JP2013180369A - Adaptive machine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a machine with a learning function having adaptability more superior than that of a conventional machine.SOLUTION: An adaptive machine includes a motion mechanism for performing motion associated with an object, a motion standard acquisition unit for acquiring a motion standard according to the object based on external information, and a control driving unit for controlling and driving the motion mechanism based on the motion standard acquired by the motion standard acquisition unit.

Description

  The present invention relates to an adaptive machine with adaptability.

  Patent Document 1 below discloses a teaching playback type continuous unloader. There are roughly two types of continuous unloaders: the teaching playback method and the programming method, but the teaching playback type continuous unloader actually moves the scraping part in the hold along the desired movement path by the driver's manual operation. The trajectory of the scraping part by moving (teaching) and the operation amount of each movable part for moving the scraping part are stored as teaching information, and the above teaching is used when actually loading the cargo in the hold By controlling each movable part based on the information, a playback operation is performed in which the scraping part is moved along the same movement path as in teaching.

JP 08-217256 A

  By the way, the teaching playback type continuous unloader stores the trajectory of the scraping unit and the operation amount of each movable unit for moving (moving) the scraping unit as a teaching result, and faithfully reproduces the teaching result. Is. Therefore, this continuous unloader has a fixed movement of the scraping part and is limited in terms of adaptability. For example, the status of the cargo in the hold changes according to the progress of landing, so that when the scraping unit is moved (moved) according to the teaching information, a situation where the cargo cannot be scraped accurately may occur. Therefore, originally, it is preferable to appropriately correct the movement (movement) of the scraping portion according to the load situation in the hold.

  In addition, this continuous unloader is a machine that specializes in the landing of the load, although it memorizes the movement of the scraping part acquired by teaching, and the range of application is the landing of the load (bulk load like grain) From this point of view, it is limited in terms of adaptability.

  This invention is made | formed in view of the situation mentioned above, and aims at providing the machine (adaptive machine) excellent in adaptability rather than the conventional machine.

  In order to achieve the above object, in the present invention, as a first solving means, an exercise mechanism that exercises on an object, an exercise norm acquisition unit that acquires an exercise norm according to the object based on external information, And a control drive unit that controls and drives the motion mechanism based on the motion code acquired by the motion code acquisition unit.

  In the present invention, as the second solving means, in the first solving means, the movement norm acquisition unit outputs a mechanism detection signal indicating the movement state of the movement mechanism when the movement mechanism is manually moved to external information. As the learning function unit, the learning function unit that learns the motion norm corresponding to the object based on the mechanism detection signal is adopted.

  Further, in the present invention, as a third solving means, in the first solving means, further comprising a human body motion detecting unit for detecting the movement of the human body, wherein the motion norm acquisition unit is a human body detection signal indicating a motion state of the human body. From the human motion detector as external information and a mechanism detection signal indicating the motion state of the motion mechanism from the motion mechanism as external information. Based on the human body detection signal and the mechanism detection signal, the motion of the human body and the motion corresponding to the object Adopt the means that it is a learning function part to learn the norm.

  In the present invention, as the fourth solution means, in the third solution means, the motion mechanism performs the motion related to the object so as to support the motion of the human body, and the control drive unit is the motion of the human body. And a means of controlling and driving the motion mechanism so as to support the motion of the human body based on the motion norms according to the object.

  According to the present invention, as a fifth solution means, in the first solution means, the motion norm acquisition unit takes in a mechanism detection signal indicating the motion state of the motion mechanism from the motion mechanism as the external information and controls the drive unit. Is a learning function unit that takes in a drive signal output to the motion mechanism as external information and learns a motion rule corresponding to the object based on the mechanism detection signal and the drive signal.

  According to the present invention, based on an exercise mechanism that exercises an object, an exercise norm acquisition unit that acquires an exercise norm according to the object based on external information, and the exercise norm acquired by the exercise norm acquisition unit Since the control drive unit that controls and drives the motion mechanism is provided, it is possible to provide a machine that is more adaptable than a machine in which a motion standard (control standard) is fixed or limited as in the related art.

