DE102015223258A1 - Method for editing the surface of a three-dimensional object - Google Patents

Abstract

A method according to the invention for processing the surface of a three-dimensional object, e.g. a body part, wherein at least one tool for processing, e.g. an ink jet printhead or dryer, or the object from a manipulator, e.g. An articulated arm robot is moved along at least one machining path in such a way that a relative movement is generated between the tool and the object, and wherein the manipulator has a plurality of axes with variable axis positions, characterized in that an a priori axis course for machining paths is calculated (Step 7).

Description

The invention relates to a method having the features of the preamble of claim 1.

State of the art

The invention is in the technical field of printing on surfaces of three-dimensional objects by means of inkjet technology (inkjet). Either the inkjet print head or the object or both is moved: in the case of large objects, such as body parts, the print head is preferred; For small objects such as balls, the object prefers. The movement is preferably effected by a robot such as an articulated arm robot. For the movement when printing any, arbitrarily shaped and / or changing objects, the planning of a (moving) orbit may be required. In the German patent applications of the same patent applicant DE 10 2014 011 301 . DE 10 2014 221 103 and DE 10 2014 008 470 generic methods are described.

From the state of the art, the problem of web planning for alternating printing and drying webs or even the common arrangement of print head and dryer at different angles on a carrier on the robot flange has hitherto not been known. In the case of surface printing or surface coating, either sawtooth-shaped or meander-shaped paths are provided, depending on whether laterally adjoining pressure paths extend approximately in the same or the opposite direction.

However, there are patents and literature on the general problem of path planning:
WO 2005/049284 A1 : "Method for optimizing the performance of a robot". It is about a method, a system, a computer program and a computer readable medium or means for performance optimization of a robot, more precisely optimization of the position and orientation of a task relative to a robot. Different experiments with different arrangement of the task relative to the robot including the robot kinematics are calculated and the best arrangement selected. The aim is to optimally arrange a task to be performed with a robot in the working space of the robot. The cycle time for a specific task is thereby to be reduced. The method distinguishes between the necessary criterion for accessibility ("possible task placement criterion"), whether all points are achievable and sufficient criterion of feasibility ("admissible task placement criterion"), that also the level crossings between them can be realized. The examination of level crossings essentially means that no singularities occur. A polynomial function or other suitable function is used to approximate the functional relationship between the location of the task and the cycle time in order to determine the minimum (optimum) for the position on this parametric function. Provided are quadratic polynomials which are fitted at three points. A fully factorial design experiment on 3 variables requires 27 tests. Dependencies between the spatial directions in the approximation function are also considered. In the case of Box Behnken design, 3 variables will pass 12 tests. In addition to the necessary criterion, additional criteria can be used, eg parts of the robot cell are occupied or collision check with other robots or devices. The tests are provided with a simulated robot control which accurately determines the cycle time. It is also determined whether the movement is executable at all. If not, the associated test is removed from the experiment. For fitting the method of least squares is provided. Instead of polynomials, spline functions or Fourier series can also be used. Depending on the complexity of additional restrictions to be observed, for example, SQP methods and gradient methods are proposed as optimization methods. The optimum is only an aptitude candidate and it still has to be checked whether the sufficient criterion of path realizability is given. If not, by sequential backward shifts the closest possible position in the direction of the optimum vector can be found to be optimum. After all the optimum candidates have been tested, the minimum of the actual tests is selected as the real optimum. The document does not deal with the possibility of trajectory variation! Only PTP or straight Cartesian tracks are named as a possibility. In fact, many more paths are possible between given points.

The method does not describe the concrete execution of the path planning between two points, but only mentions the two possibilities of the PTP movement (axis-synchronous movement) and the continuous path movement (movement planned in the Cartesian space). The method also provides for the precise simulation of planned paths for different arrangements of the task relative to the robot by means of a robot control, and consequently repeatedly indicates that this takes a long time, i. the process is slow.

Ramer, C., S. Reitelshöfer and J. Franke: Automated path generation and collision monitoring for six-axis industrial robots through 3D camera-based environmental detection. VDI REPORTS No. 2209, pp. 143-146, Dusseldorf: VDI Verlag, 2013 , The data from a time-of-flight camera is used to create a robot's environment model and "for automated generation a collision-free robot path used. During the robot movement, collision monitoring based on the current working space model and the known robot position performs hierarchical distance checks on dynamic objects and thus ensures compliance with a minimum distance. "For the collision check, the objects are approximated by simple enveloping bodies in order to reduce the computational effort. The path planning is based on the approximate cell decomposition, in which the space is successively decomposed into octrees. It is first checked whether a direct collision-free path between start and target position can be realized. The target position can be entered manually or via an intuitive input device. You can select whether PTP or continuous path movements are to be executed. If the direct route is possible, the determined axis values are transmitted to the robot controller. Otherwise, the workspace is decomposed into cube-shaped cells to identify multiple collision-free path vertices. Reference is made to the original idea of the applied principle of approximate cell decomposition in [ Lozano-Perez T (1981) Automatic Planning of Manipulator Transfer Movements. IEEE Transactions on Systems, Man and Cybernetics 11 (10), pp 681-698 ] and as other possible path planning approaches the potential field method and the frequently used sample-based planners ([ Latombe JC (1991) Robot motion planning, Kluwer, Boston ] and [ LaValle SM (2006) Planning Algorithms, Cambridge University Press, Cambridge ]).

The process of octrees generally serves to plan a collision-free robot path from a start to a target position. It requires a considerable amount of computation for the single web. In addition, it does not solve the problem of different possible Achsstellungsmöglichkeiten for the same start and finish position. In addition, this method does not ensure that a solution for the planning task is found, even if one exists.

