3D safety device and method for securing and operating at least one machine

The invention relates to a 3D security apparatus and a method for securing and operating at least one machine in a cooperative work according to the preamble of claim 1 and 15th

In the production of industrial goods is in addition to the fully automated a wide range of semi-automated solutions for assembly. Here jobs are provided, for example, where some operations not mechanized, but are done by hand. Such applications are known as hybrid assembly and serve a division of functions between human and automatic machine functions to take advantage of their specific advantages: the machine with its uniformity and repeatability of functions and the man with his ability to adapt to changes and disruptions and its sensory and tactile skills ,

Especially with new products is often started with a simple hand assembly, which is then further automated and can grow with it to adjust the quantity of demand and the life cycle. But even in mature stages of production, there are tasks which are very difficult to automate, so that manual workstations remain in the manufacturing process. This combination can be found for example in interlinked assembly lines where on workpiece carriers optionally repeated rounds between automatic stations and manual workstations are needed.

The goal is a flexible mounting in combination with semi-automatic solutions respectively. Semi-automatic operations such as joining processes or the application of adhesive beads are given for reasons of quality or through the construction and may pose risks to the worker by crushing or shearing points when executed by automatics. In current practice, therefore hedging is in the form of a distancing of the operator with respect to the danger point. These optoelectronic sensors are often used as safety light curtains or safety laser scanner. The workplace is used either by the operator or by the machine, and a direct cooperation does not take place in practice.

Cooperative collaboration of humans and robots "Team @ work" of the Fraunhofer Gesellschaft has been demonstrated, for example, in a project. The safety of the operator is ensured by a distance monitoring, determined by multiple cameras to each shortest distance between operator and robot, and the robot movement stops when falling below the safety minimum allowed distance. In the<patcit id="pcit0001" dnum="DE102004043514A1"><text>DE 10 2004 043 514 A1</text></patcit> surrounds a protective field, the machine, which is adapted to the movement of the machine.

However, these considerations is the fact that the two participants, so humans and machines largely run their operations independently and in parallel. Effective communication between humans and machines, and thus a real cooperation is lacking. The only communication interface consists in screens and switching units of the machine. With such switches, the operator of the execution of a defined activity and starts the movement of the machine. Switching elements have moved along with the workspace or be performed several times in order to stay within reach of the working process for the operator.

Here, of cuts in workflow which require interaction between humans and machines to meet relatively rigid requirements to ensure the safety, for example, the transfer of a workpiece requires that absichernden sensors temporarily tolerate a close approximation between man and machine. The coverage may not be based solely on a fixed timing, because this would exclude any correction in the operator or the workpiece errors.

Thus, the operation will force additional steps or bring additional disadvantages. are known hand controls, the operation of which is just as possible that no hand of the operator is in the danger zone. Even Foot switches are mounted, but they reduce the stability, the skeleton unilateral load and adjust their supply lines pitfalls are. Screens for displaying information machines are often located at the edges of the machine, if any recognizable representations are given. This separation of the interaction region of the workpiece and the information and control devices of the machine, especially when they show the work steps, insert directions or other necessary information leads, because of the constant head movement and accommodation of the eye fatigue.

Jobs to machines or manual assembly stations must meet a number of ergonomic standards and guidelines. Therefore, many jobs have adjustment, about Adjustment of seat and gripping heights. Partial but an individual adaptation of machines impossible and the workplace is designed for an average human. In each case, the adjustment requires a manual intervention. For the execution of various operations for larger components or semi-finished products but may have a different working height be optimal in order to optimally utilize biomechanical leverage and to avoid premature fatigue. Then manual adjustments are time consuming and cumbersome.

Ultimately, the man of the machine or the work tasks must therefore adapt and not vice versa. Especially in piece production the necessary downtimes are avoided for adapting the workplace because of the high cost. This approach is contrary to the principle of ergonomics and a healthy and effective working.

It is known to produce for securing a machine against accidents with operators via one or more optical sensors three-dimensional image data, and based on which to recognize dangerous situations. For example, in the<patcit id="pcit0002" dnum="US6297844B"><text>US 6,297,844</text></patcit> an optical security curtain disclosed generates the image data with a stereoscopic camera system.

Such 3D sensors are usually designed as a camera. A stereoscopic camera system takes two or more images of a scene from different perspectives on, assigns structures of different pictures to each other and triangulated from the disparity, ie the apparent displacement of the structures, and based on the known prospects their removal. At a term camera active, a light signal is emitted and determines for each pixel, for example, using photomix detecting the running time until reception of the remitted from the scene light. Because of the known and constant speed of light and this provides a three-dimensional distance map. There are known image sensors in the intelligent pixels the maturity determination is integrated.