It is a block diagram which shows the function structure of the cargo handling apparatus A which concerns on one Embodiment of this invention. It is the front view and top view which show an example of the transfer mechanism 1 in the cargo handling apparatus A which concerns on one Embodiment of this invention, and cargo handling work. It is a flowchart which shows operation | movement of the cargo handling apparatus A which concerns on one Embodiment of this invention. It is a top view which shows the other example of the cargo handling work in the cargo handling apparatus A which concerns on one Embodiment of this invention.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
As shown in FIG. 1, the cargo handling apparatus A according to the present embodiment includes a transfer mechanism 1 (motion mechanism), a control drive unit 2, an operation unit 3, and a transfer code acquisition unit 4 (motion code acquisition unit). ing.

  The transfer mechanism 1 moves the load (object) to a predetermined location based on the drive signal input from the control drive unit 2. That is, the transfer mechanism 1 performs a transfer operation (movement) on a predetermined load platform (load receiving platform) by gripping the load supplied from the outside and moving it to another load platform (loading platform). It is an exercise mechanism that performs.

  Such a transfer mechanism 1 is realized by, for example, an articulated manipulator as shown in FIG. The multi-joint manipulator (transfer mechanism 1) includes a base 1a, first to third arms 1b to 1d, first to third joints 1e to 1g, and a hand portion 1h. The base 1a is a pedestal of an articulated manipulator fixed to a floor or the like, and a turning motor for turning the first arm 1b around a vertical axis is provided therein.

  The first arm 1b is a rod-like member provided on the base 1a so as to be slightly tilted from the vertical axis as shown in the figure. One end of the first arm 1b is rotatably attached to the base 1a and the first arm 1b is connected to the other end. One joint 1e is provided. The first arm 1b turns around the vertical axis when the turning motor operates. The first joint 1e is a member that connects the first arm 1b and the second arm 1c so that the first arm 1b and the second arm 1c can freely bend and stretch. The first joint 1e adjusts the bending and stretching state of the first arm 1b and the second arm 1c. A first joint motor is provided.

  The second arm 1c is a rod-like member having one end attached to the first joint 1e and the other end provided with the first joint 1e. The second arm 1c changes its bending and stretching state relative to the first arm 1b when the first joint motor operates. The second joint 1f is a member that connects the second arm 1c and the third arm 1d so that the second arm 1c and the third arm 1d can freely bend and stretch, and adjusts the bending and stretching state of the second arm 1c and the third arm 1d inside the second joint 1f. A second joint motor is provided. The third arm 1d is a rod-like member having one end attached to the second joint 1f and the other end provided with the third joint 1g. The third arm 1d changes its bending and extending state with respect to the second arm 1c when the second joint motor operates.

  The third joint 1g is a member that connects the third arm 1d and the support shaft (hand shaft) of the hand portion 1h so as to be able to bend and extend, and the third joint 1g includes a hand shaft of the third arm 1d and the hand portion 1h. And a third motor for adjusting the bending / extension state and a rotation motor for rotating the hand portion 1h around the hand axis. The hand portion 1h is attached to the third joint 1g via a hand shaft so as to be able to bend and extend and to be rotatable. The hand unit 1h includes a gripping mechanism for gripping the load at the tip of the hand shaft and an imaging device (camera) for shooting the load (object).

  The turning motor, the first to third joint motors, and the rotation motor described above are provided with a sensor (for example, a rotary encoder) that detects its own operation amount. The transfer mechanism 1 (multi-joint manipulator) performs a load transfer operation based on a drive signal input from the control drive unit 2, while such a swing motor, first to third joint motors, and a rotation motor. The detection signal of the sensor provided in the control unit 2 is output to the control drive unit 2, and the detection signal and the video signal of the camera provided in the hand unit 1h are output to the transfer norm acquisition unit 4 as device detection signals (external information). To do.

