JPH11165287A - Position detection device, position control device and robot control device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a position detection device capable of speedily carry out positional detection in high precision by a simple system and to provide a position control device and a robot control device in simple constitution using this. SOLUTION: A robot hand 18 has inclination sensors 31, 32 to detect respective angles of rotation around orthogonal horizontal axes X, Y, three pieces of image recognizing points L1 , L2 , L3 provided on the hand 18, a CCD camera 45 to recognize these image recognizing points in pictures and images and a rotational angle computing means to compute an angle of rotation Δn of a flat surface including the three image recognizing points L1 -L3 around a vertical Z axis orthogonal with the horizontal axes X, Y in accordance with the recognized images by this camera 45, and a position of the hand 18 is detected from a computing result of the computing means and detected angles of rotation of the inclination sensors 31, 32.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a posture detecting device including an image processing system, and a posture control device and a robot control device using the same. More particularly, the present invention is suitable for controlling the posture and tip position of a hand portion of a robot. Related to the device.

[0002]

2. Description of the Related Art Conventionally, when controlling the position of a control object whose posture changes in an arbitrary direction in a three-dimensional space, for example, the position of an object such as a hand portion of a robot, the bending of an arm supporting the object and the deflection of the arm are described. Precise positioning is difficult due to the backlash of the driving reduction gear, etc.Therefore, not only internal sensors such as potentiometers and encoders provided at the joints of the robot, but also external sensors that monitor the hand part from outside the robot are used together. ing. Further, as such an external sensor, for example, a sensor for recognizing an image of a hand portion of a robot through a camera is known, and a robot having such an image processing system is manually operated, such as a master-slave type robot. There are work support robots that reduce the work load and ensure worker safety by remote control. This type of robot is expected to be developed in the medical and welfare fields in the future. Further, proposals have been made such as using two CCD cameras, image-recognizing three points of an object in accordance with the principle of triangulation, and detecting the position of the hand tip.

[0003]

However, in the conventional attitude detecting device and the conventional robot control device, when the above-described image processing system is provided, for example, the brightness of the background near the image recognition point tends to change. For this reason, the object cannot always be clearly recognized as an image, and the image processing is complicated. For this reason, it is not possible to quickly and accurately detect slight posture changes and displacements.For example, when securing the safety of workers by remote control, the posture of the control target is detected with high accuracy using a simple system, It was difficult to perform work with high precision. In addition, although a robot capable of performing a task that relies on a delicate manual operation of a worker has been expected, it has not been practically used because of the remarkable difficulty described above. However, if high-accuracy posture detection is enabled and a high-accuracy work support robot is developed, high accuracy and work speed (shortening of work time), which cannot be achieved by manual work, can be sufficiently expected. There is a need to solve such problems.

In addition, if a slight change in posture of an object can be detected quickly and accurately, it is not limited to work support robots, but may be applied to various production machines, for example, those which are partially complicated and finely moved, such as a fiber processing machine. However, high-precision and quick work which cannot be achieved by manual work can be expected. Therefore, an object of the present invention is to provide a posture detection device capable of quickly performing high-precision posture detection with a simple system, and to provide a simple posture control device and a robot control device having a configuration using the same. I do.

[0005]

Means for Solving the Problems In order to achieve the above object, the inventor of the present invention has conducted intensive studies to make it possible to clearly recognize even a slight change in the posture of an object, and to provide the following invention. Reached. That is, the first aspect of the present invention provides a first and a second inclination sensor for detecting respective rotation angles of a target whose posture can be changed around first and second horizontal axes orthogonal to each other; Based on three image recognition points (L1, L2, L3) provided on the object, image recognition means for image-recognizing the three image recognition points, and recognition information by the image recognition means, Rotation angle calculation means for calculating a rotation angle of a plane including the three image recognition points around a vertical axis orthogonal to the first and second horizontal axes, and a calculation result of the rotation angle calculation means And the rotation angle detected by the first and second tilt sensors, and the posture of the control target object in a coordinate system having the vertical axis and the first and second horizontal axes as three orthogonal axes. It is to detect.