WO2014 / 022786A2 : "Systems and methods for robotic surgery". The very extensive document relates to surgical robots for handling a surgical instrument and an energy applicator emanating from the surgical instrument, methods for their control and a surgical system. The consideration of restrictions eg for collision avoidance according to the potential field method is planned here.

The potential field method takes into account restrictions in the form of repulsive forces. This also shows no solution to the special problem that to Cartesian planned processing tracks multiple Achsstellungsverläufe are possible.

DE 10 2008 046 348 A1 : "Method, device and a corresponding computer program product for the computer-aided path planning of a movable device, in particular a medical device". The invention relates to a method for computer-aided path planning of a movable device, in particular a medical device, in a spatial area. In the method according to the invention, webs of the movable device are suitably characterized by means of risk volumes which describe the risk of collisions with other objects as the device moves along a given path. An overall risk of a collision of the movable device with objects in the spatial area is determined from the risk volumes and, based on an optimization method which contains a minimum collision risk as an optimization target, optimal paths are determined computer-assisted. In a preferred variant of the invention, the arrangement of sensors for the detection of objects is also taken into account in the path planning, so that the method furthermore also outputs an optimal sensor arrangement. Based on the optimal paths or the optimal sensor arrangement, the device can be moved appropriately in operation, thereby keeping the risk of collisions as low as possible. A preferred field of application of the invention is the movement of medical devices, for example an X-ray device in the form of a C-arm.

The method is specifically designed to schedule paths under uncertainty, as e.g. can occur in medical technology, if areas of people are taken only with a certain probability. For the present task, in which there are no people in the workspace of the robot, it reveals no advantageous solution.

invention

It is an object of the invention to provide a comparison with the prior art improved method, which makes it possible to perform automated, fast and reliable path planning for the manipulator.

This object is achieved by the method having the features of claim 1. A method according to the invention for machining the surface of a three-dimensional object, wherein at least one tool for machining or the object is moved by a manipulator along at least one machining path such that a relative movement between the tool and the object is generated, and wherein the manipulator has a plurality of axes with variable axis positions characterized by the fact that an a priori Axis position curve for machining paths is calculated (step 7 ).

In particular, ink jet printing heads, dryers, in particular UV radiation dryers and / or plasma nozzles can be used as tools.

The a priori Achsstellungsverlauf for Bearbeitungsbahnen can also be a hypothetical A-priori Achsstellungverlauf (definitions: see the following four paragraphs).

The term "axis course" in this application describes the time course of the position of each axis - the n axes, n preferably 1, 2, 3, 4, 5 or 6 - of the manipulator between an initial and final axial position, e.g. at constant intervals. An axle position can be calculated as a set of n angle values for a manipulator with exclusively axes of rotation, depending on the axle restriction, e.g. between -350 ° and 350 °. The axis constraints may be due to mechanical stops and / or software barriers and may be different for each axis. An axis position can be specified in a manipulator with only linear axes as a set of n length values. Even "mixed" axis positions (angle values and length values) are possible with appropriately equipped manipulator.

An example of an axis position from an axis position curve of a manipulator with six axes of rotation: [-106.27, -80.88, 140.40, -222.13, 78.08, -246.21]. Each value in the parentheses describes the angular value of the position of one of the axes of the manipulator.

The term "a priori" in connection with the term "Achsstellungsverlauf" describes in this application the fact that an Achsstellungsverlauf is calculated, which can serve as the basis or output Achsstellungsverlauf for the further steps of the method according to the invention.

The term "hypothetical" in connection with the term "A-Priori Achsstellungsverlauf" describes in this application the fact that the A-Priori Achsstellungsverlauf in the course of performing further process steps as - in practice; because of existing restrictions - is not feasible and therefore discarded or modified.

An advantage of the invention is that it decomposes the complex task of complete fully automatic path planning for robots or other machines for imaging 3D objects into subtasks so that the path planning can be carried out quickly and reliably without adversely affecting the processing paths. As a result, it quickly finds a very advantageous or even the best path planning solution. Instead of sequentially planning the tracks according to the execution order, according to the invention, only the path planning of the processing paths and then the transition paths takes place.

Another advantage is that in contrast to the prior art, the invention provides that upon collision of a tool with the surface on a transition path by superimposing the original web on a parabolic or circular displacement away from the surface, a collision-free transition path is planned so that at the point of closest approach away from the start and end point of the web just the predetermined minimum distance is maintained.

Further advantages of the method according to the invention are:
The totality of the planned paths between an initial position and an end position of the robot / manipulator consists of machining paths and transition paths between the starting position, the machining paths and the end position. In the case of the processing paths, eg printing webs or drying trajectories, the Cartesian trajectory (position and orientation course) of the tool is directly relevant to the process (and influences, for example, the printing result), at the transition paths at most indirectly (eg faster executable transition paths between the printing and traversing paths Drying Trails Larger web lengths with the same drying time of the ink and a smooth course of the transition paths reduces the risk of incomplete printing due to nozzle failures).

The kinematics of the robot as well as boundary conditions due to axle restrictions, collision restrictions and possibly line restrictions are present, restricting the Cartesian possible movements and are taken into account by the method.

The webs are not sequentially planned according to their execution order, but only the Cartesian course of the processing paths is planned, then an advantageous (possibly the optimal) combination of Achsstellungsverläufen the processing paths planned or selected and then the transition paths are planned. In this case, steps can also be executed several times if, after their execution, it is determined that no feasible paths can be planned with their result in the later steps.

For the determination of the optimal or an advantageous combination of Achsstellungsverläufen the processing tracks is not the takes into account the actual course of the transition lanes (which is not yet known), but a criterion which includes passing through singularity positions (change of axle configuration) and the maximum axle angle change on the respective transition lane.

