JPH0482683A - Robot hand grip control method - Google Patents

Abstract

PURPOSE:To shorten the teaching time by allowing all joints of each finger to reach a first holding force of previously teached holding forces on plural stages which correspond to all the joints of each finger, thereafter allowing all the joints of each finger to generate a holding force corresponding to a next stage, and finally generating a specified holding force by repeating the operation. CONSTITUTION:After an arm is moved to a previously teached position, a first joint 211 of each finger is moved to a position P1 at which a first holding force out of predetermined holding forces on plural stages is reached, while the holding force to which the first joint 211 is subjected is monitored. After holding forces to which first joints of all fingers are subjected reach the first holding force, a second joint 212 of each finger is moved to a position P2 at which the first holding force is reached. After all joints of each finger reach the first holding force by repeating the same operation, the first holding force is switched to a second holding force, and the same operated repeated. Further, the operation repeatedly performed on all joints 211-213 of all fingers until a final holding force is detected.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a grasping control method for a robot hand, and more specifically, to a method for grasping and grasping an object with a multi-jointed, multi-fingered robot hand. The present invention relates to a grasping control method for a robot hand that makes it possible to shorten motion teaching time even when using a robot hand.

[Conventional technology]

When a robot is to perform a certain action, it is common to teach the order and content of the action in advance, store it in a storage device within a control device, and then sequentially read out the stored content. However, the location of the object of work performed by the robot is not necessarily the same as in the teaching stage;
It often has some deviation. Particularly in cases where the workpiece is brought in by an adjacent robot or belt conveyor, such as in an automated production line, it is important from the perspective of improving productivity to take into account some degree of deviation and use equipment that can handle it. It is.

In addition, robots are often used as so-called hand robots that grasp and move objects, and various types of multi-joint robot hands are used.

For example, Japanese Patent Application Laid-Open No. 62-124891 r Automatic Gripping Device discloses a robot hand that has three fingers arranged at each vertex of an isosceles triangle and is capable of grasping objects of various shapes. ing.

[Problem to be solved by the invention]

However, when grasping and grasping with the above-mentioned robot hand, a problem arises in that it takes a lot of time to teach the movement.

In other words, there are two ways to grasp an object with a robot hand: ``pinch'' it with the fingertips of the robot hand, and ``grasp it'' in the fingers. Although this is well known, in the above-mentioned robot hand, it is necessary to teach the movement of each joint of the master finger, and the time required for teaching becomes long.

Furthermore, if the position of the object to be grasped deviates from the taught position, not only will it not be possible to grasp a fixed portion of the object to be grasped in an accurate posture, but there is a possibility that the robot hand or the object to be grasped may be damaged. also occurs.

Therefore, in view of the above-mentioned problems, the present invention provides an articulated, multi-fingered robot hand that is capable of shortening the teaching time and grasping the object in the correct posture even when grasping the object by grasping it. The purpose of this invention is to provide a grip control method.

[Means to solve the problem]

The present invention applies the next stage of gripping force after the fingers of the robot hand follow the object and lightly grasp it, even if the object's position deviates from the taught position. Each time the gripping force of the joint near the palm of the master finger reaches the first gripping force of the multiple levels of gripping force taught in advance, control is repeated to operate the next joint, and the gripping force of the joint near the palm of the master finger is repeated. After all the joints reach the first gripping force of the plurality of gripping forces taught in advance corresponding to all the joints of the master finger, the next stage of gripping force is generated in all the joints of the master finger. By repeating the operation, a predetermined gripping force is finally generated.

[For production]

The gripping control method for such a multi-joint, multi-fingered robot hand involves moving the arm to a pre-taught position and then gripping the first joint of the master finger while monitoring the gripping force applied to the first joint. The grip is moved until the first grip force is reached among the predetermined multiple stages of grip force.

After the gripping forces applied to the first joints of all fingers reach the first gripping force, the second joint of the master finger is moved until it reaches the first gripping force.

Thereafter, after all the joints of the name fingers reach the first gripping force through the same operation, the gripping force is switched to the second gripping force and the same operation is repeated.

This operation is then repeated until the final gripping force is detected at all joints of all fingers.

Therefore, by setting the first gripping force to a value that allows detection of contact with the object, the first finger that touches the object can detect the first gripping force when all the remaining fingers touch the object. The object can be reliably gripped by waiting until the grip is reached and then increasing the gripping force sequentially.

〔Example〕

FIG. 1 is a diagram showing an example of a robot hand attached to the tip of a robot arm.

This robot hand has three joints 211°21
The first finger 21 and the second finger 22 are arranged so that their central axes are parallel to each other, and the third finger 23 is sandwiched between them. The central axis is arranged at 20 to 60 degrees with the central axes of the first finger and the second finger, and the fingertips are arranged to face the fingertips of the first finger and the second finger.

Figure 2 shows the famous robot hands 21 and 22 mentioned above.
.

The configuration of a control device for each of 23 joints 211, 212, and 213 is shown.

Further, FIG. 3 shows the configuration of the entire control device for the robot hand shown in FIG. 1.

