Safety device for an industrial robot

Abstract

The invention relates to a safety device for an industrial robot, comprising a detection device which is designed to detect a potential hazardous condition for a body part of a person who is in the vicinity of the industrial robot, due to a component of the industrial robot or an object moved by the industrial robot. The invention is characterized in that the detection device comprises an image-capturing device which is designed to capture an image in a hazardous region which comprises the industrial robot and a spatial section surrounding the industrial robot, and the control unit is designed to carry out the evaluation on the basis of a chronological sequence of images with image data captured by the image-capturing device.

Description

[0001] Safety device for an industrial robot

[0002] The invention relates to a safety device for an industrial robot, comprising a detection device configured to detect a potential hazard to a body part of a person who is near the industrial robot, caused by a component of the industrial robot or an object moved by the industrial robot; a safety action device configured to execute a safety measure by which the hazard is reduced or avoided; and a control device that is signal-connected to the detection device and the safety action device and configured to perform an evaluation of a potential hazard detected by the detection device based on predetermined criteria and, depending on the evaluation, to control the safety action device to execute a first safety measure.

[0003] Safety devices for industrial robots serve to prevent collisions between the robot and people, vehicles, or other objects. Due to the large range of motion an industrial robot can cover and the multiple axes of movement it often possesses, the requirements for such safety systems are high. A common approach is to design a safety system using one or more sensors to detect whether a person is within the robot's range of motion, thus indicating a potential collision risk. If the sensors detect such a potential hazard, the robot's movement can be stopped, thereby preventing a collision.A disadvantage of this safety concept, however, is that a relatively large area covered by the industrial robot is monitored, and the robot's safety stop is triggered when someone enters this area, regardless of whether the robot is actually moving into the area where the person is located. This severely limits the possibilities for human interaction with a robot, as the safety system always requires a certain distance between people and the robot's potential range of motion.

[0004] A safety system is known from US patent 2021 / 053227 A1 that enables a person to work within the vicinity of a robot in a more efficient manner. This safety system incorporates multiple sensors that capture images of a portion of the work area and perform object recognition to detect both the person and the robot. This object recognition is used to identify the person and the robot, as well as their movements, and consequently to detect when the person approaches the robot. In the event of such an approach, the robot's movement is stopped to maintain a safe distance between the robot and the person. A similarly functioning safety device is also known from German patent DE 10 2017 221 305 A1.While this type of safety device requires significantly more computing power to monitor dangerous approaches between a person and a robot in real time and to stop the robot's movement before contact occurs, it has the advantage that a person can generally enter the robot's movement zone without immediately stopping the robot, as long as the robot does not approach the person critically within this zone. However, a disadvantage of this safety system is that interaction between the person and the robot is limited, because falling below a certain safety distance between the person and the robot results in the robot's movement being stopped.

[0005] From DE 10 2010 017 857 A1, a 3D safety device for securing and operating a machine at a collaborative workstation is known. This safety device comprises a 3D sensor for capturing three-dimensional image data of the workstation, the machine, and an operator at the workstation, and an evaluation unit designed to determine movement patterns such as poses or postures of the operator within a tolerance range and to classify such movement patterns, i.e., to assign them to one of several possible movement patterns. The evaluation unit is designed to derive user inputs to the 3D sensor or the machine from the recognized movement patterns. If a hazard to the operator is detected, a safety command is issued to the machine.For this purpose, the position of the hands and head is monitored to ensure that these body parts are not endangered during a subsequent work step. This hazard monitoring is carried out by intervening in a protective field. However, a disadvantage of this known device is that reliable hazard monitoring cannot be achieved because the safety device lacks predictive capability due to the recorded position data, and thus body parts are at risk during dynamic processes. Furthermore, the detection of hazardous situations based on classified movement patterns fails when the operator's movements are atypical and unclassifiable.

[0006] A method for controlling a robot is known from WO 2014 / 008929 A1. The robot can be operated in a working mode in which a part of the robot moves at a speed that poses a risk of injury to a person. This working mode is deactivated if a safety device detects that the person is entering the working area of ​​the moving part. Close cooperation between the person and the robot is to be enabled. For this purpose, a sensor unit determines the person's position and posture while the person is outside the working area of ​​the part. A prediction unit determines the person's working area. A collision monitoring unit checks whether the two working areas overlap.A disadvantage of this solution is that, based on position determination and the predicted potential action area, a safety trigger can occur even when no body part has actually entered the robot's action area. This leads to false triggers as soon as a person is near the robot's action area. Furthermore, this solution cannot detect and therefore cannot prevent safety hazards arising from greater movement dynamics.

[0007] From US patent 2017 / 0210017 A1, a robot safety system is known with a safety controller for a robot viewed by a worker. Based on the worker's position or field of vision, the controller determines the distance between the robot and the worker using positional information. The safety controller generates a virtual shield image, the display mode of which changes depending on the distance to the robot. This image indicates the boundary position of the robot's working area based on the determined distance. The generated shield image is displayed on a glasses-like display device worn by the worker, thus overlaying the worker's field of vision. While this known safety system can effectively indicate certain hazard zones to the user, it does not reliably prevent the user from being at risk of injury.The system also has the disadvantage that hazardous situations caused by dynamic processes cannot be reliably detected.

[0008] The invention is based on the objective of providing a safety system for industrial robots that enables better interaction between the industrial robot and a person.

[0009] This task is solved by a safety device of the type described above, wherein the detection device comprises an image acquisition device which is oriented to perform image acquisition in a danger zone which includes the industrial robot and a spatial section surrounding the industrial robot, and the control device is configured to perform the evaluation on the basis of a temporally successive sequence of images acquired by the image acquisition device with image data, wherein the control device is programmed to recognize a body part of the user in each image of the sequence of images on the basis of the image data, and to determine the position of the recognized body part within a spatial reference system.to calculate the direction and speed of movement of the detected body part by comparing its position in two or more consecutive images; to determine the position, direction, and speed of a potentially hazardous component of the industrial robot and, if applicable, of an object moved by the industrial robot; to calculate, by comparing the determined position and calculated direction and speed of movement of the detected body part with the position, direction, and speed of the potentially hazardous component of the industrial robot, whether contact between the detected or another body part of the user and the potentially hazardous component of the industrial robot is to be expected within a predetermined or calculated time period, and if so,to generate a first hazard signal and send it to the safety control device, the safety control device being designed to execute the first safety measure upon receiving the first hazard signal.

[0010] The safety device according to the invention basically comprises a detection device, a safety measure device, and a control device. The detection device serves to detect a potential hazardous situation, which is understood to be a situation caused by the position, movement, and acceleration of a part of a person's body or the entire person on the one hand, and a component of the industrial robot on the other, which could lead to injury to the person through contact with the component of the industrial robot. The term "component of the industrial robot" here refers, on the one hand, to the industrial robot itself in its structural components, i.e., typically housing-enclosed robot arms, joint arrangements, actuators, or the like, which constitute the actual industrial robot.Furthermore, such components of the industrial robot can also include tools that the industrial robot guides or objects that the industrial robot grasps and that, due to the movements of the industrial robot, can come into close proximity to a person and cause injury. In the safety device according to the invention, the detection device comprises an image acquisition device, which may be a device for capturing images in the visible light range, infrared, by means of active emission and reception of electromagnetic radiation, or by passive reception of ambient radiation. The image acquisition device is specifically designed to continuously and in real time detect the danger zone.The image acquisition device can be designed with a single sensor that surveys the entire area; however, the image acquisition device preferably comprises several sensors that capture the danger area from different positions and / or with different detection directions, wherein the areas captured by the individual sensors may partially or completely overlap for redundancy purposes, and by combining the several image acquisition areas to produce an overall image of the danger area.

[0011] A safety device is a device that, when a condition occurs that would lead to injury, initiates a safety measure suitable for preventing this injury. The safety device can be implemented in a specific control system or programming of the industrial robot's controller, which, for example, controls a stop or a slowing of the industrial robot's movement, thereby preventing a potentially dangerous situation.The safety device can also take alternative or additional measures, for example, emit warning signals such as optical or acoustic signals that are perceptible to the user and warn him of the potential injury; furthermore, alternatively or additionally, active protective measures can also be taken by the safety device, for example, shielding of components that pose a risk of injury, such as sharp-edged tools, or partially stopping certain movements of individual components of the robot, for example, rotating tools of the robot or crushing gripping elements of the robot.The safety control device can also, as a safety measure, control an active movement of the robot that moves one or more components of the robot in such a way that a hazardous situation is avoided or eliminated. This can, for example, be done in conjunction with the detection device to determine, based on image data, a movement pattern of the industrial robot or a specific component of the industrial robot that leads to the avoidance of the hazardous situation, and then to control or execute this movement pattern using the safety control device.

[0012] A third fundamental component of the safety device according to the invention is a control unit that is signal-linked to the detection device and the safety device. This control unit is designed to evaluate the data captured by the image detection device and thereby detect a potential hazard in real time. For this purpose, the control unit can perform corresponding image evaluations and image recognition to, on the one hand, recognize a person or parts of a person's body and determine their position and direction of movement, and on the other hand, recognize the industrial robot and its components, as well as any tools or objects guided by the industrial robot, and again determine their position and form of movement.