  The control drive unit 2 includes a motor control circuit and a motor drive circuit, and drives the transfer mechanism 1 (articulated manipulator) based on an open loop control method. That is, the motor control circuit generates a control signal for open-loop control of the transfer mechanism 1 (articulated manipulator) based on the control target input from the transfer norm acquisition unit 4 and outputs the control signal to the motor drive circuit. . The motor drive circuit is, for example, an inverter circuit, generates a drive signal (pulse drive signal) based on the control signal, and outputs the drive signal to the turning motor, the first to third joint motors, and the turning motor.

  Here, the cargo handling apparatus A has two operation modes, a learning mode and a work mode, and these two operation modes are switched by an operation signal input from the operation unit 3. In the learning mode, the control drive unit 2 performs open-loop control of the transfer mechanism 1 (articulated manipulator) based on an operation signal input from the operation unit 3, and the transfer norm acquisition unit 4 transfers the transfer mechanism 1 (multiple mechanism). This is an operation mode for learning the transfer motion of the joint manipulator. In the work mode, a control target based on the package transfer code acquired by the transfer code acquisition unit 4 in the learning mode is output to the control drive unit 2, and the control drive unit 2 transfers the transfer mechanism 1 based on the control target. This is an operation mode in which the (multi-joint manipulator) is controlled by open loop.

  The operation unit 3 is a man-machine interface for inputting a manager's instruction to the material handling apparatus A. That is, the operation unit 3 outputs an operation signal indicating an instruction to switch between the learning mode and the work mode to the control drive unit 2 and the transfer norm acquisition unit 4 based on an operation instruction from the administrator, while in the learning mode. , The operation signal for instructing the drive amounts of the turning motor, the first to third joint motors, and the rotation motor is output to the control drive unit 2.

  The transfer norm acquisition unit 4 loads the package (external information) input from the transfer mechanism 1 (articulated manipulator) and the drive signal input from the control drive unit 2 in the learning mode. Acquire transfer norms (exercise norms) according to the object. That is, the transfer norm acquisition unit 4 is a kind of computer that performs information processing based on a predetermined learning algorithm, for example, and the transfer operation (transfer motion) of the transfer mechanism 1 (articulated manipulator) in the learning mode. ), That is, the difference amount (control) between the operation amounts of the swing motor, the first to third joint motors and the rotation motor given by the device detection signal (external information) and the drive amount of the control drive unit 2 given by the drive signal The amount of error) is classified into a plurality according to the form (outer shape, color, etc.) of the luggage (object) captured by the camera of the hand unit 1h, and the control error amount is stored (learned) for each category.

  For example, the transfer norm acquisition unit 4 classifies the control error amount related to each motor with respect to two attributes of “outer shape”, “size”, and “color” in the form of the package (object). That is, the transfer norm acquisition unit 4 determines the “outer shape”, “size”, and “color” as the feature points of the baggage (target object) by performing image processing on the image captured by the camera. 2), for example, “shapes with corners (cubes, etc.)” are classified into the first group, and “shapes without corners (spheres, etc.)” are classified into the second group. The control error amount is stored (learned) in each group.

  In this case, baggage consisting of polyhedrons (triangles (tetrahedrons) and cubes (hexahedrons), etc.) whose outer shape is composed of a plurality of planes is classified into the first group. It is classified into the second group. As for the “size” of the package (object), for example, “larger than the standard size” is classified into the third group by comparison with the standard size specified in advance, and “smaller than the standard size” is classified into the third group. Further, for the “color” of the package (object), for example, “black” is classified into the fifth group, and “white” is classified into the sixth group.

  In this way, the transfer norm acquisition unit 4 assigns the control error amount related to each motor to a plurality of groups (for example, the above first to sixth groups) according to the form (outer shape, color, etc.) of the load (object) in the learning mode. By classifying and storing (learning) into groups, a control error amount relating to each motor according to the form of the package (object), that is, a package transfer standard (control standard) of the package is established (acquired).

  Moreover, the transfer norm acquisition unit 4 sets a control target based on the transfer norm (control norm) and the device detection signal input from the transfer mechanism 1 (articulated manipulator) in the work mode. . For example, the transfer norm acquisition unit 4 identifies a group corresponding to the form (outer shape, color, etc.) of the package (target object) based on the video signal input from the camera in the work mode, and controls the control error of the group. A control target (that is, a difference amount between the control error amount and the detected amount) is set based on the amount and the detected amount of each motor.

