DE102022118180B3 - Control method for the teleactuator, actuator system and robot - Google Patents

Abstract

Control method for tele-actuating for controlling an actuator system, the actuator system having: a real actuator in a real environment, and a model of the real actuator in the real environment having a model actuator and a model environment. The control method has the steps: commanding a control signal to the real actuator for controlling the real actuator, detecting an interaction of the real actuator with the real environment, detecting an interaction of the model actuator with the model environment, - determining a reference signal from the detected interaction of the model actuator with the model environment, and adjusting the control signal based on the reference signal. Also an actuator system and a robot.

Description

The invention relates to a control method for teleactuating to control an actuator system, an actuator system and a robot.

Coupled actuated systems consist of a first actuator connected to a second actuator via a transmission channel. Movements of the first actuator are to be transmitted to the second actuator by means of the transmission channel. The first actuator and the second actuator can be controlled in a master-slave configuration, so that a movement applied to the first actuator, for example by an operator, is transmitted via the transmission channel to the actuator that executes the operator's movement. The first actuator serves as the master and the second actuator as a slave. Such coupled actuator systems are used in particular in robotics, for example in medical robotics. Thus, it is no longer necessary for the operator to be at the site of the operation. Rather, the movements of the operator are recorded by the first actuator and then transmitted by means of the transmission channel to the second actuator, which carries out the movements of the operator to carry out the operation on a patient. The first actuator can be an input device and/or the second actuator can be a robot.

However, real transmission channels exhibit a certain latency or a variable time delay or packet loss during transmission. This time delay can sometimes lead to system instability. However, interaction forces must be transferred exactly. For this purpose, the “time-domain passivity approach” (TDPA) is known from the prior art, e.g. from references [16], [17], whereby the stability or passivity of the actuator system is achieved by dissipating excess energy.

Reference [18] describes another approach with the same passivity controller as in [16], [17], but using a different reference energy on the operator side.

Previous approaches to force-reflecting teleoperation have z. B. the disadvantage that they have a greatly delayed reaction of the robot with high communication latencies. For example, the force reflection in some approaches is very opaque. This means that the forces reflected, i.e. displayed on the input device, are difficult for the operator to interpret, since e.g. B. high additional damping forces are introduced by the controller to stabilize the system. In the case of the conventional passivity-based control approach TDPA (cf. [16], [17]), the robot moves slower and, in particular, less, i.e. shorter, distances than the input device under high communication latencies. In the case of the approach from [18], there is a disadvantage in that the robot only reacts very late to a contact commanded by the human operator. That is, upon detection of contact, the robot remains stationary until the operator has exerted an interaction force.

Prior art to the present invention

Credentials:

• [1] NY Lii, Riecke C, Leidner D, Schatzle S, Birkenkampf P, Weber B, Krueger T, Stelzer M, Wedler A, and Grunwald G. The robot as an avatar or coworker? An investigation of the different teleoperation modalities through the Contour-2 and METERON SUPVIS Just in space telerobotic missions. In The International Astronautical Congress (IAC). IAF, 2018 .
• [2] Ambrose R, Nesnas IAD, Chandler F, Allen BD, Fong T, Matthies L, and Mueller R. NASA technology roadmaps: Ta 4: Robotics and autonomous systems. In National Aeronautics and Space Administration (NASA), NASA Headquarters, 2015 .
• [3] Jack O Burns, Benjamin Mellinkoff, Matthew Spydell, Terrence Fong, David A Kring, William D Pratt, Timothy Cichan, and Christine M Edwards. Science on the lunar surface facilitated by low latency telerobotics from a lunar orbital platform-gateway. Acta Astronautica, 154:195-203, 2019 .
• [4] Kjetil Wormnes, William Carey, Thomas Krueger, Leonardo Cencetti, Emiel den Exter, Stephen Ennis, Edmundo Ferreira, Antonio Fortunato, Levin Gerdes, Lukas Hann, et al. Analog-I is the first part of an analogue mission to guide esa's robotic moon exploration efforts. In Global Space Exploration Conference (GLEX 2021), 2021 .
• [5] Michael Panzirsch, Harsimran Singh, Thomas Krüger, Christian Ott, and Alin Albu-Schaffer. Safe Interactions and Kinesthetic Feedback in High Performance Earth-To-Moon Teleoperation. In IEEE Aerospace Conference, 2020
• [6] Ribin Balachandran, Jordi Artigas, Usman Mehmood, and Jee-Hwan Ryu. Performance comparison of wave variable transformation and time domain passivity approaches for time-delayed teleoperation: Preliminary results. In Proceedings of 2016 IEEE/RSI International Conference on Intelligent Robots and Systems, pp. 410-4 17. IEEE, 2016 .
• [7] Jee-Hwan Ryu, Jordi Artigas, and Carsten Preusche. A passive bilateral control scheme for a teleoperator with time-varying communication delay. Mechatronics, 20(7):812-823, 2010 .
• [8th] Takashi Imaida, Yasuyoshi Yokokohji, Toshitsugu Doi, Mitsushige Oda, and Tsuneo Yoshikawa. Ground-space bilateral teleoperation of ets-vii robot arm by direct bilateral coupling under 7-s time delay condition. IEEE Transactions on Robotics and Automation, 20(3):499-5II, 2004 .
• [9] Michael Panzirsch and Harsimran Singh. Position synchronization through the energy-reflection based time domain passivity approach in position-position architectures. IEEE Robotics and Automation Letters, 6(4):7997-8004, 2021 .
• [10] Thomas Hulin, Michael Panzirsch, Harsimran Singh, Ribin Balachandran, Andre Coelho, Aaron Pereira, Bernhard M Weber, Nicolai Bechtel, Cornelia Riecke, Bernhard Brunner, et al. Model-augmented haptic telemanipulation: Concept, retrospective overview and current use-cases. Frontiers in Robotics and AI, 8:76, 2021 .
• [11] Thomas Hulin, Ricardo Gonzalez Camarero, and Alin Albu-Schaffer. Optimal control for haptic rendering: Fast energy dissipation and minimum overshoot. In 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 4505-4511. IEEE, 2013 .
• [12] Harsimran Singh, Dominik Janetzko, Aghil Jafari, Bernhard Weber, Chan-Il Lee, and Jee-Hwan Ryu. Enhancing the rate-hardness of haptic interaction: Successive force augmentation approach. IEEE Transactions on Industrial Electronics, 67(I):809-819, 2019 .
• [13] Xiao Xu, Burak Cizmeci, Anas Al-Nuaimi, and Eckehard Steinbach. Point cloud-based model-mediated teleoperation with dynamic and perception-based model updating. IEEE Transactions on Instrumentation and Measurement, 63(11):2558-2569, 2014 .
• [14] Michael Panzirsch, Harsimran Singh, Martin Stelzer, Martin J Schuster, Christian Ott and Manuel Ferre. Extended predictive model-mediated teleoperation of mobile robots through multilateral control. In 2018 IEEE Intelligent Vehicles Symposium (IV), pp. 1723-1730. IEEE, 2018 .
• [15] Michael Panzirsch, Marek Sierotowicz, Revanth Prakash, Harsimran Singh, and Christian Ott. Deflection-domain passivity control of variable stiffnesses based on potential energy reference. IEEE Robotics and Automation Letters, accepted, 2022 .
• [16] B. Hannaford, J.-H. Ryu, Time-domain passivity control of haptic interfaces, Transactions on Robotics and Automation 18 (1) (2002) 1-10
• [17] U.S. 7,027,965 B2
• [18] DE 10 2020 115 294 A1

