DE102019001207A1 - Collaborative robot and method for operating a collaborative robot and use for machining a workpiece - Google Patents

Abstract

A collaborative robot (10; 110) for machining a workpiece (300) comprises a robot tool (30; 130) for machining the workpiece (300) and a mode selection device (51) for transferring the collaborative robot (10; 110) to a learning mode; in a working mode. The collaborative robot (10; 110) is configured to learn, in the learning mode, a working movement of the robot tool (30; 130) to machine the workpiece (300) and to perform the work movement learned in the learning mode. The learning mode is configured such that the collaborative robot (10; 110) learns the working movement of the robot tool (30; 130) on the workpiece (300) to be machined in the learning mode.

Description

The invention relates to a collaborative robot and a method for operating a collaborative robot for processing a workpiece.

A collaborative robot is an industrial robot that works with people and is not separated from them by guards in the production process. Collaborative robots can be equipped with one or more robot tools that can be used to machine workpieces. Examples of this are e.g. Robot tools for milling, filing, grinding, etc. In order to have a working movement performed by the collaborative robot, it must first be taught the working movement. The teaching of the working movement is basically done by robot programming.

The robot programming is very complex in itself, but there are ways to simplify this, for example by generating robot programs from a simulation of the process. The collaborative robot carries out this programmed working movement and thus processes one or more workpieces.

Since the exact calculation of the working movement can be complex, the use of a collaborative robot has hitherto been profitable mainly in industrial manufacturing, namely in the processing of a very high number of identical workpieces by the same working movement. The high cost of calculating the work movement makes the use of a collaborative robot for processing a small or medium number of workpieces or even a single workpiece usually unprofitable.

The invention has for its object to simplify the operation of a collaborative robot, in particular to make the operation of a collaborative robot with a smaller number of workpieces more profitable.

This object is solved by the subject matters of the independent claims. Preferred embodiments are the subject matters of the appended claims.

One aspect relates to a collaborative robot for machining a workpiece with a robot tool for machining the workpiece and a mode selection device for transferring the collaborative robot into a learning mode and a working mode. In this case, the collaborative robot is configured to learn a working movement of the robot tool for working the workpiece in the learning mode. Furthermore, the collaborative robot is configured to perform the work learned in the learning mode in the working mode. The learning mode is configured so that the collaborative robot in learning mode learns the working movement of the robot tool on the workpiece to be machined.

The collaborative robot may include at least one movable robotic arm to which the robotic tool is attached. The robot tool can be interchangeable and, for example, as a milling cutter, grinding tool, polishing tool, cutting tool, o. Ä. be educated. The robot tool may be designed in particular for physically processing the workpiece. The robot tool may in particular be attached to a working end of the robot arm. The collaborative robot is designed to be able to move the robot tool, preferably in all three spatial directions. In addition, the collaborative robot can be designed to realize and / or exert pressure forces on the workpiece with the robot tool. Furthermore, the collaborative robot can be designed to perform rotations and / or rotations of the robot tool about at least one adjustable and / or adjustable spatial axis. In this case, the spatial axis can be adjusted and / or adjusted as desired in three-dimensional space.

The collaborative robot may include a controller and / or plant controller that controls the robot to perform the working movement with the robot tool. The working movement is carried out in working mode. In other words, in working mode, the robot can machine the workpiece as previously learned in the learning mode. The working movement can be stored in a memory of the collaborative robot and / or the (plant) control. Therefore, the working movement can be repeated on a structurally identical and / or similar workpiece. Thus, with the learned working movement a plurality of similar and / or identical workpieces can be processed with the same, learned working movement. Furthermore, the learned working movement can be repeatedly performed on the same workpiece, e.g. a polishing movement with a polishing tool. For example, the robot may be used and / or formed as a polishing robot for an injection molding tool as a workpiece. Here, the workpiece which the robot processes is simultaneously a tool in the injection molding area, e.g. the mold used in injection molding.

The robot tool can be designed as a polishing tool for processing an injection mold as a workpiece, and / or for Processing of an injection-molded workpiece produced with an injection mold.

Collaborative robots make it possible to simplify the programming process by manual guidance (e.g., by using a "teaching by demonstration" function of the collaborative robot). The robot can be manually guided to specific positions and the associated robot coordinates can be recorded.

The handrail function can accelerate the operation of the plant manufacturer, but so far has been useful only for an end user (an industrial company that uses the collaborative robot for its automation, but does not have expertise in robot plant engineering). This is mainly because the safety of a collaborative robot system depends on the robot movement. If the robot movement changes, the safety of the robot system must be reevaluated, which i.d.R. high expenditure generated.

The mode selection device is designed both for transferring the collaborative robot into the learning mode and for transferring the collaborative robot into the working mode. Here, the collaborative robot may be operated in one mode at a time (e.g., either in the learning mode or in the working mode or in another mode). For example, the mode selection device may include an interface such as a keyboard and / or a touch screen, and / or a processor having appropriate control software installed and executed thereon.

To learn the working movement, the collaborative robot is transferred to the learning mode by means of the mode selection device. In the learning mode, the collaborative robot, in particular its robot arm, is at least partially movable, so that the collaborative robot learns the working movement to be performed by moving the robot arm and / or the robot tool. Here, the robot arm and / or the robot tool may be e.g. be moved by manual guidance of an operator along the working movement. The learned working movement may be stored in a memory of the collaborative robot, in particular in a volatile memory, e.g. a RAM memory, for immediately subsequent retrieval of the learned working movement.

After learning the working movement, the collaborative robot can be transferred to the working mode with the mode selection device. In working mode, he carries out the learned working movement (in particular repeatedly). In working mode, the collaborative robot may perform the work movement, for example, for a predeterminable number of repetitions and / or until the robot is actively stopped. For this purpose, for example, at the mode selection device, the work mode can be ended and the collaborative robot can be put into a sleep mode by no longer performing any movement (other than, for example, a drive to a rest position).

In learning mode, the working movement is learned. This means that the collaborative robot stores at least one tool position. Preferably, the collaborative robot stores a plurality of different tool positions. In this case, in addition to at least one tool position, a pressing force, an orientation, a tool rotation speed and / or a tool translation speed can be stored. In addition, the collaborative robot can also be designed so that it also stores time points and / or time periods associated with the individual tool positions, which are assigned to the individual tool positions. In addition, a movement of the robot tool, in particular a movement path and / or a movement speed, can be stored as (for example part of) the working movement.

To learn the working motion, the collaborative robot (e.g., its robotic arm) is moved to move the robotic tool to the desired tool position. In the desired tool position, an operator may confirm, for example by means of a button, that the collaborative robot is correctly located and that the selected position as part of the work movement is to store the current tool position, in particular including one or more of the additional information listed above.

The working movement can be stored by the collaborative robot in a memory.

The learning of the working movement takes place directly on the workpiece to be machined. The exact working movement of the collaborative robot therefore does not have to be elaborately programmed and / or derived from a software simulation of the processing scenario, but is taught to the collaborative robot directly on the workpiece. The learning mode may also be referred to as a teaching mode of the collaborative robot. In the learning mode, the collaborative robot is taught at least one tool position in which the robot tool is in direct contact with the workpiece. Preferably, the collaborative robot is taught a plurality of such tool positions, in which the robot tool in direct contact with the contact Workpiece is. For this purpose, the collaborative robot is moved so that the robot tool is arranged in the desired tool position.

In addition to the tool positions applied directly to the workpiece, a learning aid (which can also be referred to as a teaching aid) can be provided in the learning mode, in particular a software-assisted learning aid. Here, the learning of the working movement can be additionally supported by a software. In contrast to conventional collaborative robots, however, the learning mode does not only teach software, but only supports the learning of the working movement through software.

In addition, the learning mode is configured so that the working movement is taught directly on the workpiece to be machined.

This has the advantage on the one hand that errors in the software calculation of the working path are excluded because the collaborative robot is taught directly on the workpiece. The robot is particularly suitable for performing repetitive working movements, such as for polishing. On the other hand, the training can be done immediately before editing, as can be continued immediately after learning with the working mode. Following the teach-in during the learning mode, the collaborative robot independently performs the learned working movement and repeats it as often as desired until the processing can be completed, for example, until the desired polishing result is achieved.

It is also possible for the robot to carry out an automated identification drive as part of the learning mode. Thus, measurement data derived from the identification run can be used to create a mathematical representation of parts of the workpiece surface.

For example, the user may define a boundary of a surface of the workpiece by a few vertices, i. Learn a few tool positions for the collaborative robot. As a result, the (plant) controller automatically creates a web over the entire area. The created track corresponds to the work movement. The track points are distributed as evenly as possible over the area. The web creation is based on predefined templates (e.g., zig-zag patterns). The robot automatically shuts down the created web. This is done under pressure in predeterminable direction to ensure component contact. The tool coordinates during the robot movement are recorded by the robot and used as bases for creating a multi-dimensional regression function. This regression function can now image in particular curved component surfaces.

The mathematical area representation can, for example, be used for area-related path planning. This can mean, for example, that a defined path is distorted in accordance with the curved surface course and / or an automatic alignment of the polishing tool corresponding to the surface orthogonal takes place at each path point.

For example, the collaborative robot relieves a skilled worker from a repetitive part of his job, i. a frequent repetition of his movement, for example a polishing movement. As a result, the work capacity of the skilled worker can be freed up or used differently. Furthermore, the working time of the skilled worker can be significantly reduced. In particular, the skilled worker must spend less time in unergonomic attitudes or is a much shorter time vibrations of the robot tool o. Ä. exposed. This can reduce a health burden for the skilled worker.

It can be a division of the activities of skilled workers and robots, which is adapted to their respective skills: The skilled worker takes over the definition of process movement, ie the labor movement. This should be flexibly adapted to the complex framework conditions of the process and requires high qualification and experience of the skilled worker. The robot undertakes repetitive repetition of the process movement until a desired polishing result is achieved. As a result, the skilled worker is relieved of repetitive tasks and is released for further work commensurate with his qualifications. Based on this division of tasks between human and robot, automation is achieved for the first time as part of a polishing process that is economical and requires little hardware and software. Thus, the robot system is not used for the full automation of the polishing process, but as an assistance system that supports the skilled worker in the process execution. The described learning mode allows the skilled worker to transfer his highly complex process capabilities to the robot.

According to one embodiment, the collaborative robot is designed to learn a working movement in the learning mode for processing an injection mold as a workpiece. The injection mold may be metallic. In working mode, the collaborative robot processes the injection mold with the learned working movement. The machining operation may be a polishing process, the robot tool being used as a polishing tool can be trained. When polishing an injection mold, high quality standards must be met, which can be met by a lengthy polishing process. The collaborative robot repeatedly performs a learned in learning mode polishing movement as a working movement until the desired polishing result is achieved.