  Next, the operation of the cargo handling apparatus A according to the present embodiment configured as described above will be described in detail along the flowchart shown in FIG.

  First, the cargo handling apparatus A obtains a classification function based on the transfer of a package by establishing (acquiring) a package transfer standard (control standard) in the transfer standard acquisition unit 4 through a learning mode. That is, the cargo handling apparatus A does not exhibit a specific motor function as a machine, since the package transfer norm has not been established (acquired) before the learning mode. Therefore, according to the present cargo handling apparatus A, it is possible to change the transfer norms relating to the load according to the learning content in the learning mode, and thus it is possible to transfer (classify) the load with a high degree of flexibility.

  As shown in FIG. 3, the cargo handling apparatus A learns the load transfer movement in the learning mode (step S1), and as a result, the load transfer code (control code) is stored in the transfer code acquisition unit 4. ) Is established (acquired) (step S2). For example, as shown in FIG. 2 (b), the transfer norm acquisition unit 4 sets “corner shape (first group)” for the “outer shape” of the load (object) on the load receiving stand D1 captured by the camera. ”And“ shapes without corners (second group) ”, and the packages belonging to“ shapes with corners (first group) ”are moved from the receiving platform D1 to the loading platform D2, For loads belonging to the shape having no corners (second group), the transfer operation (control error amount related to each motor) for moving from the loading platform D1 to the loading platform D3 different from the loading platform D2 is stored (learning). )

  In the cargo handling apparatus A, when the operation unit 3 is operated and the operation mode is switched from the learning mode to the work mode, the transfer norm acquisition unit 4 controls the control target based on the transfer norm (control norm). Output to the drive unit 2, and as a result, the control drive unit 2 performs open loop control of the transfer mechanism 1 (articulated manipulator) based on the control target and the detection signal, so that the load based on the transfer rule (classification rule) This sort operation is executed (step S3).

  For example, when the spherical load W1 is placed on the load receiving stand D1 in FIG. 2B, the transfer norm acquisition unit 4 determines that the load W1 has no corner based on the video signal input from the camera. As a result, the control target for moving the load W1 on the load receiving table D1 to the unloading table D3 is output to the control drive unit 2. On the other hand, when the legislative load W2 is placed on the load receiving platform D1, the transfer norm acquisition unit 4 converts the load W2 into a “cornered shape (first group) based on the video signal input from the camera. As a result, a control target for moving the load W2 on the load receiving stand D1 to the unloading stand D2 is output to the control drive unit 2.

  Further, when the load W2 ′ in which one corner portion of the legislature is damaged (missed) is placed on the load receiving table D1, the transfer norm acquisition unit 4 loads the load W2 based on the video signal input from the camera. 'Is classified into "shape with corners (first group)", and as a result, a control target for moving the load W2' on the load receiving platform D1 to the loading platform D2 is output to the control drive unit 2.

  That is, the transfer norm acquisition unit 4 does not learn in the learning mode in the work mode, but even if the shape of the luggage W2 ′ that is not learned in the learning mode is supplied on the receiving platform D1, the outer shape of the luggage W2 ′ has a corner. Are classified into “shapes having corners (first group)”. Therefore, the transfer norm acquisition unit 4 generates a control target in a flexible manner even for a load W2 'having an unlearned shape.

  Since the control drive unit 2 performs open loop control of the transfer mechanism 1 (articulated manipulator) based on the control target input from the transfer norm acquisition unit 4 as described above, the load W1 is unloaded from the load receiving table D1. A drive signal for moving the table D3 is output to the transfer mechanism 1 (articulated manipulator), while a drive signal for moving the loads W2, W2 'to the loading table D2 is output to the transfer mechanism 1 (articulated manipulator). To do. As a result, in the transfer mechanism 1 (articulated manipulator), each motor is driven in accordance with a drive signal input from the control drive unit 2, so that the load W1 is moved from the load receiving table D1 to the loading table D3. W2 and W2 'are moved from the receiving platform D1 to the unloading platform D2.