It can be dispensed with time savings on a complex simulation run and a complete database even on a test run of the generated lanes.

A development of the invention can provide that possible Achsstellungsverläufe be determined for the machining paths (step 8th ).

A development of the invention can provide that it is checked whether all processing paths can be realized (step 9 ).

A further development of the invention can provide that preferred axis position profiles are selected for each machining path (step 10 ).

A development of the invention can provide that transition paths between a starting position and a machining path, between at least two machining paths or between a machining path and an end position of a tool or several tools are planned (step 11 ).

A development of the invention can provide that it is checked whether all transition paths are found (step 12 ).

A development of the invention can provide that a control program for the manipulator is generated from the machining paths (step 13 ).

A development of the invention can provide that the control program is executed (step 14 ).

A development of the invention can provide that 3D surface data and image data are previously provided (step 1 ).

A further development of the invention can provide that imaging parameters are previously defined (step 2 ).

A further development of the invention can provide that planning parameters are previously determined (step 3 ).

A development of the invention can provide that the Cartesian course of the machining paths is planned beforehand (step 4 ).

A development of the invention can provide that it is checked beforehand whether the path planning is permissible (step 5 ).

A development of the invention can provide that in the case of the test result "no" is checked whether the planning parameters are optimal (step 6 ).

A development of the invention may provide that the manipulator has at most six axes (degrees of freedom of movement).

embodiments

The invention and its advantageous developments are described below using the example of specific embodiments and with reference to the figures. The figures show:

1 a flowchart of a preferred embodiment of the method according to the invention; and

2 to 9 perspective views of a motor vehicle door and a robot (shown schematically and in various positions) for moving a print head and a dryer.

In the preferred embodiment, the invention is used for automatic path planning and robot program generation for a 3D surface ink jet printer. The ink jet printer consists at least of a robot or manipulator and an ink jet print head (ink jet head) for spraying ink (ink, paint) on the surface to be imaged, but may also contain other components, such as additional printheads for multicolor printing, pinning Modules, UV dryers, IR dryers, primers, plasma or flame nozzles, ink supplies. The robot can move the printhead, printheads, or other tools relative to the 3D surface. It is possible that the print head and possibly other tools are moved by the robot over the 3D surface or the robot moves the object with the 3D surface to be imaged relative to the print head and possibly other tools. The robot is preferably a robot with 6 axes (degrees of freedom of movement; 23.1 to 23.6 of the 2 ), such as a 6-axis articulated robot or a linear robot with 3 linear axes and 3 additional rotary axes. Thus, the relative position and relative orientation of the tools to the surface can be selected.

According to the invention, based on a 3D model of the surface of the object to be imaged, the task definition for the placement of an image On the surface as well as the data describing the image, a robot program automatically generates, which describes the movements of the robot and contains control commands for the components of the inkjet printer, such as for synchronization of ink-jet system and dryer with the robot movement. This program can then be executed by the robot to image the surface of the object. The term image also includes writing, so that the invention also includes the labeling of 3D surfaces.

In the examples shown, separate printing webs were designed for printing with a printhead and drying webs with a radiation dryer wherein printhead and dryer are mounted in direct radiation between dryer and printhead preventing angles on a common support on the flange of a robot.

The path planning is based on 3D surface data of the surface to be imaged or treated. These may e.g. in the form of triangles in STL format, by NURBS or as CAD data. In order to be able to plan the robot paths for imaging the 3D surface, the digital image data must still be provided and the position of the image on the surface and, if necessary, further imaging parameters defined. In addition, the position of the 3D surface relative to the robot represents a significant planning parameter, which can be adjusted by moving and rotating the 3D surface relative to the robot.

In the preferred embodiment, the Cartesian planning of the processing tracks with the same in the German patent applications the same patent applicant DE 10 2014 011 301 . DE 10 2014 221 103 and DE 10 2014 008 470 described method.

In 1 an example flow of imaging 3D surfaces with restriction-based path planning is shown. The following paragraphs describe process steps that serve to prepare for the process steps relevant to the invention.

Provide 3D surface data and image data (step 1 ): Initially, the 3D data describing the surface, eg in the form of an STL file with triangles defining the surface, and the image data are provided. With exchangeable image content, the image data describe at least the size of the image or a placeholder for the image.

Set imaging parameters (step 2 ): The mapping parameters determine the task to be performed. They describe the desired position and orientation of the image on the surface and possibly further properties of the image, such as distortions. These determinations can be made interactively by the operator, for example, on the basis of the visualized 3D surface data, or automatically on the basis of general specifications, which are substantiated on the basis of properties of the 3D model. The result defined in this way can already be visually visualized with a suitable software before the path planning for control in virtual form.

Determine planning parameters (step 3 ): Since the path planning has numerous degrees of freedom, the planning parameters which determine the extent of these degrees of freedom are first of all determined. Together with the path planning algorithm, the 3D surface data and the image data, unique paths result from the planning parameters. The planning parameters influence the quality of the processing paths, in particular the print quality in the printing webs. The planning parameters can be set manually by the operator in the simplest case. It is also possible to automatically determine at least part of the planning parameters, eg by the in DE 10 2014 008 470 (Planning optimal paths in 3D inkjet) algorithm. Examples of planning parameters are the distances of the tools to the surface to be maintained and the nature of the inclination adjustment to the surface on the machining paths, the collision safety distance, the maximum values for speed, acceleration and jerk of the tools on the processing and transition paths, the ink's running time between pressure and Drying, the pressure strip widths, the number of pressure paths, the main movement direction on the processing paths, the position of the 3D surface to the manipulator. In addition to the slightly variable planning parameters, there are numerous other data that influence the path planning, such as the kinematics of the manipulator, 3D models of the surface, the manipulator and possibly other for the collision avoidance to be considered objects. As a special feature, changing the position of the 3D surface to the manipulator may include manual steps to move objects.