But laser scanners or light grid can educate themselves to three-dimensional sensors. For this purpose, the laser scanner, which typically scans only one plane, is also moved in the perpendicular to the normal scanning movement. A light curtain can be formed groping, therefore consists of distance-measuring light sensors instead of simple light barriers. For distance measurement in this case be used as per the procedure described for a term camera in both the laser scanner as well as the light curtain light transit time method.

Although all of these safety devices allow monitoring of three-dimensional protective fields to undue interference. However, for the operator, the scope not seen, and therefore, it always comes back to nuisance tripping of the safety device. This leads to sometimes considerable production problems because in addition to the loss of time by the restart of the machine and the workpiece is unusable for a safety-related emergency stop or at least needs to be reworked.

To visualize a three-dimensional image data are data glasses or HMDs (Head Mounted Displays) known. You work partly with at least semi-transparent display, which are arranged in the field of view of the user in order to superimpose the artificial representations of a natural setting. A more advanced variant VRDS (Virtual Retinal Display), sitting on a pair of glasses and project the image directly onto the retina. This type of visualization is not found in scientific studies or for military pilots use, however, in industrial production.

From the <patcit id="pcit0003" dnum="DE102006048166A1"><text>DE 10 2006 048 166 A1</text></patcit> is a multi-camera system for monitoring a person in an industrial environment, in which consecutively a virtual 3D model is adapted to the captured image data of a person to determine the position and the movement behavior of the person. From this, a risk potential is estimated and if this exceeds a threshold, acting on the motion control of a machine in the environment of the person. The conventional camera system gives the person without the capacity to interact with the camera system or the machine. The movement behavior is evaluated solely in terms of a potential hazard.

The <patcit id="pcit0004" dnum="EP2048557A1"><text>EP 2048557 A1</text></patcit> discloses an optoelectronic sensor for monitoring a region of space, be configured in the protective fields by a special handset along the desired protective field limits is performed in the spatial domain. The configured or teilkonfigurierten protective fields are displayed in one embodiment, the data display. In the conventional procedure for a the particular handset is required. Moreover, the interaction relates solely to the sensor and a configuration phase before the actual manufacturing operation.

It is therefore an object of the invention to improve the cooperation between man and machine in a collaborative work with full protection against accidents.

This object is achieved by a 3D safety device according to claim 1 and a method for hedging and use at least one machine in a cooperative work according to claim 15th The invention is based on the concept of to use three-dimensional image data of a safety sensor not only for hedging purposes but also for the operator. Certain patterns of movement of the operator are detected and interpreted as an input to the 3D sensor or the machine and further processed. For example, the 3D sensor is instructed to change to another mode, to select a different viewing angle to disable itself or the like. Instead of separating as conventionally man and machine by a safety device of one another, so a real cooperation is achieved.

Movement patterns are interpreted widely and also include se static elements such as a pose or a body position. Both the poses and the movements are detected with a tolerance margin because no identical movements and postures are expected of a human operator. The detection of a movement pattern for example, by classification, so the targeted assignment to a plurality of possible patterns of movement.

Under a cooperative work, a range is understood, in which at least one operator and at least one machine, such as a robot, an assembly station or the like, simultaneously operate, having exchanged at least in a part of the production process work results between man and machine or completed operations together will.

Because some user input, ie security critical or at least process-critical critical to the work product, it is conceivable that some or all movement patterns will be accepted only after a special authentication as user input. This authentication can be implemented very conventionally as A numerical code, password or reading an access card. Alternatively also be used for authentication sensor data in the form of a person recognition or a particular movement pattern to release.

The invention has the advantage that the operator, ie the human-machine interface, is fully integrated using 3D technology in the safety solution. The cooperation of the operator with the machine is understood as a 3D interface. A simple, intuitive communication for the operator is possible. Downtimes and sideline to set up and adapt the workplace to be reduced. The productivity is increased and the ergonomic adjustments easier. Thus, both the quality of the product as well as job satisfaction improved.

