A method for an industrial object in an industrial system

Abstract

The present invention relates to a method for modeling an industrial object in an industrial system, the method including obtaining measurement data pertaining to a measurement of at least in parts a physical state of the object as well as time data pertaining to a point in time of the measurement of the physical state of the object, bounding a latency for transmission of the measurement data and of the time data relative to the point in time of the measurement by a latency bound, based on a model of the object considering the measurement data, the time data and the latency bound, determining time-adjusted data pertaining to the physical state of the object reflective of the latency relative to the point in time of the measurement.

Description

[0001] A method for an industrial object in an industrial system

[0002] Technical field

[0003] The invention concerns a (computer-implemented) method for handling latency pertaining to an industrial object in an industrial system, a data processing apparatus, a computer program, a control method, a control unit and an industrial system comprising the data processing apparatus and / or control unit.

[0004] Background

[0005] Real-time control in industrial systems may depend on low-jitter measurement communication. Increased jitter may degrade control performance or cause system failure in many applications, such as motor control, e.g. in robotics and automation. Motor controls, e.g. in robotic, typically rely on low jitter between each measurement and its use in a controller (control unit). Achieving low jitter may be achieved by either a dedicated network (which may however, increase hardware costs, energy consumption and complexity) or transmitting on a fixed schedule (which may preclude the use of several advanced control strategies like event-based control).

[0006] Industrial applications, such as robotic and automation applications, may require high motor control performance, e.g. to achieve dynamic and coordinated movements across several axes. Fixed-frequency controllers with a high enough frequency may be employed to achieve the performance required - requiring a high measurement frequency.

[0007] In many applications, however, high performance is not always needed. At times, e.g. one or more axes of a robot may be at standstill or may be moving at slow speed. In those times, the high measurement frequency may lead to high energy consumption and resource utilization with no benefit. Such event-based control may help mitigating issues by transmitting measurement and control data “only when needed”. Non-fixed-frequency control like event-based control may adapt the frequency of the controller and measurement system to the performance needed. Such time-related control may increase performance when it is demanded, and reduce energy consumption when it is not. Such control may have significant benefits, e.g. in robotics and automation specifically, given the mix of dynamic motion and standstill typically encountered. At the same time, there is a need for improved handling of time-related aspects of measurement data, e.g. faster transmission of measurement signals / data.

[0008] Summary of the invention

[0009] The present invention solves this problem and relates to a (computer-implemented) method for handling / using / processing (compensating at least partially) latency pertaining to (measurements relating to) an industrial object in an industrial system, the method including obtaining measurement data pertaining to a measurement of at least in parts a physical state of the object as well as time data pertaining to a point in time of the measurement of the physical state of the object, bounding the latency (for transmission) of the measurement data and of the time data (relative to the point in time of the measurement) by a latency bound, based on a model of the object considering the measurement data, the time data and the latency bound, determining / generating time-adjusted (measurement) data pertaining to the physical state of the object reflective of the latency (relative to the point in time of the measurement).

[0010] The idea underlying the invention is to support to consider / handle latency, in particular to reduce latency (effects). This may provide a basis for more flexibility as to measurement / data transmission timing. This may be a basis to accelerate data transmission, e.g. communication between sensors for measurements and a controller and / or controller and actuators, in particular if such communication may be infrequent.

[0011] In a first step, the latency is bounded, i.e. a limit of the latency during transmission is defined, at least for one of the measurement data and the time data. In a second step, the latency bound is at least indirectly taken into account by the model to determine time-adjusted (measurement) data, so as to compensate for the latency.

[0012] The idea underlying the invention may be, in one embodiment, consideration, e.g. compensation, of latency by means of a model of the object. For example, time-synchronized networking between the measurement and the control system / unit may be used, combined with a model of the system under control (industrial object), to compensate for (e.g. variable) latency between the measurement time and its use in the control loop / unit.