According to one embodiment, the movement speed of the robot tool and / or the collaborative robot in the learning mode and / or in the working mode is throttled to a maximum speed of maximally 50 mm / s, preferably to a maximum speed of maximally 30 mm / s, particularly preferably to a maximum speed of maximum 20 mm / s. The reduction of the movement speed increases the safety at the workplace for the operator and can lead to the CE certification of the machining process.

According to an embodiment, the collaborative robot is configured to learn a plurality of tool positions on a workpiece surface of the workpiece in learning mode, project the tool positions onto a process surface to obtain tool points corresponding to tool positions on the process surface by interpolating between the auxiliary points within To determine the process surface additional work points in the process area and to work in work mode, the workpiece surface on the basis of the auxiliary points and the determined operating points. In this case, the tool positions can contain both a location information and a robot orientation. The tool positions are projected onto the process surface, e.g. to a (straight or curved) process level. In this case, the tool positions can be projected onto the process surface, in particular in the normal direction, in order to obtain the auxiliary points u. The auxiliary points can contain location information on the process surface as well as the robot orientations of the respectively associated tool positions. Furthermore, the auxiliary points can also contain the location information of the respectively associated tool positions. The interpolation between the auxiliary points for determining the additional operating points takes place within the process surface and can be done as a surface interpolation, that is, as a 2D interpolation. The operating points may contain location information on the process surface and interpolated robot orientations between adjacent auxiliary points.

According to a first development, the collaborative robot is configured to process the workpiece surface in working mode on the basis of an implicit projection of a work path planned within the process surface onto the workpiece surface. In this case, a target working area of the robot tool lies within the process area, not on the actual workpiece surface. The planned work path can only be created with knowledge of the taught example positions on the tool positions, i. without explicit and / or further (e.g., modeled) information about the surface contour of the workpiece surface. There is thus a contour-adapted surface treatment of the workpiece.

According to an alternative second development, the collaborative robot is configured to process the workpiece surface in work mode on the basis of an explicit projection of a projected work path onto a modeled surface model of the workpiece surface. The collaborative robot creates the surface model of the workpiece surface from the learned tool positions. A target working area of the robot tool lies within the modeled surface model, which differs as little as possible from the actual workpiece surface.

According to an embodiment, the learning mode is configured such that the collaborative robot in the learning mode learns the working movement by manual guidance of an operator. In other words, an operator guides the collaborative robot, more specifically the robot tool attached to the collaborative robot, along at least a portion of the work motion to be performed. The operator moves the collaborative robot to the desired target position. The robot can thus be moved in the learning mode by the operator by direct pressure application by hand. The operator stands with his or her hand / hands in direct contact with the collaborative robot, e.g. the robot tool. In doing so, the operator, for example a skilled worker, transfers his process capabilities to the collaborative robot by manual guidance. So far, a teach-in function for end users is hardly usable because in MRK, ie the human-robot collaboration, a CE certification depends on the robot movement. As a result, learning by hand does not require certification during processing. However, when learning by hand, the experience of the operator can be transferred to the working movement of the collaborative robot.

According to one embodiment, the collaborative robot further comprises an enabling button for the learning mode, wherein the enabling button is operable with the foot. By pressing the consent key, the collaborative robot is signaled that the robot tool is arranged at the actuation time in a desired tool position, which the collaborative robot should approach in the course of the working movement. In other words, when learning the collaborative Robot guided by the operator along the desired working movement, the enabling button is actuated. Here, a three-stage enabling button can be used. The three levels differ from each other in how far the consent button is depressed by the person doing the teaching. In this case, the enabling button is in a first position unconfirmed and the collaborative robot can not be moved. In a second position of the consent button at a pressure point of the collaborative robot is movable, ie it can be moved along the working movement. In this case, the collaborative robot can be taught in particular the movement to be learned. In addition, there is a panic function as the third position of the enabling button, for example a complete pressing of the enabling button, in which the robot movement is abruptly stopped.

Here, the enabling button can also be actuated both by hand and with the foot. The enabling button is flexibly positionable. This means that the enabling button is both locatable on the ground where it can be operated by foot and spaced from the ground (e.g., on the collaborative robot and / or a trolley) where it can be manually operated. The enabling button may e.g. via a cable and / or via a radio link in signal exchange with the (plant) control of the collaborative robot stand. Depending on the working movement, the enabling button can be positioned by the operator in such a way that he can safely actuate the enabling button either by hand or by foot

The consent button can be operated by a foot of the operator, whereby the operator has both hands free to teach and / or guide the collaborative robot in the learning mode. The enabling button may e.g. be arranged on the floor. This improves the handling in the learning mode, since the operator has both arms and hands free when learning, thus enabling a high-precision learning movement of the collaborative robot. Under certain circumstances, this can only bring the learning function through the end user to economic use. Through an ambidextrous leadership not only rough processes are learned, but also the finest process nuances. The learning process is thus very precise and accurate.

According to one embodiment, in the learning mode and / or in the working mode, at least one pressing force of the robot tool can be switched to the workpiece in a determinable direction. In this case, for example, an arbitrarily determinable direction of the pressing force can be switched on. The connection of the pressure force can be done for example via a (plant) control. Here, the desired contact pressure on the workpiece can be determined very accurately. The pressure force can be defined as part of the working movement, which are switched on at one or more specific working positions, times and / or periods of the working movement of these.

In one embodiment, the collaborative robot is configured to automatically determine a (eg, at least partially planar and / or at least partially curved) component surface of the workpiece, preferably automatically locating a path of the robot tool to the determined component surface and / or automatically aligning a component Determined robot tool to the determined component surface and / or automatically determines a starting movement to the workpiece. The component surface, the robot can detect, for example, optically and / or haptically by means of at least one corresponding sensor. In this case, not only a single sensor has to be provided, but it is also possible to use e.g. During the haptic determination, measurement data from torque sensors from several axes of the collaborative robot can also be combined. Alternatively or additionally, the component surface can be communicated to the collaborative robot also via a (plant) control. As part of the working movement, the collaborative robot may, for example, determine coordinates in three-dimensional space, both the component surface and the robot tool relative to the workpiece. In this embodiment, a control program of the collaborative robot is provided with a learning aid (also called a teaching aid), which supports and / or supplements the learning of the working movement in the learning mode.

In this case, the determination of the component surface can be made on the basis of an automatic identification drive of the collaborative robot. The starting movement can be designed so that shear points are avoided. The path of the robot tool can be generated by moving, scaling and / or distorting templates on the basis of taught-in support points, in particular the templates can be projected onto the surface and their curvature can be adjusted. The path of the robot tool can produce a surface movement of the robot tool, which can be matched to the determined component surface of the workpiece.

According to one embodiment, the collaborative robot is configured to automatically complete a motion sequence of the robotic tool based on positions of vertices learned in the learning mode. In other words, individual bases are first learned in the learning mode. These vertices can have positions and / or align and / or align the robot tool in three-dimensional space, eg, tool positions. Here, the collaborative robot (eg, depending on the particular application), for example, need a predetermined number of bases. For example, for one application, five vertices may be sufficient, for another ten, etc. In particular, the collaborative robot may require at least two vertices, at least four vertices, or at least ten vertices to complete the motion of the robotic tool. The completion of the movement sequence is carried out automatically due to predefined movement sequences. For example, the working movement can be supplemented by straight-line connection of the support points, or a circular, elliptical, or the like. Supplement of the individual bases. In particular, the working movement can also be supplemented in a three-dimensional form, for example along the shape of a spherical surface or the like. In this case, the support points can not only be connected, but predefined, eg complex movement patterns (also called templates) can be applied and moved on the basis of the support points on the component surface, scaled and / or distorted according to the surface curvature.

According to one embodiment, the collaborative robot has a force / torque sensor for determining at least one process force in the learning mode. The force-moment sensor may, for example, be arranged on a flange of the robot, in particular immediately adjacent to the robot tool. The force-moment sensor may be designed as a load cell. Hereinbei, for example, a pressing force of the robot tool can be determined on the workpiece as a process force. The process force measured by the force-moment sensor is stored and can be stored and / or used as part of the working movement. In working mode, the collaborative robot will then use exactly the process force measured by the force-moment sensor to machine the workpiece. The process force can thus relate in particular to a pressing force of the robot tool on the workpiece. The force-moment sensor is first used once for force recording. As a result, the work process and / or the work movement is learned and depicted more realistically, since not only the movement itself but also process forces associated with the movement are taken into account.

Alternatively, the force-torque sensor can also be used to improve the hand guidance of the robot. By using the force-moment sensor, the learning of the working movement can be improved and / or detailed.

The force-moment sensor, for example on the flange, detects the process forces and records, learns and / or stores them in correlation to the working movement of the collaborative robot. In addition, a weight and / or inertial force compensation of the robot tool attached to the robot can take place.

According to one embodiment, the collaborative robot is arranged on a trolley. In this case, the robot can be moved by moving the trolley in a desired working position. The robot can in particular be mounted on a surface of the trolley to learn from there from the work movement and perform. The trolley may be provided with a parking brake so as to reduce rolling of the trolley while working. The trolley can be moved freely by an operator. The operator can move the robot flexibly to a large component to be machined or to a small component to be machined, which is arranged for example on a workbench. A locking of the trolley is preferably carried out magnetically to fix the trolley particularly efficient. Alternatively, a mechanical fixation may be provided.

In an alternative embodiment, the collaborative robot is disposed on a magnetic lock. The Magnetarretierung may be formed as part of the collaborative robot. The Magnetarretierung can be realized on a frame which can be moved, for example by means of a crane. Magnetic locking allows the collaborative robot to be locked to all magnetic and / or sufficiently stable elements in horizontal, vertical and / or mixed alignment. In this case, the robot can be locked, for example, to third-party racks or scaffolding. In particular, the collaborative robot can also be locked to the workpiece to be machined itself, if this has a certain size. The Magnetarretierung can also be done partially on a frame and at the same time partially on the workpiece. This allows a highly flexible positioning of the collaborative robot. The positioning possibilities compared to established systems are extended. As a result, surfaces of large components can be achieved more easily, but also walls and / or facades. In addition, the use of a Magnetarretierung is designed such that the technical complexity of positioning can be reduced compared to established solutions. In particular, the workpiece to be machined can also be used as a robot base.