[0013] The control device is then designed to make a future-oriented assessment based on the person, body parts, industrial robot, components of the industrial robot, or objects or tools thus detected. This assessment, based on positions, directions and speeds of movement, and, where applicable, directions and heights of acceleration, determines whether contact between a component of the industrial robot and a body part of the person that could lead to injury is to be expected. Furthermore, based on this acquired position and movement data of the person and the industrial robot, the control device can determine how much time is required to prevent this injury to the person through an appropriate safety measure, and whether the time remaining from the time of detection until the expected time of injury is still sufficient to effectively implement the safety measure.This can include, for example, the ability of the control device to determine whether the industrial robot's movement can still be slowed down sufficiently to avoid injury, but it can also include more sophisticated control methods, such as determining whether the injury can still be avoided by an evasive movement of the robot and in what direction and at what speed this evasive movement must be carried out.For the purpose of determining when the safety measure must be implemented to effectively prevent injury, the control device can access relevant technical data characterizing the industrial robot, which may be stored accordingly, such as the mass and position of the center of mass of individual components of the industrial robot, the acceleration capabilities of individual components of the industrial robot, such as forces and torques that can be generated by the actuators of the industrial robot and their effect on possible accelerations of the components of the industrial robot, latency times in the control of the industrial robot, and other technical data relevant to the respective safety measure that affect the time period in which a safety measure can be carried out.The control device sends the first hazard signal at the latest when the time period within which a potentially harmful contact is to be expected corresponds to the period within which the injury can be avoided by the safety measure, or when this period is greater than the expected duration of the injury by a predetermined safety period, and the safety device triggers the corresponding safety measure upon receiving this first hazard signal.

[0014] This specific detection of an impending hazardous situation, on the one hand, and the subsequent analysis by the control unit, which incorporates both the movement of the industrial robot and that of the person into the determination of a hazardous situation and includes the necessary time to prevent injury through a safety measure in the control for sending the hazard signal, enables differentiated control of the safety measures. For example, in the case of slow movements that would lead to an injury, the hazard signal can be sent later than if the same movement occurs with a fast movement and therefore requires a longer time for deceleration or evasive action due to inertial forces.Furthermore, if a potentially harmful contact is anticipated between a component section exhibiting high inertia relative to the necessary change in movement, the hazard signal can be emitted earlier than if the potentially harmful contact is caused by a component section exhibiting low inertia relative to the necessary change in movement to avoid injury. This also allows the operator to work closer to the industrial robot and prevents premature or unnecessary triggering of safety measures.

[0015] According to a first preferred embodiment, the control device is programmed to calculate, based on the calculated direction and speed of movement and, optionally, the direction and acceleration of the detected body part and the direction and speed of movement and, optionally, the direction and acceleration of the injury-causing component of the industrial robot, a severity of the expected contact between the detected or other body part of the user and the injury-causing component of the industrial robot, wherein the severity of the contact describes at least the magnitude of the expected acceleration of the body part due to the contact, to compare the calculated severity of the expected contact with a predetermined first limit value, and if the calculated severity of the expected contact exceeds the first limit value,to generate the first hazard signal. According to this embodiment, the control device is designed to distinguish between contact between the industrial robot and the person that poses a risk of injury and contact that does not, and to emit the hazard signal only if contact that poses a risk of injury is to be expected. This differentiation is achieved by calculating the severity of the expected contact. This severity of the expected contact characterizes the expected acceleration of the body part of the person that comes into contact with the industrial robot as a result of this contact. The control device can take into account various properties of the body part and the industrial robot, as well as parameters of the respective movement situation that leads to the contact. For example, it can consider the moment of inertia, and in particular the mass and the point of inertia,The system assesses the acceleration of the body part caused by the potentially dangerous contact. This allows for a distinction between contact with a limb that is distant from and near a joint. Contact with a limb that is distant from a joint can be evaluated differently due to the accelerating evasive reaction or rebound that can be triggered with less force. Contact with a limb that is near a joint can be assessed differently, as such an evasive movement would only be generated by a greater force and consequently a higher risk of injury. The control system can also take into account the anatomical characteristics of the body part, particularly its cushioning soft tissue thickness and structure. In this way, it can differentiate between contact with a body part covered in soft tissue, such as a lower leg, thigh, forearm, or upper arm, and a body part not significantly covered by soft tissue, such as a person's head.The control device can differentiate between contact with the head and assess it as a potentially dangerous level of injury at lower accelerations than contact with a soft tissue-covered body part. Furthermore, the control device can consider the vulnerability of specific body parts and, for example, classify contact with sensory organs or the person's head as potentially dangerous at lower accelerations and contact forces than similar contact with the person's extremities. Accordingly, the predetermined first threshold value can vary based on these properties and parameters considered by the control device and can be individually predefined for the specific contact situation and position with respect to the body part and component.

[0016] This inventive design of the control device fundamentally avoids triggering a safety measure for every type of contact between the robot or an object guided by the robot and a person. Instead, it becomes possible for such contact to occur and for the robot to continue its movement unchanged if the control device assesses the contact as not posing a risk of injury.This allows the robot to perform actions requiring contact between a person and the robot while the safety device is continuously monitoring the system. This might occur, for example, when the person holds onto or leans against the robot to perform certain actions, or when the person wants to guide robot components by applying force (e.g., during training processes), or when the person wants to receive objects guided by the robot, hand an object to the robot, or when the person has unintentionally made contact with the robot, provided that the contact is not severe enough to cause injury. In all these situations, the control system's differentiated assessment prevents any interruption or alteration of the robot's movement and avoids triggering any safety measures.

[0017] It is further preferred that the control device is programmed to compare the calculated severity of the expected contact with several predetermined limits, and if the calculated severity of the expected contact exceeds a first limit of the several predetermined limits, to generate the first hazard signal and send it to the safety action device, and if the calculated severity of the expected contact exceeds a second limit of the several predetermined limits, to generate a second hazard signal and send it to the safety action device, wherein the safety action device is configured to execute a second safety action, different from the first safety action, upon receiving the second hazard signal.According to this embodiment, the control device is designed by electronic programming to differentiate between two different ranges of contact severity based on two different limit values. Depending on the severity range in which the expected contact lies, a first or a second hazard signal is generated, and consequently, a first or a second safety measure is triggered, which differ from each other. The different severity levels can be caused, in particular, by different accelerations of one and the same body part. However, the different severity levels can also be caused by different contact points on a body part, different body parts that would come into contact, or different contact points on the component of the industrial robot.Other influencing factors can also play a role, such as the surface contour of the contact point on the industrial robot component, the elastic compliance of the contact points, and other influences. The different safety measures triggered by the various hazard signals can include, for example, issuing a warning signal perceptible to the user, slowing down the industrial robot, stopping the industrial robot, initiating a counter-movement or evasive action suitable for avoiding contact, or other safety measures that reduce or prevent the hazard.It should be understood that the effectiveness of the safety measure to prevent the hazardous situation should be adapted to the severity of the contact; therefore, if the severity of the contact is higher, a correspondingly more effective safety measure should be implemented.It is even more preferred if the control device is programmed to compare the calculated severity of the expected contact for a first body part with the first limit value, and to compare the calculated severity of the expected contact for a second body part with a second limit value, and if the calculated severity of the expected contact of the first body part exceeds the first limit value, to generate the first hazard signal and send it to the safety action device, and if the calculated severity of the expected contact of the second body part exceeds the second limit value, to generate the first hazard signal and send it to the safety action device, wherein the control device is further programmed not to generate the first hazard signal if the calculated severity of the expected contact of the second body part lies between the first and the second limit values.According to this preferred embodiment, the assessment of the risk of injury is differentiated according to at least two different body parts, the severity of which is assessed using two correspondingly different threshold values. This assessment and corresponding determination of threshold values ​​can take into account both the sensitivity of the respective body part and the soft tissue cushioning that dampens the impact of contact. In this way, an effective distinction can be made between contact with an operator's head, the severity of which is assessed using a lower threshold value, and contact with, for example, a user's forearm, which is assessed using a higher threshold value, and a safety measure adapted to each case can be implemented.

[0018] It is even more preferred if the control device is programmed to compare the calculated severity of the expected contact between the first body part and the first hazardous component with the first limit value, and to compare the calculated severity of the expected contact between the first body part and a second hazardous component with the first limit value, and if the calculated severity of the expected contact of the first body part with the first hazardous component exceeds the first limit value, to generate the first hazard signal and send it to the safety device; if the calculated severity of the expected contact of the first body part with the second hazardous component exceeds the first limit value, not to generate the first hazard signal.According to this embodiment, the control device is designed to differentiate between contact between two different components of the industrial robot and a part of the user's body. In the event of contact with a first component, a safety measure is triggered; in the event of contact of the same body part with a second component of the industrial robot, this safety measure is not triggered, and consequently, either no safety measure is triggered at all or a different safety measure is triggered.This training is particularly advantageous when, on the one hand, especially hazardous components of the industrial robot need to be protected in such a way that they trigger a safety measure even at a low threshold of contact severity. This might be the case, for example, because these components are sharp-edged, exhibit intrinsic movement such as rotation, shearing action, or the like, and therefore pose a greater risk of injury compared to other components that, as blunt or padded components, present a lower risk of injury. Such components with a low risk of injury may also be intentionally contacted by the user, and therefore such contact should not trigger a safety measure that would impair the operation of the industrial robot.