Further prior art relevant to the present invention can be found in DE 10 2021 107 532 A1 , DE 10 2020 113 409 B4 , DE 10 2020 107 612 B3 , DE 10 2015 209 773 B3 , DE 10 2009 017 104 B4 and U.S. 2022/0 184 803 A1 described.

The object of the invention is to create a control method for teleactuating to control an actuator system, an actuator system and a robot, the control of the actuator system being improved, in particular accelerated.

The object is achieved according to the invention by a control method for teleactuating for controlling an actuator system according to claim 1, an actuator system according to claim 13 and a robot according to claim 15.

• - Kommandieren eines Steuerungssignals an den Realaktuator zur Steuerung des Realaktuators,
• - Erfassen einer Interaktion des Realaktuators mit der Realumgebung,
• - Erfassen einer Interaktion des Modelaktuators mit der Modelumgebung,
• - Ermitteln eines, insbesondere energie- und/oder leistungsbasierten, Referenzsignals aus der erfassten Interaktion des Modelaktuators mit der Modelumgebung, und
• - Anpassen, insbesondere Beschränken, des Steuerungssignals basierend auf dem Referenzsignal.

The control method according to the invention is a control method for teleactuating to control an actuator system. In particular, it is a control procedure for telerobotics. The control method is preferably designed to control an actuator system. The actuator system has a real actuator in a real environment, and a model actuator and a model environment having, in particular a virtual, model of the real actuator in the real environment. The model actuator can, in particular, involve the modeling of a single element or multiple elements of the real actuator. It is possible, for example, that the model actuator is a modeling of the end effector of the real actuator or a modeling of the entire real actuator. The control procedure consists of the following steps:

• - commanding a control signal to the real actuator to control the real actuator,
• - detecting an interaction of the real actuator with the real environment,
• - detecting an interaction of the model actuator with the model environment,
• - Determining a reference signal, in particular based on energy and/or power, from the detected interaction of the model actuator with the model environment, and
• - Adapting, in particular limiting, the control signal based on the reference signal.

The steps are preferably performed in the above order. It is preferred that the interaction of the real actuator with the real environment is detected by means of at least one sensor. The at least one sensor is arranged in the vicinity of the real actuator and/or connected to the real actuator. The at least one sensor is preferably a force-torque sensor or a visual sensor. The adjustment of the control signal based on the reference signal preferably takes place continuously, in particular continuously in terms of values and in terms of time. The adjustment of the control signal based on the reference signal particularly preferably takes place continuously in the control cycle of the controller. Adjusting the control signal based on the reference signal preferably includes dissipating and/or canceling the control signal. The real actuator can preferably be a robot, in particular a tele-robot. The model was in particular a 3D model and/or a point cloud. The real environment is, for example, an object or multiple objects. When the reference signal is determined, the reference signal is in particular calculated. It is preferred that the control signal is specified by a user.

In a preferred embodiment, the reference signal is a reference energy E ₁ and/or a reference power. The control signal can in particular be an energy and/or power-based signal. In particular, the control signal is control energy and/or control power.

In a preferred embodiment, when the real actuator interacts with the real environment, energy is transferred between the real actuator and the real environment with a maximum energy of E ₁ , in particular E ₁ . It is preferred that energy is transmitted between the real actuator and the real environment with an energy of exactly E ₁ .

In a preferred embodiment, the control signal is an input speed v _I and/or a force.

In a preferred embodiment, when the control signal is adjusted, the control signal is eliminated if it has been determined that the recorded interaction of the real actuator with the real environment does not correspond to the recorded interaction of the model actuator with the model environment. In particular, when adjusting the control signal, the control signal is eliminated if an interaction of the real actuator with the real environment was detected, but no interaction of the model actuator with the model environment was detected at the same time. Eliminating the control signal means in particular a partial or complete restriction, in particular dissipation and/or cancellation of the control signal. Eliminating preferably means setting the control signal to zero. In particular, a power and/or energy of the control signal is set equal to zero during the elimination.

In a preferred embodiment, when the control signal is adapted, the control signal is forwarded unchanged if no interaction of the real actuator with the real environment was detected.

In a preferred embodiment, the reference signal is determined from a model interaction force F _VR detected by the model and a provided input speed v _I .

In a preferred embodiment, the control signal commanded to the real actuator is regulated, in particular by means of cancellation and/or dissipation, in such a way that an energy emitted by the real actuator to the real environment is less than or equal to the reference energy E ₁ , in particular the delayed reference energy E _1,del . It is preferred here that the eradication and/or dissipation takes place on the real actuator side, in particular on the robot side.