According to one embodiment, the collaborative robot has at least one optical sensor for detecting a surface roughness of the workpiece in the working mode. In particular, in polishing processes, when a desired target surface roughness is determined by the optical sensor, the working movement and / or the working mode can be terminated, since then the workpiece and / or its (partial) surface can be sufficiently processed. Alternatively, only a part of the working movement can be ended and, for example, a polishing can be continued on another part of the workpiece. The optical sensor is arranged on the collaborative robot in such a way that it can visually record and / or check the process surface. As a result, for example, a meaningless further polishing of a workpiece surface can be avoided since further polishing can in some cases even damage the workpiece, ie a minimum roughness of the workpiece can also be maintained. In addition, a manual check by an operator can be superfluous. In addition, an optical sensor can also be configured to provide further data for processing, eg for a distance measurement, for detecting the component surface, etc. The optical sensor can be designed for the localization of the workpiece and / or of subregions of the workpiece. The optical sensor can be designed to record images and / or videos from the working area of the robot tool.

• - Bereitstellen eines Roboterwerkzeugs für einen kollaborativen Roboter, welches zum Bearbeiten des Werkstücks ausgebildet ist;
• - Anwählen eines Lernmodus des kollaborativen Roboters, in welchem dem kollaborativen Roboter eine Arbeitsbewegung des Roboterwerkzeugs zum Bearbeiten des Werkstücks gelehrt wird;
• - Anwählen eines Arbeitsmodus des kollaborativen Roboters, in welchem der kollaborative Roboter die im Lernmodus erlernte Arbeitsbewegung durchführt.

One aspect relates to a method of operating a collaborative robot for machining a workpiece, comprising the steps of:

• Providing a robot tool for a collaborative robot, which is designed to process the workpiece;
• Selecting a learning mode of the collaborative robot in which the collaborative robot is taught a working movement of the robot tool for machining the workpiece;
• - Selecting a working mode of the collaborative robot, in which the collaborative robot performs the learned in learning mode work movement.

In the learning mode, the collaborative robot is taught the working movement of the robot tool on the workpiece to be machined.

The method can be carried out in particular with the above-described collaborative robot. Therefore, all comments on the collaborative robot also affect the method and vice versa. When selecting the learning mode, the collaborative robot is transferred to the learning mode. When selecting the working mode, the collaborative robot is put into working mode. The method is performed directly on the workpiece to be machined, so in particular under touching contact of the robot tool with the revised workpiece.

In one embodiment of the method, an operator moves the collaborative robot in the learning mode by means of manual guidance along the working movement. In this case, the operator can lead and / or move the collaborative robot, in particular with both his hands. This increases the accuracy and precision of the work movement to be learned.

In one embodiment, the learning mode is performed on the collaborative robot at operating temperature. This may be enabled, for example, by warming up the collaborative robot prior to performing the learning mode, by warming it up, or by performing a movement itself until it reaches a desired working temperature. Thus, even in the learning mode, the components, in particular lubricating oils etc., are already in the desired working temperature of the collaborative robot. This increases the precision of the working movement, in particular when the working movement is repeated. For this purpose, the collaborative robot can, for example, have a warm-up mode or be transferred into this warm-up mode, for example by a mode selection device. Alternatively or additionally, the collaborative robot can have an operating temperature measuring device with which the current and / or average and / or usual working temperature of the collaborative robot can be determined. In this embodiment, the robot is only transferred into the learning mode when the collaborative robot has reached the settable and / or predeterminable operating temperature.

According to an embodiment, the learning mode is performed at most a predetermined period of time before the machining of the workpiece. Preferably, the learning mode is performed immediately before the working mode, that is, the machining of the workpiece. As a result, on the one hand downtimes of the collaborative robot can be reduced, on the other hand it can be ensured that the processing is carried out with approximately the same working temperature of the collaborative robot in which learning of the working movement has also taken place. As a result, the work of the collaborative robot can be specified and / or improved. The predetermined time period may be, for example, a maximum duration of 15 minutes, preferably a maximum duration of 5 minutes.

According to one embodiment, in the learning mode, at least one pressing force of the Robot tool connected to the workpiece in a determinable direction, in particular in an arbitrarily determinable direction. The direction is preferably directed to the workpiece. In this case, therefore, only the working movement can first be taught in without process forces which, for example, are then switched on after learning and / or are superimposed on the working movement. Furthermore, a contact between the robot tool and the workpiece can be ensured in the context of Einlembewegung.

In one embodiment, the collaborative robot automatically determines a component surface of the workpiece, in particular it automatically determines an orientation of the robot tool to the determined component surface, and / or it automatically determines a starting movement to the workpiece. One or more of these learning aids, also called teaching aids, can be switched on, for example, software-controlled. This improves the learning of the working movement, in particular when the collaborative robot uses the data determined on the component surface to optimize the learned working movement in dependence thereon.

In one embodiment, the collaborative robot automatically completes a motion sequence of the robotic tool based on locations of vertices learned in the learning mode. In this case, a complete working movement can be learned and / or performed on the basis of a few support points.

One aspect relates to the use of a collaborative robot for machining a workpiece, wherein the collaborative robot comprises a robot tool for machining the workpiece and a mode selection device for transferring the collaborative robot to a learning mode and a working mode. In the learning mode, the collaborative robot is taught a working movement of the robot tool for processing the workpiece, and the collaborative robot, in working mode, carries out the work movement learned in the learning mode. The collaborative robot learns in the learning mode, the working movement of the robot tool on the workpiece to be machined.

The use can be carried out in particular with the above-described collaborative robot. For this reason, all statements regarding the collaborative robot and the method also relate to the use and vice versa.

The collaborative robot can be used in particular for surface machining of a component surface of the workpiece, e.g. for polishing. Here, in the learning mode, the teach-in of the robot to configure a surface-oriented machining process. The surface treatment differs from a pure line of the robot tool along a programmed trajectory. In surface machining, not only is a pure trajectory traced, but the working motion may include and / or be at least partially defined by a process force (such as a nudging force). Furthermore, first a web surface can be learned, then at least parts of the surface treatment are automatically calculated.

The collaborative robot can be used for a finishing application, for example for polishing injection molds and / or injection molds. In this case, the collaborative robot can be used in particular for the partial automation of the finishing application.

In the learning mode, skills and / or experiences of an operator (such as a skilled worker) can be transferred to the robot. Traditionally, the teach-in function of the collaborative robot has only been used with the aim of faster robot programming, but not to transfer the skills and / or experience of the skilled worker to the robot. The conventional Teaching by Demonstration function is traditionally not practicable by the user due to the recertification requirement when changing the robot movement. Thus, conventionally, there is no possibility that the user transfers his knowledge of experience directly to the robot, but at most via the detour of cooperation with a robot expert. However, this means a new robot programming and subsequent renewal of the safety consideration. This is usually not practical.

However, the inventive collaborative robot may have a hedging concept that remains valid regardless of the robot movement.

Here, e.g. the working movement of the robot be throttled, a non-hazardous robot tool used and / or a component contact ensured. In doing so, the collaborative robot is stopped as the touch contact (e.g., sensor monitored) between the robotic tool and the component surface of the workpiece is interrupted.

Furthermore or alternatively, a sensor-based protection can take place, depending on the proximity of the human to the collaborative robot hazards (in particular by the robot and / or the robot tool) reduced to a safe level or completely off.

Further or alternatively, prevention of contact with a human can be accomplished by mechanical guards.

In one embodiment, only such movements are learned in the learning mode, in which the robot tool is in contact with the component surface of the workpiece. This allows a considerable avoidance of crushing and / or shearing points. In conjunction with the reduced robot speed, the safety of the entire system increases despite any robot movements within the spatial area of the component surface.

In one embodiment, a variable and / or flexible connection of process forces in determinable directions, for example, to ensure a continuous and / or continuous component contact between the robot tool and the workpiece.

In one embodiment, the teach-in function of the robot in the learning mode is used to configure a robot-human collaboration.

In one embodiment, a derivation of a mathematical representation of partial surfaces of the workpiece takes place on the basis of taught-in and / or automatically detected surface coordinates of the workpiece. In this case, the automatic detection of the surface coordinates of the workpiece can take place within the framework of an automated identification travel of the robot.

In one embodiment, an automatic adaptation of the process takes place at curvatures of the workpiece surface.

In one embodiment, a completion of learning inputs recognized in the learning mode is performed by automatic processes.

In the context of this invention, the terms "substantially" and / or "approximately" may be used to include a deviation of up to 5% from a numerical value following the term, a deviation of up to 5 ° from one on the other Term following direction and / or from an angle following the term.

Terms such as above, below, above, below, etc., unless otherwise specified, refer to the reference system of the earth in an operating position of the subject invention.

• 1 in einer schematischen Darstellung ein Ausführungsbeispiel eines kollaborativen Roboters mit einem Rollwagen;
• 2 in einer schematischen Darstellung ein Ausführungsbeispiel eines kollaborativen Roboters in einem Lernmodus;
• 3 in einer schematischen Darstellung den in 2 gezeigten kollaborativen Roboter in einem Arbeitsmodus;
• 4 in einer schematischen Darstellung den in den 2 und 3 gezeigten kollaborativen Roboter in einem Kontrollmodus;
• 5 in einem 3D-Diagramm ein Ausführungsbeispiel einer zu bearbeitenden Werkstückoberfläche;
• 6A in einem 3D-Diagramm eine Prozessebene, welche der Roboter als eine Hilfsebene zur Bahnplanung verwendet;
• 6B in einem 2D-Diagramm, wie eine Robotersteuerung einen einhüllenden Polygonzug ermittelt;
• 7A in einem 2D-Diagramm ein erstes Ausführungsbeispiel einer abzufahrenden Arbeitsbahn in der Prozessebene;
• 7B in einem 2D-Diagramm ein zweites Ausführungsbeispiel einer abzufahrenden Arbeitsbahn in der Prozessebene;
• 8A in einem 3D-Diagramm ein Prozessrechteck mit Arbeitspunkten, an denen eine Roboterorientierung zu ermitteln ist;
• 8B in einem 2D-Diagramm das Prozessrechteck transformiert in die Prozessebene mit den Arbeitspunkten, an denen eine Roboterorientierung zu ermitteln ist;
• 9A in einem 2D-Diagramm ein erstes Ausführungsbeispiel eines Verfahrens zur flächenbasierten Interpolation von Roboterorientierungen;
• 9B in einem 2D-Diagramm ein zweites Ausführungsbeispiel eines Verfahrens zur flächenbasierten Interpolation von Roboterorientierungen;
• 10 in einem 3D-Diagramm ein Ausführungsbeispiel einer expliziten Projektion der Arbeitsbahn;
• 11 in einem 3D-Diagramm einen ersten Verfahrensschritt zur Erstellung eines Oberflächenmodells;
• 12 in einem 3D-Diagramm einen zweiten Verfahrensschritt zur Erstellung eines Oberflächenmodells;
• 13 in zwei 3D-Diagrammen einen dritten Verfahrensschritt zur Erstellung eines Oberflächenmodells; und
• 14 in zwei 3D-Diagrammen aus unterschiedlichen Richtungen einen Vergleich zwischen einem aus Tangentialebenen ermittelten Oberflächenmodell und einer tatsächlichen Werkstückoberfläche.