[0019] It is still preferred if the control device is programmed to recognize a second body part of the person, different from the first, and to calculate its position, speed, and direction of movement, and, if applicable, its acceleration and direction of acceleration. According to this advanced training method, the control device is configured to recognize and calculate the movement of a second body part. This can be done, for example, by separately recognizing the image of this body part and calculating its movement. Alternatively, a dependent calculation can be performed based on another body part, such as the first body part, if the first and second body parts are connected by one or more joints and the movement of the second body part results from or is partially dependent on the movement of the first body part.

[0020] According to a further preferred embodiment, the safety device is further developed by means of stationary markings in the movement area of ​​the industrial robot, in particular markings formed on a floor mounting surface, which can be detected optically by the image acquisition device, and is characterized in that the control device is designed to recognize the markings and to enter the position of the body part and / or the industrial robot in a stationary coordinate system based on the markings and to evaluate this position based on the stationary coordinate system. A fundamental problem in the safety monitoring of industrial robots using image acquisition lies in the fact that a clear spatial assignment of the user's body parts to the components of the industrial robot must be made. This spatial assignment must take into account the three-dimensional spatial relationships.Reliable identification can be impaired by changes in the position of the industrial robot itself, but especially by changes in the positions or angular orientations of the sensors used for image acquisition, for example, if these are attached to moving components of the industrial robot. Equipping the safety device with detectable, fixed markers and making them recognizable by the control unit's image recognition ensures robust detection of hazardous conditions, as these fixed markers always allow for a reliable determination of relative positions based on the entries in a fixed coordinate system.

[0021] Detectable, fixed markers can be those that also serve as signals for the user, such as limit points of the robot's maximum movement area. A number of markers can be used that meets or exceeds the minimum of three required for unambiguous spatial assignment, thereby including redundant markers and compensating for any obscured markers. These markers can be two-dimensional, i.e., planar, elements such as stickers, colored markings, or the like, or three-dimensional markers such as spheres, cuboids, pyramids, or similar shapes.

[0022] It is further preferred if the control device is programmed to calculate, based on a comparison of the respective position of the detected body part and / or component in three or more successive images, an acceleration direction and acceleration of the detected body part, or to determine an acceleration direction and acceleration of the hazardous component of the industrial robot, and to perform the calculation of whether contact between the detected or the other body part of the user with the hazardous component of the industrial robot is to be expected within the predetermined time period, also based on the acceleration direction and acceleration of the detected body part and the acceleration direction and acceleration of the hazardous component of the industrial robot.In principle, determining the likelihood and severity of contact can be based solely on the positions, speeds, and directions of movement of the industrial robot component and the user's body part. Using this data, and assuming constant motion, a prediction can be made as to whether contact is to be expected and, if so, its severity. However, both the industrial robot components and the user's body parts can be accelerated in a given direction, thus changing their direction and velocity. Therefore, considering this acceleration in terms of direction and magnitude can enable a more reliable and longer-term prediction of an expected contact.In particular, this allows for the consideration of potential injury-avoiding reactions by the user, as well as potential reflexes that increase the risk of injury. The appropriate safety measure can then be selected based on these reactions to reliably prevent injury even at such accelerations. This can generally be determined directly using three consecutive images; however, the corresponding algorithm can also indirectly rely on such three images by calculating the acceleration, for example, based on two velocities measured consecutively over time.

[0023] It is even more preferred if the control device is configured to detect an operational input variable of the industrial robot based on the image data of an image, in particular the provision of an object to be grasped by the robot, the provision of a storage location for an object grasped by the robot, and that the control device has an interface to a controller of the industrial robot and is configured to transmit the detected operational input variable via this interface. According to this embodiment, the control device is additionally configured to detect operational input variables for controlling the industrial robot, such as the detection of objects that the industrial robot is to grasp or the detection of storage locations where the industrial robot is to place an object.Through a suitable interface to the control of the industrial robot or through a suitable integration of the properties of the control unit necessary for the safety device into the control of the industrial robot, the insights gained from this image acquisition and evaluation by the control unit can be directly integrated into the control of the movement sequence of the industrial robot.The safety device according to the invention fulfills, on the one hand, an improved safety function by taking into account such objects and storage locations or other input variables, by means of which the industrial robot is controlled, when determining a possible hazardous state, and on the other hand, the functionality of the safety device according to the invention is also used to improve the operational function of the industrial robot, for example by controlling access to objects and their storage in an improved manner.

[0024] According to a further preferred embodiment, the first safety measure and, if applicable, the second safety measure are selected from

[0025] • a reduction in the movement speed of the industrial robot,

[0026] • an emergency stop of the industrial robot's movement,

[0027] • an emergency stop of the industrial robot's movement and a subsequent control of the industrial robot into a passive mobility state, in which the industrial robot's actuators are controlled in such a way that weight forces are compensated by the robot's components and, if necessary, by an object held by the robot, and the robot's components are movable by the action of external forces,

[0028] • a change in the movement of the component of the industrial robot into an evasive movement, wherein the control device is programmed to calculate the direction and preferably the speed of the evasive movement based on the position, the calculated direction and speed of movement and, if applicable, the direction and acceleration of the detected body part and the position, direction and speed of movement and, if applicable, the direction and acceleration of the component of the industrial robot that poses a risk of injury, in such a way that the evasive movement avoids contact between the body part and the component of the industrial robot that poses a risk of injury.

[0029] According to this embodiment, different reactions of the safety device are possible as a first and, if applicable, a second safety measure. For example, the speed intended for the normal operation of the industrial robot can be reduced if a hazard involving potential contact with the user is anticipated, thereby reducing the severity of the contact to a level that does not cause injury. This allows the industrial robot to continue its operation, albeit at a slower speed, while simultaneously alerting the user to the hazardous situation they have created through the perceptible reduction in speed, enabling them to eliminate the situation through their own actions.Likewise, an emergency stop of the industrial robot's movement is possible as a safety measure, which is particularly relevant if a level of contact that is immediately classified as potentially dangerous is to be expected and this can only be avoided by a corresponding emergency stop.

[0030] The emergency stop may then be followed by locking the industrial robot's movement to maintain the position reached after the emergency stop. In another variation of this safety measure, the emergency stop may also be followed by controlling the industrial robot's actuators. These actuators merely compensate for the weight of the individual components of the industrial robot, but allow movements generated by external forces acting on these components. This control method, also used for training processes on the industrial robot, enables a user to move out of a position on the industrial robot where they are prevented from their desired movement or escape by components of the robot. To this end, the user can also move components of the industrial robot by pushing, sliding, pivoting, or similar means.Finally, the movement of the industrial robot component can also be changed into an evasive movement. This evasive movement may be necessary, for example, if an emergency stop can no longer reliably prevent the potentially dangerous contact, but a change in movement can still achieve this; or also if an emergency stop can avoid potentially dangerous contact, but the user is subsequently restricted in their movement or even trapped by the industrial robot, and the evasive movement can remove this restriction on the person's ability to free themselves.

[0031] It is further preferred if the control device is designed to determine the position, speed, and / or acceleration of the industrial robot component based on data processing with artificial intelligence used in the evaluation of the image data, or a marking on the industrial robot component that can be detected during the evaluation of the image data, or an operator input, or position data received from a transmitter on the component. According to this embodiment, the position and motion data of the industrial robot are determined by means of specific data or data evaluations, thereby achieving a reliable determination of the current position and motion data as well as a reliable prediction of the future position and motion data of the individual components of the industrial robot.This can be achieved, on the one hand, by analyzing the image data captured by the image acquisition device, and on the other hand, by employing artificial intelligence in this analysis. This type of analysis is particularly suitable because industrial robots typically perform repetitive movements, and therefore, with a sufficiently long observation period, artificial intelligence can train itself. Subsequently, based on the learned data, it can make reliable predictions of future positions and movements once a pattern of the current position and movement has been recognized by the image acquisition system.

[0032] For reliable position detection of the industrial robot component, a marker can be applied to the component and detected by the image acquisition device. Such a marker eliminates the need for image analysis to identify the component and the associated computing power. Consequently, it is easier to detect and track, and enables computer-aided real-time image recognition and tracking of the component's movement with lower computing power.

[0033] In principle, the control unit can also have an interface for receiving control data from the industrial robot or have such control data stored. Based on this control data, the position and motion data of the industrial robot can be directly determined and predicted for a specific time, thus enabling a reliable forecast of position and motion data. Furthermore, the position of a component of the industrial robot and its motion data can be recorded by operator input via a user interface on the control unit, as occurs, for example, when programming the movement of the industrial robot and can be derived from such programming.Finally, a transmitter can also be arranged on a component of the industrial robot and used to determine the position and movement data of the component, for example by taking a bearing of the position based on the direction of the emitted transmission beams, a time-of-flight measurement or the like.