Cartesian course of the processing paths (measuring tracks, pre-treatment tracks, printing lines, drying tracks, etc.) plan (step 4 ): With the 3D surface data, the image data and the planning parameters, the processing paths are planned, ie calculated according to a defined algorithm. Initially, only the Cartesian position and orientation course of the tools in a basic coordinate system along the machining paths relative to the 3D surface is planned. One of these six values (three Position coordinates and three orientation values) is often called a frame in robotics. At this point, only the frames history is calculated, whereas the associated axis value curve does not need to be calculated yet. Because the calculation of the Achswerteverlaufs from the frames history, ie the calculation of the inverse kinematics, often very expensive, thus considerable computational effort is avoided.

If the planning parameters, eg as part of an optimization according to the above DE 10 2014 008 470 , at least partially determined automatically and the Cartesian course of the processing paths has already been calculated within the framework of the planning parameter optimization, this step is reduced to the selection of the path planning resulting with the optimal planning parameters.

Check whether the path planning is permissible (step 5 ): If with the planning parameters already on the basis of the frames no permissible path planning results, eg because in order to avoid collisions the maximally permitted nozzle distance would be exceeded, then the planning parameters must be changed. If the planning parameters have been determined manually, the operator can adjust the planning parameters interactively until a permissible path planning results. Very advantageous is the short possible cycle time of these iterations, since up to this point no inverse kinematics needs to be calculated.

Check whether the planning parameters are optimal (step 6 ): Were the planning parameters automatically determined as part of an optimization, eg according to the above DE 10 2014 008 470 , then the railway planning will be inadmissible only in exceptional cases. This can occur, for example, if due to the dimensions of the print head in connection with a problematic surface topology, the areas to be imaged are not reachable at all. In this case, even with the optimal planning parameters, no permissible path planning for the given task is possible. By choosing other imaging parameters, eg the imaging of problematic surface areas can be avoided. Even with manual definition of the planning parameters, this may be necessary if even with the best found planning parameters no permissible path planning results.

The following paragraphs describe process steps which represent the process steps essential to the invention:
(Hypothetical) Calculate a priori axis position curve for the machining paths (step 7 ): The print quality is mainly determined by the Cartesian course of the processing paths calculated up to this point. It was not necessary to consider the axis positions, whereby a complex calculation of the inverse kinematics was avoided and the print quality could be optimized very efficiently. However, complete automatic path planning also requires consideration of axis limitations, avoidance of collisions of the robotic arm with other objects, and possibly restrictions due to lines to moving tools, eg, for ink supply and control of a printhead. This requires knowledge of the respective axis positions in the frames. With the help of the inverse kinematics axis positions can be calculated to the frames in the working space, whereby these are usually not clear. Thus arise, for example, in the often encountered symmetrical arrangement of the axes 4-6 (see reference numerals 23.4 to 23.6 of the 2 ) with 6-axis articulated robots usually several possibilities for positions of the axes 4-6 in the same frame, which can be quickly converted into each other. Thus, for example, the eight axle positions correspond to [-106.27, -80.88, 140.40, 222.13, 78.08, -246.21], [-106.27, -80.88, 140.40, -222.13, 78.08, 113.79], [-106.27, -80.88, 140.40, - 42.13, -78.08, -66.21], [-106.27, -80.88, 140.40, -42.13, -78.08, 293.79], [-106.27, -80.88, 140.40, 137.87, 78.08, -246.21], [-106.27, -80.88 , 140.40, 137.87, 78.08, 113.79], [-106.27, -80.88, 140.40, 317.87, -78.08, -66.21], [-106.27, -80.88, 140.40, 317.87, -78.08, 293.79] all the same Cartesian position in the Space (ie the same frame).

Since there are usually several axis position profiles which are suitable for each machining path and, in principle, there are infinite possibilities for permissible path paths for the intermediate transition paths, complete automatic path planning represents a complex planning problem whose solution is the prerequisite for an economical lot size. Production is. Not only due to the required knowledge of the axis position curves and the often great effort for calculating the inverse kinematics, the calculation and the comparison of all possible path combinations are usually not possible or associated with high planning times.

For solving the complex planning problem according to the invention, the path planning is subdivided into subproblems which make it possible to largely reduce the number of calculations required for inverse kinematics and thus the computing time.

For the Cartesian course of each machining path, an associated (hypothetical) a priori Achsstellungverlauf is determined using the inverse kinematics first. It is the compliance of Restrictions irrelevant, since it is usually not the final planned Achsstellungsverlauf.

Determine possible axis position curves for the machining paths (step 8th ): From the (hypothetical) Achsstellungsverlauf can be derived without calculation of the inverse kinematics, the other belonging to the same Cartesian course Achsstellungsverläufe. When traversing a singularity position on a machining path, the path accuracy would be impaired and the web speed would fluctuate greatly. The embodiment therefore provides that machining paths with such Achsstellungsverläufen be classified as unrealizable. As long as no singularity positions are passed through, due to the continuity of the axis position progression in a 6-axis kinematics, such as an articulated arm robot, an axis position is clearly defined by the associated frame and the immediately preceding axis position. From the initial axial position of the (hypothetical) Achsstellungsverlaufs and the axis and line restrictions, all restriction-containing initial axial positions per processing path can be calculated. The associated further axis position curves can also be calculated with the (hypothetical) axis position curve. Possible Achsstellungsverläufe a machining path are then all those who comply with the restrictions on the entire train and also go through no Singularitätsstellung.