[0013] The measurement system carrying out the measurement of the physical state of the object may optionally time the measurement accordingly (e.g. at an ideal moment in time), ideally without (substantially) degrading the performance of the control loop. In other words, the invention may be implemented to time a measurement to take place when the power signal disturbances are less severe. This changes the latency between when the measurement is performed and when it is used in the control unit, which, as such (if not compensated for) may degrade the control performance, as jitter may be increased.

[0014] The method of the invention may be carried out on a single computer or, in parts, on several computers.

[0015] The invention may be seen, in one embodiment, as allowing the measurement system to determine / control the timing of measurements, and for the controller to at least partially compensate the resulting (variable) latency and jitter in the measurements. The compensation may be done using a time-synchronized network to allow the (variable) latency to be quantified for each measurement, and a model (in the controller) which may at least partially compensate for the (variable) latency.

[0016] In prior art methods, the measurement latency may be assumed to be constant and may e.g. be compensated offline by tuning the controller.

[0017] The invention offers various improvements as to timing aspects in connection with measurements, in particular cost-effective embodiments. E.g. in robotics and automation, specifically in motor control, expensive cabling solutions may replaced to reduce measurement disturbances by the system of the invention. A high control performance may, by means of the invention, be maintained even in the presence of disturbances.

[0018] The invention also concerns a corresponding data processing apparatus, computer program, control method and a controller (control unit) as well as an industrial system.

[0019] The latency bound, i.e. maximum latency or latency threshold, may at least indirectly be considered in the model. As an example, the bound may be 1 ms.

[0020] Time-adjustment (or time-adaption) of the data and / or the measured physical state may mean to shift the data / measurement to a different timing, e.g. to the past or the future. E.g. time adjustment may allow for time-compensation, such as latency compensation.

[0021] In one embodiment, a method of the invention may be realized as an event-based control, e.g. based on the Carrier Sense Multiple Access (CSMA) scheme with two additional features: 1. Use a priority scheduling algorithm (e.g. “Quality-of-Service”) to bound the measurement latency, e.g. even in a CSMA scheme. (The jitter may still be high, even if the latency may be bounded. In other words, a Carrier Sense Multiple Access (CSMA) scheme may be used, in which case the measurement jitter is both high and variable, per se resulting in a degradation of the control performance).

[0022] 2. Compensate the high jitter, e.g. resulting from CSMA, to mitigate any control degradation, by determining time-adjusted data.

[0023] For Carrier Sense Multiple Access (CSMA), the overall transmission latency may be both due to network characteristics and physical phenomena affecting the access contention to the transmission medium and the generation of events starting a data transmission. In CSMA (Carrier Sense Multiple Access) schemes, instead of having uniquely-assigned time slots also the communication mechanism may be made such that all the packets containing data for event-based control are exchanged through opportunistic time slots, where more than one network node attempts to transmit its payload.

[0024] Jitter may be defined as an indication of the smallest and largest latency, and the statistical distribution of latency values occurring in the system. Transmitting a burst of data traffic at a high rate followed by an interval or period of lower or zero rate transmission may also be seen as a form of jitter, as it represents a deviation from the average transmission rate.

[0025] Latency may be defined as the period of time between the occurrence of a physical event (here: measurement of the physical state of the object) till the “arrival” of a signal (data) indicative of the occurrence of the event (i.e. measurement) at a desired site (here: control unit and / or model). Latency may include measurement and / or transmission delays.

[0026] Typical measurement intervals may be 10 ps to 1 ms, in particular 100 to 500 ps. Hence, between multiple measurements, the time may be between 10 ps to 1 ms, in particular 100 to 500 ps. The interval may be fixed or variable.

[0027] The invention is not limited to variable, i.e. flexible, irregular, latency applications. As such, the invention is also applicable for fixed, i.e. constant, latency applications.

[0028] The invention is not limited to event-based control and / or measurements, but may be applicable to fixed frequency control and / or measurements.

[0029] An object’s physical state may refer to the location of the object in the space. The model describes, at least in parts, the behaviour of the object, in particular the physical state and the object’s configuration. By way of modelling, i.e. the model, the true behaviour of the object may be estimated / reflected. The physical state of the object does not need to be modeled in its entirety, but may at least partially be described by the model.

