FI103031B - Method for controlling the crane - Google Patents

Description

103031

Uppfinningen avser förfarande och anord-Ning för stränggjutning av gjutformat tned ungefärl slutmätt för vidare kallbear-betning, blazing gjutformat & av icke-järn-metaller, speciellt band, stänger eller rör av koppar eller kopparlegeringar. För-farandet uppvisar feljande steg: (a) smäl-tan gjuts continuerligt i en vertikalko-kill, (b) den stelnade strängen avförs cen-triskt styrd och träs direkt in i ytbe-handlingsmaskin, (c) strängytan avdras, d) därpä avskiljs den account ytan färdiga strängen pä tvären och pä förhand bestämt provision, och e) slutligen uttas de mättrik-Tiga strängstyckena ur anläggningen. För genomföring av förfarandet har anordningen en med ett fundament (51) förbunden Grund-stomme (21), som efter varandra har en ut-dragningsanordning (22), en ytbehandlings-anordning (23), en tväravskiljningsanord-Ning (24) och en transportanordning (25 eller 41).

103031

METHOD FOR CONTROLLING THE CRANE

The invention relates to a method for controlling crane-dependent load fluctuations and controlling the crane based on the information obtained.

The ropes on the crane move the load, placing a burden on the pendulum movement, which makes working difficult. The oscillations have been taken into account using, for example, a pacing device, whereby every change in acceleration produced is followed by another equal change, but in a different direction after a certain time. In order to dampen the oscillation, an optimum velocity profile is calculated for the transmission, so that there is no oscillation at the end of the movement and the time spent on movement is minimized. Thus, in prior art solutions, the load oscillation equation is determined by the calculated quantities. The data can be used to control the fluctuation of the burden. In fact, the control of the oscillation follows a calculated, assumed formula. No Reaa-20 There is no glare control. With current systems dampening the load oscillation according to different counter-forces, the target point may drift away from a particular point and repeating the same stopping event at the same known target point may be quite the time for the theo-25.

♦

In addition, the damping of the load oscillation also works with the lifting pillar or other structure preventing the free swinging of the load and the gripper. With a small load, the robot's lifting column is assumed to be rigid. As the mass to be moved, the columns also have load-bearing structures but do not follow the mathematical formula of harmonic oscillation because * the load-bearing structures work like springs. In such a case, arranging the damping of the oscillation with a mathematical model of 35 would require empirical tests, since the spring constants of the structures and the like vary, for example, in the bridge structure depending on the proximity of the bearing end support of the bridge. The invention also controls the above-described conditions. Even if the hoist has a mechanical load-shake suppressor, which in itself most often improves load control, machine vision can still improve the operation of this mechanical damper, and in some cases, the mechanical suppressor could be less than the original design with the machine-vision load-shielding.

U.S. Patent 3,826,380 describes an arrangement for detecting a deviation from a vertical line. The patent describes a device for emitting a beam of light and a reflector device supported by the gripper. The reflector device is also up the path of reflected light.

EP 5963330 describes a damping of cargo swing. Here, the camera is directed to certain markings on the chassis or cargo to control cargo swing.

SE-A-502609 discloses a camcorder system having a field of view for viewing different sized tag positions with light emitting diodes. When moving the gripper, either small or large markings, depending on the length of the rope, illuminate. Only butter-25 bright rays from the LED are detected. The SE publication examines the specific, readily detectable signaling associated with a cargo hold device, but does not indicate how the cargo is positioned, where the cargo should be placed, but only how to control the fluctuation.

JP 7309582 describes, like SE-publication, camera-controlled swinging of a load securing member relative to a bridge.

35

The invention is specifically aimed at real-time detection. To achieve this, the invention is characterized by; as set forth in the characterizing part of the claim.

3, 103031

According to the invention, the load fluctuation is monitored almost in real time. Avoid calculating different oscillation equations or the like. Observation knows where the burden is at any given time. Attenuation of load fluctuation seeks to reach a new dimension / utilization perspective.

The invention will now be described with reference to the accompanying drawings, in which Figure 1 illustrates a principle view of transferring a load from one location to another.