Preferably, a display device is provided, particularly a projector or data goggles to present the operator an overlaid with additional information image of the workplace, the additional information includes in particular the presentation of a workpiece, a designated position, pose or movement of the operator and / or a protective field , The display is more preferably three-dimensional. The operator recognizes it, for example the required insertion direction for a workpiece, its own correspondingly provided position and posture, or the current limits of a danger zone. A data goggles is here mentioned as a proxy for all HMDs or similar display means that are worn attached to the body, to present to the eye an image that can be superimposed on the natural field of vision. The display device preferably communicates wirelessly with the evaluation device and is integrated into a already worn helmet or goggles.

The display device is preferably configured to display controls in the workplace, in particular virtual switch or controller. The 3D sensor detects motion pattern of the operator as a user input. For operating virtual switch it therefore is at issue, to bring a body part to a certain place and there optionally perform a particular movement. This movement can be a gesture in particular from the manual alphabet. But that is where the operator knows which movement is expected of it, these controls are displayed so that the operator activates for example the positioning of the hands in the switch area indicated as such. Conventional switch on the machine are thus unnecessary, and the virtual switch can always be ergonomically displayed easily accessible for the operator. A particularly safety-relevant control element is a virtual emergency stop switch, which is shown, for example, as a visual ring around a specific area. Also conceivable is an optical hand controls.

The evaluation unit is preferably adapted to detect a difference between a desired state and an actual state and display by the display device. A target state is, for example, the required type and shape of a workpiece, the position and orientation, the position and posture of the operator or its motion for a work to be performed step. The display can be a color marking. For color-blind both the selectable color and a different structure is possible. The operator recognizes as soon whether the workpiece or the workflow is something wrong, and this can be corrected at an early stage.

The evaluation unit is more preferably adapted to refuse an operation of the machine as long match to target state and current condition. In order to have a quality assurance. So is about checking whether the correct workpiece in the right way is inserted. Only then a triggering of the subsequent automatic working step is possible. The conformity is required in each case only in a practically relevant degree. Another possibility consists of verifying the completeness before a clock triggering for example when joining or assembly tasks.

The evaluation unit is preferably adapted to detect a movement pattern gestures or postures, in particular hand movements. For example, certain gestures are associated with virtual switches. This type, a switch, for example, to use means of wiping can be used alternatively or cumulatively for insertion of switching elements by the display device. A gesture control can be analogous to the sign language and complex commands give. A movement pattern is compared with a gesture of overarching concept, because a pattern of movement in addition to gestures and a position change and the body includes inclusive movements. The closer the detected movement patterns are based on well-known from the everyday gestures, the intuitive is the operator. All gestures mentioned in this description are only examples, the invention is not limited. The direct interaction with the machine enables manual handling of the workpiece and therefore the saving of support means, because the usual racks or moldings can be omitted. The workpiece is inserted and fixed manually and then the processing operation triggered by a specific gesture, such as a particular spatial hand position for both hands.

The evaluation unit is preferably adapted to detect movement patterns for the raising or lowering of a machine part or the setting of a distance between operator and machine part and move the machine part accordingly. Such movement patterns consist, for example similar to the hands of a marshaller from a lowering or raising of both downward or upward directed palms or the display of a distance between the facing palms.

The evaluation unit is preferably adapted to detect movement patterns for triggering or end of an automatic work step and to control the machine according to run or stop the work step, wherein a movement pattern is especially a Notausgeste or posture to operate a virtual two-hand control. These are, for example, gestures such as the upward or downward thumb or the formed between the thumb and forefinger OK sign. There may be a special emergency gesture can be defined for an emergency stop, which should be particularly robust execute recognizable and particularly simple and at the same time preclude the operator performs this movement patterns randomly. One example is the defensive prestretching both hands. A hand controls can be interpreted as movement patterns. The position of the hands and possibly also of the head is monitored, so as to ensure that at least those parts of the body are not compromised by the subsequent step.

The evaluation unit is preferably adapted to detect movement patterns for pushing back a machine part and move the machine part accordingly. The purpose of a particular movement pattern, for example, a gesture analogous to shoo a fly. So instead trigger an emergency stop when approaching the source of danger, this enables the machine part, about an axis of a robot to move to retreat. As long as no hazard is detected and no protective field is violated, the retraction of the machine part can be delayed to first complete automatic operation.

The difference between Shoo and emergency stop can resolve the conflict between process safety and risk situation. Accesses the operator within the scope of a, an emergency stop is required, which can lead to quality problems. The procedure is performed at times but to perform corrections. For this can be due to the abschlossen Shoo gesture of partial step, the axis then withdraws, and the correction can be made. Subsequently, the operation is started and completed again by means of a starting gesture. This purpose, a beckoning. Should d change their position ie axle, as will be pointed with his index finger in the appropriate direction. The duration of this gesture will determine if the distance traveled.