The invention will be described in more detail with reference to exemplary embodiments shown in FIGS. Herein, the same or similar reference numerals may designate the same or similar features of the embodiments. Individual features shown in the figures may be implemented in other embodiments. Show it:

• 1 in a schematic representation of an embodiment of a collaborative robot with a trolley;
• 2 a schematic representation of an embodiment of a collaborative robot in a learning mode;
• 3 in a schematic representation of the in 2 shown collaborative robot in a working mode;
• 4 in a schematic representation of the in the 2 and 3 shown collaborative robot in a control mode;
• 5 in a 3D diagram an embodiment of a workpiece surface to be machined;
• 6A in a 3D diagram, a process level which the robot uses as an auxiliary plane for path planning;
• 6B in a 2D diagram, how a robot controller detects an enveloping traverse;
• 7A in a 2D diagram, a first embodiment of a work path to be traversed in the process level;
• 7B in a 2D diagram, a second embodiment of a work path to be traversed in the process level;
• 8A in a 3D diagram, a process rectangle with operating points at which a robot orientation is to be determined;
• 8B in a 2D diagram, the process rectangle transformed into the process level with the operating points at which a robot orientation is to be determined;
• 9A in a 2D diagram, a first embodiment of a method for area-based interpolation of robot orientations;
• 9B in a 2D diagram, a second embodiment of a method for area-based interpolation of robot orientations;
• 10 an embodiment of an explicit projection of the working path in a 3D diagram;
• 11 in a 3D diagram, a first process step for creating a surface model;
• 12 in a 3D diagram, a second process step for creating a surface model;
• 13 in two 3D diagrams, a third process step for creating a surface model; and
• 14 in two 3D diagrams from different directions, a comparison between a surface model determined from tangent planes and an actual workpiece surface.

1 shows a schematic representation of an embodiment of a collaborative robot 10 on a trolley 60 is mounted. The trolley 60 can as part of the collaborative robot 10 be formed or formed as a separate component. The collaborative robot 10 has a robot arm 20 on, essentially from a pedestal 22 up to a flange end 21 is formed running. The flange end 21 can as a working end of the robot arm 20 be educated. Along its course are on the robot arm 20 several joints 23 only a few of them 1 Marked are. At the joints 23 is the robotic arm 20 movable, for example pivotable and / or rotatable. This will allow the robot arm 20 can perform a complex three-dimensional working movement through the room, in particular with its flange end 21 , The robot arm shown in the embodiment 20 has seven degrees of freedom, namely, three translational workpiece positions, three rotary workpiece positions, and one elbow (or even zero space) movement of the robot arm 20 ,

At the flange end 21 is a robot tool 30 on the robot arm 20 attached. The robot tool 30 is used for editing a not in 1 represented workpiece. At the flange end 21 can be a force-moment sensor 40 for measuring a pressing force of the tool 30 be arranged on the workpiece.

On the trolley 60 is a control on a bracket 50 of the collaborative robot 10 and / or the entire facility arranged. A controller of the collaborative robot 10 Can at least for the most part in the trolley 60 be integrated. This control, not shown, may be designed as a system control, with which not only the robot arm 20 , but also the other parts of the plant can be controlled, such as the robot tool 30 , a (consent) button, a traffic light, etc. The control element 50 may be formed as part of the plant control and or communicate with it.

The (plant) control and / or the operating element 50 can / may have a microprocessor and / or a user interface such as a touch screen. On the microprocessor of the control 50 becomes a control software for controlling the collaborative robot 10 executed. The (plant) control and / or the operating element 50 includes a mode selection device 51 , The mode dialer 51 may be part of the control software that is on the (plant) control and / or the control 50 is performed. The mode dialer 51 Used to select a mode of operation of the collaborative robot 10 , By pressing the mode dial 51 , for example, by pressing the touch screen of the (plant) control and / or the control element 50 , the collaborative robot can 10 be converted into a desired operating mode. The collaborative robot 10 can by means of mode selection device 51 optionally into at least one of the following operating modes: a learning mode, a working mode, at least one sleep mode (eg, a standby mode and / or an on / off mode), and / or a control mode.

Furthermore, either the (plant) control and / or the operating element 50 and / or another component of the collaborative robot 10 a memory, in particular a volatile memory (such as a RAM memory) in which the learning learned in the learning mode working movement can be stored.

The trolley 60 can preferably be moved freely by an operator to a workpiece to be machined. Alternatively, the trolley can be moved along a rail track, eg automatically. In the first case, the operator shifts the collaborative robot 10 flexible to a workpiece, for example, a large component to be polished or a small component to be polished, which is arranged on a workbench. Alternatively, small components as workpieces on the trolley 60 yourself filed and edited there. The trolley 60 can be magnetically locked, for example. The working movement, for example a polishing movement, of the robot tool 30 becomes the collaborative robot 10 taught in the learning mode. Here, the robot arm 20 , in particular the flange end 21 and / or an adjacent portion of the robot arm to the flange in FIG 20 be guided by an operator, for example by manual guidance along the working movement. In particular, the operator can guide the robot in learning mode to individual bases, which the operator as Want to save part of the work movement. The bases include information and / or data about the position of the robot tool 30 at the learned base, eg tool positions.

In working mode, the collaborative robot performs 10 the learned working movement, for example a polishing movement, by itself and repeats this until a desired work result, for example a desired polishing result, is reached. The work result can be monitored by a sensor and / or an operator, for example.

The operator, for example, a skilled worker, can during the automatic work process, for example, during the polishing process, ie during the collaborative robot 10 in working mode, stop in the working area of the robot.

At the same time, the speed of the collaborative robot 10 , in particular the speed of movement of the flange end 21 and / or the robot tool 30 , a maximum of a predetermined value. The speed of movement of the collaborative robot 10 can thus be throttled, especially in the learning mode and / or in the working mode. A predetermined maximum speed value may be, for example, a maximum of 50 mm / sec, preferably a maximum of 25 mm / sec. Such a limitation of the working speed of the collaborative robot can guarantee a relatively safe workplace for the operator. Furthermore, by such a limitation of the working speed of the collaborative robot 10 a CE certification for different types of tools and / or any process movements remain valid. Thus, no recertification after modification of the process is needed. The operator can monitor the work result, for example the polishing result, and / or intervene in the work process, for example applying a polishing paste, and / or the movement of the collaborative robot 10 correctively correct and / or carry out further activities in the immediate vicinity of the robot and the polishing tool.

The collaborative robot is particularly suitable for machining an injection molding tool, which must be polished. The required injection mold requires very high surface finishes, which can be achieved conventionally by manual polishing on partially relatively large areas. Here, a skilled worker requires a high level of qualification to carry out the polishing work. The polishing work requires a lot of time of the skilled worker, the polishing movement must be repeated often until the desired surface quality is achieved. The overall polishing time for an injection mold may be several days. Automation of the polishing process requires complex process control that reflects and / or incorporates the high quality and experience of the skilled worker. Furthermore, each injection mold can be unique, especially in "lot size 1". This requirement may be the collaborative robot 10 who in 1 is shown, especially when it is taught by a skilled worker as the operator and the robot tool 30 is designed as a polishing tool.

2 shows a schematic representation of an embodiment of a collaborative robot 110 , which is operated in a learning mode. The collaborative robot 110 is essentially similar to the one in 1 shown collaborative robot 10 , This is also the case of the collaborative robot 110 a robot arm 120 up, extending from a stand 122 up to a flange end 121 extends and which at several joints 123 is mobile. At the flange end 121 is a robot tool 130 arranged, which for machining a workpiece 300 is trained. The workpiece 300 can be arranged on a workbench. At the workpiece 300 For example, it may be an automobile rim made as an injection molded part, which may be replaced by the robot tool 130 must be polished. The workpiece 300 may for example be designed as an injection mold, for example for molding a car rim.

An operator 200 first transfers the collaborative robot into its learning mode and then moves the flange end 121 and in particular the robot tool 130 in a desired working position relative to the workpiece 300 , In particular, the operator 200 a polishing movement as working movement of the robot tool 130 on a surface of the workpiece 300 carry out. The movement of the robot tool 130 relative to and / or on the workpiece 300 becomes in a memory of the collaborative robot 110 stored. This may in particular be a volatile memory.

In the learning mode, the operator can use the robot tool 130 especially ambidextrous along the workpiece to be machined 300 to lead. Here, the operator 200 Movements and / or individual bases of the working movement via a consent button, not shown in the figures store and / or confirm, which, for example, with a foot of the operator 200 is operable. That's why the skilled worker, so the operator 200 , both hands free to guide the robot tool 130 and thus for learning the work movement. Here, the operator 200 either directly that robot tool 130 Touch with his hands and move to the desired working position, and / or the robot tool 130 by means of a handle or an intermediate piece of the robot arm 120 , which or about at the flange end 121 is designed to move to the desired working position.

The collaborative robot 110 allows a simple and ergonomic operation, which leads to a constant result. Furthermore, by learning the working movement directly on the workpiece to be machined 300 a quick learning done, since the working movement (at least partially) not cumbersome in advance must be calculated by software, but is tuned directly to the workpiece to be machined.

3 shows in a schematic representation of the collaborative robot 110 in working mode. This is the workpiece 300 with the previously learned in learning mode working movement of the robot tool 130 processed. The operator does not have to be required anymore. However, the latter can observe and / or monitor the operation from near or far and, if necessary, intervene in the processing or carry out other activities.

4 shows the collaborative robot 110 in an optional additional operating mode, namely in a control mode. Here is the editing tool 130 exchanged for an exchange tool 131 , The exchange tool 131 is at the flange end 121 of the robot arm 120 attached. In other words, the robot tool 130 generally interchangeable at the flange end 121 be attached. This applies to both the robot tool 130 as well as for the replacement tool 131 , The exchange tool 131 may comprise a sensor, in particular an optical sensor such as a camera, with which the work result is controlled. In this case, in particular, the quality of a polishing process can be checked, and / or errors and / or cracks on the workpiece 300 be detected. Thus, the replacement tool 131 eg used for quality control.