[0030] For bounding latency, the following may be used: “Quality of service” (QoS) may allow for traffic prioritization and resource reservation control mechanisms. Quality of service is the ability to provide different priorities to different applications, users, or data flows (here: measurement and / or time data), or to guarantee a certain level of performance to a data flow. In the context of computer networks, packet jitter or packet delay variation (PDV) is the variation in latency as measured in the variability over time of the end-to-end delay across a network. A network with constant delay has no packet jitter. Packet jitter may be expressed as an average of the deviation from the network mean delay.

[0031] By way of the invention, e.g. event-based control, in particular in robotics and automation, may be possible without or at least with reduced degradation of the control performance. Specifically, it may be possible to compensate jitter resulting from shared networks without using a fixed transmission schedule, by using timestamped measurement data and a model of the system being controlled.

[0032] A control unit may be part of controller or may separately be provided. The control unit may be provided in the controller and / or the measurement system may be provided in the object, such as in a robot arm.

[0033] The following timestamped network protocols may be applied for a time-synchronized network, i.e. determination of the latency: IEEE 1588 (2019) and / or IEEE 802.1AS.

[0034] Optionally, the measurement data and / or time data is prioritized during transmission via the network.

[0035] Optionally, priorization includes bounding the latency to a predetermined range.

[0036] Optionally, a carrier sense multiple access (CSMA) scheme is used for transmitting data in the network. Optionally, the method represents a part of an event-based control and / or an event-based measurement.

[0037] Optionally, multiple measurement data pertaining to multiple measurements, respectively, are obtained at variable time intervals relative to each other so that the latency per measurement is variable.

[0038] Optionally, the latency, optionally the variable latency per measurement, is quantified by means of the point in time of the measurement.

[0039] Optionally, the model includes modelling the object dynamics.

[0040] Optionally, the time-adjusted data reflective of latency compensation is obtained based on a time-synchronized network.

[0041] Optionally, bounding the latency includes consideration of the network topology and / or of the behaviour of nodes in the network.

[0042] The invention is also directed to a data processing apparatus comprising means for carrying out the method of the invention.

[0043] The invention is also directed to a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of the invention.

[0044] The invention is also directed a method for controlling an industrial object in an industrial system, wherein at least in parts a physical state of the object is measured as well as a point in time of the measurement of the physical state of the object is measured, (and respective measurement data and time data is transmitted via a network), a latency (for transmission) of the measurement data and / or time data relative to the point in time of the measurement is bounded by a latency bound, based on a model of the object considering the measured physical state, the point in time and the latency bound, a time-adjusted physical state of the object reflective of the latency is determined, and based on the time-adjusted physical state, an action of the object is controlled.

[0045] The invention is also directed to a control unit configured to carry out the method of the invention. The invention is also directed to an industrial system comprising the industrial object, in particular a robot and / or an automated object, and including the control unit and / or the data processing apparatus of the invention.

[0046] The present invention is not limited to industrial applications such as robotics. However, an advantage of robotics applications may be that in the field of robotics, appropriate models of the system are available, so that the robotics system can appropriately be reflected by means of a model. Additionally or alternatively, it may be advantageous if the topology is fixed, i.e. known.

[0047] The invention also relates to the following:

[0048] A computer-readable medium having stored thereon the computer program of claim 12. A control method of claim 13, optionally wherein: the measurement data is prioritized during transmission via the network; priorization includes bounding the latency to a predetermined range; a carrier sense multiple access (CSMA) scheme is used for transmitting data in the network; an event-based control and / or an event-based measurement is carried out; multiple measurement data pertaining to multiple measurements, respectively, are obtained at variable time intervals relative to each other so that the latency per measurement is variable; the latency, optionally the variable latency per measurement, is quantified by means of the point in time of the measurement; the model includes modelling the object dynamics; the time-adjusted data reflective of latency compensation is obtained based on a time-synchronized network; bounding the latency includes consideration of the network topology and / or of the behaviour of nodes in the network.