Figure 2 shows the determination of the crane height with a laser beam. Figure 3 shows a method for defining an image area.

Figure 4 illustrates a method for detecting a transition in an image area.

Figure 5 shows a laser light source located in the transfer carriage.

Figure 6 illustrates the identification by means of the product key.

Figure 7 further illustrates specific product identification.

20 Figure 8 shows a schematic view of a data management system. Figure 9 shows a flow diagram of the damping of the swing from detection to crane or robot control.

The bridge crane shown in Fig. 1 has a bridge 3 and a horizontal lifting carriage 4 on top of it 25. The lifting carriage 4 is moved in the direction of the horizontal movement of the carriage, which causes oscillation. The drive mechanism of the wagon consists of an electric motor, an electric-controlled brake, and a suitable transmission mechanism, not shown. The wagon has a lifting mechanism which controls the movement of the rope. When examining the transfer of one container of a normal hand pallet truck from stack A to ship's hold C as shown in Figure 1 5, the operator must execute several repair and forward movement commands. Depending on the structure and 'goodness' of the control system, there will be a delay of 20-500 ms between the start of the wagon issued and the actual start of the wagon. When the driver makes a change to the control command, the lift car always follows the delay. Actually after the infection, lifting takes place and the horizontal transfer begins towards the paint point. Since during transport the gripper 21, 22 not only swings according to the mathematical hi-15 Lur but also according to the wind forces and other auxiliary forces, the camera provides a new opportunity especially when approaching the paint point with the gripper 21, 22 and the product. In the prior patent application, it was attempted to state that, by means of difference dimensions and storage points, machine vision system 20 has more known information about the target area when compared to a completely unknown target environment.

When the crane transfer carriage arrives at the programmed paint point, a situation is sought where the oscillation 25 of the gripper 21, 22 would be attenuated before the carriage stops. In this case, the target of the target is first approached by utilizing the camera to dampen the load fluctuation and, most recently, to target the target to the actual target. The load is detected by the front-facing camera 1, 2.

30 From the diagrams, the upper one describes the speed as a function of the distance traveled. The middle diagram illustrates when the gripper 21, 22 remains oscillating without damping the oscillation. The optimum driving instruction of the automation system should be predicted when driving to the finish point of the gripper 21, 22, as shown in the diagram below 35. The changes in the swing angle of the gripper 21, 22 are also shown in Figure 1 as the movements of the load. The stacks between which the transfer occurs are indicated by the letters A, B, and C. Various reference points related to the phase of motion are shown in tx-t9.

The damping of the load oscillation without height information h «h1-h2 can be performed during horizontal movement in the time interval t: ... t7 5 by rotating the camera 1, 2 of the clamp 21, 22 upwards and examining the crane's main carriers (or main carrier), bridge, or by placing a known shape above a gripper 21, 22 at a known position. The actual export to the paint point-10 would take place in the time interval t5 ... t9. Although there is an overlap in the above times, the starting point of the approach can be freely selected if the system includes multiple cameras. Existing camera systems can connect up to four cameras to the same control computer. Thereby one 15 could be directed upwards and the other 15 would be in the jaws of the gripper 21, 22. Alternatively, for example, cameras may be located in the clamps, in addition to which one camera is mounted on the lifting trolley for studying the clamp and load fluctuation. Knowing the level of the upper surface of the floor or the products on the floor-20a and the features of the gripper, the exact movement of the gripper relative to the floor can be determined. In this embodiment, the lift has a numerically positioned height of the gripper.

It is not essential that the damping of the load oscillation occurs within the actual area below the gripper 21, 22 but that the damping of the load oscillation is generally taken into account.

