization information can already contain information on a degree of dispersion or an error resulting from a measurement inaccuracy when it is created, or the planning level and / or the correction device themselves determine deviations or discontinuities in temporally successive localization information. According to defined criteria, the correction module can therefore be made to carry out additional correction determinations.

Due to the various aforementioned configurations of parts of the system to move the correction module to determine a correction, sudden changes in localization information, such as those caused, for example, by an environmental or system-related correction of accumulated sensor errors, or by recalibrating or reinitializing sensors, can be made possible arise, are absorbed in their effects on the control commands.

Optionally, the planning level is designed to generate the first location information at a first rate, and the execution level is designed to generate the second location information at a second rate, the first rate being different from the second rate.

Exemplary embodiments also relate to an object which comprises a system of the type described above in order to plan and execute movements in an automated manner.

• - Regelung einer Pose des Fahrzeugs,
• - Regelung einer Geschwindigkeit des Fahrzeugs,
• - Regelung einer Beschleunigung des Fahrzeugs,
• - Regelung einer Fahrdynamik,
• - Regelung einer Fahrtrajektorie,
• - Übergangsregelung in ein sicheres Anhalten,
• - einen Leitwartenbetrieb.

This object is optionally a vehicle, and the movements are at least partially used to automate normal operation of the vehicle. In particular, the movements can cause one or more processes from the following list:

• - regulation of a pose of the vehicle,
• - regulation of a speed of the vehicle,
• - regulation of an acceleration of the vehicle,
• - regulation of driving dynamics,
• - regulation of a driving trajectory,
• - Transitional regulation in a safe stop,
• - a control room operation.

The vehicle can in particular be an automated motor vehicle such as a passenger or truck, a bus or an industrial truck, but also a rail vehicle.

• - ein mobiler Roboter,
• - ein Fluggerät wie beispielsweise ein Flugzeug, eine Drohne oder eine Rakete,
• - ein Wasserfahrzeug,
• - ein Haushaltsgerät, wie beispielsweise ein automatisierter Staubsauger oder Rasenmäher.

Optionally, the object can also be a device from the following list:

• - a mobile robot,
• - an aircraft such as an airplane, a drone or a rocket,
• - a watercraft,
• - a household appliance, such as an automated vacuum cleaner or lawn mower.

• - Bestimmung von ersten Lokalisierungsinformationen, zweiten Lokalisierungsinformationen und Planungsdaten;
• - Auffinden einer ersten und einer zeitgleichen zweiten Lokalisierungsinformation;
• - Feststellen einer Abweichung zwischen den zwei Lokalisierungsinformationen;
• - Durchführen einer Korrektur der Planungsdaten oder mindestens einer Lokalisierungsfunktion unter Verwendung der Abweichung zwischen den zwei Lokalisierungsinformationen.

The present invention also relates to a method for correcting deviations in localization information of a planning level and an execution level, wherein the planning level is designed to repeatedly determine first localization information and based thereon to determine planning data for controlling an object, and wherein the execution level is designed is to repeatedly determine second location information. The process is characterized by the following steps:

• - Determination of first location information, second location information and planning data;
• - Finding a first and a simultaneous second localization information;
• - Detecting a discrepancy between the two location information items;
• - Carrying out a correction of the planning data or at least one localization function using the discrepancy between the two localization information items.

Embodiments of the method also include, in particular, a method for a movement system that plans a trajectory in a machine-controlled manner and executes the movement for it, and divides the planning and execution functions into different modules. Independent, independent localization functions are used for planning and execution. The method is characterized in that an execution consistent with the planning is guaranteed by a permanent comparison of possible deviations between the localization functions.

In exemplary embodiments, there are control deviations between planning data in the form of a target trajectory on the one hand and localization information in the form of actual poses on the other hand, and the method takes into account the positional difference in the correction with reference to the time of the last trajectory planning. The comparison can in particular modify the target trajectory transferred by the planning module, but independently of this also the second localization information, that is to say the actual values determined by the localization function of the execution. In particular, both planning data and the second localization information can be changed in the method, for example by the comparison transferring the target trajectory transferred by the planning module and the actual values determined by the localization function of the execution to a third reference system. In exemplary embodiments of the method, comparison or correction and measurement updates for the localization function of the execution are also carried out in a synchronized manner.