[0034] It is even more preferred if the control device has pre-stored anatomical data and is programmed to calculate, based on the position, speed of movement and direction of movement of the first detected body part and an assignment of the first body part to a second body part different from the first, as defined in the anatomical data, a position, speed of movement and direction of movement of the second or a third body part of the person and / or the control device is set up to determine calibration image data by means of calibration image acquisition by means of acquiring one or more body parts or an entire person and to carry out the evaluation based on the temporally successive sequence of the images acquired by the image acquisition device, in particular the calculation of the direction and speed of movement of a body part taking into account the calibration image data.The fundamental problem is that while some body parts of a person can be clearly identified through image analysis during image capture, this is not the case for all body parts. This is due, on the one hand, to the different geometric characteristics of body parts, and on the other hand, to the fact that body parts may be obscured and therefore cannot be identified. According to this embodiment, this problem is overcome by storing anatomical data for several adjacent body parts in the control unit. This makes it possible to determine the position and movement data of a neighboring body part from one of the captured body parts, whose position and movement data is available through image capture and analysis.This can be done, for example, based on joint data from a joint connecting the two body parts, using anatomical data describing the length of the first, second, and possibly third and further body parts, their surrounding structures, and the like. Alternatively or additionally, the control device can also be designed to perform calibration image acquisition in a step preceding subsequent safety monitoring and to determine calibration image data from one or more body parts in this process.In this way, relevant anatomical data in the form of these calibration image data are read into the control unit, which ideally correspond to the person who is to be monitored by the safety device and can be used in the same way as before to evaluate and recognize the body parts recorded in the calibration from image data of an image acquisition, if necessary to infer positions and movement data of neighboring body parts and consequently to determine position and movement data of the body parts quickly and reliably.

[0035] It is even more preferred if the control device is configured to determine the position, direction of movement, speed of movement and, if applicable, acceleration and direction of acceleration of the component of the industrial robot and / or an object moved by the industrial robot from received program data of the industrial robot, and / or from an image acquisition by the industrial robot using the image acquisition device, wherein the image acquisition takes place in real time parallel to the image acquisition of the first body part of the person or the image acquisition takes place in a temporally prior learning process in which the industrial robot performs a repeating sequence of movements and the control device stores the individual positions, speeds of movement and directions of the injury-prone component of the industrial robot and preferably other components of the industrial robot.to determine the positions, speeds, and directions of movement of the component in a monitoring process following the learning process, preferably prior to real-time image acquisition or supplementing the learning process, calibration image data is acquired by capturing one or more components of the industrial robot and / or an object to be moved by the industrial robot, and to determine the position, direction of movement, speed, and, if applicable, acceleration and acceleration direction of the component of the industrial robot and / or the object, taking the calibration image data into account. According to this embodiment, the position and motion data of the industrial robot are determined from received program data of the industrial robot itself.which control the movements of the industrial robot and therefore define the position and movement of the individual components of the industrial robot at any given time, making them determinable and predictable, or which are based on image acquisition by the industrial robot, which determines the position and movement data of the components of the industrial robot by means of image acquisition and analysis and, as explained above, can optionally predict these using trained software with artificial intelligence. The image acquisition takes place in real time, parallel to the image acquisition of one or more body parts of the person, thus enabling real-time monitoring of any collisions that could cause injury. A training process can precede the image acquisition in order to use image acquisition of the industrial robot while it performs a repetitive sequence of movements,The aim is to store the position and motion data of the individual components of the industrial robot and, based on this stored data, to determine and predict the position and motion data of the individual components of the industrial robot in a monitoring process following the learning process. It is therefore particularly preferred to capture the components of the industrial robot with an image acquisition system in a calibration process preceding the monitoring, and optionally also to capture objects or tools moved by the industrial robot, thereby determining calibration data. This data enables reliable image evaluation and recognition as well as motion prediction of these components and objects during the subsequent monitoring process. In particular, such evaluation and recognition can be achieved with reduced computational effort based on the calibration data, because distinctive,During the calibration of determined geometric properties, a time-reducing detection method is possible.

[0036] It is even more preferred if the control device is designed to calculate a future position and / or direction of movement and / or speed and / or acceleration and / or direction of acceleration of the component of the industrial robot based on a position, direction of movement, speed of movement and, if applicable, acceleration and direction of acceleration of the component, and to include this in the comparison of whether contact between the detected or another body part of the user with the potentially hazardous component of the industrial robot is to be expected, and / or based on a position, direction of movement,The system uses the movement speed and, if applicable, the acceleration and direction of acceleration of the user's body part to calculate a future position and / or direction of movement and / or movement speed and / or acceleration and / or direction of acceleration of the body part and to include this information in the comparison of whether contact between the detected or another body part of the user and the potentially hazardous component of the industrial robot is to be expected. According to this embodiment, a future potential collision between the component and the body part is determined from the position and movement data of the industrial robot component and / or the user's body part, acquired in real time. At least the position and movement data for a prediction of the future position of the component or the body part, ideally both the component and the body part, are used.This enables the necessary pre-calculation of movement and positions for adequate safety measures and can be used in particular to initiate a safety measure in a timely manner, to issue a warning signal, and especially to avoid having to undertake an emergency safety measure that does not necessarily completely interrupt the work process of the industrial robot.

[0037] It is even more preferred if the control device is designed to determine the orientation of a longitudinal axis of the body part based on the image data of an image and, by comparing this with pre-stored correlation data in which the orientation of the longitudinal axis is correlated with the future direction of movement of the body part, to determine a future, expected direction of movement of the body part and to include this in the comparison as to whether contact between the detected or another body part of the user with the potentially hazardous component of the industrial robot is to be expected, and / or to determine the orientation of the component of the industrial robot and / or an object moved by the industrial robot and, by comparing this with pre-stored correlation data in which the orientation is correlated with the future direction of movement of the component and / or the object,to determine a future expected direction of movement of the component and / or object and to include this in the comparison of whether contact between the detected or another body part of the user and the potentially hazardous component of the industrial robot and / or object is to be expected. According to this embodiment, based on the determination of the orientation of a body part or a component of the industrial robot and on the basis of data that correlate such orientation in space with an expected direction of movement of the corresponding body part or component in space,A prediction of a future movement and the resulting position of the body part or component is calculated based on this determined longitudinal orientation. This embodiment therefore utilizes the probability of certain movement patterns, which often arises from the integration of such body parts and components into articulated arrangements with corresponding movement restrictions and which must or are likely to result from an orientation of the longitudinal axis, in order to make a prediction for the future position and movement data of the body part or component. It is understood that this embodiment can be used according to the invention, either alone or in addition to other calculation methods, for such a prediction, and the control device can be specifically designed to determine the prediction of the position and movement data of a body part or component using several calculation methods and probabilities.This involves weighting individual calculation methods and, if different forecasts with error probabilities occur, making the best estimate.

[0038] According to a further preferred embodiment, the detection device comprises smart glasses with an image capture device which is oriented to perform image capture in the field of vision of the wearer of the smart glasses. For the purposes of the invention, smart glasses are understood as a head mount for attachment to the user's head and therefore do not necessarily have to have a transparent area in front of the user's eyes in the sense of a lens.Such an arrangement of the image capture device on a head mount is particularly well suited for carrying out safety monitoring on an industrial robot, since it allows the numerous different perspectives resulting from the accessibility of the industrial robot from several sides and the numerous different hazard locations resulting from the high mobility of the industrial robot to be captured effectively in the area where the user interacts with the industrial robot and which the user therefore typically has in their field of vision.

[0039] The control unit can either control a single safety measure or select and control a suitable safety measure from several different options, based on the detected potential hazard. This selection of a suitable safety measure can also be based on predetermined criteria.

[0040] According to the invention, the detection device comprises a head mount for attachment to the user's head and an image capture device attached to the head mount, which is oriented to capture images in the user's field of vision while wearing the head mount. The detection device is therefore designed to detect a potential hazard by means of image capture, and the necessary image capture device is arranged on a head mount worn by the user. The image capture device is oriented such that it captures images in the user's field of vision; typically, an image capture device is attached to the head mount, oriented in the user's straight-ahead viewing direction, and has an image capture area that corresponds at least to that which a user perceives with their eyes when looking straight ahead.The image capture device can also be aligned and designed to capture a larger area than the area captured by the user in a straight-ahead view, in order to be able to determine a potential hazard state also on the basis of capturing components or body parts that pose a risk of injury and are outside the user's actual line of sight.