In a preferred embodiment, the control signal commanded to the real actuator is regulated, in particular by means of cancellation and/or dissipation, based on the reference energy E ₁ , in particular the delayed reference energy E _1,del , and a detected actuator energy E ₃ , in particular detected delayed actuator energy E _{3 , del} , between real actuator and real environment. It is preferred here that the eradication and/or dissipation takes place on the real actuator side, in particular on the robot side. In this regulation, the commanded control signal is preferably regulated to a maximum value from reference energy E ₁ and actuator energy E ₃ or a weighted sum of reference energy E ₁ and actuator energy E ₃ .

In a preferred embodiment, the actuator system also has a control device, preferably having an input device, for controlling the real actuator and/or the model actuator. It is preferable that at least the control signal is inputted with the control device. Possible it is in particular that the control device consists of an input device. It is preferably a haptic input device and/or an input device for telemanipulation.

In a preferred embodiment, the control device is designed to transmit the detected interaction between the model actuator and the model environment. In particular, the control device can be designed to reproduce, preferably as a force, the interaction between the model actuator and the model environment, in particular to a user. The control device is in particular a force feedback control device.

In a preferred embodiment, the reference signal, in particular the reference energy E ₁ , and/or a detected actuator energy E ₃ is adjusted if no interaction of the model actuator with the model environment and no interaction of the real actuator with the real environment was detected. In particular, when the reference signal is adjusted, the reference signal is reset. It is preferred that the adaptation takes place on the real actuator side, in particular on the robot side, preferably by means of cancellation and/or dissipation. The reset or resetting of the reference signal takes place in particular if no interaction of the model actuator with the model environment and no interaction of the real actuator with the real environment was detected and/or if the position of the input device corresponds to the position of the real actuator.

In a preferred embodiment, when the reference signal is adapted, in particular the reference energy E ₁ and/or the detected actuator energy E ₃ , the reference energy E ₁ and/or the actuator energy E ₃ is set to zero. Alternatively, when the reference signal is adapted, the reference energy E ₁ and the actuator energy E ₃ are in particular equated, or the reference energy E ₁ and the actuator energy E ₃ are brought closer together. The reference signal is set to zero in particular if no interaction of the model actuator with the model environment and no interaction of the real actuator with the real environment was detected and/or if the position of the input device corresponds to the position of the real actuator.

In a preferred embodiment, the actuator system also has a transmission channel, in particular a time-delayed and/or lossy transmission channel, between the real actuator and the control actuator, preferably for transmitting the control signal. If within the scope of the invention, e.g. B. above, a delayed signal, such as a delayed energy, is defined, it is in particular a signal transmitted through the time-delayed and/or lossy transmission channel.

In particular, the control method according to the invention can advantageously be implemented in such a way that the user or operator controls a virtual, possibly simplified model of the robot in this virtual environment, while the movement of the input device is transmitted to both the virtual and the remote real actuator, in particular robots. is commanded. The operator preferably feels the interaction forces between the virtual actuator, in particular the virtual robot, and the virtual environment on the input device. From the speed of the input device and the force applied, the mechanical power that the operator intentionally transmits to the environment can be determined. In contrast to the prior art, this power and/or energy is preferably sent to the real actuator, in particular a robot, with the movement command, so that it can directly apply power to the environment. If, for example, an unexpected, especially unmodeled, contact occurs, the real actuator, especially the robot, will still not be able to apply any force without 'release' by the operator. In the prior art, for example, a force measured by the real actuator, in particular a robot, must first be sent to the operator. The operator must e.g. B. apply a power against this force and this power is then sent back to the real actuator, in particular the robot, so that it can apply a force that leads to a maximum of this power in particular in its environment. In particular, the model, e.g. the virtual environment, should be implemented in such a way that the interaction power corresponds to that in the real environment as best as possible. The model, in particular the virtual environment, can preferably be designed to be adaptive so that changes in the real environment can be taken into account. For example, objects newly placed in the environment can be modeled as well.

• - Kommandieren eines Steuerungssignals an den Realaktuator zur Steuerung des Realaktuators,
• - Erfassen einer Interaktion des Realaktuators mit der Realumgebung,
• - Erfassen einer Interaktion des Modelaktuators mit der Modelumgebung,
• - Ermitteln eines, insbesondere energie- und/oder leistungsbasierten, Referenzsignals aus der erfassten Interaktion des Modelaktuators mit der Modelumgebung, und
• - Anpassen, insbesondere Beschränken, des Steuerungssignals basierend auf dem Referenzsignal.

The actuator system according to the invention is in particular an actuator system for teleactuation. The actuator system has a real actuator in a real environment, and a model actuator and a model environment having, in particular a virtual, model of the real actuator in the real environment. The model actuator can in particular involve the modeling of a single element or multiple elements of the real actuator. It is possible, for example, that the model actuator is a modeling of the end effector of the real actuator or a modeling of the entire real actuator. The actuator system is designed for:

• - commanding a control signal to the real actuator to control the real actuator,
• - detecting an interaction of the real actuator with the real environment,
• - detecting an interaction of the model actuator with the model environment,
• - Determining a reference signal, in particular based on energy and/or power, from the detected interaction of the model actuator with the model environment, and
• - Adapting, in particular limiting, the control signal based on the reference signal.

The actuator system according to the invention is, in particular, a regulated transmission system for teleactuating. In particular, the actuator system according to the invention is a controller.

It is preferred that the actuator system according to the invention is designed to carry out one or more steps of the control method according to the invention. The actuator system according to the invention preferably has one or more features of the control method according to the invention.

The robot according to the invention is in particular a telescopic robot. The robot has an actuator system according to the invention and/or the robot is set up to carry out the control method according to the invention.

The invention is explained in more detail below on the basis of preferred embodiments with reference to the attached drawings.