Check whether all lanes can be realized (step 9 ): A machining path can be realized if there is at least one possible axis course. If not all processing paths can be realized, then the planning parameters or possibly the imaging parameters must be adapted. If a processing path leaves the working space of the manipulator, a change in the position of the 3D surface to the manipulator (or the tools to the manipulator with a moving surface) may result in a more favorable placement of the processing paths within the working space of the manipulator, which implies an alteration of the Planning parameter means. By the method according to the invention, the feasibility of the processing paths is already known at an early stage, so that the planning time remains low even with iteratively required parameter changes.

Select preferred axle position curve for each machining path (step 10 ): If different Achsstellungsverläufe for the realization of the Cartesian course of a machining path are possible, then the preferred course is selected in the embodiment. This step makes it possible to significantly reduce the complexity of further path planning by selecting a combination option and thereby speeding up the path planning.

Criteria for selecting the preferred axle position curves are the avoidance of axle configuration changes (avoiding the passage of singularities also on the transition tracks) and a total track trajectory that is as short as possible. Since the length of the processing paths is already fixed, a consideration of the transition paths is sufficient. Their initial and final axial position is given by the final and initial axial position of the machining paths and the start and Endachsstellungen the overall planning, from which the manipulator starts at the beginning and ends at the end again.

In the first step, an attempt is made to reduce the amount of the axis position curves to those with a uniform axis configuration, if possible with the axis configuration of the start position. The axis configuration of the axis 5 (see reference numeral 23.5 of the 2 ) of two axis positions of a conventional 6-axis articulated robot agrees, for example, if the angles of the axis 5 have the same sign. With the same configuration, no singularity position needs to be traversed during the drive between the axle positions.

In the second step of the preferred embodiment, the best rated combination of axis position curves of the machining paths is selected. Since the concrete course of the transition lanes will be planned later, the evaluation is done on the basis of already known properties, such as the frequency required to pass through singularity positions, the maximum required Achsstellungsänderung on the transition lanes and possibly the maximum line load. The line load is relevant when lines, eg for ink supply, lead to moving tools through the manipulator. Even if the line restrictions are adhered to, lines tend to be more heavily loaded if they are twisted or bent more. Line loading and compliance with line restrictions can be coarsely determined with suitable routing, for example via the relative position of the axes 4 and 6, or more accurately calculated using a more comprehensive mechanical model of the manipulator including the routing. With the maximum required Achsstellungsänderung a transition path between the last axis position of the previous machining path and the first Achsstellung the subsequent machining path increases, assuming a PTP movement and the time period for the transition path. Since the exact trajectory of the transition paths is not yet established at this time, this assumption is an essential component, to reduce the complexity of the path planning by a quality-based preselection of the axis course of the processing tracks. If a separate drying path follows each printing web, for example to dry printed UV ink within a limited running time window with a UV lamp, the duration of the transition path between the printing web and the drying web is relevant for the process. For this purpose, in the preferred embodiment, the weighting of the maximum required change in position of these transitional paths in the quality function is higher than that of the other transition paths whose length only affects the total duration of all paths, but not the process quality.

In order to select the best axis position curves for all machining paths with regard to the quality function, all combinations of possible axis position curves must be taken into account. In the case of more extensive planning tasks, however, a great many combination possibilities can result, for 10 printing presses with subsequent drying paths, each with four possible axis position curves, e.g. already over a trillion (exactly 4 ^ 20 = 1,099,511,627,776). According to the invention, the extensive planning problem is therefore decomposed at this point again into independently solvable sub-problems. Depending on the kinematics of the manipulator, the arrangement of the tools on the manipulator and the position and the orientation of the surface to the manipulator (or moving surface of the arrangement of the surface on the manipulator and the position and the orientation of the tools to the manipulator), arises For some machining paths often only one possible Achsstellungsverlauf, while for others many possibilities exist. The invention therefore provides to form blocks of consecutive machining paths with multiple possible Achsstellungsverläufen between machining paths with only one possible Achsstellungsverlauf and to determine separately in these blocks the best combination of Achsstellungsverläufen. In the above example, e.g. the number of possible combinations to be examined by a factor of 838,861 to 4 ^ 9 + 4 ^ 10 = 1,310,720, if there is only one possible Achsstellungsverlauf for the fifth dryer web. Often even blocks of length 1 result if, for example, by the arrangement of the different tools on the manipulator for all drying paths only one Achsstellungsverlauf is possible. Nevertheless, there can also be very many possible combinations within the blocks. The invention therefore provides that only up to a limit value (for example 1024) are all possible combinations (brute force) within the blocks taken into account. If the number of possible combination possibilities exceeds the limit value, then heuristically based only those are selected, to which experience has shown that the best combination possibility belongs. For several approximately in the same direction extending processing paths, as they result in the construction of a print image of laterally adjacent tracks, only those are examined in the preferred embodiment, in which the axis positions of adjacent tracks for the same tool are most similar. In the case of alternating pressure and drying paths, it is then sufficient to examine all combinations of possible axis position curves of the first pressure path and of the first drying path of a block and to combine for the further processing paths of the block only that axis position curve in which the magnitude of the maximum axle position difference with respect to the preceding pressure curve. or drying path is smallest at the initial axial position. In the above example, the number of possible combinations to be examined would then be reduced by the factor 68,719,476,736 to 16, so that a short planning time is made possible, whereby a nearly optimal and with a high probability even the optimal combination of the axis position courses is found.