  Here, in the conventional teaching playback type machine, since it is not possible to deal with other than the teaching result, it is possible to determine which of the loading platforms D2 and D3 should be moved by recognizing the unloaded memory W2 ′. However, according to the present cargo handling apparatus A, excellent adaptability is exhibited without causing such an operation stop. Therefore, according to the present embodiment, it is possible to provide an adaptable machine that is more adaptable than conventional machines.

  Further, in the above description of operation, only whether the package (object) belongs to the “outer shape”, that is, belongs to “shape with corners (first group)” or “shape without corners (second group)”. The case of transferring (classifying) a package (object) based on the above has been described, but the package (object) may be transferred (classified) in consideration of other third to sixth groups. .

  In addition, for example, based on whether the package (object) belongs to “size”, that is, “larger than standard size (third group)” or “smaller than standard size (fourth group)” When classifying packages (objects), a detection signal indicating the operation amount of the hand unit 1h may be used as the device detection signal (external information) instead of the video signal of the camera. That is, since the operation amount of the hand unit 1h that holds the load (object) varies depending on the size of the load (object), each motor (the turning motor, the first motor) is based on the detection signal indicating the operation amount of the hand unit 1h. The control error amount related to the first to third joint motors and the rotation motor may be learned.

  Further, the camera provided in the hand unit 1h is used as a human motion detection unit for detecting the motion of the worker (human body) in order to assist (assist) the worker's cargo handling work, and the worker (human body) cargo handling work. Is transferred from the camera to the transfer norm acquisition unit 4 as one of external information, and is based on the detection signals of the human body detection signal and the rotation motor, first to third joint motors, and rotation motor sensors. Thus, an assist standard according to the cargo handling work (exercise) and the load (object) of the worker (human body) may be learned.

  Regarding the support (assist) of the cargo handling work using the transfer mechanism 1 (multi-joint manipulator), for example, there is a cargo handling work as shown in FIG. That is, the transfer mechanism 1 (articulated manipulator) transfers the load W (target object) on the first belt conveyor together with the worker onto the second belt conveyor, thereby reducing the load on the operator with respect to the weight of the load. To do (power assist).

  In this case, the transfer norm acquisition unit 4 classifies the worker's motion into a plurality of motion types based on the worker's image captured by the camera in the learning mode, The control error amount related to the third joint motor and the rotation motor is stored (learned). This operation type is, for example, “gripping the luggage W (object)”, “lifting the grasped luggage W (object)”, “moving the lifted luggage W (object) horizontally”, “moving horizontally” A series of operations of an operator who transfers the load W (target object) on the first belt conveyor onto the second belt conveyor, such as “lower the load W (target object)”.

  Then, the transfer norm acquisition unit 4 acquires an assist norm for assisting the cargo handling work of the worker (human body) in such a learning mode, and sets a control target in accordance with the assist norm in the work mode to the control drive unit. Output to 2. The control drive unit 2 performs feedback control of the transfer mechanism 1 (articulated manipulator) based on a control target that complies with the assist standard, so that the load W (target object) on the first belt conveyor is the operator. And the transfer mechanism 1 (multi-joint manipulator) are transferred onto the second belt conveyor.

In addition, this invention is not limited to the said embodiment, For example, the following modifications can be considered.
(1) In the above embodiment, the present invention is applied to a machine that performs a cargo handling operation, but the present invention is not limited to this. By performing (learning) different from the above-described learning content, the transfer norm (control norm) other than the transfer (classification) of the load is established (acquired) in the transfer norm acquisition unit 4 to transfer the load ( Various tasks (exercises) other than (classification) can be performed.

(2) In the above embodiment, the transfer norm of the transfer mechanism 1 (articulated manipulator) is acquired in the transfer norm acquisition unit 4 by classifying and learning the mechanism detection signal. However, the present invention is limited to this. Not. Since various learning algorithms are known, a loading standard for cargo handling work may be obtained from known learning algorithms. Further, the transfer norm may be obtained by using an artificial intelligence technique such as case-based reasoning.