1. (1) Ein Lokalisierungssenor zur Lokalisation des Werkstücks, durch welchen trotz möglicherweise (z.B. leicht) geänderter Positionierung des Rollwagens und/oder des Werkstücks eine abgespeicherte Prozessbewegung automatisiert appliziert werden kann.
2. (2) Ein Überwachungssensor, welcher z.B. als eine Kamera ausgebildet sein kann, zum Übertragen eines Videos des Bearbeitungsprozesses an eine mobile Bedienerschnittstelle wie z.B. ein Smart Phone. Dadurch kann eine Bedienperson den Prozessfortschritt aus größerer Entfernung überwachen. Der Überwachungssensor kann am Roboterflansch und/oder am Rollwagen befestigt und/oder frei positionierbar ausgebildet sein.

Furthermore, at least one of the following two sensors can be provided:

1. (1) A localization sensor for locating the workpiece by which a stored process movement can be automatically applied despite possibly (eg slightly) changed positioning of the trolley and / or the workpiece.
2. (2) A monitoring sensor, which may be formed, for example, as a camera for transmitting a video of the processing process to a mobile user interface such as a smart phone. This allows an operator to monitor the progress of the process from a greater distance. The monitoring sensor can be attached to the robot flange and / or on the trolley and / or be designed to be freely positionable.

Furthermore, as a replacement tool and a pointer can be used. This makes it possible to teach in the process movement and in particular to carry out automatic identification runs without damaging the workpiece surface. Alternatively, the pointer can also be firmly anchored to the process tool.

In an alternative embodiment, the sensor, in this case the optical sensor, may also not be designed as an exchange tool, but rather as a permanently mounted component, in which case it is in addition to the actual machining tool 130 on the robot arm 120 be attached. Thus, the workpiece during processing, for example, in the working mode, recorded and / or checked.

In the embodiments shown, learning can be done on the real component, ie on the workpiece 300 , take place, with the need for the working movement robot tool 30 or. 130 , Teaching in the learning mode takes place shortly before the actual work process, ie the machining of the workpiece. This may mean shortly before that the learning mode is terminated at most 15 minutes before the machining of the workpiece, that is, the learning of the working movement is completed. More preferably, the learning of the working movement is terminated a maximum of about 5 minutes before the machining of the workpiece. In this case, optionally, a warm-up movement of the robot arm 20 or. 120 take place, so as to a predetermined operating temperature of the collaborative robot 10 or. 110 to reach. This can increase the accuracy of Einlemens.

In contrast to a conventional teach-in using offline programming eliminates sources of error such as CAD model errors, errors due to different robot operating conditions, sensor errors in sensor-based learning, calibration errors, etc. By learning the workpiece to be machined thus allowing a more accurate work, as sources of error reduced become. Thus, the repeat accuracy of the collaborative robot can be taught.

Furthermore, the collaborative robots allow 10 and 110 the learning of partial working steps of the working movement one after the other. So can For example, first part of the component surface of the workpiece 300 be polished by means of a first partial work movement. After a desired polishing result is achieved, another part of the component surface of the workpiece 300 can be polished by a second partial working movement, etc. The partial working movements can be learned immediately prior to performing the respective partial working movement in the working mode, so for example immediately before polishing the respective part of the component surface of the workpiece 300 ,

The plant control of the collaborative robot 10 and / or the collaborative robot 110 may contain one or more of the following learning aids, also called teaching aids:

A first learning aid enables a variable and / or flexible stiffness of degrees of freedom of movement of the collaborative robot 10 or. 110 , Thus, depending on the expected work movement, either individual joints 23 or. 123 of the robot arm 20 or. 120 be stiff in order to facilitate the learning of the working movement. For example, learning a two-dimensional polishing motion along a (substantially non-curved) component surface may block movement out of that plane, or the like. In this case, software-supported rough work movements and / or work areas can be predefined and / or predefined. In contrast to the programming of conventional working movement, in this case the exact working movement is only roughly predetermined, that is, for example, a coarse movement corridor is defined along the working movement. Then, at least in the learning mode, the freedom of movement of the robot tool 30 or. 130 limited to this predetermined and / or predefined movement corridor. This means that the robot is controlled in such a way that the robot tool 30 or. 130 is not moved out of the predetermined and / or predefined movement corridor. This can simplify the learning of the working movement in the learning mode.

The movement corridor can, for example, be at most 10 cm away from a provisionally estimated working movement in each spatial direction, preferably at most 5 cm, particularly preferably at most 1 cm. The movement corridor can be determined to be different in size for the individual spatial directions and / or support points.

A second learning aid enables a variable and / or flexible connection of process forces in determinable directions, for example to ensure a continuous and / or uninterrupted component contact of the robot tool 30 or. 130 with the workpiece 300 , In this case, in particular, a pressing force of the robot tool 30 or. 130 on a component surface of the workpiece 300 be switched on, for example, to set a polishing pressure.

A third learning aid can be an automatic variation of a zero space position of the collaborative robot 10 or. 110 enable. For example, the null space location may be from unneeded degrees of freedom of the collaborative robot 10 or. 110 derived and / or determined. For example, singularities can be avoided by varying the null space position. Here, the null space position of the collaborative robot is changed compared to a default or random null space position. Furthermore, by varying the null space position, the accuracy of a process force determination can be increased and / or maximized. Due to the variation of the zero space position, working space limits can be further avoided, a rigidity and / or a vibration behavior of the collaborative robot can be improved, in particular optimized.

A fourth learning aid allows a plausibility check and / or correction of the learned work movement. In this case, implausible contour components of the working movement are removed on the basis of model assumptions of the working movement. This may, for example, relate to unnecessary twitching or passing through a working space without component contact on the workpiece. Furthermore, it can be the orientation of the robot arm 20 or. 120 and / or the orientation of the robot tool 30 or. 130 corrected and, for example, be adapted to the orientation of the component surface.

A fifth learning aid allows for (eg automatic) generation of a spline trajectory of work movement from a cloud of vertices. Here, the spline can include a smoothing and / or jitter compensation of the working movement.

A sixth learning aid makes it possible to determine (for example automatically) an alignment of the component surface of the workpiece to be machined 300 , The orientation can be determined, for example, on the basis of the occurring process forces and / or optically. Subsequently, depending on this, a validation of the learned interpolation points can take place, for example based on 3D coordinates of the interpolation points. Alternatively or additionally, the orientation of the robot tool 30 or. 130 determined to the component surface and / or adjusted and / or optimized and / or validated. Likewise, a force vector of a process force (such as a pressure force) relative to the component surface of the workpiece 300 adjusted and / or optimized and / or validated. Depending on the determined component surface can continue to be defined the work area in which the working movement is arranged. Furthermore, depending on Component surface can be a starting movement to the workpiece 300 determined and / or planned. This can be, for example, a movement of the robot tool 30 or. 130 concern from the end point of the working movement to the starting point of the working movement, in particular without collision with the workpiece 300 or a movement of the robot tool from the default point to the starting point of the working movement.

The component surface can be determined, for example, by determining an orthogonal vector on the component surface. For example, the orthogonal vector may be derived from a mathematical representation of the component surface or a portion thereof. Such a mathematical representation can be achieved, for example, by an automatic measuring run of the collaborative robot 10 or. 110 be generated along the component surface, in particular under a predefined Andruckkraftregelung. The pressure force control can be predefined in the learning mode as part of the sixth learning aid and / or predefined by an operator. The approach movement to the workpiece 300 can be planned to avoid a shearing situation. Here, first, a movement at a first, greater distance from the workpiece 300 or whose component surface take place. Subsequently, a lowering of the robot tool 30 or. 130 on the component surface of the workpiece 300 take place, in particular in the direction of the surface normal (ie the orthogonal vector) on the component surface.

A seventh learning aid allows completion of learning inputs recognized in the learning mode by automatic processes. Here, on the basis of a few predetermined interpolation points, in connection with motion templates predefined in the (plant) control, it is possible to define paths, for example straight lines and / or ellipses and surfaces, for example a rectangular area and / or a circular area. The surfaces can be filled by predefined movement patterns, for example parameterizable triangular, sawtooth or spiral tracks. The movement patterns can be determined in such a way that the distribution of the track points over the total area is as uniform as possible. Furthermore, a shift, scaling and / or distortion of the predefined motion templates based on taught-in bases can be done. Thus, a circular motion can be distorted into an elliptical motion. Furthermore, a curvature of a two-dimensional motion template can take place and / or be provided. In particular, the distortion can be an automatic projection of predefined motion templates onto curved component surfaces of the workpiece 300 include. For example, the bends may be identified by a fitted CAD model and / or determined CAD-free by automated identification rides of the robot. Furthermore, complex movement patterns can be superimposed on learned working movements. This is also possible in several stages: eg. can be generated from individual trained bases in a first stage connecting these bases web. In a second step, this path can be overlaid with complex further movement patterns. Thus, for example, can result from the superposition of the web with a circular motion a spiral movement.

The seventh learning aid allows a faster learning of complex webs, a CAD-free learning of the work movement and / or an increase in the efficiency of learning.

One, several or all of the learning aids can be implemented as software and / or as part of the (plant) control and / or software-supported.

The learning aids combine a "Teaching by demonstration", ie learning the work movement by hand guidance, with automated assistance functions, eg. B. during Einlemens and / or after learning the working movement. Furthermore, a partial learning of the working movement is made possible, for example, a learning of some bases and / or support tracks, the rest of the working movement can be supplemented automatically. Furthermore, a working movement can be adapted to the concrete component contour in the context of the MRK and / or Anlemens. Disabling degrees of freedom, for example, can facilitate learning of two-dimensional work motions. By combining the learning of the working movement by hand guidance with a software-based path planning by one or more of the learning aids, an increase in the security of the application is made possible.

• - Der Prozess kann anders geregelt werden, z. B. kann eine Poliergeschwindigkeit und/oder eine Andruckkraft verändert werden.
• - Weiterhin kann der Prozess aufgrund der Daten optimiert werden, z.B. die Arbeitsbewegung auf einen Bereich der Bauteiloberfläche mit relativ hoher Oberflächenrauigkeit konzentriert werden.
• - Der Prozess kann beendet werden, z. B. bei Erreichen einer Zielgenauigkeit des Polierens. Hierbei kann z. B. der Arbeitsmodus beendet werden.
• - Eine Dokumentation des Arbeitsprozesses kann erstellt werden. Hierbei kann zum Beispiel festgehalten werden, welcher Rauigkeitsgrad nach wieviel Polierbewegungen erreicht wird.

According to one embodiment, the collaborative robot 10 or. 110 used for a polishing process. Here, a surface roughness of the workpiece 300 sensor-based, especially during the collaborative robot 10 or. 110 works in working mode. On the basis of the acquired sensor data, one or more of the following work reactions can be carried out automatically, in particular automatically:

• - The process can be regulated differently, eg. For example, a polishing speed and / or a pressing force can be changed.
• Furthermore, the process can be optimized on the basis of the data, for example the work movement can be concentrated on a region of the component surface with a relatively high surface roughness.
• - The process can be terminated, eg. B. upon reaching a target accuracy of polishing. This z. B. the working mode are terminated.
• - A documentation of the work process can be created. It can be noted, for example, which degree of roughness is reached after how many polishing movements.