According to the first aspect of the present invention, a rotation angle about a vertical axis orthogonal to the first and second horizontal axes is calculated for a plane including three image recognition points, and the calculation result and the first rotation angle are calculated. The posture of the target object is detected based on the rotation angle detected by the second tilt sensor. Therefore,
The processing of the image data only needs to calculate the rotation angle of the plane including the three image recognition points around the vertical axis, and quick attitude detection is possible.

Further, if the image recognition points are constituted by light emitting elements attached to the object and emit light, the three image recognition points can be clearly focused and focused. And wherein said first and second tilt detecting means are attached to said object, respectively, and wherein said first and second tilt detecting means are attached to said object, respectively, around said first and second horizontal axes. If two inclination sensors for detecting the respective rotation angles are provided, the rotation in the inclination direction with respect to the horizontal plane can be directly detected with a simple configuration with high accuracy, and the detection accuracy can be improved.

According to a fourth aspect of the present invention, there is provided a control object capable of controlling a posture, a control means for controlling a posture of the control object to control a tip position of the control object, and And a posture detecting means for detecting the posture and feeding it back to the control means, wherein the posture detecting means rotates each of the controlled objects around first and second horizontal axes orthogonal to each other. First and second tilt sensors for detecting angles, three image recognition points (L1, L2, L3) provided on the control object, and image recognition for recognizing the three image recognition points Means and the first and second means based on the recognition information by the image recognition means.
Rotation angle calculation means for calculating a rotation angle of a plane including the three image recognition points around a vertical axis orthogonal to the horizontal axis of
Based on the calculation result of the rotation angle calculation means and the rotation angles detected by the first and second inclination sensors, the vertical axis and the first and second horizontal axes are set to three orthogonal axes. The attitude of the control object in the coordinate system is detected.

According to the present invention, the rotation angle in the inclination direction directly detected by the first and second inclination sensors and the calculation result of the rotation angle calculation means are provided as feedback information to the control means. By doing so, the processing of the image data only needs to calculate the rotation angle of the plane including the three image recognition points about the vertical axis, and the rotation of the plane in the inclination direction with respect to the horizontal plane can be performed by the first and second inclination sensors. And can be directly detected with high accuracy by a simple configuration. Therefore,
Quick and accurate attitude control is possible.

Further, according to the present invention, the posture of the arm and the hand supported by the arm is controlled to move the tip of the hand to a target position and to control the hand to the target posture. In a robot control device comprising: a control unit; and a posture detection unit that detects a posture of a tip portion of the hand and feeds back the control unit to the control unit, the posture detection unit may include a first and a second horizontal axis orthogonal to each other. First and second inclination sensors for detecting the respective rotation angles of the hand around, three light emitting elements attached to the hand and emitting light, and light emitting points (L1, L2, L3), and the three light emitting points around a vertical axis orthogonal to the first and second horizontal axes are determined based on information recognized by the image recognition means. Rotation angle calculation means for calculating the rotation angle of the plane, and the vertical axis is calculated based on the calculation result of the rotation angle calculation means and the rotation angles detected by the first and second tilt sensors. And detecting the attitude of the controlled object in a coordinate system having the first and second horizontal axes as three orthogonal axes.

According to the fifth aspect of the present invention, by accurately recognizing the image of the light emitting points of the three light emitting elements, the rotation angle of the plane including these light emitting points around the vertical axis is accurately calculated. The result and the rotation angle in the tilt direction directly detected by the first and second tilt sensors are fed back to the control means. Therefore, the processing of the image data only needs to calculate the rotation angle of the plane including the three image recognition points about the vertical axis, and the rotation of the plane in the inclination direction with respect to the horizontal plane includes the first and second inclination sensors. Since the detection can be directly performed with high accuracy using a simple configuration, the attitude control can be performed quickly and with high accuracy. Further, by directly detecting the rotation in the inclination direction with respect to the horizontal plane by the inclination sensor, it is possible to compensate for the bending of the arm and the backlash of the drive system, etc., and it is possible to perform more accurate posture control. In addition, as set forth in claim 6, the first
When the second tilt sensor and the second tilt sensor are attached to the hand, a simple robot control device that does not require a complicated external sensor is obtained.