The following describes alternatives for determining the height of the crane gripper 21, 22 from the desired platform. Without the height information of the crane gripper 21, 22, the combination of camera / cameras 1, 2 and laser beam 32 of Figure 2 can determine the selected target area of the video image, thereby defining the selected target area of two consecutive or very close image samples of a machine vision camera (RGB, CCD camera). converting relative displacement to crane control information. The laser beam source 31, e.g., the semiconductor laser 6 103031 and the camera 1, 2 are at a known angle α with respect to their center axes. Since the laser beam does not decompose in the same way as a normal light beam, the shape of light reflected from the reflection point of the laser beam to the camera 1, 2 is very standard and easy to find even in large footage. In Fig. 2, the distance h of the gripper 21, 22 from the subject can be calculated as the reflection of the laser light from the camera-provided image material on the computer 5. Depending on the deviation of the reflection from the selected point in the camera image area, the distance of 10 light points from the gripper can be accurately calculated, since the angle a is known. Zoom Lens, the Zoom Lens performs calibration during system setup, after which the height of the gripper can always be calculated from the subject when the zoom ratio of the zoom lens 15 is taken into account.

Utilizing known product dimensions or the auto focus function of some machine vision systems, by knowing the focal length of the lens, such as the Cognex 20 Auto-Focus function, also allows for load swing reduction without the positioning of the lifting device and no laser light source as in the previous example. The height information h can thus be calculated using a program. When the first full-resolution video pixel area 25 is 'trimmed' e.g. the pixel area of a 64x64 pixel, it can be localized to the next or subsequent images with a delay of about 30-50 ms per image to be examined (Figure 3, 4). This results in a situation where the offset of the cropped image area in each newly selected video image can be determined relative to the original or comparable image. By knowing the position of the pixels relative to each other, the direction and speed of crane grab wobble relative to the ground, the trolley and / or the bridge can be determined. The measurement can be started at any time, and the known time interval between comparative image shots is then the information brought by practical tests.

For example, if a product or the actual size of a feature is known from the image region that is typical of exporting standard type and size products to or from storage 7 103031, height measurement is not required because image pixels can be converted to relative motion specified grab distance to target area. If a combination of camera and laser-5 light source is used, the above video image capture method takes a sample of the reflection of the laser light and compares the transition of this cut area between successive and very close-up shots.

The information obtained from the camera image is illustrated in Figures 3 and 4. In Figure 3, the camera 1 or 2 placed in the gripper 21, 22 takes a picture of the area below the camera 100, whereby the camera shows part of the area below. From the image displayed by the camera, a searchable digital image area 102, which is part of 101, is determined and stored in the memory of the computer's digitization card 10. The area to be searched for must be positioned so that the crane or robot does not have time to move out of the screen or the angle of view to the area does not change too much so that the shadows due to the lighting can change substantially. The deviations of the retrieved image area, e.g., X 'and Y', from a point on the image area of the camera, e.g., an angle point 104, are calculated.

After a moment, e.g. 10-500 ms, a new image 101 is taken with the same camera 1 25 or 2 as the one used to take the previous image.

: Depending on the movement of the crane or robot, the digital image area 102 stored in memory has moved relative to camera 1 or 2. The computer digitization card can be used to search for the searchable area 102 on the full screen or only part of the camera screen. After the machine vision system has searched for all possible digital image area information within the defined search area, the system informs each new object found in image 4 area 101 of the original image 3 digital image area 102. The search area has to be unique and large enough to search. , there are practically no two or more options. If the system declares more than one area to be searched for, always select the most suitable 8 103031 area. Once found, the location of the region in the camera image area 101 can be calculated to obtain X "and Y". If the relative displacement of X "and Y" relative to the X 'and Y' dimensions in the previous figure is above a specified threshold and more of the 5 found areas described above are selected for goodness, etc. Eventually, if results are not sensible, again (Figure 3).

Since the distance to the target is known by either the crane or Robo-10 t digital positioning system or by laser light 32 mounted in a known oblique angle in the immediate vicinity of the camera, the direction, speed and acceleration of the cameras 1 and 2 and thus the gripper can be determined. and Y 'and Y "by relative motion travel differences by counting pixels.