Overall, unnecessary polishing work can be reduced and / or eliminated. Furthermore, objective and reproducible polishing results can be achieved because the process is independent of human variability. Another advantage is a possible automatic documentation of the polishing results and / or an automated optimization of the polishing process. Hand-held teach-in can be combined with a sensor-based process control.

The following explains how to use the collaborative robot 10 can work in the learning mode.

5 shows an embodiment of a workpiece surface in a 3D diagram 400 which the collaborative robot 10 to edit. The workpiece surface 400 is shown in a Cartesian coordinate system of the 3D diagram in which the x-axis, the y-axis and the z-axis are labeled.

• - einer vorbestimmten Geschwindigkeit,
• - einer vorbestimmten Andruckkraft und/oder
• - einer vorbestimmten Ausrichtung zur Werkstückoberfläche 400

möglichst homogen über einen definierten Flächenabschnitt der Werkstückoberfläche 400 geführt werden. Dabei kann die Bearbeitungsbewegung des Roboterwerkzeugs 30 über einen vorbestimmten Zeitraum und/oder eine vorbestimmte Anzahl Zyklen zu wiederholen sein.The target should be on the workpiece surface 400 a surface treatment can be realized (eg a polishing), ie the robot tool 30 should with

• a predetermined speed,
• - A predetermined contact pressure and / or
• - A predetermined orientation to the workpiece surface 400th

as homogeneous as possible over a defined surface section of the workpiece surface 400 be guided. In this case, the machining movement of the robot tool 30 be repeated over a predetermined period and / or a predetermined number of cycles.

One challenge can be the operator 200 to be able to self-teach the surface treatment (eg the polishing application) efficiently, ie within a few minutes if possible, and thereby the process knowledge of the operator, which has grown over years 200 on the robot 10 transferred to. Here, the processing of any (eg a priori unknown) surface contours such as the workpiece surface 400 and / or freely adjustable surface sections are made possible.

Here, the teaching of machining applications for workpiece surfaces 400 by a combination of a hand guiding function of the robot 10 So, a Teaching by Demonstration, with software-based assistance functions. Teaching takes place on the real workpiece 300 with the real robot tool 30 , This can cause transmission errors between the robot 10 and possible sensors and / or CAD data are reduced and / or eliminated. The accuracy of the overall process can only by the repeatability of the robot 10 be determined.

Details of the learning process and / or path planning of the collaborative robot 10 are with reference to the following 6 to 14 described. The described method steps can be supported, planned and / or performed by a robot controller, in particular a robot controller with a processor. The robot controller may be configured, configured and / or programmed to be the collaborative robot 10 created a working track for carrying out the surface treatment, on the basis of which the robot 10 by means of an implicit and / or explicit projection of the planned work path onto a surface contour of the workpiece surface 400 this workpiece surface 400 can edit.

The collaborative robot 10 is from the operator 200 to a plurality of tool positions x on the workpiece surface 400 guided. The operator moves 200 the robot tool 30 to discrete points, namely the tool positions x , on the surface contour to be machined, ie the workpiece surface 400 , In this case, the operator 200 a mechanical contact between the robot tool 30 and the workpiece surface 400 produce. These discrete points of tool positions x form the basis for a predetermined traverse to be created which defines and / or demarcates the processing area.

Alternatively, the robot can 10 Automatically teach-in complex surface contours (eg as partial boundaries of the machining area) by automatically moving along these surface contours (under pressure against these surface contours) and storing the resulting path of movement.

Determination of the process level P

Depending on the taught-in tool positions x becomes first a process level P determined.

6A shows such a process level in a 3D diagram P which the robot 10 used as an auxiliary level to plan his work path. The process level P can be arbitrarily positioned and / or oriented in three-dimensional space. The process level P can eg in the taught-in tool positions x be fitted, for example by minimizing the square distances between the tool positions x and the process level P ,

The robot controller may design the work path (also referred to as path planning) for the robot tool 30 along the workpiece surface 400 in two-dimensional space, namely within the process level P ,

By projection of the tool positions x to the process level P along a normal N to the process level P become auxiliary points f within the process level P determined. The auxiliary points f are in 6A marked with a small circle. This corresponds to the auxiliary points f in the process level P the tool positions x on the workpiece surface 400 ,

At the process level P can be an enveloping process rectangle R be defined, which all auxiliary points f contains. The process rectangle can be R be minimized so that in each case at least one auxiliary point f on a sideline of the process rectangle R is positioned, but not a single auxiliary point f outside the process rectangle R , The process rectangle R can be due to geometric conditions of the workpiece 300 and / or the workpiece surface 400 be aligned, for example, along principal axes of inertia of the workpiece surface 400 be oriented.

The process rectangle R may be rectangular with a first major axis H1 _P , and a second major axis H2P , Here are the two main axes H1 _P and H2 _P parallel to each other opposite sides of the process rectangle R through a center M of the process rectangle R therethrough. The first major axis H1 _P may be formed as a shorter major axis, while the second major axis H2 _P may be formed as a longer main axis.

This approach of projecting the taught-in tool positions onto a process surface, the position and shape of which are adapted to the taught-in tool positions, and within which the path planning takes place, can alternatively take place by means of different surface forms. In other words, the path planning may also be within a different geometric shape than within the process rectangle R occur, such as within a circle and / or a polygon within the process level P , Furthermore, the path planning does not necessarily have to be within the process level P Instead, for example, a cylindrical, spherical or otherwise curved process surface can be used. A Selection of the Suitable The basic shape of the process surface, eg, process level, process cylinder, process sphere, can for example be adapted to the workpiece geometry. This can be done either automatically or by the operator 200 respectively. If rotationally symmetrical workpieces are to be machined, for example, the selection of a cylindrical process surface can be advantageous in order to simplify the machining and to minimize errors which arise due to the projection onto the process surface.

Determination of the enveloping traverse

6B shows in a 2D diagram, how the robot control the auxiliary points f within the process level P to one of the auxiliary points f enveloping traverse PZ combines. The polygon train PZ lies completely within the process rectangle R or at most on its sidelines. In this case, the polyline PZ at least one intersection with each sideline of the process rectangle R respectively. The polygon train PZ connects all auxiliary points f and is more convex. Alternatively, non-convex polygons can be used and, for example, be useful to describe complex boundary profiles of surface sections.

Alternatively, the traverse PZ also as an outer, eg convex, frame through some outer auxiliary points f be defined, where the traverse PZ several auxiliary points f encloses, which are arranged in its interior. The exterior of the traverse PZ is in any case free of auxiliary points f educated.

Planning the work path

7A and 7B show in each case a diagram, as the work path to be departed B in the process level P in the process rectangle R and within the traverse therein PZ is defined. Pre-defined prototypical movement patterns such as zigzag, loops, spirals, etc. are used as templates. The movement patterns for the work path B can by the operator 200 be freely parameterizable. For example, the operator 200 select a pitch of the zigzag patterns, a loop width, etc. as parameters freely. This is the work path B along the sides of the wrapping process rectangle R in the process level P clamped.

6A shows in a 2D diagram an embodiment, wherein a zigzag track as a work path B is determined. This will be parallel to the first main line H1 _P of the process rectangle R Guides shown in dashed lines H _i used, where i corresponds to a numbering. The guides H _i have a constant distance to each other, which as a parameter of the working path B from the operator 200 can be determined and / or selected.

An intersection of the traverse PZ with an end-side guide H1 , that is, an auxiliary line which is adjacent to a (here lower) lateral line of the process rectangle R can be arranged as a starting point S the work path B be determined. At the starting point S cuts the end first guide H ₁ the polyline PZ either on a left or right side of the surface of the traverse PZ , In the embodiment shown on a left side. From the starting point S becomes the work path B determined to an intersection of the adjacent, second auxiliary line H ₂ with an opposite (right side) side of the traverse PZ , from there to an intersection with an adjacent, third auxiliary line H ₃ with the opposite (left side) side of the traverse PZ etc. to the last auxiliary line H _i at the upper end of the traverse PZ , From there, the work path B determined in opposite direction between the intersections with the traverse back and forth.

The direction of the work path B is marked with arrows. The determined work path B serves as the basis of the robot movement during the working mode.

6B shows in a 2D diagram an embodiment, such as a spiral path as a work path B is determined. It will be within the polygon PZ a spiral path determined. Intersections of the spiral path with the polygon PZ can also be based on (in 7B not shown) auxiliary lines are determined, which exactly as in 7A shown guides H _i parallel to one of the main lines of the process rectangle R can be configured and parameterized, in particular parallel to the shorter first main line H1 _P of the process rectangle.

This is a two-dimensional work path B defined in the process level.

The following describes possibilities such as this work path B on the surface contour, ie the workpiece surface 400 can be projected and used there.

This can be done, for example, either by an implicit projection of the work path B on the workpiece surface 400 or by an explicit projection of the work path B on the workpiece surface 400 ,

Implicit projection of the work path

In an implicit projection of the work path B can be a sensitive robot as a collaborative robot 10 be used. It is programmed to be the one at the process level P defined work path B departs. At the same time the robot becomes 10 in a direction of equilibrium, eg in the direction of the normal N to the process level P (see also 6A) , yielding. In this case, a co-force is applied in this direction of equilibrium, the force on the workpiece 300 is directed and works. Alternatively, the direction of equilibrium can also be in a different direction than the normal N be directed, eg in the direction of impact of the tool o.Ä.

Thus, the real robot pose of the programmed robot pose (in the work path B ), in the direction of the normal N to the process level P until a mechanical contact between the robot 10 and the workpiece 300 will be produced. Translational tool coordinates automatically adapt to the surface contour, ie the workpiece surface 400 even though this is the robot 10 actually is not known and he has planned his working movement only in the two-dimensional process level.

In other words, the robot becomes 10 so controlled, as if he were a two-dimensional working movement along the work path B execute in the process level, whereby only the equal force is switched on.

A purely translator-specific adaptation of the robot / tool pose for machining applications can often not be sufficient. Therefore, in addition, rotary robot and / or tool coordinates can also be continuously adjusted. This can be done so that the robot tool 30 always at a constant angle to the workpiece surface 400 stands.

In the context of this invention, these robot and tool coordinates, which also contain robot and tool orientations, are summarized under the term robot orientations.