[0041] The head mount can be, for example, a pair of glasses to which one or two cameras are attached as image capture devices, such as on the sides of the temples. Such glasses then serve to capture images in the direction of the wearer's facial orientation. In other configurations, the head mount can also be a helmet, headband, cap, or similar device worn on the user's head, to which the image capture device is attached. This can again consist of one or two cameras, for example, two video cameras attached to the sides of the helmet or headband.In principle, the detection device designed in this way has the advantage that it performs image capture in the user's field of vision, thus replicating the user's head movements and swivels, and is therefore, on the one hand, favorably oriented towards a danger area that the user observes with their eyes, and on the other hand, allows conclusions to be drawn about the user's direction of gaze, thus also making it possible to determine if the user turns their head away from the danger area in a potentially dangerous way.

[0042] In the safety device according to the invention, the control unit is designed to assess whether a potential hazard exists due to the situation detected by the detection device, based on the images captured by the image capture device on the data glasses.

[0043] It is preferred if the control device is designed to determine a section of the captured image that corresponds to a predetermined field of view or section of the field of view, in particular a section corresponding to the fovea of ​​the field of view, of a user wearing the data glasses, and the control device is designed to determine, by means of image evaluation, whether the component is located within the image section and to take into account, as a further parameter when generating the first hazard signal, whether the component is located inside or outside the image section.

[0044] The predetermined field of view can correspond to the user's field of vision or a portion thereof. According to this embodiment, the image acquisition device or control device captures an image area that corresponds to a predetermined field of vision or a portion thereof. In particular, this can correspond to the fovea of ​​the field of vision, i.e., the area that a user has approximately in the center of their field of vision and that they can monitor with concentration. The image acquisition device is therefore focused on or encompasses precisely such an image area, which the user can also reliably perceive with their eyes. Image analysis can then determine whether the hazardous component is located within this image area.In this case, it is ensured that when the component is detected within the image area, the user also has the component within their focused field of vision, and a decision can then be made as to whether a potential hazard exists or not. However, if the component is not detected within the image area, the situation must be assessed as meaning that the user is not adequately monitoring the component, and in this case, according to this training, the control system is programmed to trigger a safety measure.

[0045] It is preferable that the smart glasses have an eye-tracking device, and that the image area is determined based on the user's gaze direction as determined by the eye tracking. Eye tracking can also determine the direction in which the user's eye is looking, i.e., whether the user is looking to the side or not straight ahead. This allows it to be determined whether the user, although their head is positioned so that their central, straight-ahead gaze is directed towards the hazardous component, deviates from this main axis due to their gaze angle and therefore does not have the component in their line of sight. This, in turn, would indicate a potential hazard and trigger a safety measure.

[0046] It is even more preferred if the control unit is arranged on the head mount and the head mount also includes a signal transmission unit for wireless signal transmission and is signal-technically coupled to the safety device by means of this signal transmission unit. With such a signal transmission unit, the head mount can be worn comfortably by the user, and the data determined by the control unit on the head mount, and in particular the data content for triggering a safety measure, can be transmitted to the safety device.

[0047] Alternatively, the control unit is arranged on the industrial robot, and the head mount also includes a signal transmission unit for wireless signal transmission, which is coupled to the control unit via this signal transmission unit. According to this embodiment, the control unit is not arranged on the head mount but on the industrial robot, and in this case, the data acquired by the image acquisition device on the head mount is transmitted to the control unit via the signal transmission unit.

[0048] In addition to these two possibilities for arranging the control device on the head mount or on the industrial robot, another design is also advantageous in certain embodiments, in which the control device comprises a first control unit arranged on the head mount and a second control unit arranged on the industrial robot, and the head mount further comprises a signal transmission unit for wireless signal transmission, and the first and second control units are coupled signal-technically by means of this signal transmission unit, wherein the first control unit is designed to perform a first part of the evaluation and to transmit a result obtained therefrom to the second control unit, and the second control unit is designedto perform a second part of the assessment based on this determined result and, depending on the assessment, to control the safety device to execute a safety measure. Here, the assessment of the situation captured by the image acquisition device for a potential hazard state is carried out by a first control unit on a head mount and a second control unit on the industrial robot. Such a division of the assessment allows parts of the assessment to be carried out on the head mount that reduce the amount of data to be transmitted, thus enabling an efficient real-time assessment.

[0049] It is even more preferred if the head mount also includes an image output device for outputting image and text information to one or both eyes of the user wearing the head mount, in particular by means of

[0050] Display of image and text information on a screen, or

[0051] Projection of image and text information onto the retina of one or both of the user's eyes.

[0052] According to this training method, the head mount is also equipped to display an image that is seen by the user, for example, via a screen attached to the head mount or by projecting an image onto the retina of both the user's eyes. This allows, on the one hand, simple or differentiated warning signals to be displayed to the user via the head mount. On the other hand, additional information unrelated to a hazardous situation can be shown to the user via the image output device.It is particularly preferred if the head mount has a data interface for data transmission, in particular a data transmission unit for wireless data transmission, and is designed to receive image and / or text information via the data interface and to display it on the image output device, which includes a sequence of predetermined work processes for machining a workpiece, a representation of predetermined orientations and / or movements of a tool or a workpiece for carrying out a machining step, and a designation of machined workpieces before and / or after machining, wherein preferably the image or text information is displayed based on image acquisition with the image acquisition device and an image evaluation carried out therefrom to determine the position of the tool or the workpiece in an area that is assigned to the tool or the workpiece in the user's field of vision.This embodiment also uses the head mount for the targeted, location-specific output of information to the user that is helpful for processing a workpiece. For example, a workpiece requiring multiple processing steps with the industrial robot can be shown to the user via the image output device on the head mount with regard to the necessary orientations and movements of the workpiece or tool. A sequence of the processing steps to be performed successively can also be displayed as predetermined workflows. Furthermore, workpieces produced in this way can be labeled accordingly via the image output device on the head mount, for example, as finished parts that the user should place, as intermediate parts that still require further processing, or as offcuts.

[0053] This information can be presented visually or textually, preferably with the image or text information being displayed to the user in such a way that they see it within the respective workpiece, similar to augmented reality, and can therefore clearly identify it for processing. It should be understood that, in an alternative embodiment, these spatially resolved display functions of the image output device on the head mount can also be implemented without the image capture device located on the head mount.According to a further preferred embodiment, the control device is coupled to an electronic storage device via a signal connection and continuously transmits the captured image data to the electronic storage device. The electronic storage device is configured to store the image data and, upon reaching its capacity limit, overwrites the oldest image data with the most recently transmitted image data. According to this improved embodiment, the image data is stored in the storage device, thereby electronically preserving image data for a specific period so that it is accessible for later evaluation. The storage period can be one hour, several hours, or even one or more days or weeks.This approach allows the stored image data to be used for further training of artificial intelligence, thereby increasing the accuracy of the analysis and reducing the probability of errors. Furthermore, the image data enables an assessment of security measures or security incidents that occurred within the period in which the image data was stored.

[0054] It is even more preferred if the control device is designed to distinguish between a low and a higher hazard level and to execute a first safety measure, in particular a warning signal, at the low hazard level, and to execute a second safety measure, different from the first, at the higher hazard level. According to this embodiment, at least two different hazard levels are differentiated and trigger different safety measures.The two or more different hazard levels can differ from each other by different properties, for example, by different distances between body part and component, by different approach speeds, by different predicted time periods until contact between body part and component, by different injury sensitivities of individual body parts stored in the control unit, or by different injury risks from different components of the industrial robot stored in the control unit.The differentiated triggering of various safety measures helps to interrupt the operation of the industrial robot only in genuinely serious hazardous situations, while at the same time enabling preventive safety measures to be triggered in the event of minor hazards. These measures help to avoid the occurrence of a serious hazard by warning of its impending occurrence and thus allowing the user to react in a way that prevents the occurrence of a serious hazard.

[0055] It is even more preferred if the image acquisition device is designed to capture images with depth information, in particular by designing the image acquisition device to capture runtime information about image content, or by designing the image acquisition device to determine depth information from two temporally offset images with different recording directions by means of an image evaluation of the position of detected objects relative to each other, or by designing the image acquisition device to comprise a first and a second image acquisition unit which are arranged at a distance from each other, for example attached to the head support on the left and right sides of the user's head, and is designed to determine depth information from a stereoscopic image evaluation of simultaneously captured images of the first and second image evaluation unit.

[0056] According to this embodiment, the image acquisition device is designed to capture depth information about the objects captured in an image, thereby enabling the determination of the distance of these objects from the image acquisition device. This depth information can be obtained point by point for each captured pixel or object by object. In principle, various configurations for such depth information acquisition are possible, for example, based on the time-of-flight information of the signal from an object to the image acquisition device, or by capturing two images with a single image acquisition device at different times and determining depth information by comparing objects captured in both images that have different positions due to a movement of the image acquisition device between the two time-shifted images.Furthermore, stereoscopic depth information can also be determined if two spaced-apart image acquisition units are arranged on the head mount, which have at least partially overlapping image acquisition areas and form the image acquisition device. Based on an object that is simultaneously captured by both image acquisition units, depth information can then be calculated stereoscopically from the angular position of the object in each of the two image acquisition areas.

[0057] Another aspect of the invention is an industrial robot that includes a safety device of the type described above.