According to a seventh aspect of the present invention, there is provided an attitude control means for controlling the attitude of an arm and a hand supported by the arm, moving the tip of the hand to a target position, and controlling the hand to a target attitude. And a detecting means for detecting the position and attitude of the tip of the hand and feeding it back to the attitude control means, wherein the detecting means is attached to the hand and emits light. And the light emitting points (L
1, L2, L3) from a predetermined direction, and an image of a triangle formed by the three light emitting points based on the image recognized by the image recognition means. The position of the distal end of the hand in the predetermined direction is detected by comparing the reference position with a triangular recognition image formed corresponding to the three light emitting points.

According to the present invention, the light emitting points of the three light emitting elements can be accurately image-recognized from a predetermined direction, and a triangular recognition image formed by the three light emitting points at the current position and a hand Since the recognition image of the triangle formed at the intermediate reference position is substantially similar to the triangle, the ratio of these sizes and the known distance from the image recognition means to the intermediate reference position, The position of the hand can be easily detected.

[0014]

Embodiments of the present invention will be described below with reference to the accompanying drawings. 1 to 11 are views showing an embodiment of a posture detection device and a robot control device according to the present invention, and show an example in which the present invention is applied to a dual-arm articulated robot. First, the configuration will be described. In FIGS. 1 and 2, reference numeral 10 denotes one arm of a double-arm articulated robot used as a work support robot. 1
1 and a plurality of movable arms 12, 13, 14, 15, 1 provided to be connected to each other so as to extend in an arm shape from the base 11.
6 and 17 and a hand 18 supported by a movable arm 17 on the distal end side. Articulated robot 10
Has a total of eight degrees of freedom, four degrees of freedom for turning in the arm swing direction, three degrees of freedom for rotation in the arm torsion direction, and one degree of freedom for the gripping direction of the hand 18.
As shown in the figure, the movable arms 12, 13, 15, and 17
Swing in the arm swing direction about a swing center axis orthogonal to a center axis in the arm length direction (hereinafter, referred to as an arm center axis), and the movable arms 14 and 16 and the hand 1
8 rotate around respective rotation center axes in the arm length direction. The hand 18 has a pair of left and right gripping fingers 21 and 22 that can move in parallel so as to approach and separate from each other. It is reciprocable to grip and release.

The first and second joints from the base section 11 to the movable arm 13 form a staggered gimbal mechanism by separating their turning center axes 15, 16 while being orthogonal to each other. Although not shown in detail, the pivot shafts 15a, 17 on the base end side of the movable arms 15, 17 are not shown.
The fourth and sixth joints each having a joint axis a are driven by a worm wheel after being decelerated by a spur gear, and are driven by the movable arms 14 and 16 and the hand 18.
The 3rd, 5th and 7th joints on the proximal end side are internal gears (large-diameter internal gear and pinion inscribed in it)
This drives the movable arms 14, 16 and the hand 18 to rotate.

The reason why the turning portion is constituted by a worm wheel is that a large reduction ratio can be obtained with a simple mechanism.
The reason why the rotating portion is constituted by the internal gear is that the rotation center of the robot can be coaxial with the drive gear, which is advantageous in terms of load load. In addition, a normal gimbal mechanism has two axes orthogonal to each other on the same plane. However, the staggered gimbal mechanism used for the first and second joints in the present embodiment has a movable area that aims at a distance between two orthogonal axes. Until it is obtained. The gimbal mechanism can be configured by a link, but the link mechanism requires three movable areas.
It is difficult to set the angle to 0 degrees or more, and a gimbal mechanism excellent in rigidity and movable area is preferable. Assuming that the distance between the two orthogonal axes is 0, circular movements are performed in two directions. Therefore, it is better that the distance is short, and in this embodiment, the distance is, for example, 50 mm. The body of arms 12-17 has CFRP (Carbon Fiber Reinforced Plastic)
s) That is, the weight is reduced by using an acrylic-carbon fiber composite material obtained by curing carbon fiber with an acrylic resin, and the bending compressive strength is secured.