The evaluation unit is preferably made for a configuration mode, taught in the movement patterns of a particular operator and assigned to controls or user input. This is a kind of "3D Customizing". By adjusting the "look and feel", ie the procedures for design, color and structural elements that realized by the invention 3D interface to individual preferences of the operator is customizable. But also individual features with a particularly large impact on the ergonomics, as handedness, or height or physical impairments are taken into account. The movement patterns are individually be taught in order to achieve a robust detection, and for each operator intuitive and simple control. Finally, specific control options can be activated or deactivated.

The evaluation unit is preferably configured for a calibration mode in which the workplace and the machine is adapted to the operator, in particular by the hand height or the head height is determined and then an ergonomic height adjustment of the workplace and / or an ergonomic presentation of workpieces through the machine to the operator takes place. Through this balance between the presentation of the workpiece, the worktable and the footprint of the operator postures are avoided and used biomechanical lever the operator optimally. Because an adjustment of the table height makes the workpiece Logistics, an adjustment of the height of the stand space is alternatively possible. Calibration is carried out, for example, based on the hand position by the current level of hand hands presented is measured and the amount of elements of the workplace be adjusted accordingly. It is also conceivable to use the body size based on the addendum. Calibration can be performed at start of shift or be activated at any time to ensure an optimal adaptation to the stature of the operator currently active. The storage in a user profile is possible.

Not only the workstation itself, but also the presentation of the workpieces is adapted to the operator. Especially with heavy workpieces lifting aids are used, it will interact with robots, or the Assembly is rolling workbenches. In this case, the workpiece should always be set to an ergonomic working height, which depends on the operator and is known of the safety device due to the calibration. If even the preferred operating manual specifies, so the workpiece presentation can be optimized, for example through an angle of attack to the operator on.

The evaluation unit is preferably adapted to detect a hazard to the operator on the basis of an impermissible interference with a scope within the workplace. A dangerous situation is thus always detected when a procedure is performed in a protective box. This facilitates hazard identification. From improper and permissible interventions are to be distinguished, for example, by machine or workpiece. Can assign protective field interventions such known objects, the safety device and such a protective field is in accordance with the process, so no protection must be done.

The protection zones are thereby more preferably dynamically adapted to a currently processed workpiece. The dynamic adjustment therefore includes not only the moving machine part, but also the work that could swing open, for example. On the other hand, a reduction of the protection zone is possible for small workpieces, whereas the protective field to be interpreted in accordance with the maximum case without consideration of the workpiece. The smaller the protective fields at still guaranteed security, the less the operator is restricted by this, and the less unnecessary shutdowns reduce availability.

The display device is preferably configured to the operator to present the protective fields. This applies to a configuration phase, in which the protection fields are set up, as well as for the actual operation. During configuration, which is initiated by a more complex authentication gesture, the hand is moved along the desired protective field boundaries with a particular finger position. By a further gesture, such as double-tapping with the index finger, an existing sensing zone can be selected and subsequently altered by hailing or Fort pushing or virtual touch and positioning. The display means are each visual feedback on the progress of work and the current input. During operation, the active protection fields appear to reduce accidental intervention by the operator. Especially with dynamically adapted to the current situation protective fields would be without such a display is not always clear which areas trigger the safety function, so that the operator complies with unnecessary safety margins and thus restricted in their freedom of movement.

The evaluation unit is preferably adapted to contours of objects that engage inadmissible in a protective box, highlight visually by the display device. For this purpose reference is made, for example, with a flashing function on entire objects or interfering contours, which are still in scope. With these clear instructions can be quickly the cause of a safety shutdown find and eliminate, and availability improves.

The evaluation unit is preferably adapted to store the three-dimensional image data at issue a shutdown command. This documents how it could come to the emergency in order to reconstruct an accident happened or to optimize the flow in the future. The images can of course be beyond constantly recorded and visualized for example, in a security center.

The inventive method can be further developed in a similar manner and shows similar advantages. Such advantageous features are exemplary, but not exclusive in the adjoining the independent claims dependent claims.