• 1 Signalflussdiagramm eines herkömmlichen Regelungsverfahren gemäß TDPA,
• 2 Netzwerkdarstellung des Regelungsverfahrens TDPA,
• 3 Signalflussdiagramm eines herkömmlichen Regelungsverfahren gemäß TDPA-HD,
• 4 Netzwerkdarstellung des Regelungsverfahrens TDPA-HD,
• 5 Signalflussdiagramm einer Ausführungsform eines erfindungsgemäßen Regelungsverfahrens,
• 6 Netzwerkdarstellung einer Ausführungsform eines erfindungsgemäßen Regelungsverfahrens, und
• 7 Netzwerkdarstellung einer weiteren Ausführungsform eines erfindungsgemäßen Regelungsverfahrens.

Show it:

• 1 Signal flow diagram of a conventional control method according to TDPA,
• 2 Network representation of the TDPA control procedure,
• 3 Signal flow diagram of a conventional control method according to TDPA-HD,
• 4 Network representation of the control method TDPA-HD,
• 5 Signal flow diagram of an embodiment of a control method according to the invention,
• 6 Network representation of an embodiment of a control method according to the invention, and
• 7 Network representation of a further embodiment of a control method according to the invention.

In the practice paper described below, the 1-6 one of the background of the present regulatory procedure ( 1-4 ) and preferred embodiments of the control method according to the invention ( 5-6 ) described. In particular, the descriptions of the control method according to the invention ( 5-6 ) are implemented accordingly within the framework of an actuator system according to the invention and/or a robot according to the invention.

I. INTRODUCTION

Recent plans by the major space agencies envisage scheduled exploration using mobile robots [I]. However, the capabilities of autonomous robots are still limited and, above all, the sensor functions are not entirely robust. As a workaround [2], and to extend the capabilities of the robots, astronauts in an orbiting spacecraft are equipped with a teleoperation interface to control robotic rovers and manipulators [3], [4]. In such scenarios, the delay in communication between the robot and the haptic interface will exceed the round-trip delay of around 800 ms in a geostationary connection from Earth to the ISS.

In order to be able to safely carry out manipulation tasks with contacts in particular, the display of force feedback on the input device is of the utmost importance. In addition to the force information for the astronauts themselves, the transmission of interaction forces also makes it possible to avoid hard impacts on the environment [5] (CITE Science), as will be explained in more detail later. The delayed communication represents an active element that can cause instability through energy injection. Various control approaches have been developed, including the wave variable method [6] and the Time Domain Passivity Approach (TDPA, [7]) or Llewellyn criteria [8]. However, the transparency of the teleoperation setup (i.e. the operator's sense of immersion in the robot's environment) is greatly reduced as the delay increases. In some approaches, the quality of position tracking is reduced [6], while in most approaches, the quality of force feedback is also heavily damped or suffers from excessive damping (position-position architectures [8], [9]). Frequency-domain-based approaches such as the Raisbeck, Llewellyn or Routh-Hurwitz criterion do not use adaptive but constant damping, which usually leads to very conservative parameterization and thus poor performance. But the performance of the conventional TDPA [7], one of the most common among them, deteriorates sharply with increasing delay.

In [5] a control concept according to TDPA-HD for high delays is proposed, which achieves high position tracking accuracy and secure interactions regardless of the communication delay. Although [5] presented the successful performance of a number of different applications from sample collection and injection to maintenance-related tasks at a delay of around 3s (Earth-Moon communication), the quality of the force feedback was affected by the control method. In addition, the time until a force can be applied to the environment (time-to-interact) is increased by the control approach.

In order to increase the quality of the force feedback and reduce the time-to-interact, a combination of TDPA-HD and the model-extended telemanipulation concept (MATM, [10]) is undertaken. The MATM generally provides for the integration of a local (operator-side) and a remote (robot-side) virtual model of the robot's environment. The local model supports, among other things, visual or haptic augmentations, while the remote model provides common control functionalities. A local model, which can be a known model or a point cloud scanned in the distant environment, is used to determine a reference energy for the robot's interaction boundary with the environment. Satellite maintenance, e.g. investigated in the DLR AI-In-Orbit-Factory 4.0 project, represents a potential application that requires interactions with structured or known environments at high communication delay. Therefore, the operator interacts with the local VR (Virtual Reality) model and perceives a rendered force so that the (reference) energy that the operator exerts on the modeled environment can be observed. This reference energy can be immediately applied to the distant environment while still preventing effects on unwanted interactions.

This practice paper (with reference to the 1-6 ) is structured as follows: In Section I (cf. above), the background to the control method according to the invention, in particular in the context of the present practice paper (with reference to the 1-6 ) described. In Section II, the principles of conventional TDPA, TDPA-HD and the disadvantages of the prior art are presented as the basis for the present invention. Preferred embodiments of the control method according to the invention, in particular an integration of the TDPA-HD into the MATM frame, are presented in Section III.

II. PRINCIPLES AND PROBLEM DEFINITION

The control circuit of a standard teleoperation setup is in 1 shown. A human operator uses an input device to control a robot in a remote environment (environment). A coupling controller (Ctrl.) with spring-damper characteristics ensures that the two devices follow each other's positions. For this purpose, a force (F _c ) is calculated from the difference in position of the devices. This force is applied locally to the robot and displayed on the input device (Force Feedback, FF). The position and velocity references of the input device v ^I and FF are delayed in the communication channel (CC) by time delays Tf and _Tb , respectively. The Time Doman Passivity Approach (TDPA) is used to stabilize the closed control loop despite communication delays. To do this, he uses passivity observers (PO) to observe the energy introduced by the CC (which can lead to instability) and dissipates excess energy via passivity controllers (PC) with variable damping α and β. To do this, the PC β impedance type dampens the force feedback, while the PC α admittance type adjusts the velocity command.

oder von rechts nach links $\left(P_{R2L}^i\right)$

fließt. Durch diskrete Zeitintegration lassen sich die Energien $E_{L2R}^i\text{ und }{E}_{R2L}^i$

berechnen. Damit kann die Energieerzeugung des CC in jeder Energieflussrichtung bestimmt und für die Passivitätsregelung berücksichtigt werden. 2 shows the network representation of the in 1 described teleoperation system. This diagram consists of 1 and 2 port subsystems connected via port interfaces where a force (force u) and a flow (velocity u) can be measured. Thus, at each port i, a power flow P ⁱ can be calculated as P'(k)=F ⁱ (k)v ⁱ (k) at each time step k. With regard to the sign of the power P ⁱ it can be determined whether the respective power in the direction from left to right $\left(P_{L2R}^i\right)$

or from right to left $\left(P_{R2L}^i\right)$

flows. Through discrete time integration, the energies can be $E_{L2R}^i\text{and}{E}_{R2L}^i$

calculate. This means that the energy generation of the CC can be determined in every energy flow direction and taken into account for the passivity control.