In FIG. 2, P1, P2, R3, P4,
R5, P6 and R7 are driving DC servo motors for driving the first to seventh joints, and these are the left arm motor driver 25 and the right arm motor driver 26 shown in FIG.
The driving is controlled by the computer 27 (control means, rotation angle calculation means) via one of the above. Computer 2
7 determines the target position and attitude of the hand 18 according to a predetermined program stored in an internal memory device, and determines the servo motors P1, P2, R3, P4, R
5, a drive amount of each motor is calculated based on the feedback information from P6 and R7 and the information on the target position and attitude, and a command (command signal) for specifying the drive amount of each motor is output. Then, according to this command, for example, the motor driver 25 causes the DC servo motors P1, P
2, R3, P4, R5, P6 and R7 were converted to PWM (Pull
The motors P1, P2, R3, P4, R5, P6 and R
7 is driven forward and reverse, whereby the movable arms 12 to 17 and the hand 18 are each turned or rotated via a speed reducer (not shown).

When calculating the motor drive amounts by the computer 27, the motors P1, P2, and P2 are controlled in the same manner as the known control for driving and positioning the arm of the articulated robot.
A basic drive amount for R3, P4, R5, P6 and R7 is determined, and a specific motor, for example, a motor is controlled by a specific attitude detection as described below and an attitude control using the detection result as feedback information. P1, P
An angle correction value (a correction value for a deviation from a target value) corresponding to each minute rotation amount of each of R2 and R3 is obtained, and tip position control with fine attitude control of the hand 18 is performed to obtain a gripping target object. Is positioned with high precision. Therefore, the basic control regarding the driving and positioning of the arm will not be described in further detail, and the attitude control of the hand 18 and the method of correcting the angle therefor, the grip control of the hand 18 and the like will be described below.

First, a configuration for controlling the attitude of the hand 18 will be described. As shown in FIG. 4, two torque balanced type tilt sensors 31 and 32 are attached to the hand 18 and these tilt sensors are provided. Reference numerals 31 and 32 denote rotations of the hand 18 about two axes of the three absolute coordinate axes except the vertical axis (around each of the horizontal axes X and Y passing through the center point O of the hand 18 in FIG. 4). The corner can be detected. Further, rotation of the remaining one axis for knowing the posture of the hand unit, that is, rotation about the vertical axis (around the axis of the vertical axis Z orthogonal to the horizontal axes X and Y through the center point O of the hand 18 in FIG. 4). Since the angle cannot be detected by the inclination sensor, it is detected by the image processing system 33 described below.

The image processing system 33 includes three LEDs (Light-em) for detecting the position and posture attached to the hand 18.
itting Diode; light emitting diode) 41, 42, 43,
And a CCD (Charge Coupled Device) camera 45 fixedly installed at a predetermined position at a predetermined distance from the hand 18 on the mounting surface side of the LEDs 41 to 43 with respect to the hand 18. The CCD camera 45 is connected to a computer 47 via an image processing board (Image Processing Board) 46. Image recognition means for processing the output data of the CCD camera 45 by the image processing board 46 and taking it into the computer 47 is provided. It is configured. Further, the computer 47 operates in accordance with a predetermined program stored in an internal memory device.
7, the target posture of the hand 18 is determined based on the three-dimensional CAD data prepared in advance, and the image is displayed on the monitor 48 as shown by a virtual line in FIG. Forty-five recognition images (images of the hand 18 shown by solid lines in the figure) are displayed in an overlapping manner. Here, assuming that the rotation angle of the hand 18 around the X axis in absolute coordinates is a rolling angle, the rotation angle around the Y axis is a pitching angle, and the rotation angle of the hand 18 around a vertical axis is a yawing angle, the rolling angle and the pitching angle are A / D (Analog-t) is detected by the inclination sensors 31 and 32, respectively.
It is always fed back to the computer 47 via an o-Digital (analog / digital) converter 49. Also, the deviation of the hand 18 from the target posture, that is, the X axis,
Assuming that the rotation angles of the hand 18 to be corrected around the Y axis and the Z axis are correction values Δl, Δm, and Δn, respectively, the correction values Δl,
Δm can be immediately calculated by the computer 27 or 47 (rotation angle calculation means) as the difference between the detection values of the inclination sensors 31 and 32 and the target rotation angle.