When the method is carried out in exemplary embodiments, the system also includes a module for self-observation with the task of a functional check. If predefined criteria are exceeded, the self-observation module is notified or alerted to the correction or to a discrepancy between the localization information.

• In einem technischen System, insbesondere in einem automatisierten Fahrzeug, das maschinell gesteuert eine Trajektorie plant und die Bewegung dafür ausführt, und bei dem Planung und Ausführung dieser Steuerung auf unterschiedlichen Ebenen verteilt und lediglich über eine oder mehrere Schnittstellen verbunden sind, werden in beiden Ebenen jeweils Sensoren (beispielsweise GPS und Kamera) zur Schätzung der eigentlichen Position bzw. Pose verwendet, die aufgrund ihrer Eigenschaften auf unterschiedliche Ergebnisse führen können. Eine Vorrichtung der hier vorgestellten Art verhindert Regelabweichungen durch Herausrechnen der Fehler. Außerdem können unerwünschte Effekte der Sensordatenfusion der Lokalisierungsfunktion auf die ausführende Ebene verhindert werden.

In training examples, the present device offers, inter alia, the following advantages:

• In a technical system, in particular in an automated vehicle, which plans a trajectory under machine control and executes the movement for it, and in which the planning and execution of this control is distributed on different levels and only connected via one or more interfaces, in each case Sensors (for example GPS and camera) are used to estimate the actual position or pose, which, due to their properties, can lead to different results. A device of the type presented here prevents system deviations by calculating the errors. In addition, undesired effects of the sensor data fusion of the localization function on the executing level can be prevented.

Furthermore, changes in a module of the device or a level can easily be integrated into the overall system due to the modular structure of the system. In particular, in the event of the smallest changes, all components and functions of the system must be checked for security with a corresponding expenditure in terms of costs and personnel. Due to the modular structure, this test can be concentrated on one module or, in the case of modules not affected by the changes, at least be limited considerably.

By separating different localization functions that use different sensors, the failure safety and redundancy within the overall system can also be increased.

In exemplary embodiments, the device converts all poses into a uniform reference system. This makes it possible, for example, to use a localization function with a lower data rate for the planning level and a localization function with a higher data rate for the executive level. In addition, the pose offset between the localization functions can be monitored and transmitted to the self-monitoring of the overall system for function monitoring.

The synchronized comparison of the deviations in the localization functions can enable consistent and jump-free trajectory control even when planning and execution are separated. Measurement updates that lead to a change in the difference between the localization functions are synchronized in such a way that they only become effective when the target trajectory is updated.

Figure list

• 1 zeigt ein Schema für ein Ausführungsbeispiel der vorliegenden Erfindung.
• 2 zeigt zusätzliche Details für eine Korrekturvorrichtung im Rahmen eines Ausführungsbeispiels in einem automatisierten Fahrzeug.
• 3 zeigt ein Verfahren für eine Korrektur von Lokalisierungsinformationen aus einer Planungs- und solchen aus einer Ausführungsebene.

The exemplary embodiments of the present invention will be better understood with reference to the following detailed description and the accompanying drawings of the various exemplary embodiments, which, however, should not be understood in such a way that they restrict the disclosure to the specific embodiments, but merely serve for explanation and understanding.

• 1 shows a scheme for an embodiment of the present invention.
• 2 shows additional details for a correction device within the scope of an exemplary embodiment in an automated vehicle.
• 3 shows a method for correcting localization information from a planning level and such from an execution level.