The digital image area information 102 retrieved must not contain (or would be out of range) light reflection of the laser light source 32 caught in the gripper 21, 22 because this high intensity light moves relative to the retrieved image area 102 with cameras 1 and 2 and degrades search success. If the crane has a numerical positioning system, the movement speed of the grab 21, 22 relative to the crane or robot can be determined and 25 real-time grab pendulum corrections can be made by adjusting the speed of the crane or robot. At higher traveling speeds of the crane or robot, the relative motion of the gripper is attenuated with respect to the movement speeds of the crane or robot bearing structures, e.g., the bridge, and at low travel speeds relative to the ground and the actual paint point. Specifically, if the crane or robot does not have a numerical positioning system, the relative difference in movement speeds between the bridge and / or trolley and the clamp and the swing angle of the clamp can be determined 35 relative to the Saima image area 100 from above the lifting trolley with reflected laser light 61 or 62 (cf. Figure 5). Then, from this reflection of the laser light, the above dimensions X ', 9 103031 Υ', X "and Υ" are calculated. The result obtained is directly the relative speed of movement of the gripper relative to the corresponding directions of movement of the crane. Since, when the load is swinging, the speed of the swinging movement of the transfer carriage or other load carrier can be equal to the speed of the transfer carriage at both top dead centers (maximum swing angle a) and without the load upon reaching the image area 100 of the sensors 1 and 10 2, the load swing angle can be accurately determined and the aforementioned swing states having the same situation speed but representing different situations can be distinguished. At the same time, the angle of swing of the load relative to the overhead bearing structures can be accurately determined by the angular amplitudes of the small bullet. As shown in Fig. 5, the Ia servalon source is located in the transfer carriage 4. The swing angle α in the upper figure is different from zero and in the lower figure 5 is zero. In each of Figures 5, the wagon has a running speed vtroUey in relation to the ground. The velocity of the gripper has a velocity vgripper of 20, which in the upper image varies between vgripper and zero, which is achieved by the angle of rotation a.

In some cases, assigning fixed identifiable characters or lines to the area below can make it easier to find the digital image information that you are looking for.

If the crane is not equipped with a numerical positioning system, the method also achieves better positioning by hand than by, for example. the harmonic oscillation training systems based on the mathematical pendulum formula (T0 = 2πΊ (1 / g), where T0 = oscillation time (s), 1 = lifting rope length (m), g = 9.807 m / sa) are achieved. In manual cranes, the 35-oscillation damping methods based on mathematical formulas and assumptions have most of the typical over- or under-positioning of them during manual travel to a predetermined point. The invention does not have such a feature because when the driver issues a stop command, the swing is already 10 103031 relative to the actual target area and the stopping point is within the small image area 101 shown by cameras 1 and 2. Cameras 1 and 2 and thus the grippers 21, 22 according to the empirical study 5, the crane corresponding to the crane of the corresponding crane without damping the load can be accurately reproduced from the moment of stopping and at the same time the damping of the load can be damped.

10

The information provided by the camera system can be used to determine the instantaneous relative speed of movement of the gripper relative to the selected object (floor level or, for example, an overhead hoist). The motion speeds and directions are obtained in both sil-15 and trolley directions. Similarly, the pivot angle of the pivoting gripper may be determined relative to the selected target.

With a container crane, the load swing reduction device 20 can also identify a series of numbers at the end of the container to secure the contents of the container and the loading address in the hold of the ship. Automation increases safety because the position of containers on the ship is predetermined by the stability of the ship. The same loading information can be utilized at the port of unloading by 25 similar equipment directly into the receiving port: - the information system, which improves the handling of the material at both the sending and receiving ports.

Figure 6 shows the identification by product key. The product identification label 37 is placed over the product or on a page by hand or with the aid of an automatic applicator. The label may be permanently painted in the cargo container. Label 37 is labeled with product identification methods to permit automatic product identification. The text type commonly used by the camera is of the SEMI type. Barcodes can also be recognized by the camera, but in particular the barcode is intended to serve inventory and other functions in the production environment.

If there is no numerical information about the height of the gripper to control the camera, a light source producing 5 laser lights can be mounted on the grip surface of the gripper at an oblique angle in the open / closed direction of the jaws. As the gripper approaches the product from top to bottom, the camera 1 or 2 on the gripping jaws 23 or 24 detects the movement of laser light on the top surface of the product. At one of the thresholds, the laser light 36 disappears or exceeds the threshold of a given image area 10 in the field of view of the second camera, whereupon the laser light 36 in question. the camera reads a pre-programmed area from the top of the product. A number of product identifiers are printed on label 37 so that the position of the label is not very critical. The automated system or the user can accept the read product identifier 15 or reject the information for later use.