The yawing angle of the hand 18 is given by L
A plane SL including the light emitting points L1, L2, and L3 of the EDs 41 to 43.
Can be obtained by calculation using the angle ΔA (see FIG. 6) formed by the plane SL and the known reference plane ST in absolute coordinates, for example,
Actually, it suffices to set a plane corresponding to the target posture and perform the calculation processing of the correction value Δn required for the posture control.

In the present embodiment, in the tip position control of the hand 18, the position of the hand 18 is detected, and the inclination posture in the absolute coordinates of the plane SL including the light emitting points L1, L2, L3 is brought closer to the target posture. Is executed to perform correction. First, as shown in FIG.
Based on the CAD data, the hand 18 sets a target at a known intermediate reference position in the direction (predetermined direction) of the CCD camera 45 with respect to the recognized image of the triangle L1 L2 L3 formed by the light emitting points L1, L2, L3 of 1 to 43. C when in posture
Light emitting points L1, L2, L recognized by the CD camera 45
A triangle P1P2P3 is set as the recognition image of No. 3, and the virtual triangle P1P2P3 is compared with the triangle L1L2L3. At this time, the triangle L1L2L3 and the triangle P1P2P3 are rotated about a vertical axis by performing feedback control to constantly follow the target value based on the detection values of the inclination sensors 31 and 32 for errors in the posture of the hand 18 in the inclination direction. (Yawing) A state where the error of the angle should be corrected based on the image information recognized by the CCD camera 45 is established.

The correction will be described. First, assuming that the tip coordinate of the robot 10 is R (xi, yi, zi),
From the joint angles of the first to third joints (the angles of the joints at which the motors P1, P2, R3 rotate), they can be expressed geometrically as follows. _{R (xi) = (C 2} · C 3 · C 4 + S 2 · C 4) L + S 2 · l 3 ··· (1) R (yi) = {- (C 1 · S 2 · S 3 + S 1 · C ₃ ) S ₄ + C ₁ · C ₂ · C ₄ ｝ L + C ₁ · C ₂ · l ₃ (2) R (zi) = ｛-(C ₁ · S ₂ · S ₃ + S ₂ · C ₃ ) S ₄ + C ₁ · C ₂ · C ₄ ｝ L + C ₁ · C ₂ · l ₃ (3) where C is cos θ, S is sin θ, and L is the distance from the movable arm 14 to the tip of the hand 18. total length (l ₄ to l ₈ in FIG. 2)
And the subscripts of C and S are the motors P1, P2, R3,
P4, R5, P6, R7 intended to represent the joint number corresponding to the subscript of l is the length from the wrist of each link length l ₄ to l ₇ or hand 18 of the movable arm 14-17 shown in FIG. 2 fingertips it is intended to be representative of l _8.

Using this, the rolling angle Δl, pitching angle Δm, and yawing angle Δn of the hand 18 with respect to the target attitude can be obtained as correction values from the following equations. Δl = cos ⁻¹ (z / l ₈ ) −cos ⁻¹ (R (zi) / l ₈ ) (4) Δm = cos ⁻¹ (y / l ₈ ) −cos ⁻¹ (R (yi) / L ₈ ) (5) Δn = cos ^-1 (x / l ₈ )-cos ^-1 (R (xi) / l ₈ ) (6) where x, y, and z are each is a coordinate component of the coordinates of the tip of the hand 18 to be identified from the CAD data and the recognition image of the CCD camera 45, l ₈ is the length from the wrist of the hand 18 to the hand.

As described above, since the rotation angles Δl and Δm of the hand 18 in the tilt direction are constantly corrected to the target values based on the feedback information from the tilt sensors 31 and 32, (4) and ( The calculation using the expression 5) becomes unnecessary. In addition, the hand 18 may also be affected by load weight (bending of the arm due to this), drive system backlash, and the like.
Although the position of the tip changes, feedback control based on the detection information of the inclination sensors 31 and 32 is performed,
It is possible to eliminate the effects of the bending of the arm and the backlash of the drive system.