In this case, an interpolation of robot orientations can take place via lines or paths, eg around the robot orientation at the starting point S the work path B continuously into the robot orientation at the end point of the work path B to convict. This can also be for start and end points of a section of the work path B apply, eg between two taught-in tool positions x and / or the corresponding auxiliary points f ,

The robot controller thus configured can perform interpolation of robot orientations not only along a line but also in a plane. In this case, a surface interpolation is performed. Based on a limited number of distributed sample orientations at the auxiliary points f The robot orientations are at arbitrary points within the process area, eg the traverse PZ , derived and / or determined. The robot orientations correspond to the auxiliary points f the robot orientations at the associated, taught-in tool positions x , The robot orientations interpolated in this way are used, in particular, in the context of the determination and / or definition of the work path B used.

8A and 8B show in diagrams, at which points the robot orientations are determined. It shows 8A the process rectangle R while in a 3d diagram 8B the process rectangle R transformed into the process level in a 2D diagram P shows.

The learned robot orientations are at the auxiliary points f each shown by a stylized, bold coordinate system sign. For the railway planning of the work path B The robot orientations will be at additional work points a needed. The robot orientations at the working points a are each shown by a stylized, sparingly shown coordinate system character and are determined by interpolation of the learned robot orientations. Here are not all, but only some of the working points a indicated by a reference numeral.

The operator 200 Gives the possibility to manually teach in the robot orientations and thus the tool orientations at dedicated points of the machining area. From the orientations at the auxiliary points, there can be an interpolated orientation for each point within the two-dimensional traverse PZ be calculated.

If between the taught and / or determined working points a an approximately uniform curvature of the surface contour of the workpiece surface 400 is present, by the interpolation an approximately automatic adaptation of the robot orientation to the curved surface contour of the workpiece surface 400 reached. This can apply to any point within the editing area.

Thus, on the basis of only less einzulernender example points, namely the Werkzeugpositiionen x and the derived auxiliary points f , a contour-matched both translational and rotational alignment of the robot tool 30 be achieved. Thus, within the implicit projection of the work path B only with knowledge of the taught example poses, in particular without explicit information about the surface contour of the workpiece surface 400 , a contour-adapted surface treatment of the workpiece 300 allows.

Number and distribution of the taught example poses can be done by the operator 200 determine the accuracy of the contour adjustment to the process requirements.

The area-based interpolation method, compared to conventional methods, significantly reduces the technological and practical effort for the operator. Furthermore, this makes it possible to reduce aberrations of, for example, sensor CAD robots, since with the real robot tool 30 on the real workpiece 300 is working.

First exemplary embodiment of a surface-based interpolation of robot orientations for an implicit projection of the working path

By way of example, two possible methods for a surface-based interpolation of robot orientations are explained below.

9A shows in a 2D diagram a first embodiment of a method for area-based interpolation of robot orientations at operating points in a polygon PZ , The polyline is based on this PZ eg on the auxiliary points f1 to f10 ie at these eg ten auxiliary points f1 to f10 are the robot orientations learned and thus known.

In this method, those at the polygon vertices of the polygon draw PZ stored robot orientations used as a basis to interpolated robot orientations at any operating points within the polygon PZ derive.

1. 1. Es wird eine Interpolationslinie I durch den Arbeitspunkt fS gelegt, und zwar parallel zur (z.B. kleineren) Seitenkante des Prozessrechtecks R
2. 2. Es werden Schnittpunkte der Interpolationslinie I mit dem Polygonzug PZ ermittelt, in als Schnittpunkte fL1 und fL2 gekennzeichnet.
3. 3. Die Roboterorientierungen an den Schnittpunkten fL1 und fL2 werden z.B. durch lineare Interpolation der Roboterorientierungen an den entlang des Polygonzugs PZ benachbarten, vorbekannten Hilfspunkten f ermittelt. Dies sind für den ersten Schnittpunkt fL1 die Hilfspunkte f1 und f2, für den zweiten Schnittpunkt fL2 die Hilfspunkte f8 und f9.
4. 4. Die Roboterorientierung an dem Arbeitspunkt fS wird durch z.B. lineare Interpolation der Roboterorientierungen an den Schnittpunkten fL1 und fL2 ermittel.

An algorithm for deriving an interpolated robot orientation at an operating point fS in the traverse PZ in the process rectangle R can work as follows:

1. 1. It becomes an interpolation line I through the working point fs parallel to the (eg smaller) side edge of the process rectangle R
2. 2. There are intersections of the interpolation line I with the traverse PZ determined in as intersections f L1 and f L2 characterized.
3. 3. The robot orientations at the intersections f L1 and f L2 eg by linear interpolation of the robot orientations along the traverse PZ adjacent, previously known auxiliary points f determined. These are for the first intersection f L1 the auxiliary points f1 and f2 , for the second intersection f L2 the auxiliary points f8 and f9 ,
4. 4. The robot orientation at the working point fs is achieved by, for example, linear interpolation of the robot orientations at the intersections f L1 and f L2 ermittel.

This area-based interpolation method allows the robot orientation to be located at any operating point within the polygon PZ determine.

Second exemplary embodiment of a surface-based interpolation of robot orientations for an implicit projection of the working path

9B shows in a 2D diagram a second embodiment of a method for area-based interpolation of robot orientations at operating points within a process rectangle R , Within the process rectangle R are the robot orientations at several auxiliary points f but not only along the enveloping traverse PZ are distributed, but arbitrary in the traverse PZ can be arranged.

Also in this method, the at auxiliary points f stored robot orientations used as a basis to interpolated robot orientations at any operating points within the traverse PZ derive.

1. 1. Der gesamte Bearbeitungsbereich wird in Dreiecke unterteilt, deren Eckpunkte die eingelernten Hilfspunkte f bilden.
2. 2. Innerhalb jedes Dreiecks wird die im Zusammenhang mit 9A beschriebene flächenbasierte Interpolation durchgeführt.

It can robot orientations on arbitrarily distributed auxiliary points f be used as a basis for the area-based interpolation. An algorithm for deriving an interpolated robot orientation at an operating point in the traverse PZ can work as follows:

1. 1. The entire editing area is divided into triangles whose vertices are the taught-in auxiliary points f form.
2. 2. Within each triangle is related to 9A area-based interpolation described.

Being after related 9A described method, an interpolated robot orientation is calculated from a maximum of four adjacent, taught robot orientation, while the robot orientation is calculated from a maximum of three adjacent, taught-robot orientation.

Alternatively, interpolated robot orientations can also be calculated by weighted averaging of adjacent example orientations. The weights may be e.g. result from area-based membership functions (e.g., radial basis functions) centered around each example point.

This area-based interpolation method also makes it possible to orient the robot at any operating point within the polygon PZ determine.

Explicit projection of the work path

10 shows in a 3D diagram an embodiment of an explicit projection of the work path B , At the same time, this becomes the process level P defined work path B on the surface contour of the workpiece surface 400 projected, which is present or calculated as a mathematical model. By projection of the work path B on the workpiece surface 400 , eg along the normal N to the process level P , results in the 3D work path B '. Instead of at the process level P determined work path B becomes the robot 10 on the mathematically calculated workpiece surface 400 projected workspace B ' guided.

In the explicit projection of the work path B on the workpiece surface 400 are the target coordinates of the robot 10 on the surface contour of the workpiece surface 400 , at least according to the mathematical surface model.

In the case of the implicit projection of the work path, the target coordinates of the robot are located 10 in the process level P ,

Thus, in the case of explicit projection usually curved target tracks of the robot arise 10 , even in the case of a railroad within the process level P , That's why the in 10 shown section of the projected work path B ' curved, although the corresponding section of the working path B in the process level P is straightforward.

1. 1. Eine automatische Positionierung des Roboterwerkzeugs 30 auf der Oberflächenkontur der Werkstückoberfläche 400.
2. 2. Eine automatische Ausrichtung des Roboterwerkzeugs 30 in einem definierten Winkel zur Oberfläche (z.B. unter einem Arbeitswinkel von 90°).
3. 3. Eine verbesserte Vorhersehbarkeit der Roboterbewegung, da der Abstand zwischen realer und programmierter (projizierter) Arbeitsbahn B' durch den Abstand zwischen Oberflächenmodell und realer Werkstückoberfläche 400 bestimmt wird und größere Abweichungen zwischen Modell und Realität nicht zu erwarten sind. Im Falle der impliziten Projektion sind dagegen größere Abstände zwischen realer und programmierter Arbeitsbahn B zu erwarten.
4. 4. Eine Homogenisierung des Bahnverlaufs, da durch Verzerrung der projizierten Arbeitsbahn B' eine homogenere Verteilung selbiger Bahn über die modellierte Werkstückoberfläche 400 erzielt werden kann.
5. 5. Eine Homogenisierung der Bahngeschwindigkeit durch eine dynamische Einstellung der Bahngeschwindigkeit, so kann bei einer großen Bahndichte eine große Geschwindigkeit verwendet werden, während bei einer kleinen Bahndichte eine kleine Geschwindigkeit verwendet werden kann.

Explicitly projection-based methods allow:

1. 1. An automatic positioning of the robot tool 30 on the surface contour of the workpiece surface 400 ,
2. 2. An automatic alignment of the robot tool 30 at a defined angle to the surface (eg at a working angle of 90 °).
3. 3. Improved predictability of robot movement, as the distance between real and programmed (projected) work path B ' by the distance between surface model and real workpiece surface 400 is determined and larger deviations between model and reality are not to be expected. in the In contrast, the implicit projection has larger distances between the real and the programmed work path B expected.
4. 4. A homogenization of the trajectory, as by distortion of the projected work path B ' a more homogeneous distribution of the same web over the modeled workpiece surface 400 can be achieved.
5. 5. Homogenizing the web speed by dynamically adjusting the web speed, a high speed can be used with a high web density, while a small speed can be used with a small web density.

It can in particular by the Homogenisierungsansätze 4 , and 5. a uniform distribution of the residence time of the machining tool 30 over the entire workpiece surface to be machined 400 and thus a uniform work result - despite the curved surface contour - can be achieved.

The generation of a surface model is conventionally based on sensor and / or CAD data. This leads to disadvantages such as a high technical effort, multiplied error sources and transmission errors between CAD data, sensor data and robots 10 ,

Therefore, the robot according to the invention 10 one of the methods described below for mathematical model calculation of the workpiece surface 400 use, so to create a surface model.

Surface model based on haptic scanning for an explicit projection of the working path

A surface model of the workpiece surface 400 can be created by haptic scanning. In this case, before the process execution, the workpiece surface 400 be scanned by a probe haptic.

Alternatively, a haptic scan can be performed during process execution. During the process execution, the robot slides 10 Process-related with his robot tool 30 on the surface contour of the workpiece 300 from. The associated poses, ie robot orientations, are determined by the robot 10 recorded and for the derivation of a mathematical surface model or for the continuous refinement of a mathematical surface model of the workpiece surface 400 used. A mathematical surface model can, for example, be defined and / or be based on splines. Adaptation to recorded robot / tool poses, ie robot orientations, can take place, for example, by fitting in the sense of a minimal quadratic distance between the surface model and recorded poses.