DETAILED DESCRIPTION

1 shows a scheme for a device 300 to correct discrepancies between localization information 115 a planning level 100 on the one hand and localization information 215 an execution level 200 on the other hand. Within the planning level 100 determines a first localization function 110 - for example from data from the planning level 100 assigned sensors - repeats initial localization information 115 . Analog creates another, the execution level 200 assigned second localization function 210 also repeats second localization information from sensors assigned to it 215 . Within the planning level 100 will be the first localization information 115 in one planning unit 120 used to planning data 125 or to create a target trajectory for a controller. The planning data 125 as well as both localization information 115 , 215 are connected to a receiving module 310 in the device 300 transfer. The receiving module 310 saves the location information 115 , 215 continuously, for example in a ring buffer. The device 300 then includes an evaluation module 320 , which is designed to provide simultaneous localization information 115 , 215 to find and discrepancies between two localization information 115 , 215 ascertain. There can also be a discrepancy between two first pieces of localization information 115 or between two second location information items 215 to be determined. The device 300 also includes a correction module 330 , which is designed to use at least one determined discrepancy between a first localization information 115 and a simultaneous second location information 215 a correction for the planning data 125 and / or for at least one of the second location information 215 to determine. In addition, there is the correction module 330 trained to make correction based on planning data 125 or to second localization information 215 apply. The planning data 125 or the second localization information 215 are considered corrected data 335 one of the execution level 200 assigned execution unit 220 passed on. The execution unit 220 determined from the localization information 215 and the corrected data 335 Control commands for an actuator (not shown here) that causes the movement.

An exemplary embodiment of an object with a system of the type shown in the figure comprises an automated or autonomous motor vehicle with a service-oriented architecture (SOA). In service-oriented architectures, certain functions are provided by individual services. These services are clearly delimited (modularized) and have defined interfaces with other services. In particular, these services include those on a planning level 100 as well as an execution level 200 correspond.

A highly available and precise localization of the vehicle is essential for automated vehicle guidance. In the present exemplary embodiment, localization is understood to mean the estimation of a current pose, that is to say a current position and orientation of the vehicle. Various types of sensors, such as inertial measuring units, sensors of a global navigation satellite system (GNSS), sensors for estimating the position based on data from the propulsion system of the vehicle (odometry) or camera sensors can be used for localization. Each type of sensor has specific advantages and disadvantages, which enable different levels of accuracy of the pose estimation in different environments (e.g. canyons, autobahns, country roads). The individual sensors can be fused with one another as part of a sensor data fusion (eg in a Kalman filter) so that the advantages of several sensors can be combined and a larger number of different possible architectures for the localization functions 110 , 210 arises.

The behavior of the vehicle is largely determined by the two different levels of the planning level 100 and the execution level 200 influences: The planning level 100 plans the desired behavior while the execution level 200 implements this target behavior by generating control commands for the actuators in the vehicle. The device is located between the two levels 300 located as a deviation or offset correction, with the help of which a difference between the two poses 115 , 215 determined and the target behavior 125 or poses used for the target behavior 215 corrected accordingly.

The added value of this architecture should now be illustrated using this exemplary embodiment of an automated motor vehicle. The planning level generates behavior and trajectory planning 100 of the vehicle's planning data 125 which in particular include a target trajectory that the vehicle has to travel. This target trajectory comprises a sequence of poses, that is to say of translational positions in space and the orientation of the vehicle, together with time stamps that describe when the respective pose should be present. Current first localization information lies in the target trajectory 115 in the form of an actual pose of the vehicle, mostly synchronized at the time of the last environmental measurement by environmental sensors. The localization information 115 the first localization function is used to plan the trajectory 110 (for example with GNSS or camera sensors).

Without the device 300 would get the execution level 200 the target trajectory to be driven through a direct transfer of the planning data 125 . The execution unit 220 in the execution level 200 calculates setting commands for the actuators in order to achieve the desired target behavior. Controllers are often used here, which determine the deviation between the target and actual pose and generate positioning commands on the basis of this control deviation. Since the execution level 200 especially with its own localization function 210 poses from different sources are compared with one another. The two poses determined - a piece of localization information from the planning data transmitted directly 125 and one by the second location function 210 determined second localization information 215 - can differ due to the different sensors used.

The problem here can be illustrated using the example of the vehicle at a standstill. Is the current actual pose (localization information 115 ) defined as the target pose, the vehicle should be stationary. However, finds the second localization function 210 due to systematic sensor errors, a different actual pose for one of the second localization information 215 , a control deviation arises in the controller solely due to this sensor error. Without a corresponding correction, the consequence would be that control commands are determined for the actuators that correct this deviation. After the control interventions, the vehicle would, at best, be offset from the actually desired pose by a distance associated with the sensor errors.