The deviation of the yaw angle from the target attitude of the hand 18 is repeatedly performed by the computer 27 or 47 using the correction value Δn obtained from the equation (6) by the above-described image processing. And the tip coordinates of the robot 10 are quickly and accurately matched. As a result, the hand 18 takes the target posture. Incidentally, assuming that the detection accuracy of the inclination sensors 31 and 32 is, for example, ± 0.088 degrees and that of the rotation angle of the triangle L1L2L3 is ± 0.4 degrees, the accuracy of the attitude control is ± 0.8 degrees. realizable.

Further, by correcting the yawing angle of the hand 18, the recognized image of the triangle L1L2L3 formed by the three light emitting points at the current position and the recognized image of the triangle P1P2P3 corresponding to the target attitude of the hand 18 are almost equal. Since the shape becomes similar, the position of the hand 18 with respect to the intermediate reference position is determined by the ratio of these sizes, for example, the ratio of the length of each side, and the known distance S1 from the CCD camera 45 to the intermediate reference position. For example, the distance from the CCD camera 45 to the light emitting point L1 can be accurately detected. Therefore, the hand 18 is maintained while continuing the above-described posture control.
The required tip position can be achieved by bringing the position of the target closer to the target position.

Next, the grip control system will be described with reference to FIGS. The grip control system 60 of the grip fingers 21 and 22 performs grip control independently of the main computer 27. As shown in FIG. It has two pairs of strain gauges 81, 82, 83, 84 attached.
These strain gauges 81 to 84 are connected to each other so as to form a 4-gauge bridge circuit 61 together with a power supply battery 62, as shown in FIG. 9B. The output voltage of this bridge circuit 61 is a strain gauge amplifier. The signal is amplified at an amplification rate of about 2000 times by the H.63 and becomes a distortion signal. This distortion signal is converted into a high-speed 8-bit A / D (Analog-to-Di-
The digital signal is converted by a gital (analog / digital) converter 64, and a microprocessor (Central Processi
ng Unit) 67. The CPU 67 calculates the gripping force of the gripping finger portions 21 and 22 based on the distortion signal, compares the calculation result with a preset gripping force programmed in advance, and instructs the motor driver 68 to increase or decrease the gripping force (command). Signal). This motor driver 6
In step 8, the DC servomotor 69, which is a micromotor, is driven by PWM according to a command from the CPU 67 to perform normal rotation and reverse rotation, and the gripping finger portions 21 and 22 are opened and closed via the speed reduction mechanism 71. Then, such a control program is looped at a high speed, so that the gripping fingers 21 and 22 grip the gripping target object (suture needle) with a constant gripping force as shown in FIG. 10, for example. In FIG. 10, the vertical axis represents the gripping force (N), and the horizontal axis represents the time (seconds). 3) shows characteristics when gripping control is performed. Also, in FIG.
This shows characteristics when a needle having a diameter of 0.5 (mm) is similarly gripped and controlled.

In the above-described apparatus, the plane SL including the three image recognition points L1, L2, and L3 has a Z axis (vertical axis).
Is calculated, and the attitude of the hand 18 is detected based on the calculation result and the rotation angles (rolling angle, pitching angle) detected by the inclination sensors 31 and 32. Therefore, the processing of the image data recognized by the CCD camera 45 mainly calculates the deviation angle (value to be corrected) of the plane SL including the three image recognition points L1, L2, and L3 from the target posture around the Z axis. The speed of the posture detection can be increased by quick calculation. Also, the image recognition point has three LEDs 41,
Since these are constituted by 42 and 43, the three image recognition points L1, L2, and L3 are always clearly defined by adjusting the focus of the CCD camera 45, etc., even if the background color or the brightness of the use environment changes. Images can be recognized. Furthermore, since the inclination sensors 31 and 32 are attached to the hand 18, the rotation angle of the inclination direction of the hand 18 with respect to the horizontal plane can be directly detected with a simple configuration with high accuracy, and a complicated external sensor is not required. A simple and highly accurate robot controller can be realized.