[0058] Another aspect of the invention is a pair of smart glasses for a safety device of the type described above, which is characterized by an image capture device that is oriented to perform image capture in the direction of the wearer of the smart glasses' gaze, and preferably further comprising a control device that is coupled to the image capture device via a signal and is configured to perform an evaluation of a potential hazard state detected by the image capture device based on predetermined criteria, wherein the smart glasses preferably have a data interface, in particular a wireless data transmission device, and are configured to control the safety measure device to execute a safety measure depending on the evaluation via the data interface.

[0059] The data glasses can be further trained, in particular according to the features of the previously described data glasses.

[0060] Another aspect of the invention relates to a method for preventing injuries on machine tools, such as industrial robots, comprising

[0061] • Teaching a control device by reading in data in which several different situations in which contact between a body part of a user and a component of the machine tool is to be expected is depicted in image data and

[0062] • Assigning situations to hazardous situations,

[0063] • Recognizing correlations between image data and the associated hazard situation using a neural network, and

[0064] Storing these relationships in the control device, whereby • the learning status of the control device is evaluated after passing through positions a) to c) and compared with a predetermined triggering safety level, which describes a probability with which the control device triggers a safety measure when hazardous situations occur, and

[0065] • where, when the comparison shows that the learning status has reached the predetermined trip safety level, the saving of the relationships in the control unit is completed and subsequently the control unit is operated with this learning status without further changes to the learning status.

[0066] According to this aspect of the invention, the control unit of the safety device according to the invention is programmed by a learning process that uses artificial intelligence to assign image data acquired with the detection device, depicting different situations in which a body part is endangered by a component of the industrial robot, to a specific hazard situation. The control unit thereby acquires a programmed learning state that can be improved by various influences, e.g., by reading in more image data, reading in further data describing the different situations, reading in data and image data describing further situations, and defining the correlation between this read-in data, image data, and situations to hazard situations.This continuously improved learning status is compared with a trip safety level, which describes whether there is already a sufficient probability for operational safety that the control device will correctly and promptly detect a hazardous situation in order to prevent injury through a safety measure. The trip safety level can also include further criteria, such as a probability of error with which the control device detects a hazardous situation even though none exists. According to the invention, a learning status that has reached or exceeded the predetermined trip safety level is then stored and serves to control the control device.It is crucial that this initial learning status of the control unit is not further altered during the operation of the industrial robot, and in particular, not modified by additional data acquired during operation. This ensures that the positively established operational safety of the control unit is not altered by subsequent modifications, such as those commonly used in the context of artificial intelligence, and thus potentially adversely affected, thereby compromising the system's safety.

[0067] The programming and installation of such a control unit for a safety device is regularly carried out with the aim of obtaining product approval or acceptance, i.e., having the control unit approved by a test regulated by laws, ordinances, or provisions, and thereby establishing and documenting consumer safety. For this approval process, it is advantageous if the control unit exhibits immutability with regard to its safety-relevant functions. However, to achieve a high level of safety, it is advantageous if artificial intelligence is used in a progressive and iterative learning process for programming the control unit. This is achieved by the method according to the invention.The method according to the invention does not preclude the possibility that, in a subsequent step outside of the control device's operation, the control device is re-learned in a kind of update procedure, and that a new learning state is then frozen and saved, which forms the basis for the operation of the control device. However, the crucial aspect of the method according to the invention is that it serves only to achieve a learning state, which is then saved, and that this learning state is no longer changed during the operational use of the control device by the consumer.

[0068] Another aspect of the invention is a method for preventing injuries to industrial robots, comprising:

[0069] • Detecting a potential hazard to a user's body part caused by a potentially hazardous component of the industrial robot using a detection device,

[0070] • Implementing a safety measure by which a hazardous condition for the user of the machine tool is reduced or avoided by means of a safety device,

[0071] • Evaluating a potential hazard state detected by the detection device based on predetermined criteria using a control device that is signal-technically coupled to the detection device and the safety measure device, and controlling the safety measure device to execute a safety measure depending on the evaluation, in which

[0072] • the detection of the potential hazard state is carried out by means of an image capture device, in particular an image capture device arranged on data glasses, and

[0073] • The evaluation is carried out using images captured by the image capture device of the data glasses, with the following steps: o Detecting a body part of the user in each image of the sequence of images based on the image data, o Determining the position of the detected body part within a spatial reference system, o Calculating the direction and speed of movement of the detected body part by comparing the respective position of the detected body part in two or more consecutive images, o Determining the position, direction, and speed of movement of a potentially hazardous component of the industrial robot, o Calculating whether contact between the detected or another body part of the user and the potentially hazardous component of the industrial robot is to be expected within a predetermined time period,by comparing the determined position and the calculated direction and speed of movement of the detected body part with the position, direction and speed of movement of the hazard-causing component of the industrial robot to, and o sending a first hazard signal to the safety device if contact is expected, o executing a first safety measure by the safety device upon receipt of the first hazard signal.

[0074] This aspect of the invention relates to a method for preventing injuries to industrial robots, which can be implemented in particular with the safety device described above, a correspondingly equipped industrial robot and / or correspondingly equipped smart glasses, and which in particular performs an assessment of the potential hazardous state by means of a control device that has been programmed with the previously described methods for preventing injuries to industrial robots.It is understood that the aspects and advantages previously described with regard to the safety device, the industrial robot, the smart glasses, and the injury prevention procedure are applicable and usable for this procedure accordingly, and that the procedure is further developed with process steps that correspond to the device and process characteristics of the previously described safety device, the industrial robot, the smart glasses, and the method for teaching the control unit. In particular, the assessment of whether a hazardous condition exists that would trigger a safety measure can be carried out by determining the severity of the expected contact, and an appropriate safety response can thus be achieved, as previously described.

[0075] Preferred embodiments of the invention are described in more detail with reference to the accompanying figures.

[0076] Preferred embodiments of the invention are described in more detail with reference to the accompanying figures. These show:

[0077] Fig. 1 shows a schematic perspective view of a first embodiment of a safety device according to the invention for an industrial robot.

[0078] Fig 2 shows a schematic, perspective view of a second embodiment of a safety device for an industrial robot.

[0079] Figure 3 shows a look-up table for assessing the severity of contact between body parts and components of the industrial robot. Referring first to Figures 1 and 2, an exemplary industrial robot has mobility about four axes 11, 21, 31, and 41, which connect a base 10 with an intermediate arm 20, an end arm 30, and a gripper 40. The industrial robot can thus move and pivot the gripper 40, which is attached to the end of the end arm 30, in three-dimensional space and perform a gripping movement with the gripper 40.

[0080] In the embodiment shown in Figure 1, a plurality of video cameras 50a-d are arranged on a frame 50 positioned above the industrial robot and are connected to a control unit 60 via signal lines. The video cameras 51a-d are arranged at the four corners of the square frame 50 and each monitors a cone-shaped area below the frame. The combined areas cover the range of motion of the industrial robot. The individual areas monitored by the video cameras 51a-d partially overlap and therefore provide redundancy for image recognition in certain parts of the space covered by the industrial robot.

[0081] The industrial robot has markings 20a and 30a on its intermediate and end arms, respectively. These markings are implemented as QR codes and are detected by the video cameras 51ad. The QR codes can be recognized and evaluated from the image data of the video cameras by the control unit 60, thereby enabling a unique assignment of the position of the components of the industrial robot.

[0082] Markings 70a-d, designed as tetrahedrons, are arranged in the floor area around the industrial robot and are also detected by the video cameras. This allows for the reliable assignment of a user's position 100 in relation to the industrial robot and the position of its components.

[0083] The control device 60 is connected to a controller 80 of the industrial robot by means of a signal connection 61, which in the embodiment shown in Figure 1 is wired.

[0084] The video cameras 51 ad continuously monitor the room in which the industrial robot moves. If a person 100 enters this room and the control unit 60 detects, based on the evaluation of the image data, that contact between a body part and a component is imminent due to the position, speed, direction, and acceleration of the individual components of the industrial robot and the user's body parts, the severity of the contact is determined based on the movement data of the user and the industrial robot. This severity is evaluated using a table shown in Figure 3 and, in the embodiment shown in Figure 3, can be differentiated according to contact on the user's head, on an arm of the user, or on the torso of the user, as symbolically represented.The component of the industrial robot is further differentiated, as it can be one of the two arms of the industrial robot, the gripping element located at the end of the arm, or a milling head that can be mounted in place of the gripping element. It should be understood that this differentiation can include further elements, subdivisions of the body parts and components.

[0085] The contact between a body part and a component can be classified into severity levels A and B, although further subdivisions are possible. In cases of mild severity level A, safety measures I, II, III, or IV are implemented depending on the affected body part and component; the same applies to severe severity level B contact.

[0086] As can be seen in Figure 3, different safety measures are taken depending on the body part involved, the component involved, and the severity of the incident. For example, in the case of contact with the user's head, safety measure III must always be implemented in the case of a minor incident and IV in the case of a major incident to reliably prevent injuries. If the milling head comes into contact with any part of the user's body, a correspondingly effective safety measure IV must always be implemented to prevent any injuries from this sharp-edged component. In other cases, particularly when contact is intended to be permitted, such as contact between the hand and one of the arms of the industrial robot, only a minor safety measure I is implemented.