The device 300 now offers a solution to problems of this kind. On the one hand, it receives the planning data 125 , in particular the planned target trajectory in each time step, and on the other hand localization information 115 , 215 from the two localization functions 110 , 210 . The pose from the first localization function is relevant here 110 , on the basis of which the current planning data 125 were created, and a corresponding pose in a second localization function to be determined 210 at the same time. In exemplary embodiments, the planning data include 125 In addition to the target trajectory, the localization information used for planning 115 , for example in the form of a pose with an associated time stamp. The second location information 215 (which also include poses and time stamps) can be stored in a ring buffer in the receiving module 310 be held up. The time stamp makes it possible to determine a corresponding pair of poses. A systematic deviation or offset can then be calculated from the two poses. All poses contained in the target trajectory are corrected by this offset, so that there are deviations in the two localization functions 110 , 210 do not lead to a control deviation and thus to control commands for the actuators. Alternatively, the actual values or localization information can also be used 215 the second localization function 210 corrected by the offset or both values transferred to a third reference system, which can lead to an identical result.

The deviations in the localization information 115 , 215 can differ from the true value with each recalculation using the two localization functions 110 , 210 change, which is why there is a need for synchronization in order to avoid jumps in the regulation. Changes in the deviations from the true value can, especially in cases where at least one of the localization functions 110 , 210 Location information 115 , 215 at a higher data rate than the correction by the device 300 outputs, jumps in the output actual pose of this localization function occur until the correction is determined again. If such a jump occurs in particular with two chronologically successive second localization information items 215 this can lead to major errors and undesirable behavior on the execution level 200 to lead. This is done by means of a synchronization between new correction determinations by the correction module 330 and the localization functions 115 , 215 prevented. If between two temporally consecutive first localization information 115 or second location information 215 If there is a jump or a difference to be assessed as discontinuous, this is on the one hand based on the course of the actual pose and possibly further information such as, for example, the associated standard deviation, which is also contained in the localization information 115 , 215 included or in the device 300 can be calculated and held on the basis of their course, recognizable. On the other hand, there are jumps at the output of one of the localization functions 110 , 210 in certain steps in the processing sequence, such as during a measurement update in the Kalman filter or when the filter is re-initialized, with an increased probability, and can therefore - for example after determining an actually occurred deviation or as a general rule - be marked by the relevant localization function and the device 300 are displayed. Thus, the device lies 300 the information about a jump at the output of one of the localization functions 110 , 210 before. By (early) redetermining a correction before processing the pose in the execution level 200 undesired behavior of the actuators can be prevented. In the corresponding step, for example, the jump in the actual pose can be made by correcting the target pose in the planning data 125 be compensated so that the target / actual comparison in the execution level 200 remains unaffected by the jump in the actual pose.

An example of such a jump in the actual pose can be found in the second localization function 210 arise in a motor vehicle when exiting a tunnel. When the second localization function 210 For example, data from a GNSS receiver and an inertial measurement unit (IMU) are processed in a Kalman filter and the navigation calculation of the IMU data is used as the basic system, there are no measurement updates in the tunnel due to the restriction of GNSS reception. During the journey in the tunnel, the errors of the IMU are therefore added up or integrated in the navigation calculation. When exiting the tunnel, this error appears when localization information is generated again 215 corrected, and the actual pose at the output of the localization function 210 changes in leaps and bounds. If this occurs before the desired positions still to be reached in the corrected data 135 themselves have been corrected for the consequences of this jump, it leads to undesirable behavior on the part of the execution level 200 . With the help of the device 300 can cause this undesirable execution level behavior 200 due to different reference systems of the target and actual values. The offset correction can be done in two steps and runs with the same data rate as localization information 215 in the execution level 200 away. The first step is to check whether there is any new planning data 125 or a jump in the second location information 215 exist and therefore there is a need for action for the offset correction. A jump in the second location information 215 can be detected, for example, by monitoring the estimated uncertainty. In the case of a measurement update that involves a jump in the localization information 215 As a result, the estimated uncertainty will suddenly decrease, which is in contrast to the constant or slowly increasing regular behavior (e.g. as it can occur frequently or usually within certain limits) of the estimated uncertainty. If there is a case of new planning data 125 or a jump in the second location information 215 a, is in the correction module 330 a new pose offset or a new correction is calculated. The poses of the target trajectory in the planning data 125 and the second localization function 215 can be transferred to a common or uniform reference system.