Further, in the present embodiment, the rotation angle in the tilt direction directly detected by the first and second tilt sensors 31 and 32 and the calculation result of Δn by the computer 47 (rotation angle calculation means) To computer 47
By providing the feedback information to the (control means), the correction values Δl and Δm are simply obtained from the detection values of the first and second tilt sensors 31 and 32 and the rotation angles corresponding to the target attitude, as described above. As a result, quick image processing and high-accuracy posture detection can be performed, thereby enabling quick and high-accuracy posture control. Moreover, by directly detecting the rotation angle in the inclination direction with respect to the horizontal plane by the inclination sensor, it is possible to compensate for bending of the arm, backlash of the drive system, and the like. In addition, using the image information at the time of the attitude control, 3
Recognition image of a triangle formed by three light-emitting points and hand 1
The position of the hand 18 can also be easily detected by comparing with a recognized image of a triangle corresponding to the target posture of No. 8;
It is possible to provide a robot control device that can execute quick and accurate tip position control.

[0031]

According to the first aspect of the present invention, the processing of the image data is sufficient to calculate the rotation angle of the plane including the image recognition point around the vertical axis, and the quick attitude detection is performed. Can be. According to the second aspect of the invention, three image recognition points constituted by light emitting elements can be clearly recognized.

According to the third aspect of the present invention, the rotation in the inclined direction with respect to the horizontal plane can be directly detected with a simple configuration with high accuracy, and the detection accuracy can be increased. According to the invention according to claim 4, the rotation angle in the tilt direction directly detected by the first and second tilt sensors and the calculation result of the rotation angle calculation means are used as feedback information to the control means. Therefore, the processing of the image data only needs to calculate the rotation angle of the plane including the three image recognition points about the vertical axis, and the rotation of the plane in the inclination direction with respect to the horizontal plane is the first and second.
With a simple configuration composed of the inclination sensor described above, detection can be directly performed with high accuracy. As a result, quick and accurate attitude control can be performed.

According to the fifth aspect of the invention, by accurately recognizing the image of the light emitting points of the three light emitting elements, the rotation angle of the plane including these light emitting points about the vertical axis is accurately calculated. Since the calculation result and the rotation angle in the tilt direction directly detected by the first and second tilt sensors are fed back to the control means, quick and highly accurate attitude control can be performed. Compensation for backlash and the like of the drive system can also be made.

According to the sixth aspect of the present invention, there is provided a simple robot control device which does not require a complicated external sensor. According to the invention according to claim 7, the recognition image of the triangle formed by the three light emitting points at the current position and the recognition of the triangle formed corresponding to the three light emitting points at the intermediate reference position of the hand. By comparing with the image, the position of the hand with respect to the intermediate reference position can be easily detected.

[Brief description of the drawings]

FIG. 1 is a schematic perspective view of an entire articulated robot showing an embodiment of a posture detection device, a posture control device, and a robot control device according to the present invention.

FIG. 2 is an explanatory diagram showing a link connection and a joint operation direction of the articulated robot in FIG. 1;

FIG. 3 is an overall configuration diagram of a robot control device that controls the robot of FIG. 1;

FIG. 4 is a perspective view of a posture detection device that detects the posture of the hand unit of the robot in FIG. 1;

FIG. 5 is an operation explanatory view of the attitude detection device shown in FIG. 4;

FIG. 6 is an explanatory diagram of another operation of the attitude detection device shown in FIG.

FIG. 7 is a perspective view of a device for detecting a posture and a tip position of a hand unit of the robot in FIG. 1;

FIG. 8 is a block diagram of a grip control system of the articulated robot of FIG. 1;

FIG. 9 is a perspective view of a gripping finger portion of a hand portion of the articulated robot of FIG. 1;

FIG. 10 is a circuit diagram of a gripping force detection circuit mounted on the gripping finger unit of FIG. 9;

FIG. 11 is a characteristic diagram showing characteristics at the time of grip control of the hand portion in FIG. 9;

[Explanation of symbols]