• • der Zeit t₁, zu der dem Roboter von der Fernkopplungssteuerung eine Kraft F_c befohlen wird,
• • der Zeit t₂ = t₁ + T_b, zu der diese Kraft (des jeweiligen Zeitschritts) am Eingabegerät angezeigt wird (woraus der Energieeintrag $E_{L2\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">R</figure-callout>}}^3$
  
  des Bedieners ermittelt wird) und
• • der Zeit t₃ = t₂ +Tf, zu der der Energieeintrag $E_{L2\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">R</figure-callout>}}^3$
  
  bezogen auf F_c(ti) auf der fernen Seite ankommt.

As previously discussed, even the performance (force feedback quality and position tracking) of the conventional TDPA critically degrades with increasing deceleration. With such passivity-based approaches, the problem arises from the large time difference between

• • the time t ₁ at which the robot is commanded a force _Fc by the remote coupling controller,
• • the time t ₂ = t ₁ + T _b , at which this force (of the respective time step) is displayed on the input device (from which the energy input $E_{L2\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">R</figure-callout>}}^3$
  
  of the operator is determined) and
• • the time t ₃ = t ₂ +Tf at which the energy input $E_{L2\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">R</figure-callout>}}^3$
  
  relative to F _c (ti) arrives on the far side.

Put simply, in the conventional TDPA [7] the operator commanded velocity is heavily dampened by the passivity controller PC _R at t = [t ₁ , t ₃ ] since the far side has not been told with what force the robot is interacting with the environment may. With increasing deceleration, the performance of the conventional TDPA decreases significantly, especially since the calculated controller forces F _c (which are also present with free movement) are taken into account for the passivity control.

auf der fernen Seite bei t₃ angekommen ist. Dies führt zum einen zu sehr sicheren Interaktionen und der Vermeidung von harten Stößen. Darüber hinaus zeigt die Positionsverfolgung des Roboters eine sehr hohe Leistung in der freien Bewegung, da der Admittanztyp PC2 in diesen kraftlosen Phasen keine Energie abführen muss. Andererseits kann der TDPA-HD nicht verhindern, dass der Roboter erst nach f₃ - t₁ = t_f + t_b (Time-to-Interact t_TI = t₃ - t₁) eine gewünschte Interaktion mit der Umgebung beginnt. Im TDPA-HD begrenzt der PC2 die Ausgangsenergie an Port 6 gemäß dem Energieeintrag $E_{L2R}^3\left(t-T_{ƒ}\right)$

in Energieflussrichtung L2R. Dabei wird $E_{L2R}^3$

aus der Eingabegerätegeschwindigkeit v^I und der verzögert gemessenen Kraft F_E,del auf der Bedienerseite beobachtet.This problem has been immensely reduced by the TDPA-HD for high delay telemanipulation [5]. The signal flow diagram in 5 and the network representation in 4 explain how the TDPA-HD works. In contrast to the conventional TDPA, the force F _e measured by a force-torque sensor on the robot's end effector is used here for passivity monitoring and control. Thus, the robot follows perfectly in free motion (when F _e = 0), but stops upon contact with the environment (t ₁ ) until the operator has perceived the contact (t ₂ ) and the desired input/interaction performance $E_{L2\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">R</figure-callout>}}^3$

arrived at t ₃ on the far side. On the one hand, this leads to very safe interactions and the avoidance of hard impacts. In addition, the position tracking of the robot shows a very high performance in free movement, since the PC2 admittance type does not have to dissipate any energy in these powerless phases. On the other hand, the TDPA-HD cannot prevent the robot from starting a desired interaction with the environment only after f ₃ - t ₁ = t _f + t _b (time-to-interact t _TI = t ₃ - t ₁ ). In the TDPA-HD, the PC2 limits the output energy at port 6 according to the energy input $E_{L2R}^3\left(t-T_{ƒ}\right)$

in the energy flow direction L2R. In doing so $E_{L2\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">R</figure-callout>}}^3$

observed from the input device speed v ^I and the delayed measured force F _E,del on the operator side.

III. PROPOSED APPROACH

In the following, preferred embodiments of the control method according to the invention are described within the scope of the proposed approach outlined. The main focus of the proposed approach is to reduce the time-to-interact t _TI while maintaining the safety aspects of the TDPA-HD in case of unexpected collisions.

A. Concept Description

Therefore, the interaction with a virtual model (VR) of the robot environment on the operator side is integrated into the control loop of the TDPA-HD (see signal flow diagram of 5 ). The position or speed of the input device remains the motion reference for the robot-side coupling controller (Ctrl), while the force feedback to the human operator is replaced by the interaction force f _VR represented in VR. The passivity observer PO _L and the passivity controller damping β are used for passive interaction with the VR. The basic concept of the model-extended energy flow reference is presented here. A further possibility is a fusion of f _VR and f _E as force feedback and other stability control mechanisms for the VR interaction [11], [12], which will not be discussed in detail here.

für die lokale 1-Port-Passivitätssteuerung der Interaktion mit der VR und als Energieeintragsreferenz für PC2 in L2R-Richtung verwendet.As in the respective network representation in 6 illustrated, the energy input into the VR $E_{L2\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">R</figure-callout>}}^3$

used for 1-port local passivity control of interaction with the VR and as energy input reference for PC2 in L2R direction.