Plan transition paths between start position, machining paths and end position (Cartesian and axis position progression) (step 11 ): Since now besides the Startachsstellung and the Endachsstellung, which occupies the manipulator at the beginning and end of the program, the Achsstellungsverlauf all machining paths, so that the start and Endachsstellungen all transition paths are set so that they now independent of each other and on multiprocessor systems can be planned in parallel. Independence, however, simplifies the complexity of the railway planning considerably, since no combination of possible transition railways with their numerous degrees of freedom has to be considered. As a result, the planning of the transition railways can be done faster and also better, because there is scope for the use of advanced and expensive procedures that require more computational effort. In principle, a transitional track, while maintaining the axis, line and collision restrictions arbitrarily between their initial and the Endachsstellung run so that it has infinitely many degrees of freedom. In order to achieve high productivity, however, the shortest possible (or fast) transitional path is advantageous. Without consideration of the tool and the dynamic restrictions to be complied with, a PTP (Point To Point) movement normally allows the fastest movement with manipulators. The axes between an acceleration and a braking phase simultaneously move at a uniform speed between their start and end positions, whereby the dynamic limits of speed, Acceleration and jerk are given. In the preferred embodiment, as a starting point of a possible transition path, the course of such a PTP movement is assumed, which is defined by a sufficient number of nodes on which the axes have covered the same percentage between the initial and final axis position. With the aid of direct kinematics, the corresponding frame can be calculated for each interpolation point, ie the Cartesian position and orientation at the TCP (Tool Center Point), which results at the respective axis position. If collisions of the tool with the surface would occur on this transitional path, the Cartesian trajectory becomes superimposed on the Cartesian trajectory with a circular, parabolic, polynomial or spline-based or other continuous and possibly twice-continuously differentiable displacement function between the start and end point moved so far away from the surface that a collision is avoided while maintaining a safe distance. In the beginning and end of the path, the shift by the shift function is 0. In most cases, this results in realizable transition paths, which are very similar to a TCP movement and thus are short and can be executed quickly. Also potential collisions of the tool with other detected objects as the surface can be avoided by such displacement tracks, the displacement direction is then adjusted if necessary. In order to ensure realizability, the resulting paths are checked for compliance with all axis, line and collision restrictions, including those of the manipulator with the surface and other detected objects. It is only at this late stage of the restriction test that it is necessary to know the axis positions of the transition tracks, so that costly calculations of the inverse kinematics are reduced to a minimum and the path planning can be carried out quickly. If no displacement is required, a calculation of the inverse kinematics at this point even completely omitted, because the axis positions are indeed known as the starting point of the direct kinematics. If in rare cases restrictions are violated, the lanes can be adjusted with shift functions or alternative lanes can be selected.

If the tool is subject to strong dynamic constraints, e.g. due to the risk of printing problems after large accelerations of the printhead, the duration of the transition paths is determined primarily by Cartesian dynamics limits, rather than the dynamic limits of the axes. For the tool Cartesian briefly planned lanes are then faster executable than PTP movements. In the preferred embodiment, therefore initially short transition paths between the initial position and the end position are calculated, tested for feasibility and compliance with the required Endachsstellung and selected in the positive case. This results in some degrees of freedom, which lead to alternative trajectories, which can also be tested until a suitable track was found. When several tools for different machining paths are moved simultaneously by the manipulator, e.g. a printhead and a dryer, cartesian planning can in principle be done for any of these tools, but best for a tool point which minimizes stress on the most sensitive component, e.g. of the printhead. A short transition results from a straight connection of the initial position with the end position, along which support points can be calculated at a suitable distance. The rotation of the tool from the Cartesian start to the end orientation is carried out accordingly, so that a constant movement and rotation speed results between an acceleration and a braking phase. For the Cartesian rotation arise different directions of rotation and also rotations by more than 360 ° are possible. Permitted webs transfer the initial axial position into the correct final axis position and maintain collision, axis and possibly line restrictions. If collision restrictions are violated, then the tracks can be adapted by means of shift functions as described above for the axis-based transition paths.

The violation of axis position or line restrictions may possibly also be eliminated by means of shift functions, e.g. if the axle restrictions are adhered to on a circular transition track, but not on a straight transition track. Since there are very many possibilities of adaptation and a suitable adaptation depends not only on the robot kinematics, but also on the position of the surface (or the position of the tools in a moving surface) to the manipulator, the implementation of the adaptation is complicated. The preferred embodiment therefore provides that initially unmatched Cartesian straight straight transition paths are planned and if none of them complies with all restrictions, then the above-described planned on the basis of the axis positions and possibly adapted with shift functions transition paths are planned.

Check if all tracks are found (step 12 ): So far, suitable transition paths have always been found in the practical use of this method. If, for at least one transitional path, no solution would nevertheless be found, in principle by changing the Planning parameters or mapping parameters of the course of the webs in the working area of the robot can be changed so that transition paths are found for this changed parameterization. A strong influence allows in particular a change in position of the 3D surface to the manipulator (or the tools to the manipulator with a moving surface), if there is a cheaper placement of the webs in the working area of the manipulator. In practical terms, this means, for example, a displacement of a car to be imaged, which is reflected in the planning parameters as translation of the 3D object in the robot coordinate system.

The following paragraphs describe method steps that serve to implement the method steps essential to the invention or the results obtained therefrom on a robot.

Create control program from planned paths (step 13 In the preferred embodiment, when realizable traces have been found for all transition paths, a control program is generated in the preferred embodiment, which contains all paths to be executed for the movement of the manipulator as support points of movement commands, preferably spline commands. The support points are preferably given as frames, which summarize the positions and orientations of a tool or alternatively as axis positions. If trajectories pass through or come very close to singularity positions, these trajectories in the preferred embodiment are specified by axis positions, at least in the area of the singularity positions, and are defined by means of axis-based motion commands, such as PTP commands in robots. In addition, the control program contains additional commands for controlling the tools and possibly other components, such as for synchronizing the printheads and the dryer with the robot movement. The language of the control program depends on the controller, which controls the manipulator. In particular, this may be a G code program or a robot program if the manipulator is part of a robot.