Reference Signs List 10 double-armed articulated robot (work support robot) 12, 13, 14, 15, 16, 17 movable arm 18 hand 21, 22 gripping finger 27 computer (control means, rotation angle calculation means) 31, 32 tilt sensor 41, 42, 43 LED (light emitting element) 45 CCD camera (image recognition means) 46 image processing board (image recognition means) 47 computer (image recognition means, rotation angle calculation means) L1, L2, L3 Light emission point (image recognition point) P1 , P2, R3, P4, R5, P6, R7 DC servo motor for driving

Claims (7)

[Claims] 1. A first and a second inclination sensor for detecting respective rotation angles of a target whose posture can be changed around first and second horizontal axes orthogonal to each other, and provided on the target. Three image recognition points (L1, L2,
L3), image recognition means for image-recognizing the three image recognition points, and based on information recognized by the image recognition means, the image recognition means comprising: a vertical axis orthogonal to the first and second horizontal axes; Rotation angle calculation means for calculating a rotation angle of a plane including three image recognition points, and a calculation result of the rotation angle calculation means and a rotation angle detected by the first and second tilt sensors. A posture detection device for detecting the posture of the object in a coordinate system in which the vertical axis and the first and second horizontal axes are three orthogonal axes, based on 2. The attitude detecting apparatus according to claim 1, wherein said image recognition point is constituted by a light emitting element attached to said object and emitting light. 3. The apparatus according to claim 2, wherein said first and second inclination detecting means are attached to said object, respectively, for detecting respective rotation angles of said object about said first and second horizontal axes. The attitude detecting device according to claim 1, further comprising an inclination sensor. 4. An object whose attitude can be controlled, and control means for controlling the attitude of the object to control the tip position of the object;
And a posture detecting means for detecting the posture of the object and feeding it back to a control means. The posture detecting means, wherein the posture detecting means detects the position of the object around first and second horizontal axes orthogonal to each other. First and second tilt sensors for detecting the respective rotation angles; and three image recognition points (L1, L2,
L3); an image recognizing means for recognizing the three image recognizing points; based on the recognition information by the image recognizing means, the three image recognizing points about a vertical axis orthogonal to the first and second horizontal axes. Rotation angle calculation means for calculating a rotation angle of a plane including the image recognition point, based on the calculation result of the rotation angle calculation means and the rotation angles detected by the first and second tilt sensors. An attitude control device for detecting an attitude of the object in a coordinate system in which the vertical axis and the first and second horizontal axes are three orthogonal axes. 5. An attitude control means for controlling the attitude of an arm and a hand supported by the arm to move a tip of the hand to a target position and to control the hand to a target attitude, and a tip of the hand. A posture detecting means for detecting a posture of the hand and feeding it back to the control means, wherein the posture detecting means comprises: a rotation angle of each of the hands around first and second horizontal axes orthogonal to each other; First and second tilt sensors for detecting the position, three light-emitting elements attached to the hand and emitting light, and image recognition means for image-recognizing light-emitting points (L1, L2, L3) of the three light-emitting elements. And calculating a rotation angle of a plane including the three light-emitting points around a vertical axis orthogonal to the first and second horizontal axes based on the recognition information by the image recognition means. Angle calculation means, based on the calculation result of the rotation angle calculation means and the rotation angles detected by the first and second tilt sensors, the vertical axis and the first and second horizontal axes. A robot control device for detecting the posture of the hand in a coordinate system having three orthogonal axes. 6. The robot controller according to claim 5, wherein said first and second inclination sensors are respectively attached to said hand. 7. An attitude control means for controlling the attitude of the arm and the hand supported by the arm, moving the tip of the hand to a target position, and controlling the hand to a target attitude, and the tip of the hand. And a detecting means for detecting the position and attitude of the hand and feeding it back to the attitude control means, wherein the detecting means is attached to the hand and emits three light-emitting elements; and the three light-emitting elements Image recognition means for recognizing light-emitting points (L1, L2, L3) of the element from a predetermined direction; and a recognition image of a triangle formed by the three light-emitting points based on the recognition image of the image recognition means. The position of the tip of the hand in the predetermined direction is compared with a triangular recognition image formed corresponding to the three light emitting points at the intermediate reference position in the predetermined direction. A robot control device characterized by detecting.