• Insbesondere miss der Passivitätsbeobachter an Port 3 die Eingangs- $\left(E_{L2R}^3\right)$
  
  und Ausgangsenergie $E_{R2\mathrm{<figure-callout\; id="3"\; label="L"\; filenames="DE102022118180B3\_0040.png,DE102022118180B3\_0042.png"\; state="\{\{state\}\}">L</figure-callout>}}^3$
  
  des VR I-Port-Teilsystems. Die beobachtete Energie $W_{obs}^{PC1},$
  
  die von PCI abgeführt werden muss, wird vorzugsweise berechnet als: $$W_{obs}^{PC1}\left(k\right)=E_{L2R}^3\left(k\right)-E_{R2L}^3\left(k\right)+W_{diss}^{PC1}\left(k-1\right),$$
  
  $W_{PC1}^{diss},$
  
  unter Berücksichtigung der Energie die bereits von PCI abgeführt wurde. Insbesondere wirkt dann der Impedanztyp PCI so, dass $$F_I\left(k\right)=F_{VR}\left(k\right)+\beta \left(k\right)v_I\left(k\right),$$
  
  $$\beta \left(k\right)=\{\begin{array}{cc}\frac{-W_{obs}^{PC1\left(k\right)}}{T_s{v}_I{\left(k\right)}^2}& \text{if }{W}_{obs}^{PC1}\left(k\right)<0\\ 0& \text{if }{W}_{obs}^{PC1}\left(k\right)\ge \mathrm{0,}\end{array}$$
  
  mit Abtastzeit T_s.

Local 1-port passivity control from PC1:

• In particular, the passivity observer on port 3 measures the input $\left(E_{L2R}^3\right)$
  
  and output energy $E_{R2L}^3$
  
  of the VR I-Port subsystem. The observed energy $W_{Obs}^{PC1},$
  
  which must be paid by PCI is preferably calculated as: $$W_{Obs}^{PC1}\left(k\right)=E_{L2R}^3\left(k\right)-E_{R2L}^3\left(k\right)+W_{\mathrm{i.e}iss}^{PC1}\left(k-1\right),$$
  
  $W_{PC1}^{\mathrm{i.e}iss},$
  
  taking into account the energy already dissipated by PCI. In particular, the impedance type PCI then acts in such a way that $$f_I\left(k\right)=f_{VR}\left(k\right)+\beta \left(k\right)v_I\left(k\right),$$
  
  $$\beta \left(k\right)=\{\begin{array}{cc}\frac{-W_{Obs}^{PC1\left(k\right)}}{T_s{v}_I{\left(k\right)}^2}& \text{if}{W}_{Obs}^{PC1}\left(k\right)<0\\ 0& \text{if}{W}_{Obs}^{PC1}\left(k\right)\ge \mathrm{0,}\end{array}$$
  
  with sampling time T _s .

Since the rendered VR force F _VR is displayed on the input device instead of the measured force F _E , the control circuit of the robot and the input device is an open circuit. This means that the circuits of 4 do not have to be passive. The circuit with operator and VR can also be designed with optimal control theories [11]. fen that do without the conservative 1-port passivity control of PC1. The energy that the operator consciously puts into the VR can then still be applied to the robot side as an input energy reference.

• Insbesondere wird die beobachtete Energie $W_{obs}^{PC2},$
  
  die von PC2 abgeführt werden muss, berechnet als: $$W_{obs}^{PC2}\left(k\right)=E_{L2R}^3\left(k\right)-E_{L2R}^5\left(k\right)+W_{diss}^{PC2}\left(k-1\right),$$
  
  unter Berücksichtigung der Energie $W_{PC2}^{diss},$
  
  die bereits von PC2 abgeführt wurde.

Remote passivity control of PC2:

• In particular, the observed energy $W_{Obs}^{PC2},$
  
  which must be dissipated by PC2, calculated as: $$W_{Obs}^{PC2}\left(k\right)=E_{L2R}^3\left(k\right)-E_{L2R}^5\left(k\right)+W_{\mathrm{i.e}iss}^{PC2}\left(k-1\right),$$
  
  considering the energy $W_{PC2}^{\mathrm{i.e}iss},$
  
  which has already been discharged by PC2.

hierbei das jeweilige gewählte Referenzsignal repräsentierten. Dann wirkt der Impedanztyp-PC1 so, dass $$v_{R\text{'}}\left(k\right)=v_{I,del}\left(k\right)+\beta \left(k\right)F_E\left(k\right)$$

$$\beta \left(k\right)=\{\begin{array}{cc}\frac{-W_{obs}^{PC2}\left(k\right)}{T_s{F}_E{\left(k\right)}^2}& \text{if }{W}_{obs}^{PC2}\left(k\right)<0\\ 0& \text{if }{W}_{obs}^{PC2}\left(k\right)\ge \mathrm{0,}\end{array}$$

mit Abtastzeit T_s.Depending on the choice of reference signal can $E_{L2R}^3\left(k\right)$

here represent the respective selected reference signal. Then the impedance type PC1 acts so that $$v_{R\text{'}}\left(k\right)=v_{I,\mathrm{i.e}el}\left(k\right)+\beta \left(k\right)f_E\left(k\right)$$

$$\beta \left(k\right)=\{\begin{array}{cc}\frac{-W_{Obs}^{PC2}\left(k\right)}{T_s{f}_E{\left(k\right)}^2}& \text{if}{W}_{Obs}^{PC2}\left(k\right)<0\\ 0& \text{if}{W}_{Obs}^{PC2}\left(k\right)\ge \mathrm{0,}\end{array}$$

with sampling time T _s .