[0049] Detailed embodiments and further advantages and features related to the present invention are described in the following, wherein these examples shall not be regarded as limiting the invention.

[0050] Brief description of the drawings

[0051] Fig. 1 schematically shows steps performed in a system not having a model. Fig. 2 schematically shows steps perfumed in a latency-variable system.

[0052] Detailed description

[0053] Fig. 1 shows a typical (real-time control) system without latency compensation according to the prior art. A controller / control unit 1 includes a control action generator 2 generating a control action. The control action 3 is executed, such that the object 4 carries out the physical behavior I movement 5 in line with the control action 3. A measurement system 6 measures the physical state / movement 5 of the object 4, and provides measurement data 7 on the measured physical state to the control unit 1 .

[0054] Fig. 2 shows a (real-time control) system with latency compensation according to an embodiment of the invention. In this case, the measurement system 6 may e.g. predict when disturbances will occur and time measurements around the disturbances, i.e. by avoiding measurements during occurrence of disturbances. Accordingly, in an embodiment of the invention, no measurements are taken as long as there is a disturbance. However, disturbances do not need to be avoided completely. The latency L is variable and may differ from measurement to measurement. Hence, L is, in the embodiment of Fig. 2, not fixed or predetermined.

[0055] The following differences relative to the system of Fig. 1 are identifiable, in the system of Fig. 2: Measurement data pertaining to the measured physical state 7 as well as a point in time of the measurement indicative of the latency L are measured. The measurement data pertaining to the physical state 7 as well as the latency L are sent (transmitted via the network) to the control unit. The controller 1 additionally comprises an object model 8, by means of which the object 4 can be modelled, i.e. artificially rebuild. Based on the measurement data 7 and the model 8, time-adjusted, e.g. predicted measurement data 9 pertaining to a time-adjusted, e.g. predicted, physical state may be obtained.

[0056] To compensate for variable measurement latency L, the latency L for each individual measurement may need to be quantified for compensation. The latency L may be quantified by using a time-synchronized network, which has a common, global time. Measurements may be timestamped according to this global time, and the controller 1 may know exactly when they were taken (and thus the latency L).

[0057] With the measurement latency L quantified, it may be compensated using the (online) model 8 of the system. For the invention, such models rely on having a variable measurement latency relative to the control action.

[0058] The model may compensate the measurement latency L for each individual measurement. A simple example may be to predict the measurement value (physical state of the object 4) based on the last known measurement by adding the measured or predicted rate of change of the measurement (e.g. axis speed) multiplied by the measurement latency L.

[0059] The control action 3 may also be fed into the model 8 for consideration by the model 8 for determination of the time-adjusted data. For example, the model may be directed to a motor, allowing to predict the motor position by modelling the motor. The predicted measurement data may reflect what would be measured if a measurement were to be done at this moment in time. The control action generator may then control an action of the object based on the predicted measurement data.

[0060] A bounded measurement latency L may be applied. More specifically, many networks support priority-based transmission of data. In this scheme, each transmission is assigned a priority. High-priority transmissions are performed before lower-priority ones, thus reducing the latency of the high-priority transmissions.

[0061] A priority-based transmission in a known network may be applied to bound (i.e. limit) the latency L for a transmission between any two nodes, in particular if there is knowledge of the network topology (which nodes are connected to which), and / or the behavior of all nodes sending high- priority transmissions. For measurement networks in e.g. robotics, this is typically possible since the networks are statically defined and have relatively few nodes with high-priority traffic, which may be analyzed. A measurement latency L between the measurement system and the controller may be achieved, which is variable, but below a maximum bound.

[0062] The measurement latency may be quantified. More specifically, to compensate measurement latency, it may need to be quantified, e.g. by using time-synchronized networks. Measurements may then be timestamped according to this global time, and the controller may know exactly when they were taken (and thus what is the latency before they are used).

[0063] By means of a latency-compensated system model, the quantified measurement latency may be compensated using an online model of the system. The measurement data may be fused with the predicted value of the measurement data based on the control actions and a model of the system dynamics. Model may need to be expanded to take the variable latency into account.