The invention is exemplified below with respect to additional features and advantages by way of embodiments and with reference to the accompanying drawings. The images of the drawing show in:<dl id="dl0001" compact="compact"><dt>Fig. 1</dt><dd>a schematic three-dimensional representation of a collaborative workplace with a 3D security apparatus of the invention; and</dd><dt>FIG. 2</dt><dd>a representation according to <figref idrefs="f0001">figure 1</figref> at a different time of the manufacturing process.</dd></dl>

<figref idrefs="f0001">figure 1</figref> shows a schematic three-dimensional representation of the general structure of a 3D inventive security device 10 on a cooperative work 12. There are an operator 14 and an engine 16, in the example a robot, is concerned with a hybrid assembly, which in the illustrated operation, the machine a workpiece 18 processed.

3D security device 10 comprises a 3D-sensor 20, which is designed as a stereoscopic camera. However, the invention is independent of the type obtained with the three-dimensional image data, and also includes other 3D sensors including the aforementioned laser scanners and light runtime cameras. It is conceivable that an additional, in the<figref idrefs="f0001">figure 1</figref> not illustrated 2D camera to use to gain additional image data, the dissolved example above, be included in another spectrum or from a different perspective. Similarly, the 3D-sensor may be also mounted laterally instead of vertically.

The 3D safety device 10 is further equipped with a projector 30 are projected to the visual information in the workplace 12th An evaluation unit 40 is connected to the 3D sensor 20 and the projector 30 and is concerned with the control and evaluation of 3D security apparatus 10th This combination of both data and information can be displayed as well as control commands are triggered. It can be 30 used several 3D sensors 20 and / or multiple projectors, for example, a larger workspace cover or compensate for poor visibility and projection angle or shadowing.

In the 3D sensor 20 two camera modules 21a-b are mounted to each other in a known fixed distance and take each case on photos of the workplace 12th In each camera, an image sensor 22a-b is provided, usually a matrix-like recording chip which receives a rectangular pixel image, such as a CCD or a CMOS sensor. The image sensors 22a-b, a respective lens 23a-b associated with an imaging optical system which can be implemented in practice as any known imaging objective. The viewing angle of these lenses is in<figref idrefs="f0001">figure 1</figref> shown by dashed lines, each of which forms a pyramid of vision 24a, 24b.

In the middle between the two image sensors 22a-b, a lighting unit 25 is provided, this spatial arrangement can be understood only as an example and the lighting unit may also be arranged asymmetrically or even outside of the 3D sensor 10th This lighting unit 25 has a light source 26, for example, one or more laser or LED, and a pattern generating element 27, which is designed for example as a mask, the phase plate or diffractive optical element. Thus, the lighting unit 25 is able to illuminate the space area 12 having a structured pattern. The structured illumination pattern facilitates obtaining stereoscopic 3D image data, but should preferably be for the operator 14 invisible example in the infrared spectrum. There are also embodiments with a homogeneous illumination or those conceivable that do not require lighting unit 25 alone with the ambient light.

The two image sensors 21a-b a stereoscopic evaluation unit 28 is connected, which image data of the image sensors 21a-b receives and calculates using a stereoscopic disparity three-dimensional image data of the workplace 12th The structured illumination pattern ensures a good contrast and a clearly identifiable structure of each pixel of the illuminated workplace 12th

known and unexpected objects can be located in the monitored by the safety sensor 10 room area 12th This may be, for example, as shown by a robot arm 32, as well as any other device, an operator and acting another. The space portion 12 provides access to a source of danger, either because it is an access area or because a dangerous machine such as the robot arm 32, located in the spatial region 12 itself. To hedge this risk source, one or more virtual detection and warning zones can be configured 34. They form a virtual fence around the dangerous machine 32. Due to the three-dimensional analysis, it is possible to define protective and warning fields 34 in three dimensions, so that a great deal of flexibility.

The 3D sensor 20 outputs the three-dimensional image data to the evaluation unit 40th In a movement pattern recognizing unit 41 of the evaluation unit 40, the operator 14 is identified within the three-dimensional image data and determines their position, trajectory and posture. Depending on the requirements of the application gestures and poses are rather coarsely resolved only of the entire body to a fine recognition of individual finger movements. Possibilities of detection of 3D movement patterns are known per se and, for example, in the aforementioned<patcit id="pcit0005" dnum="DE102006048166A1"><text>DE 10 2006 048 166 A1</text></patcit> and the literature cited therein. For example, a 3D model of a person is parameterized using the current image data. By classification known movement pattern is then decided which motion patterns the person currently performs. Given not only the current frame and the past is often analyzed, but taken into account. One known method for this is the object-tracking example, based on Kalman filtering.