B. Effect analysis

Analogous to the force feedback of F _E in the TDPA-HD, the force F _VR is zero in free motion situations, resulting in zero power consumption during free motion. Therefore, the CC does not introduce energy during free movement and the PC2 does not need to dampen the velocity reference. This avoids position drift during free movement.

von Port 3 (gemäß F_VR(k - T_f) ≈ F_E(k). Der PC2 wird also keine Energie abführen. Falls der Kontakt auf der Roboterseite unerwartet stattfindet (Kontakt mit einem unbekannten Objekt, das in VR nicht modelliert ist), wird die Leistungsaufnahme an Port 3 $\left({\dot{E}}_{L2R}^3\left(k-T_{ƒ}\right)=0\right)$

gleich Null $\left({\dot{E}}_{L2R}^3\left(k-T_{ƒ}\right)=0\right).$

Dann dämpft PC2 die Referenzgeschwindigkeit vollständig ab, so dass der Roboter vor dem Kontakt stillsteht. Man beachte, dass in dem präsentierten Open-Loop-Aufbau PC2 für die Stabilisierung der verzögerten Kopplung nicht erforderlich ist, sondern nur zum Gewährleisten der Sicherheit in Interaktionen angewendet wird.In contact situations, there will be power at port 5 due to the measured force feedback. If the contact area was modeled in VR (expected contact), PC2 will receive corresponding input power $E_{L2R}^3\left(k-T_{ƒ}\right)$

from port 3 (according to F _VR (k - T _f ) ≈ F _E (k). So the PC2 will not dissipate any energy. In case the contact on the robot side happens unexpectedly (contact with an unknown object not modeled in VR) , the power consumption at port 3 $\left({\dot{E}}_{L2R}^3\left(k-T_{ƒ}\right)=0\right)$

equals zero $\left({\dot{E}}_{L2R}^3\left(k-T_{ƒ}\right)=0\right).$

Then PC2 dampens the reference speed completely so that the robot stands still before the contact. Note that in the presented open-loop setup, PC2 is not required for the stabilization of the delayed coupling, but is applied only to ensure safety in interactions.

(Port 5) und der Robotereferenzpose ${}^w{R}_R^{ref}$

(Port 6) nach Passivitätskontrolle als: $$\mathrm{\Delta }{R}_{drift}\left(k\right){=}^I{R}_W^{del}{\left(k\right)}^W{R}_R^{ref}\left(k\right).$$

The control mechanism of PC2 (velocity reference damping) results in a position drift in the reference pose x ^R' with respect to the input device pose x ^I . Similar to the telenavigation structure in [14], this positional drift in the position of the input device in the VR must be taken into account. The drift ΔR _drift is calculated from the delayed input device pose ${}^w{R}_l^{\mathrm{i.e}el}$

(port 5) and the robot reference pose ${}^w{R}_R^{\mathrm{right}ef}$

(Port 6) after passivity check as: $$\mathrm{\Delta }{R}_{\mathrm{i.e}\mathrm{right}ift}\left(k\right){=}^I{R}_W^{\mathrm{i.e}el}{\left(k\right)}^W{R}_R^{\mathrm{right}ef}\left(k\right).$$

des Eingabegeräts im VR: $${{}^w}_{R_{I\text{'}}}\left(k\right){=}^w{}_{R_I}\left(k\right)\Delta R_{drift}^{del}\left(k\right)\Delta R_{init}^{del}\left(k\right),$$

mit der Differenz $\Delta R_{init}\left(k\right)$

aus initialem Eingabegerät ${}^w{R}_I^{init}$

(Port 5) und Roboterbasis ${}^w{R}_R^{init}$

(Port 8): $$\Delta R_{init}\left(k\right){=}^I{R}_W^{init,del}{\left(k\right)}^w{R}_R^{init}\left(k\right).$$

Hence the pose ${}^w{R}_{I\text{'}}$

of the input device in VR: $${{}^w}_{R_{I\text{'}}}\left(k\right){=}^w{}_{R_I}\left(k\right)\Delta R_{\mathrm{i.e}\mathrm{right}ift}^{\mathrm{i.e}el}\left(k\right)\Delta R_{init}^{\mathrm{i.e}el}\left(k\right),$$

with the difference $\Delta R_{init}\left(k\right)$

from initial input device ${}^w{R}_I^{init}$

(Port 5) and robot base ${}^w{R}_R^{init}$

(port 8): $$\Delta R_{init}\left(k\right){=}^I{R}_W^{init,\mathrm{i.e}el}{\left(k\right)}^w{R}_R^{init}\left(k\right).$$

des Eingabegeräts im VR wird entsprechend berechnet: $${}^w{p}_{I\text{'}}\left(k\right){=}^w{p}_I\left(k\right)+\Delta p_{drift}^{del}\left(k\right)+\Delta p_{init}^{del}\left(k\right),$$

mit $$\Delta p_{drift}\left(k\right){=}^w{p}_R^{ref}\left(k\right){-}^w{p}_I^{del}\left(k\right)$$

und $$\Delta p_{init}\left(k\right){=}^w{p}_R^{init}\left(k\right){-}^w{p}_I^{init,del}\left(k\right).$$

The position ${}^w{p}_{I\text{'}}$

of the input device in VR is calculated accordingly: $${}^w{p}_{I\text{'}}\left(k\right){=}^w{p}_I\left(k\right)+\Delta p_{\mathrm{i.e}\mathrm{right}ift}^{\mathrm{i.e}el}\left(k\right)+\Delta p_{init}^{\mathrm{i.e}el}\left(k\right),$$

with $$\Delta p_{\mathrm{i.e}\mathrm{right}ift}\left(k\right){=}^w{p}_R^{\mathrm{right}ef}\left(k\right){-}^w{p}_I^{\mathrm{i.e}el}\left(k\right)$$

and $$\Delta p_{init}\left(k\right){=}^w{p}_R^{init}\left(k\right){-}^w{p}_I^{init,\mathrm{i.e}el}\left(k\right).$$

The above statements regarding the control method according to the invention can be correspondingly implemented within the framework of an actuator system according to the invention and/or a robot according to the invention.

7 shows a network representation of a further embodiment of a control method according to the invention within the scope of an embodiment of an actuator system 10 according to the invention for tele-actuation, the representation being based on the network representation 6 based. The actuator system 10 has an operator side 11 and a robot side 13, which has a delay and / lossy transmission supply channel 20 are connected to each other. The elements of the actuator system 10 are connected to one another via ports 30 _i - 30 _ix to transmit signals, in particular to transmit energy and/or power.