Run control program and handle 3D surface (pretreatment, printing, drying, etc.) (step 14 ): The finished generated control program will then be executed once or several times to depict the surface (s), depending on the lot size to be realized. If the program for imaging a plurality of identical objects repeatedly executed, they must be placed for example by a suitable holder in the same position relative to the manipulator (or their moving surfaces must be in the same position to the flange of the manipulator).

In the 3 to 9 By way of example, webs of an imaging task are illustrated, from which essential aspects of the described method can be recognized. The planning was done for a manipulator 23 a 6-axis articulated robot with purely serial kinematics. In the figures, the 3D model used for planning is the surface 20 (in the example shown: a car door), the planned printing line 21 as well as representations of the tools 26 . 27 and the manipulator 23 shown. The manipulator 23 is represented by a slender kinematic model, which gives information about the axle positions. The printhead 26 is by a smaller and the dryer 27 represented by a larger rectangle, each one of the on the machining path 21 the surface 20 facing the lower boundary surface of the tool. The slender presentation of the temporal course can be better expressed by temporally successive snapshots in the same figure. So an understanding of sample webs without animation is possible.

Due to the selected, angled arrangement of printhead 26 and dryers 27 on the robot flange 28 via a mechanical connection 24 . 25 (eg one support each) prevents direct UV radiation from the dryer from drying the ink on the printhead and thereby clogging its nozzles. This arrangement can serve as an example for the joint movement of several tools by a robot, wherein a tool change is easily realized by another robot position. Due to the greatly different orientation of the tool carrier on the robot flange on the printing lines compared to the drying paths, however, the finding of feasible paths is much more difficult due to restrictions.

2 shows the beginning and identical end position of the manipulator 23 in his coordinate system, which apply to all examples, as well as a vehicle door 20 with a sloping pressure path 21 between the two lines shown on the door 22 , printhead 26 and dryers 27 are at a sufficient distance on the flange 28 of the manipulator 23 mounted and are always moved together.

In 3 is the transition path from this initial position to the first processing path represented by successive snapshots. The planning of the illustrated Cartesian planned transitional railway was possible here without restriction violation.

4 illustrates for better clarity the same movement but without representation of the dryer 27 , In comparison, the first part of the railway was in 5 as a linear transition of the axis positions analogous to a PTP movement planned so that printhead 26 and dryers 27 be moved on the resulting curves. This movement could be the manipulator 23 without taking the dynamic limits into account faster, due to the dynamic limitations of the printhead 26 but is that in 4 nevertheless in a shorter time and gentler for the printhead 26 executable. In addition, a transitional path planned as a linear axis transition results in trajectories of the tools 26 . 27 with increased risk of collision. In order to avoid thereby required Bahnanpassungen the transition railway can also as in 5 shown composed of two partial webs. The second part of the transitional track is approaching the print head 26 in 5 orthogonal to the surface of its initial position for the first printing web. This optionally provided short additional movement can significantly reduce the likelihood of required path adjustments for collision avoidance, especially in demanding planning tasks.

6 shows a corresponding representation of the course of the printing web and 8th the drying path, here the dryer 27 instead of the printhead 26 is visualized.

The transition path between these two tracks shows 7 , This is the printhead 26 from the surface 20 Panned away and the dryer 27 swung to her. This transitional track became a Cartesian straight line for the printhead 26 planned.

9 Finally, the transition path between the drying path and the end position shows the movement of the printhead 26 and dryers 27 , Again, the path of the printhead was 26 Cartesian planned along a straight line.

The advantageousness of the method according to the invention is particularly clear below with reference to the axis region data of a very simple example with a pressure and a drying path:
For the in 2 to 9 The presented planning task results in the step 8th "Determine possible axis position curves for the machining paths" 1 for the printing web eight and for the drying path six possible variants for the same reasonable initial axial position of the axes 1 to 3 (reference numeral 23.1 . 23.2 . 23.3 ). These are each described by the beginning and end axis positions:

print web

• Possibility Number 1: [-106.27, -80.88, 140.40, -222.13, 78.08, -246.21] ... [-76.46, -81.12, 123.63, -194.97, 41.89, -243.23]
• Option 2: [-106.27, -80.88, 140.40, -222.13, 78.08, 113.79] ... [-76.46, -81.12, 123.63, -19.4.97, 41.89, 116.77]
• Option 3: [-106.27, -80.88, 140.40, -42.13, -78.08, -66.21] ... [-76.46, -81.12, 123.63, -14.97, -41.89, -63.23]
• Option 4: [-106.27, -80.88, 140.40, -42.13, -78.08, 293.79] ... [-76.46, -81.12, 123.63, -14.97, -41.89, 296.77]
• Option 5: [-106.27, -80.88, 140.40, 137.87, 78.08, -246.21] ... [-76.46, -81.12, 123.63, 165.03, 41.89, -243.23]
• Option 6: [-106.27, -80.88, 140.40, 137.87, 78.08, 113.79] ... [-76.46, -81.12, 123.63, 165.03, 41.89, 116.77]
• Option 7: [-106.27, -80.88, 140.40, 317.87, -78.08, -66.21] ... [-76.46, -81.12, 123.63, 345.03, -41.89, -63.23]
• Option 8: [-106.27, -80.88, 140.40, 317.87, -78.08, 293.79] ... [-76.46, -81.12, 123.63, 345.03, -41.89, 296.77]