When designing the localization and control architecture, in the exemplary embodiment shown there are in particular possibilities from the separation of the planning and executive level. Different localization functions can be used 115 , 215 for the planning level 100 and the execution level 200 which are based on different sensors. This can increase the reliability and redundancy within the overall system. It is also possible to use a localization function 110 with a lower data rate for the planning level 100 and a localization function 210 with higher data rate for the execution level 200 to use. In addition, the pose offset between the location information 115 , 215 monitored and transmitted to a self-monitoring of the overall system for function monitoring. Furthermore, undesirable effects of the sensor data fusion of the localization functions can occur 110 , 210 to the execution level 200 be prevented.

2 shows further details for an embodiment of the present device in an automated vehicle. In the scheme you can initially see those already in 1 shown elements of a planning level 100 , an execution level 200 and a device 300 to correct deviations between two of the localization information 115 , 215 .

Specifically, the first localization information is generated 115 as actual poses from, for example, a map-based environment localization function 110 with a data rate of, for example, 20 Hertz (Hz). The planning unit 120 in particular takes over the trajectory planning, which includes planning data comprising a target trajectory 125 outputs with a data rate of e.g. 10 Hz. In the device 300 a receiving module takes over 310 receiving and storing the first location information 115 and the planning data 125 and, in particular, second localization information comprising further actual poses 215 , which in a localization function 210 to estimate the driving dynamics state with a high data rate (for example 100 Hz). Inside the device 300 is an evaluation module 320 shown, which is designed to determine poses of the vehicle from the localization information 115 , 215 to be assigned in time and to be compared. There is also a correction module 330 which includes the planning data, for example, target poses, target speeds and target accelerations 125 and / or the second location information 215 corrected and as corrected data 335 to an execution unit 220 the execution level 200 for driving dynamics and trajectory control. The modules 310 , 320 , 330 the device 300 are not necessarily designed to be separate from one another or so that they can be separated from one another.

The localization function 210 is the execution level 200 assigned. Within the execution level 200 it transmits the second location information 215 , together with further information such as the speed and acceleration of the vehicle or the rate of rotation, for example of vehicle wheels, also to the execution unit 220 the execution level 200 . This execution unit 220 works, for example, with a data rate of 50 Hz. It receives and generates manipulated variables 225 initially for dynamic modules 410 of the vehicle. The dynamic modules have sensors, via which, for example, odometry data 415 to the second localization function 210 sent and there for the determination of the second localization information 215 be used. The dynamic modules effect the control processes 420 of the vehicle, which sensor data 425 generated for determining the first localization information 115 to the localization of the surroundings or the first localization function 115 be sent.

The offset correction by the device 300 enables movement control of the vehicle in all operating modes, especially in the areas of automation, safe stopping and control room operation. The modular structure gives rise to the advantages already mentioned elsewhere, including in particular the possibility of dividing the development and production of the parts of the system shown here between different partners (such as different companies or research institutions) who work separately from one another.

3 shows steps of a method for correcting deviations in localization information 115 , 215 a planning level 100 and an execution level 200 . Here is the planning level 100 designed to repeatedly provide first localization information 115 to be determined and planning data based on it 125 for a control of an object to be determined, and the execution level 200 is designed to repeatedly provide second location information 215 to determine. A first characterizing step of the method comprises receiving S100 from initial localization information 115 , second location information 215 and planning data 125 . This can be done at different rates. The first localization information 115 show deviations from the second localization information generated in real-time 215 on, for example due to different sensors used, different sensor errors or inaccuracies in the creation of the localization information 115 , 215 . The planning data 125 provide information for a controller, which is in the execution level 200 be implemented in control commands for an actuator of the object. Thereby additional second localization information 215 with elements of the planning data 125 compared so that the deviations lead to errors in the control.

Another step of the method includes finding S200 from initial localization information 115 and simultaneous second location information 215 . Simultaneousness is to be understood here with a tolerance, which is in particular due to different generation rates of localization information 115 , 215 is determined.