Thus, during the process execution, surface coordinates are haptically acquired and used during the running time of the processing application for the generation or improvement of the surface model.

Surface model for an explicit projection of the work path by teaching

Again alternatively, a surface model can be created by teaching. This will be the operator 200 enables a complex surface contour by manual guidance of the robot 10 (including the robot tool 30 ). The surface creation process can be accomplished by the following steps:

In a first step, the operator learns 200 the position and orientation of any selected example points by manual guidance, ie arbitrarily selected tool positions x , This is analogous to and just as with implicit projection. In doing so, the operator provides 200 at every tool position x a mechanical contact between the robot tool 30 and the workpiece 300 ago.

11 shows in a 3D diagram this first process step, namely as on the workpiece surface 400 a plurality of tool positions x with associated robot orientation are taught in three-dimensional space. There are only a few of the taught-in tool positions x marked with a reference numeral.

At every tool position x judges the operator 200 robot tool 30 in the desired and always identical angle to the surface contour, ie in the associated robot orientation.

12 shows in a 3D diagram a second process step, wherein the by the operator 200 taught-in robot orientations at the tool positions x the associated normal vectors on the surface contours of the workpiece surface 400 be determined. It can be for every tool position x based on the normal vector, an associated tangent plane T To be defined. In the 12 These are tangential planes T symbolized and marked by circular areas.

13A and 13B show in a 3D diagram from different views a third process step, the function values of the tangent planes T with the reference to the 9A and 9B described interpolation approach are mutually grinded.

The tangential planes can do this T within the polygon PZ in the process level P be surface-based interpolated. Instead of the robot orientations, the function values of the tangential planes become T at the tool positions x , more precisely at the associated auxiliary points f in the process level P , interpolated.

The result of this method is one of the exemplarily selected tool positions x reconstructed mathematical surface model 400 ' the workpiece surface 400 ,

As an alternative to this third process step, the tangential planes T is approximated by a general surface model based, for example, on splines. In this case, a weighting of the surface coordinates of each tangent plane T be taken into account, the larger the distance from the development point of the tangent plane T , ie the taught-in point, drops.

An advantage of this alternative is that the properties of the surface model can be influenced by the choice of the model structure. For example, a model based on splines can achieve a steady course of the approximated surface.

As described above, with the help of the surface model on the one hand poses and on the other hand the orientations of the robot can be adapted to the surface contour. Both adaptation methods can be used independently. For example, the translatory coordinates of the robot from the surface model and the rotatory coordinates can be independently derived from an interpolation of the learned robot orientations. As a result, even in the case of discontinuities in the surface model, a continuous orientation course over the entire surface can be achieved.

The 14A and 14B show in each case a 3D-diagram from different directions a comparison between that from the tangential planes T determined surface model 400 ' and the actual workpiece surface 400 ,

The comparison between real surface 400 and reconstructed surface 400 ' shows only minor deviations. This was only with 21 example points, ie tool positions x , a complex contour course simulated in practical accuracy.

Teaching in a surface model 400 ' Similar to the learning process, it implies an implicit projection with comparably little practical effort. At the same time, the above advantages of explicit projection can be used. As the learning process of the surface model 400 ' with the real robot tool 30 on the real workpiece 300 is done, transmission errors between a sensor, CAD and robots are avoided. Furthermore, the user can change the number and distribution of tool positions x as supporting points free and thereby the accuracy of the surface model 400 ' control and adapt to the process requirements. This makes it possible to different parts of the surface model 400 ' to teach in different ways.

Through augmented reality, the current surface model can be used during the learning process 400 ' that results from the already taught-in tool positions x results, virtually projected on the real contour. This allows the operator 200 a direct feedback about, in which depth of detail the surface model 400 ' the real contour of the workpiece surface 400 reproduces. Furthermore, the operator 200 Based on this feedback, simply decide where more tool positions x should be taught. Further points of contact for augmented reality arise in the context of the presentation of the process level, the processing area and the taught-in course.

LIST OF REFERENCE NUMBERS

10

collaborative robot

20

robot arm

21

flanging

22

stand

23

joint

30

robot tool

40

Force-torque sensor

50

operating element

51

Mode selection device

60

trolley

110

collaborative robot

120

robot arm

121

flanging

122

stand

123

joint

130

robot tool

131

exchange tool

200

operator

300

workpiece

400

Workpiece surface

400 '

modeled workpiece surface

B

working web

B '

projected work path

H _i

ledger line

H1 _P

first main axis

H2 _P

second major axis

I

interpolation

M

Focus

N

normal

P

process level

PZ

traverse

R

process rectangle

S

starting point

T

tangent

a

working

f

auxiliary point

fs

working

f L1

intersection

f L2

intersection

x

tool position

Claims (20)

Collaborative robot (10; 110) for processing a workpiece (300) with: a robot tool (30; 130) for machining the workpiece (300) and a mode selection device (51) for transferring the collaborative robot (10; 110) to a learning mode and a working mode; in which the collaborative robot (10; 110) is configured to learn, in the learning mode, a working movement of the robot tool (30; 130) for machining the workpiece (300); - the collaborative robot (10; 110) is configured to perform in work mode the work movement learned in the learning mode; and the learning mode is configured such that the collaborative robot (10; 110) learns, in the learning mode, the working movement of the robot tool (30; 130) on the workpiece (300) to be machined. Collaborative robot (10; 110) after Claim 1 wherein the learning mode is configured such that the collaborative robot (10; 110) learns the working movement in the learning mode by manual guidance of an operator (200). Collaborative robot (10; 110) after Claim 1 or 2 wherein the collaborative robot (10; 110) is adapted to learn a working movement in the learning mode for processing an injection mold as a workpiece. Collaborative robot (10; 110) according to one of the preceding claims, wherein the movement speed of the robot tool (30; 130) and / or the collaborative robot (10; 110) in the learning mode and / or in the working mode to a maximum speed of at most 50 mm / s is throttled. A collaborative robot (10; 110) according to any one of the preceding claims, wherein the collaborative robot (10; 110) is configured to learning a plurality of tool positions (x) on a workpiece surface (400) of the workpiece (300) in the learning mode, to project the tool positions (x) onto a process surface (P) in order to obtain auxiliary points (f) on the process surface (P) corresponding to the tool positions (x), - To determine by means of interpolation between the auxiliary points (f) within the process area (P) additional operating points (a) in the process area (P) and - To edit in work mode, the workpiece surface (400) on the basis of the auxiliary points (f) and the determined operating points (a). Collaborative robot (10; 110) after Claim 5 wherein the collaborative robot (10; 110) is configured to operate in work mode on the workpiece surface (400) based on an implicit projection of a work path (B) within the process surface (P) onto the workpiece surface (400). Collaborative robot (10; 110) after Claim 5 wherein the collaborative robot (10; 110) is configured to, in working mode, machine the workpiece surface (400) based on an explicit projection of a projected work path (B ') onto a modeled surface model (400') of the workpiece surface (400). A collaborative robot (10; 110) as claimed in any one of the preceding claims, including an enabling button for the learning mode, the enabling button operable with the foot; and / or wherein in the learning mode and / or in the working mode at least one pressing force of the robot tool (30; 130) is connectable to the workpiece (300) in a determinable direction. The collaborative robot (10; 110) of any one of the preceding claims, wherein the collaborative robot (10; 110) is configured to automatically determine a component surface of the workpiece (300), preferably automatically locating a path of the robot tool to the determined component surface and / or automatically determines an orientation of the robot tool (30; 130) to the determined component surface, and / or automatically determines a starting movement to the workpiece (300). The collaborative robot (10; 110) of any one of the preceding claims, wherein the collaborative robot (10; 110) is configured to automatically complete a motion sequence of the robotic tool (30; 130) based on locations of learning points learned in the learning mode; and or with a force-moment sensor for determining at least one process force in the learning mode. A collaborative robot (10; 110) according to any one of the preceding claims, wherein the collaborative robot (10; 110) is disposed on a trolley (60). Collaborative robot (10; 110) after one of the Claims 1 to 10 wherein the collaborative robot (10; 110) is disposed on a magnetic lock. A collaborative robot (10; 110) according to any one of the preceding claims, comprising at least one optical sensor (131) for detecting a surface roughness of the workpiece in the working mode. A method of operating a collaborative robot (10; 110) to process a workpiece (300), comprising the steps of: - providing a robot tool (30; 130) to the collaborative robot (10; 110) adapted to machine the workpiece (300); - selecting a learning mode of the collaborative robot (10; 110) in which the collaborative robot (10; 110) is taught a working movement of the robot tool (30; 130) for machining the workpiece (300); - selecting a working mode of the collaborative robot (10; 110) in which the collaborative robot (10; 110) performs the work movement learned in the learning mode; wherein the collaborative robot (10; 110) in the learning mode teaches the working movement of the robot tool (30; 130) on the workpiece (300) to be machined. Method according to Claim 14 in which an operator (200) moves the collaborative robot (10; 110) in the learning mode by means of manual guidance, in particular with two hands, along the working movement; and / or wherein the learning mode is performed on the collaborative robot (10; 110) at operating temperature. Method according to Claim 14 or 15 wherein the learning mode is performed at most a predetermined time period before the machining of the workpiece (300). Method according to one of Claims 14 to 16 , wherein in the learning mode at least one pressing force of the robot tool (30; 130) is switched to the workpiece (300) in a determinable direction. Method according to one of Claims 14 to 17 wherein the collaborative robot (10; 110) automatically detects a component surface of the workpiece (300), in particular automatically determines an orientation of the robot tool (30; 130) to the determined component surface, and / or automatically determines a starting movement to the workpiece (300). Method according to one of Claims 14 to 18 wherein the collaborative robot (10; 110) automatically completes a motion sequence of the robotic tool (30; 130) based on positions of vertices learned in the learning mode. Use of a collaborative robot (10; 110) for processing a workpiece (300), wherein the collaborative robot (10; 110) comprises a robotic tool (30; 130) for processing the workpiece (300) and a mode selection device (51) for transferring the collaborative Robot (10; 110) in a learning mode and in a working mode; in which - the collaborative robot (10; 110) in the learning mode, a working movement of the robot tool (30; 130) for machining the workpiece (300) is learned; - the collaborative robot (10; 110) in the working mode performs the work movement learned in the learning mode; and - the collaborative robot (10; 110) learns in the learning mode the working movement of the robot tool (30; 130) on the workpiece (300) to be machined.

Images