drying track

• Possibility Number 1: [-109.52, -83.91, 144.79, -224.91, 80.17, -5.96] ... [-75.80, -83.92, 126.29, -192.05, 41.13, -5.27]
• Option 2: [-109.52, -83.91, 144.79, -44.91, -80.17, -185.96] ... [-75.80, -83.92, 126.29, -12.05, -41.13, -185.27]
• Option 3: [-109.52, -83.91, 144.79, -44.91, -80.17, 174.04] ... [-75.80, -83.92, 126.29, -12.05, -41.13, 174.73]
• Option 4: [-109.52, -83.91, 144.79, 135.09, 80.17, -5.96] ... [-75.80, -83.92, 126.29, 167.95, 41.13, -5.27]
• Option 5: [-109.52, -83.91, 144.79, 315.09, -80.17, -185.96] ... [-75.80, -83.92, 126.29, 347.95, -41.13, -185.27]
• Option 6: [-109.52, -83.91, 144.79, 315.09, -80.17, 174.04] ... [-75.80, -83.92, 126.29, 347.95, -41.13, 174.73]

In step 9 As a result, it is determined that the pressure path and the drying path are realizable.

In step 10 "Select preferred axis position curve for each machining path" is first determined that both tracks with the same sign of the axis 5 as in the start and end position are feasible (in this case, positive), so that the other options can be discarded.

This leaves 8 possible combinations of printing and drying path. A comparison of the required Achswinkeländerungen on the transition tracks shows that the combination of option 2 of the printing web and possibility 1 of the drying path best and the combination of possibility 5 of the printing web and possibility 1 of the drying path is second best.

Because the drying time of the ink between printing and drying is limited in the example process, the maximum axis angle change of this transition path is weighted higher than that of the other transition paths.

In sequentially according to the temporal execution order of successive path planning, the choice of the possibility 5 of the printing web would be optimal, in which yes the first transition path is better than option 2 of the printing web. However, this would result in disadvantages for the process because the transition path between the printing and drying web would be longer and more stressful for the print head, which would in particular also reduce the maximum printable web length. In addition, it would only be detected very late, if there is no feasible solution for a processing path at all, so that the average overall planning time would be considerably longer.

On the other hand, if all possible path combination possibilities were simulated and compared, as in the prior art, for example WO 2005 / 049284A1 mentioned by simulation on a virtual robot control, then much longer planning periods would be required, which would prohibit more complex planning tasks with a lot more combination options due to the resulting astronomical planning time entirely.

By contrast, by comparing the possible machining path combinations with the aid of the maximum axle angle changes on the transition tracks despite a short planning time, the method according to the invention finds a very advantageous or even the best possible track course.

LIST OF REFERENCE NUMBERS

1 to 14

steps

20

Object or its surface to be printed

21

Processing / printing / drying track

22

lines

23

manipulator

23.1 to 23.6

manipulator axes

24

Compound / carrier

25

Compound / carrier

26

printhead

27

dryer

28

robot flange

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 102014011301 [0002, 0048]
• DE 102014221103 [0002, 0048]
• DE 102014008470 [0002, 0048, 0052, 0054, 0056]
• WO 2005/049284 A1 [0004, 0089]
• WO 2014/022786 A2 [0008]
• DE 102008046348 A1 [0010]

Cited non-patent literature

• Ramer, C., S. Reitelshöfer and J. Franke: Automated path generation and collision monitoring for six-axis industrial robots through 3D camera-based environmental detection. VDI REPORTS No. 2209, pp. 143-146, Dusseldorf: VDI Verlag, 2013 [0006]
• Lozano-Perez T (1981) Automatic Planning of Manipulator Transfer Movements. IEEE Transactions on Systems, Man and Cybernetics 11 (10), pp 681-698 [0006]
• Latombe JC (1991) Robot motion planning, Kluwer, Boston [0006]
• LaValle SM (2006) Planning Algorithms, Cambridge University Press, Cambridge [0006]

Claims (15)

Method for processing the surface of a three-dimensional object, wherein at least one tool for machining or the object is moved by a manipulator along at least one machining path such that a relative movement between tool and object is generated, and wherein the manipulator has a plurality of axes with variable axis positions, characterized in that an a-priori Achsstellungsverlauf is calculated for processing paths (step 7 ). A method according to claim 1, characterized in that possible Achsstellungsverläufe for the machining paths are determined (step 8th ). A method according to claim 2, characterized in that it is checked whether all processing paths can be realized (step 9 ). A method according to claim 3, characterized in that preferred Achsstellungsverläufe are selected for each machining path (step 10 ). A method according to claim 4, characterized in that transition paths between a starting position and a machining path, between at least two machining paths or between a machining path and an end position of a tool or more tools are planned (step 11 ). A method according to claim 5, characterized in that it is checked whether all transition paths are found (step 12 ). A method according to claim 6, characterized in that a control program for the manipulator is generated from the processing paths and the transition paths (step 13 ). A method according to claim 7, characterized in characterized in that the control program is executed (step 14 ). A method according to claim 1, characterized in that previously provided 3D surface data and image data (step 1 ). A method according to claim 1, characterized in that previously mapping parameters are set (step 2 ). A method according to claim 1, characterized in that previously planning parameters are determined (step 3 ). A method according to claim 1, characterized in that previously the Cartesian course of the processing paths is planned (step 4 ). A method according to claim 1, characterized in that it is checked in advance whether the path planning is allowed (step 5 ). A method according to claim 3, 6 or 13, characterized in that in the case of the test result "no" is checked whether the planning parameters are optimal (step 6 ). Method according to one of the preceding claims, characterized in that the manipulator has at most six axes.

Images