If the gripper moves very low and cameras 1 and 2 cannot read the product label on the product eg. a third camera 35 can be mounted on the underside of the gripper 20 because of the narrow angle of view of the camera 1 and 2. This laser reading camera 35 can be mounted at an angle that is optimal for label recognition. Even with a label reading camera 35, not even a laser light source 36 is needed, since product identification 25 can take place even after the product is already supported by the grips jaws 21 and 22. The label-reading camera 35 can also act as a detector of unexpected obstacles when picking up the product in an automatic system with a numerically positioned height of the gripper. In some embodiments, the product may be rotated or rotated with the product supported by jaws 21 and 22, whereby the product label may be located where it would not normally be readable without the product's rotating or rotating device.

35

Figure 7 further shows specific product identification. Horizontal cylindrical product 50 is rewound or similarly mounted in the center hole of the product ID 12 103031 carriage. Frame 49 includes a branding system recognizable by the camera system. The center of gravity of the body is at the very bottom near the lowest cylindrical wheel 42. By means of springs 45 and 46, gravity or other force is applied to press the cylindrical 5 rollers against the inner surface of the center hole. Since the wheels 40, 41 and 42 are cylindrical and parallel, the produced carriage 49 can rotate during the rotation of the roller at various stages of work but remains in the position provided by the wheels 40, 41 and 42. The product identifier on body 49 may be a fixed, perforated, or other label. If the body 49 is very close to the camera (s) on the gripper jaws 23 and 24, the identification mark may be a hole number plate.

As shown in Figure 8, the computer may be, for example, an ordinary 15 personal computer with a display and a keyboard. The computer has a permanent mass memory, such as a hard disk, which is suitable for a permanent storage of a large amount of information. The image signal produced by the cameras is directly connected to the video cards on the computer. The computer 5 has a central unit 20, the communication bus of which is connected to enable data transmission e.g. a communication card 6, a computer network card 7, a sound card 8, a video card 10 e.g. Cognex VP50 or 5000 Series, which allows searching for a desired location and comparing sequential images in desired pixels 25 at intervals of tens of milliseconds, memory drive 11, hard disk 12 and a display card 13 between which communication is possible. According to the information received from the central unit, the crane and the cameras can be controlled by the control unit 14, on the other hand also by the information received from the camera. The computer can be equipped with eg. CD drive, user interfaces (keyboard, microphone and speakers, display), mass storage and modem. A program is installed on the computer's mass memory and the computer is connected to the control system. The computer's operating system is the so-called. multitasking-35 system, so it uses multimedia equipment. The control system performs real-time monitoring of the grabber or machine in a pre-programmed manner (logic program). The control system includes logic control, actuators 13 103031 controls (front, rear, right, left, slow, fast, etc.), digital positioning system and motor drives. The logic also takes care of the real-time control and adjustment of the optical rotation elements of the cameras.

5 The crane or robot is controlled either automatically, semiautomatically or manually.

There is a high speed communication link between the computer and the logic. The computer has direct access to all data in the lo-10 memory. Through the communication bus, the camera application of the computer corrects the position of the gripper. With the speakers connected to the computer, the system can provide audio messages to the driver. The speakers are controlled via a multimedia card on the computer. This can be done either by playing the pre-recorded speech through the sound card or by converting the recorded text through the program through the sound card when there is an obstacle or due to other predetermined control. The grab can be controlled via a microphone when the audio is converted into a computer-readable 20 signal.

As the hand-crane approaches the finish area, machine vision provides a real target. The deceleration and acceleration of the crane should be such that the movable variable mass (load) does not • substantially alter the deceleration of the crane. If the weight of the load were decisive, the crane could slip on its 'skins' on the rails at too fast a stop. During acceleration, the wheels of the trolley would empty because the loading of the inertia mass 30 requires its own inertia. Known for cranes and mouth. the acceleration and deceleration values generally used in the composite are 0.1-0.7 m / s2, preferably 0.3-0.5 m / s2. Robots have higher acceleration and deceleration, typically 1-4 m / s2.