If a motion pattern is recognized which corresponds to a pre-set or taught movement pattern, the evaluation unit 40 interprets this as a user input similar to the operation of a switch or controller. The evaluation unit 40 then performs the intended for the movement pattern command or operation of, or directs a corresponding signal to the 3D sensor 20, the projector 30 or via a non-illustrated connection, which can also be implemented wirelessly or indirectly through a system controller, to the machine 16 further.

For commands to the projector 30, a display controller 42 is provided in the evaluation unit 40th This controls can projected on the workplace, reconfirmation of inputs and general data and information for the operator 14 to be visualized. An important information for the operator 14 are protective fields 50, in which can not be touched for security reasons, said in<figref idrefs="f0001">figure 1</figref> an example of a virtual wall between the operator 14 and the machine 16 is shown as a protective field 50th

This protective fields 50 are evaluated in a protective field evaluation unit 43 of the evaluation unit 40 on undue interference. The evaluation rules dictate that in protective fields 50 absolutely no object may be present, for example. More flexible evaluation rules provide to differentiate between permitted and prohibited objects, such as based on trajectories, patterns or contours, speeds or general work processes, both beforehand learned are as rated yet also based on ratings, heuristics or classifications during operation than allowed.

The protective field evaluation unit 43 knows the movement behavior of the operator 14 of the movement pattern recognizing unit 41. Similarly, also the movement pattern of the machine are detected 16 by the 3D sensor 20th Alternatively or additionally, are also data on the machine control system available for the movement pattern of the machine 16th This highly flexible protective fields 50 are possible, which are adapted to the respective situation.

Does the protective field evaluation unit 43 an impermissible interference with a protective panel 50, it is stopped by a safety output 44 (OSSD, Output Signal Switching Device) or a security network the machine 16, slowed down or moved to a safe position. To be suitable for safety-related applications, the safety device 10 is designed fail-safe. This means among other things that the 3D sensor 10 and the evaluation unit 40 can itself, also test cycles below the required response time, and that the safety output is 44 sure, for example, designed with two channels. Likewise, at least all are confident with safety evaluations units, so evaluate two channels or uses algorithms which can examine itself. Such safety regulations have been standardized for general contact protective devices in EN 61496-1 or IEC 61496, as well as standardized in the DIN EN ISO 13849 and EN 61508th A corresponding standard for safety cameras is in the preparation. The movement pattern recognizing unit 41 and the display controller 42 and the projector 30 need not be so strictly structured and monitored necessary as here, a failure at any risk of accident.

In place of a projector 30 is a data goggles or a HMD is used in a further embodiment, not shown. The data goggles is wirelessly controlled by the display controller 42 and provides the information to the evaluation unit 40 of the operator 14 is such that the natural image and the artificial information overlay. This is done either by the data goggles still allows the view of the scenery, or by another camera on the operator's head receives the workplace 12 from their perspective and electronically processed into a layered three-dimensional view. It can even for the convenience of the operator irrelevant pixels are hidden, for example, distant objects. As another example, artificially altered image data a hazardous movement of the machine 16 could be optically anticipated to achieve a natural retreat of the operator. For a correct visualization of the information on the data goggles position and viewing direction of the operator must be recorded, either by the 3D sensor 20 and the movement pattern analysis unit 41 or supports or exclusively by additional sensors the data goggles.

By 3D safety device 10 creates an intuitive and uniform 3D interface, are summarized in the information display, operator actions, visualization of operations and machine safety. So far, independent elements are thus integrated to an optimal solution for an effective, safe and ergonomic man-machine interaction, which also ensures the quality and reduces non-productive time and sideline.

The <figref idrefs="f0001">figures 1</figref> and <figref idrefs="f0002">2</figref> illustrate one aspect of this interaction. In<figref idrefs="f0001">figure 1</figref> edited the machine 16, the workpiece 18, and the operator is completely separated by the wall-like protective panel 50 of the machine 16th This situation, the conventional parallel work reflects. According to such a separation is made only for certain operations where no interaction is required or where the machining is so dangerous that the operator must keep well away.

<figref idrefs="f0002">figure 2</figref> showing a later processing step. The automatic part processing of the workpiece 18 is completed, and the machine 16 transfers the workpiece 18 to the operator 14 or holds it at a comfortable working height and position for manual processing or inspection. The protective panel 50 is dynamically adjusted, in this case so far that it has completely disappeared, because the dormant robot does not constitute a source of danger. The operator 14 is now with the gestures described above, for example, the robot arm to position changes, rotation, or, after completion of the manual partial step, the continuation of automatic cause. For all machine movements, the evaluation unit will generate 40 suitably adapted protective fields 50th