The actuator system 10 has an input device 14 via which an operator 12 can command movement commands. A first passivity controller 16 is connected to the input device 14 in a signal-transmitting manner. A virtual model 18 embodied as a VR is in turn connected to the passivity controller 16 .

The model 18 is a modeling of the environment 28 of the robot 26, in particular a telerobot, which is used to identify or predict whether the robot 26 will come into contact with the environment 28 as a result of the movement command. In the simplest case, this can be represented in particular by a boolean 1/0. In the event of an expected contact, an energy E ₁ is transmitted to the robot side 13, which the robot 26 may apply to the environment 28. A coupling controller 24 is connected in a signal-transmitting manner to the robot 26 .

On the robot side 13, by adjusting one or more signals on the robot side 13, it is ensured that no more energy than E ₁ is applied to the environment.

It is preferred that the desired energy E ₁ is calculated from v _I and F _VR . Preferably, E ₁ is just the energy flowing into VR 18 from left to right. In particular, E ₂ flows from right to left out of VR 18 and is ignored. The direction of power flow can be determined by the sign of the power.

It is preferred that on the robot side 13 is ensured by eradication/dissipation, for example by reducing the setpoint speed at port 5/6 and/or by reducing the force commanded to the robot 26 at port 8 (not shown here) that the energy , which flows in the direction of the environment 28, does not exceed E _1,del , ie the delayed energy E ₁ .

The energy that has to be dissipated on the robot side 13 is preferably determined from E _3,del and E _1,del , for example the maximum value of both energies and/or a weighted sum can be selected.

It is preferable that an energy reset is made from E ₁ to E ₃ or E ₃ to E ₁ to prevent uneven energy accumulation.

The operator 12 can preferably specify the energy E ₁ intuitively via force feedback from F _VR on the input device 14 . Alternatively or additionally, a purely visual specification is possible.

A second passivity regulator 22 is connected to the transmission channel 20 in a signal-transmitting manner.

Claims (15)

Control method for teleactuation, in particular for telerobotics, for controlling an actuator system, the actuator system having: - a real actuator in a real environment, and - a, a model actuator and a model environment having, in particular virtual, model of the real actuator in the real environment; wherein the regulation method comprises the steps: - commanding a control signal to the real actuator to control the real actuator, - detecting an interaction of the real actuator with the real environment, - detecting an interaction of the model actuator with the model environment, - Determining a reference signal, in particular based on energy and/or power, from the detected interaction of the model actuator with the model environment, and - Adapting, in particular limiting, the control signal based on the reference signal. regulatory procedure claim 1 , characterized in that the reference signal is a reference energy E ₁ and/or a reference power. regulatory procedure claim 2 , characterized in that when the real actuator interacts with the real environment, energy is transferred between the real actuator and the real environment with a maximum energy of E ₁ , in particular of E ₁ . Regulatory procedure according to one of Claims 1 - 3 , characterized in that the control signal is an input speed v _I and/or a force. Regulatory procedure according to one of Claims 1 - 4 , characterized in that when adjusting the control signal, a partial or complete elimination of the control signal upon detection of a detected interaction of the real actuator with the real environment that does not correspond to the detected interaction of the model actuator with the model environment, in particular upon detection of a detected interaction of the real actuator with the real environment at the same time missing Detection of an interaction of the model actuator with the model environment takes place. Regulatory procedure according to one of Claims 1 - 5 , characterized in that when the control signal is adapted, the control signal is forwarded unchanged if no interaction of the real actuator with the real environment is detected. Regulatory procedure according to one of Claims 1 - 6 , characterized by determining the reference signal from a model interaction force F _VR detected by the model and a provided input speed v _I . Regulatory procedure according to one of Claims 1 - 7 , characterized by regulating, in particular by means of cancellation and/or dissipation, of the control signal commanded to the real actuator in such a way that energy emitted by the real actuator to the real environment is ≤ the reference energy E ₁ , in particular the delayed reference energy E _1,del . Regulatory procedure according to one of Claims 1 - 8th , characterized by regulating, in particular by means of cancellation and/or dissipation, of the control signal commanded to the real actuator using the reference energy E ₁ , in particular the delayed reference energy E _1,del , and a detected actuator energy E ₃ , in particular detected delayed actuator energy E _3,del , between real actuator and real environment. Regulatory procedure according to one of Claims 1 - 9 , characterized in that the actuator system also has a control device, preferably having an input device, for controlling the real actuator and/or the model actuator. Regulatory procedure according to one of Claims 1 - 10 , characterized by adjusting, preferably resetting, the reference signal, in particular the reference energy E ₁ , and/or a detected actuator energy E ₃ in the absence of detection of an interaction of the model actuator with the model environment and failure to detect an interaction of the real actuator with the real environment. regulatory procedure claim 11 , characterized in that when the reference signal is adjusted, in particular the reference energy E ₁ and/or the detected actuator energy E ₃ , the reference energy E ₁ and/or the actuator energy E ₃ is set to zero, or the reference energy E ₁ and the actuator energy E ₃ , or an approximation of the reference energy E ₁ and the actuator energy E ₃ takes place. Actuator system, especially for teleactuation, with - a real actuator in a real environment, and - a model of the real actuator in the real environment having a model actuator and a model environment, in particular a virtual one, the actuator system being designed for: - commanding a control signal to the real actuator to control the real actuator, - detecting an interaction of the real actuator with the real environment, - detecting an interaction of the model actuator with the model environment, - Determining a reference signal, in particular based on energy and/or power, from the detected interaction of the model actuator with the model environment, and - Adapting, in particular limiting, the control signal based on the reference signal. actuator system Claim 13 , characterized in that the actuator system is set up to carry out the control method according to one of Claims 1 - 12 . Robots, in particular telerobots, with an actuator system according to one of Claims 13 - 14 and/or set up to carry out the control method according to one of Claims 1 - 12 .

Images