[0087] Safety measure I can, for example, consist of a warning tone, safety measure IV can consist of a complete stop of all movements of the industrial robot, and safety measure II or III can, for example, consist of a slight or significant slowing down of the movement of the industrial robot.

[0088] Figure 2 shows a second embodiment of the safety device according to the invention. In this embodiment, the industrial robot is designed in the same way as in the embodiment according to Figure 1. In contrast to the embodiment according to Figure 1, this embodiment includes an image acquisition device that is arranged on smart glasses worn by the user and that captures a monitoring area that includes and extends beyond the user's field of vision. The image acquisition device comprises two video cameras 151 a, b arranged laterally on the temples of the smart glasses, which capture and continuously monitor a stereoscopic image of the surroundings in the user's field of vision. A control unit 160 is also arranged on the headband of the smart glasses and is connected to the controller 80 of the industrial robot by means of a wireless signal connection.

[0089] In the same manner as previously described, the two video cameras 151 a,b of the image acquisition device monitor the movement area of ​​the industrial robot when the person is working within its area and the image acquisition is therefore directed at the industrial robot. The control unit can determine the position, speeds, and accelerations of these components in space and relative to the user's body parts, which are also within the field of view of the image acquisition device, based on the markings 20a, 30a arranged on the components of the industrial robot and the floor markings 70a-d. In the same manner as previously described, if a risk of contact between a body part and a component is detected, safety measures can be taken, which, again in the same manner as previously described, can be selected from a table according to body part, component, and severity.

[0090] It is generally understood that the image acquisition device arrangements shown in Figures 1 and 2, on the one hand on a frame positioned above the industrial robot and on the other hand on a user's head support, can also be combined, and a corresponding signal connection can be established between the video cameras of these image acquisition devices and the control unit 60 or 160, or both. In this way, a monitoring area covered by both stationary video cameras and video cameras moving with the user can be used for safety and injury prevention.

Claims

Claims 1. Safety device for an industrial robot, comprising a detection device configured to detect a potential hazard to a body part of a person located near the industrial robot, caused by a component of the industrial robot or an object moved by the industrial robot; a safety action device configured to execute a safety action by which the hazard is reduced or avoided; a control device that is signal-connected to the detection device and the safety action device and configured to perform an evaluation of a potential hazard detected by the detection device based on predetermined criteria and, depending on the evaluation, to control the safety action device to execute a first safety action, characterized in thatthat the detection device comprises an image capture device configured to perform image capture in a hazard zone encompassing the industrial robot and a spatial section surrounding the industrial robot, and the control device is configured to perform the evaluation based on a temporally successive sequence of images captured by the image capture device with image data, wherein the control device is programmed to recognize a body part of the user in each image of the sequence of images based on the image data, to determine the position of the recognized body part within a spatial reference system, to calculate the direction and speed of movement of the recognized body part by comparing the respective position of the recognized body part in two or more successive images, and to determine the position,To determine the direction and speed of movement of a potentially hazardous component of the industrial robot and, if applicable, of an object moved by the industrial robot, by comparing the determined position and the calculated direction and speed of movement of the detected body part with the position, direction and speed of movement of the hazardous component of the industrial robot, to calculate whether contact between the detected or another body part of the user with the hazardous component of the industrial robot is to be expected within a predetermined or calculated time period, and if so, to generate a first hazard signal and send it to the safety action device, wherein the safety action device is designed to execute the first safety action upon receipt of the first hazard signal.

2. Safety device according to claim 1, characterized in that the control device is programmed to compare the calculated severity of the expected contact with a predetermined first limit value based on the calculated direction and speed of movement and, if applicable, the direction and acceleration of the detected body part and the direction and speed of movement and, if applicable, the direction and acceleration of the component of the industrial robot that poses a risk of injury, wherein the severity of the contact describes at least one level of the expected acceleration of the body part by the contact, and to generate the first hazard signal if the calculated severity of the expected contact exceeds the first limit value.

3. Safety device according to claim 2, characterized in that the control device is programmed to compare the calculated severity of the expected contact with several predetermined limit values, and if the calculated severity of the expected contact exceeds a first limit value of the several predetermined limit values, to generate the first hazard signal and send it to the safety action device. o if the calculated severity of the expected contact exceeds a second limit value of several predetermined limit values, to generate a second hazard signal and send it to the safety action device, wherein the safety action device is configured to execute a second safety action, different from the first safety action, upon receiving the second hazard signal.

4. Safety device according to claim 3, characterized in that the control device is programmed to compare the calculated severity of the expected contact for a first body part with the first limit value, and to compare the calculated severity of the expected contact for a second body part with a second limit value, and: if the calculated severity of the expected contact of the first body part exceeds the first limit value, to generate the first hazard signal and send it to the safety action device; if the calculated severity of the expected contact of the second body part exceeds the second limit value, to generate the first hazard signal and send it to the safety action device, wherein the control device is further programmed to then, if the calculated severity of the expected contact of the second body part lies between the first and the second limit values,to avoid generating the first danger signal.

5. Safety device according to claim 4, characterized in that the control device is programmed to compare the calculated severity of the expected contact between the first body part and the first injury-prone component with the first limit value, and to compare the calculated severity of the expected contact between the first body part and a second potentially harmful component with the first limit value, and o if the calculated severity of the expected contact of the first body part with the first potentially harmful component exceeds the first limit value, to generate the first hazard signal and send it to the safety device, o if the calculated severity of the expected contact of the first body part with the second potentially harmful component exceeds the first limit value, not to generate the first hazard signal.

6. Safety device according to one of the preceding claims, characterized in that the control device is programmed to detect a second body part of the person, different from the first, and to calculate its position, speed of movement and direction of movement and, if applicable, acceleration and direction of acceleration.

7. Safety device according to one of the preceding claims, characterized by stationary markings in the movement area of ​​the industrial robot that can be detected optically by the image acquisition device, in particular markings formed on a floor mounting surface, and characterized in that the control device is designed to recognize the markings and to enter the position of the body part and / or the industrial robot in a stationary coordinate system based on the markings and to evaluate it based on the stationary coordinate system.

8. Safety device according to one of the preceding claims, characterized in that the control device is further programmed to calculate an acceleration direction and an acceleration of the detected body part by comparing the respective position of the detected body part in three or more successive images, to determine the direction of acceleration and the acceleration of the component of the industrial robot that poses a risk of injury, and to calculate whether contact between the detected or the other body part of the user with the component of the industrial robot that poses a risk of injury is to be expected within the predetermined time period, also based on the direction of acceleration and the acceleration of the detected body part and the direction of acceleration and the acceleration of the component of the industrial robot that poses a risk of injury.

9. Safety device according to one of the preceding claims, characterized in that the control device is designed to detect an operational input variable of the industrial robot on the basis of the image data of an image, in particular • the provision of an object for the robot to grasp, • the provision of a storage location for an object grasped by the robot, and that the control device has an interface to a control device of the industrial robot and is designed to transmit the detected operational input variable via this interface.

10. Safety device according to one of the preceding claims, characterized in that the first safety measure and optionally the second safety measure is selected from a reduction of the movement speed of the industrial robot, an emergency stop of the movement of the industrial robot, an emergency stop of the movement of the industrial robot and a subsequent control of the industrial robot into a passive mobility state in which actuators of the industrial robot are controlled in such a way that weight forces through the components of the robot and optionally through an object held by the robot are compensated and the components of the robot are movable by the action of external forces, a change in the movement of the component of the industrial robot into an evasive movement, wherein the control device is programmed to calculate, based on the position, the calculated direction and speed of movement and, if applicable, the direction and acceleration of the detected body part and the position, direction and speed of movement and, if applicable, the direction and acceleration of the component of the industrial robot that poses a risk of injury, the direction and preferably the speed of movement of the evasive movement in such a way that contact between the body part and the component of the industrial robot that poses a risk of injury is avoided by the evasive movement.

11. Safety device according to one of the preceding claims, characterized in that the control device is designed to determine the position, speed and / or acceleration of the component of the industrial robot on the basis of data processing with artificial intelligence used in the evaluation of the image data, or a marking on the component of the industrial robot that can be detected in the evaluation of the image data, or an input from the operator, or position data received from a transmitting device on the component.

12. Safety device according to one of the preceding claims, characterized in that the control device has pre-stored anatomical data and is programmed to calculate, based on the position, speed of movement and direction of movement of the detected first body part and an assignment of the first body part to a second body part different from the first, a position, speed of movement and direction of movement of the second or a third body part of the person defined in the anatomical data, and / or the control device is configured to generate calibration image data by capturing one or more body parts or to determine the entire person and to carry out the evaluation based on the temporal sequence of the images captured by the image acquisition device, in particular the calculation of the direction and speed of movement of a body part taking into account the calibration image data.