[0064] A latency-compensated controller may at least partially mitigate any performance degradation which occurs due to the increased measurement latency, e.g. in an adapted CMSA network according to the invention.

[0065] Using event-based control, one may react faster to changes in measurement values than using fixed-frequency control. The performance relative to the fixed-frequency benchmark is then a function of (at least) the latency bound one may achieve in the network, the quality of the model - lo used to compensate the latency and the tuning of the event-based controller itself, for example how quickly new measurements are sent when the physical quantity being measured starts to change.

[0066] Specifically, for typical robotic applications, the following may apply:

[0067] • The latency bound, e.g. for an adapted CMSA network may be low relative to the benchmark controller frequency, as there are relatively few nodes with high-priority traffic, and the nodes contain low-latency hardware.

[0068] • The models of robotic systems are accurate, so the degree of compensation of the variable measurement latency may be high.

[0069] • The event-based controller may be tuned to balance performance and energy consumption I resource utilization. In robotics, one may have advanced knowledge of when high performance will be required, therefore an online tuning of the controller performance is possible.

[0070] E.g. for robotic applications, less energy and fewer resources consumption without substantially compromising performance may be achieved.

[0071] The detailed description of the invention is provided with respect to the embodiments depicted in the drawings. Obvious variations and alternatives may occur to the skilled person, based on the summary of the invention. These variations and alternatives are part of the invention in so far they are covered by the appended claims.

Claims

Claims:1 . Computer-implemented method for handling latency pertaining to an industrial object (4) in an industrial system, the method including obtaining measurement data pertaining to a measurement of at least in parts a physical state of the object (4) as well as time data pertaining to a point in time of the measurement of the physical state of the object, bounding the latency of the measurement data and / or of the time data relative to the point in time of the measurement by a latency bound, based on a model (8) of the object considering the measurement data, the time data and the latency bound, determining time-adjusted data pertaining to the physical state (9) of the object reflective of the latency (L).

2. Method of claim 1 , wherein the measurement data and / or time data is prioritized during transmission via the network.

3. Method of claim 2, wherein priorization includes bounding the latency to a predetermined range.

4. Method of any of the preceding claims, wherein a carrier sense multiple access (CSMA) scheme is used for transmitting data in the network.

5. Method of any of the preceding claims, wherein the method represents at least a part of an event-based control and / or an event-based measurement.

6. Method of any of the preceding claims, wherein multiple measurement data pertaining to multiple measurements, respectively, are obtained at variable time intervals relative to each other so that the latency (L) per measurement is variable.

7. Method of any of the preceding claims, wherein the latency (L), optionally the variable latency per measurement, is quantified by means of the point in time of the measurement.

8. Method of any of the preceding claims, wherein the model includes modelling object dynamics.

9. Method of any of the preceding claims, wherein the time- adjusted data reflective of latency compensation is obtained based on a time- synchronized network.

10. Method of any of the preceding claims, wherein bounding the latency includes consideration of the network topology and / or of the behaviour of nodes in the network.11 . Data processing apparatus comprising means for carrying out the method of any of the preceding claims.

12. Computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of any of the preceding claims 1 to 11.

13. Method for controlling an industrial object in an industrial system, wherein at least in parts a physical state of the object is measured as well as a point in time of the measurement of the physical state of the object is measured, a latency for transmission of the measurement data and of time data reflective of the measured physical state and / or the point in time of the measurement, respectively, relative to the point in time of the measurement is bounded by a latency bound, based on a model of the object considering the measured physical state, the point in time and the latency bound, a time-adjusted physical state of the object is determined reflective of the latency relative to the point in time of the measurement, and based on the time-adjusted physical state (9), an action (3) of the object is controlled.

14. Control unit (1) configured to carry out the method of claim 13.

15. Industrial system comprising the industrial object (4), in particular a robot and / or an automated object, and including the control unit (1) of claim 14 and / or the data processing apparatus of claim 11.

Images