35

As the crane driver approaches the finish area, he slows down to near 0.6m / s (5-6m / min). The crane designer has lowered the deceleration to the above-mentioned acceleration of 14 103031. Ultimately, the final acceleration is determined by the drive system selected and thus each crane has application-specific deceleration and acceleration. When the driver is used to being a crane driver, he drives what he knows at slow 5 speeds and with his skill raises his hand from the crane's directional control at just the right moment to reach the right stop. Depending on the crane's own deceleration, the crane will continue to advance from the stopping speed given by the user along a relatively linear deceleration curve down to 10 stops if the driver does not make new repair movements.

For a crane without numerical positioning, the machine vision system allows the crane operator to teach the normal 15 crane stopping speeds and the deceleration distances they require. Once the deceleration distance has been taught, the machine vision system can reproduce the deceleration taught by the driver to a point at a specified distance from the stop command, while also attenuating the load oscillation with shift-wool load masses. Thus, the deceleration occurs by means of machine vision with numerical control even if the crane does not have the entire numerical positioning system in different transmission movements.

The crane or robot oscillation detection system is illustrated in flow chart in Figure 9.

•

The laser beam may be guided through the reflecting surface to a desired location and the cameras may be rotated in different directions by a conventional piston / cylinder means, or the areas depicted therein may be reflected through a mirror or a corresponding reflecting surface.

Numerical positioning means that each crane area has its own numeric value 35, where the crane finds this area (point), and is stored in the camera and crane monitoring system (Figure 8). Based on the value, the crane movement can be controlled repeatedly.

Claims (8)

103031 It should be noted that the invention has been described above only in the light of certain examples. This is in no way intended to limit the invention only to these solutions, e.g. the gripper 5 is any gripping member or the like, but also to those which can be carried out by one skilled in the art within the scope of the appended claims. • r 103031 A method for detecting fluctuations of a load in a hoisting device and controlling the hoisting device on the basis of information 5, wherein the hoisting device has a bridge or the like to which the gripping members (21, 22) are movably secured by the lifting arm or similar (23, 24) , 22) and at least one sensor (1, 2) is connected 10 to monitor the movement of the goods to be connected to the computer for controlling the crane to the location data and / or sensors (1, 2) stored in the memory. and whose image area can be viewed by means of specific pixels by converting the image into numeric data and transferring the data via a connection to a computer (5), characterized in that the object or region to be identified from the previous or previous image is determined. , from the previous identifiable target of the image generating numerical data 20 or region, after a given time interval, taking at least another image whose numerical data generated from the subject or region of the image determines the observed displacement of the previously identified object or region, based on known interrelationship between pixels; speed and direction with respect to the desired location, based on the transition information, provide speed and direction information from the computer to the crane control system. A method according to claim 1, characterized in that the detector is a camera, preferably a CCD or RGB camera. A method according to claim 1, characterized in that the camera (1, 2) captures an image of the area below the camera (100), wherein the camera shows a portion of the area below, defining a digital image area (102) to be searched. is part of the image area of the camera 17 103031 and stored in the computer memory, the same camera (1, 2) used to take the previous image, taking a new image (101), the computer searching for the searchable area (102) 5 from the camera image area (101), calculating the relative offset relative to the previous image area (102) considered. A trolley according to one or more of claims 1 to 3, characterized in that the position of the crane is determined, the distance of the gripper from the object is determined, the speed of the lifting carriage and / or the bridge is determined. 5. A method according to claim 4, characterized by determining the distance of the gripper from the object by describing a known object, preferably the feature being a feature of the load to be lifted, the load to be lifted itself, a separate laser light beam whose measurable dimension 20 is known. Method according to one or more of Claims 1 to 3, characterized in that the speed of the lifting carriage and / or the bridge is determined, the angle of rotation (a) is measured, the speed of the lifting carriage and the bridge are adjusted. Method according to one or more of Claims 1 to 6, characterized in that the product is identified at the same time and in addition, the gripper is guided to grip the load to be lifted. 4 Method according to one or more of Claims 4 to 6, characterized in that, in order to determine the distance of the object or area from the detector, the camera is orientated to a laser beam or similar image area disposed at a known angle to the camera. Area. 18 103031