(3) In the above embodiment, the transfer norm relating to the classification of the package is established (acquired) in the transfer norm acquiring unit 4 in the learning mode, but the present invention is not limited to this. By establishing (acquisition) the exercise norms other than the baggage classification in the transfer norm acquisition unit 4 in the learning mode, an adaptive machine that exhibits functions other than the baggage classification function can be realized.

(4) In the above embodiment, in addition to the device detection signal consisting of the detection signal of each sensor provided in the turning motor, the first to third joint motors and the rotation motor and the video signal of the camera provided in the hand unit 1h. Then, the drive signal of the control drive unit 2 is also taken into the transfer norm acquisition unit 4, and the difference amount (control error amount) between the detection amount indicated by the detection signal of each sensor and the drive amount of the control drive unit 2 indicated by the drive signal is obtained. Although learned, this invention is not limited to this. For example, the detection amount of each sensor may be classified and learned without taking the drive signal of the control drive unit 2 into the transfer norm acquisition unit 4.

(5) In the above-described embodiment, as shown in FIG. 3, a configuration is adopted in which the mode is shifted to the work mode after the learning mode. For example, the learning mode and the work mode may be performed in parallel after the learning mode. That is, after acquiring the transfer standard after learning, the load handling work may be performed while correcting the transfer standard by learning the control error amount and the detection amount while performing the actual cargo handling work (transfer work). .

(6) In the above embodiment, the transfer mechanism 1 is open-loop controlled by the control drive unit 2, but the control drive unit 2 feeds back the transfer mechanism 1 by capturing the detection signals of the respective motors in the control drive unit 2. You may control. In this case, it is considered that the transfer norm acquisition unit 4 learns the detection amount of each sensor instead of the control error amount.

  A ... cargo handling device (adaptive machine), W, W1, W2, W2 '... luggage, 1 ... transfer mechanism (movement mechanism), 1a ... base, 1b-1d ... first to third arms, 1e- DESCRIPTION OF SYMBOLS 1g ... 1st-3rd joint, 1h ... Hand part (human body motion detection part), 2 ... Control drive part, 3 ... Operation part, 4 ... Transfer norm acquisition part (Exercise norm acquisition part)

Claims (5)

A motion mechanism for performing motions on the object;
An exercise norm acquisition unit that acquires an exercise norm according to an object based on external information;
An adaptive machine comprising: a control drive unit that controls and drives the motion mechanism based on a motion rule acquired by the motion rule acquisition unit. The motion norm acquisition unit takes in a mechanism detection signal indicating a motion state of the motion mechanism when the motion mechanism is manually moved from the motion mechanism as the external information, and applies the mechanism detection signal to an object based on the mechanism detection signal. The adaptive machine according to claim 1, wherein the adaptive function machine is a learning function unit that learns a motor norm corresponding thereto. It further comprises a human motion detector that detects human motion,
The motion norm acquisition unit captures a human body detection signal indicating a motion state of a human body from the human body motion detection unit as the external information and a mechanism detection signal indicating a motion state of the motion mechanism from the motion mechanism as the external information. The adaptive machine according to claim 1, wherein the adaptive machine is a learning function unit that learns a human body motion and a motion standard according to an object based on the human body detection signal and the mechanism detection signal. The motion mechanism performs a motion related to an object so as to support a motion of a human body,
The adaptive machine according to claim 3, wherein the control driving unit controls and drives the motion mechanism so as to support the motion of the human body based on the motion of the human body and a motion rule corresponding to the object. The motion norm acquisition unit captures a mechanism detection signal indicating a motion state of the motion mechanism from the motion mechanism as the external information and a drive signal output by the control drive unit to the motion mechanism as the external information, 2. The adaptive machine according to claim 1, wherein the adaptive machine is a learning function unit that learns a motion norm corresponding to an object based on the mechanism detection signal and the drive signal.

Images