A further step of the method comprises a determination S300 a discrepancy between first location information 115 and simultaneous second location information 215 . A subsequent step includes performing S400 a correction of the planning data 125 or at least one of the second location information 215 using at least one deviation. The correction can in particular initially be determined from at least one deviation at a specific time and then used repeatedly until a next correction is determined. The correction prevents errors and jumps in the control of the object.

The features of the invention disclosed in the description, the claims and the figures can be essential for the implementation of the invention both individually and in any combination.

List of reference symbols

100

Planning level

110

first localization function

115

initial localization information

120

Planning unit

125

Planning data

200

Execution level

210

second localization function

215

second location information

220

Execution unit

225

Manipulated variables

300

contraption

310

Receiving module

320

Evaluation module

330

Correction module

335

corrected data

410

Dynamic modules

415

Odometry data

420

Control operations

425

Sensor data

S100, ..., S400

Procedural steps

Claims (12)

Device (300) for correcting deviations in localization information (115, 215) of a planning level (100) and an execution level (200), the planning level (100) being designed to repeatedly determine first localization information (115) and, based thereon, planning data ( 125) for a control of an object, and wherein the execution level (200) is designed to repeatedly determine second localization information (215), the device (300) comprises: a receiving module (310) which is designed to continuously receive the first localization information (115) and the second localization information (215) and to continuously receive the planning data (125); an evaluation module (320) which is designed to determine a discrepancy for at least some of the first localization information (115) and simultaneous second localization information (215); and a correction module (330) which is designed to carry out a correction for the planning data (125) or for at least one of the second localization information (215) using at least one determined deviation. The device (300) after Claim 1 , wherein the correction module (330) is designed to redetermine the correction according to at least one rule from the following list: - at regular time intervals, - at the request of the receiving module (310) according to predefined criteria. The device (300) according to any one of the preceding claims, wherein the correction module (330) is designed to redetermine the correction according to a predefined condition for two temporally offset second localization information items (215). The device (300) after Claim 3 , wherein the evaluation module (320) is further designed to determine and store a degree of dispersion from stored localization information (115, 215), and to take into account the degree of dispersion in the predefined condition for the two temporally offset second localization information items (215). The device (300) according to one of the preceding claims, further comprising a warning module which is designed to output a warning when a predefined condition for the deviation or for the correction is fulfilled. A system for planning and controlling the movement of an object, with: a planning level (100) which is designed to repeatedly determine first localization information (115) and to create planning data (215) for a controller; an execution level (200) which is designed to repeatedly determine second localization information (215) and to execute a control based on the planning data (125); a device (300) according to any one of the preceding claims. The system after Claim 6 , wherein the planning level (100) and / or the execution level (200) is / are designed to cause the correction module (330) to redetermine a correction according to a predefined condition of localization information (115, 215). The system after Claim 6 or Claim 7 wherein the planning level (100) is configured to generate the first location information (115) at a first rate, and the execution level (200) is configured to generate the second location information (215) at a second rate, the first Rate is different from the second rate. An object with a system according to one of the Claims 6 until 8th to plan and execute movements automatically. The object after Claim 9 , which is a vehicle, and wherein the movements include those which at least partially serve to automate normal operation of the vehicle and in particular cause one or more processes from the following list: - regulation of a pose of the vehicle, - regulation of a speed of the vehicle - Regulation of an acceleration of the vehicle, - Regulation of driving dynamics, - Regulation of a driving trajectory, - Transitional regulation in safe stopping, - Control room operation. The object after Claim 9 , which is one of the following list: - a mobile robot, - an aircraft, - a watercraft, - a household appliance. A method for correcting deviations in localization information (115, 215) of a planning level (100) and an execution level (200), the planning level (100) being designed to repeatedly determine first localization information (115) and, based thereon, planning data (125) for controlling an object, and wherein the execution level (200) is designed to repeatedly determine second localization information (215), characterized by : receiving (S100) first localization information (115), second localization information (215) and planning data (125); - Finding (S200) of first localization information (115) and simultaneous second localization information (215); - Establishing (S300) a discrepancy between first localization information (115) and simultaneous second localization information (215); - Carrying out (S400) a correction of the planning data (125) or at least one of the second localization information (215) using at least one deviation.

Images