13. Safety device according to one of the preceding claims, characterized in that the control device is configured to determine the position, direction of movement, speed of movement and, if applicable, acceleration and direction of acceleration of the component of the industrial robot and / or an object moved by the industrial robot from received program data of the industrial robot, and / or from an image acquisition of the industrial robot by means of the image acquisition device, wherein o the image acquisition takes place in real time parallel to the image acquisition of the first body part of the person or o the image acquisition takes place in a temporally prior learning process in which the industrial robot performs a repeating sequence of movements and the control device determines the individual positions,to store the movement speeds and directions of the hazardous component of the industrial robot and preferably other components of the industrial robot in order to determine the positions, movement speeds, and movement directions of the component in a monitoring process following the learning process, wherein preferably before the real-time image acquisition or in addition to the learning process, calibration image acquisition is performed to determine calibration image data by capturing one or more components of the industrial robot and / or an object to be moved by the industrial robot, and to determine the position, direction of movement, movement speed, and, if applicable, acceleration and acceleration direction of the component of the industrial robot and / or the object, taking the calibration image data into account.

14. Safety device according to one of the preceding claims, characterized in that the control device is designed to • to calculate a future position and / or direction of movement and / or speed and / or acceleration and / or direction of acceleration of the component of the industrial robot based on a position, direction of movement, speed and, if applicable, acceleration and direction of acceleration of the component, and to include this in the comparison of whether contact between the detected or another body part of the user with the potentially hazardous component of the industrial robot is to be expected, and / or • to calculate a future position and / or direction of movement and / or speed and / or acceleration and / or direction of acceleration of the user's body part based on a position, direction of movement, speed of movement and, if applicable, acceleration and direction of acceleration of the body part, and to include this in the comparison of whether contact between the detected or another body part of the user with the component of the industrial robot that poses a risk of injury is to be expected.

15. Safety device according to one of the preceding claims, characterized in that the control device is designed to use the image data of an image • to determine the orientation of a longitudinal axis of the body part and, based on a comparison with pre-stored correlation data in which the orientation of the longitudinal axis is correlated with the future direction of movement of the body part, to determine a future, expected direction of movement of the body part and to include this in the comparison of whether contact between the detected or another body part of the user with the potentially hazardous component of the industrial robot is to be expected, and / or • to determine the orientation of the component of the industrial robot and / or an object moved by the industrial robot and, based on a comparison with pre-stored correlation data in which the orientation is correlated with the future direction of movement of the component and / or the object, to predict a future direction of movement of the component and / or the object to determine and include in the comparison whether contact between the detected or another part of the user's body with the potentially hazardous component of the industrial robot and / or object is to be expected.

16. Safety device according to one of the preceding claims, characterized in that the detection device comprises data glasses with an image capture device which is aligned to perform image capture in the field of vision of the wearer of the data glasses.

17. Safety device according to claim 16, characterized in that the control device is configured to determine a section of the captured image that corresponds to a predetermined field of vision or section of the field of vision, in particular a section corresponding to the fovea of ​​the field of vision, of a user wearing the data glasses, and the control device is configured to determine, by means of image evaluation, whether the component is located within the image section and to take into account, as a further parameter when generating the first hazard signal, whether the component is located inside or outside the image section.

18. Safety device according to claim 16 or 17, characterized in that the data glasses have an eye tracking device and that the image section is determined depending on the user's gaze direction as determined by means of eye tracking.

19. Safety device according to one of the preceding claims 16-18, characterized in that the control device is arranged on the data glasses and the data glasses further comprise a signal transmission unit for wireless signal transmission and are coupled to the safety measure device for real-time data transmission by means of this signal transmission unit.

20. Safety device according to one of the preceding claims 16-18, characterized in that the control device is arranged on the machine tool and the data glasses further constitute a signal transmission unit for wireless signal transmission and is coupled to the control device via this signal transmission unit.

21. Safety device according to one of the preceding claims 16-20, characterized in that the control device comprises a first control unit arranged on the data glasses and a second control unit arranged on the machine tool, and the data glasses further comprise a signal transmission unit for wireless signal transmission and the first and second control units are coupled by means of this signal transmission unit, wherein the first control unit is configured to perform a first part of the evaluation and to transmit a result obtained therefrom to the second control unit, and the second control unit is configured to perform a second part of the evaluation on the basis of this result obtained and to control the safety measure device to execute a safety measure depending on the evaluation.

22. Safety device according to one of the preceding claims 16-21, characterized in that the data glasses further comprise an image output device for outputting image and text information to one or both eyes of the user wearing the data glasses, in particular by means of Display of image and text information on a screen, or Projection of image and text information onto the retina of one or both of the user's eyes.

23. Safety device according to the preceding claim, characterized in that the data glasses have a data interface for data transmission, in particular a data transmission unit for wireless data transmission - M - The transmission has and is designed to receive image and / or text information via the data interface and to reproduce it on the image output device, which A sequence of predetermined work processes for handling a workpiece by the industrial robot, A representation of predetermined orientations and / or movements of a tool or workpiece during the handling of the workpiece by the industrial robot. A designation of the workpieces handled by the industrial robot includes, preferably, the image or text information based on image acquisition with the image acquisition device and an image evaluation carried out therefrom to determine the position of the component of the industrial robot or the workpiece in an area that is assigned to the component of the industrial robot or the workpiece in the user's field of vision.

24. Safety device according to one of the preceding claims, characterized in that the control device is coupled to an electronic storage device via signal technology and continuously transmits the captured image data to the electronic storage device, which is configured to store the image data and, upon reaching a capacity limit of the storage device, to overwrite the oldest image data with the most recently transmitted image data.

25. Safety device according to one of the preceding claims, characterized in that the control device is designed to distinguish between a low and a higher hazard level and to execute a first safety measure, in particular a warning signal, at the low hazard level and to execute a second safety measure, which differs from the first safety measure, at the higher hazard level.

26. Safety device according to one of the preceding claims, characterized in that the image capture device is configured to capture images with depth information, in particular by the image acquisition device is designed to capture runtime information about image content, or the image acquisition device is designed to determine depth information from two temporally offset images with different recording directions by means of an image evaluation of the position of recognized objects relative to each other, or by the image acquisition device comprising a first and a second image acquisition unit which are arranged at a distance from each other, for example attached to the left and right temple of a pair of data glasses, and is designed to determine depth information from a stereoscopic image evaluation of simultaneously captured images of the first and second image evaluation unit.

27. Industrial robot comprising a safety device according to any of the preceding claims.

28. Data glasses for a safety device according to one of the preceding claims, characterized by an image capture device which is oriented to perform image capture in the direction of view of the wearer of the data glasses, and preferably further comprising a control device which is coupled to the image capture device in terms of signal technology and is configured to perform an evaluation of a potential hazard state detected by the image capture device based on predetermined criteria, wherein the data glasses further preferably have a data interface, in particular a wireless data transmission device, and are configured to control the safety measure device to execute a safety measure depending on the evaluation via the data interface.

29. Data glasses according to claim 28, characterized in that the data glasses are further developed according to the features of the data glasses according to the safety device according to one of the preceding claims 16-23.

30. Method for injury prevention on machine tools, such as industrial robots, comprising: a) teaching a control device by reading data in which several different situations in which contact between a body part of a user and a component of the machine tool is to be expected is depicted in image data; b) assigning the situations to hazardous situations; c) recognizing correlations between the image data and the assigned hazardous situation by means of a neural network, and storing these correlations in the control device; characterized in that the teaching status of the control device is evaluated after passing through positions a) to c) and compared with a predetermined trigger safety level, which describes a probability with which the control device triggers a safety measure when hazardous situations occur; and that then,If the comparison shows that the learning status has reached the predetermined trip safety level, the saving of the relationships in the control unit is completed, and subsequently the control unit is operated with this learning status without further changes to the learning status.

31. Methods for preventing injuries on industrial robots, comprehensive Detecting a potential hazard to a user's body part caused by a potentially hazardous component of the industrial robot using a detection device, Implementing a safety measure by which a hazardous condition for the user of the machine tool is reduced or avoided by means of a safety device, Evaluating a potential hazard state detected by the detection device based on predetermined criteria using a control device that is coupled to the detection device and the safety measure device via signal technology, and Controlling the safety device to execute a safety measure depending on the assessment, characterized in that the detection of the potential hazard state is carried out by means of an image acquisition device, in particular an image acquisition device arranged on smart glasses, and the assessment is carried out on the basis of images captured by the image acquisition device of the smart glasses, with the steps: Recognizing a body part of the user in each image of the sequence of images based on the image data, Determining the position of the detected body part within a spatial reference system, Calculating the direction and speed of movement of the detected body part by comparing the respective position of the detected body part in two or more consecutive images, Determining the position, direction of movement and speed of movement of a potentially hazardous component of the industrial robot, Calculate whether contact between the detected or another body part of the user and the potentially hazardous component of the industrial robot is to be expected within a predetermined time period, by comparing the determined position and the calculated direction and speed of movement of the detected body part with the position, direction and speed of movement of the potentially hazardous component of the industrial robot. Sending an initial hazard signal to the safety device when contact is expected, Execution of a first safety measure by the safety device upon receipt of the first hazard signal.

Images