That is, the robot hand shown in Fig. 1 is composed of three fingers each having three joints, but the control device for each joint is modularized, and the three joint control devices shown in Fig. 2 can be used as a single unit. A finger control device corresponding to each finger is configured, and the three finger control devices and the robot hand control device constitute a control device for the entire robot hand.

As shown in FIG. 2, the control device for each joint is such that the driving force of the electric motor 70 is transmitted from a drive pulley 72 attached to the shaft of the electric motor 70 via a reducer 71 to a wire lobe 73 serving as a driving force transmission means. The signal is transmitted to a joint pulley 74 fixed to a link mechanism 75 of the joint.

The gripping force received by the joint pulley 74 is
The torque is detected by a torque detector (not shown), such as a strain gauge, installed on a beam-like rope piece 77 that fixes the idle pulley 76, which is installed approximately halfway between the joint pulley 74 and the joint pulley 74. The rotation angle is determined by the angle sensor 7 attached to the joint pulley 74.
Detected at 8.

The output of this angle sensor 78 is compared with the angle target value 83, and the deviation between the two is multiplied by a virtual spring constant in a multiplier 84 to obtain a virtual torque command 92.

That is, in order to simultaneously control the angle and torque of the joint, a virtual spring constant is provided to adjust the ratio of angle control and torque control.

The virtual torque command is added to the torque target value 85 to become a torque command 86, and a corrected torque command 87 is calculated from the deviation from the output of the torque detector attached to the beam-like rope piece 77, and is input to a compensator 88.

The compensator 88 outputs a current command value 80 based on the corrected torque command 87 and the estimated motor rotation speed 91. Based on this current command value 80, the output current of the variable constant current source 79 is controlled, and the driving force of the electric motor is controlled.

The compensator 88 calculates the estimated motor rotation speed from the motor voltage 89 and the current command value 80 based on the following.

When the rotor inductance of the motor can be ignored, the following equation holds.

Nm = (Vg-1m - Rm) /Kv (1) where Nm: Estimated motor rotation speed Vg: Actual motor terminal voltage m: Variable current source output current Rm: Motor rotor resistance Kv: Motor back electromotive force constant Here Since the variable current source 79 has constant current characteristics, the variable current source output current Im is maintained at a constant current value proportional to the current command value 80 (Imt) even if the motor load fluctuates. Since the resistance Rm and the back electromotive force constant Kv of the motor are known, it is possible to calculate the estimated motor rotation speed from the following equation.

Nm=(Vg-a・Imt-Rm)/K
(2) where α: Proportional constant of variable constant current source By using this compensator 88, damping can be applied to the joint control system without using a rotation speed sensor such as a tachometer, which is conventionally used, and stable control can be achieved. It becomes possible to configure the system.

In the robot hand grasp control device according to the present invention, the rotation angle command of each joint output from the robot hand control device 5 and the two-stage torque target value for each joint of multiple fingers, that is, contact with the grasped object. a first torque target value for detecting that the object has been gripped, a second torque target value for reliably gripping the object to be gripped, and a beam-like rope piece 77.
A gripping control device 16 is additionally installed, which receives the output of a torque detector attached to the grip and outputs an angle target value 83 and a torque target value 85 to the control device for each joint.

FIG. 4 is a diagram showing the principle of grip control of the multi-joint, multi-fingered robot hand according to the present invention.

That is, FIG. 4 shows the grasping motion of the first finger 21 of the three fingers of the robot hand shown in FIG. 1.

In the first step, the finger 21 is moved to an initial position P (counterclockwise or counterclockwise) with the first joint (innermost joint) 211 as the center. The first joint 211 is rotated from the position P1 to the position P1 where the torque detector corresponding to the first joint 211 detects the first torque target value for the first joint.

In the second step, the finger 21 is moved in the same direction as the first step with the second joint 212 as the center at a position P2 where the torque detector corresponding to the second joint 212 detects the first torque target value for the second joint. Swirl until.

In the third step, the fingers 21 are similarly rotated from position P2 to P3 about the third joint 213, and the grasping motion is completed in the fourth step.

FIG. 5 shows a gripping control routine executed by the gripping control device, and is executed at regular intervals.

In step 501, the direction and speed of finger movement are instructed to the robot hand. This instruction can be given by a method of being taught in advance, a method of giving instructions each time, etc., and an appropriate method can be used.

In step 502, it is set to n-1 and the movement of the first joint (the innermost joint) is started.

In step 503, the finger is rotated in the designated direction around the n-th joint, and in step 504, eight torque detectors corresponding to the n-th joint are read and the process proceeds to step 505.

In step 505, it is determined whether or not the current torque detected is greater than or equal to the first torque target value determined for the n-th joint. If the determination is negative, the process returns to step 503 and continues turning the finger.

If a positive determination is made in step 505, step 506
Stop rotating the finger around the n-th joint at step 50.
Proceed to 7.

In step 507, it is determined whether the first torque target value has been reached for all joints (in this example, three joints), and if the determination is negative, n is incremented in step 508, and step 503 Return to , and rotate your fingers around the next joint.

If a positive determination is made in step 507, step 509
and change the torque target value to the second torque target value,
Generate gripping force by gripping the robot hand.

Note that the first torque target value and the second torque target value are determined as follows.

That is, at the stage of teaching the robot hand how to operate, the torque required to reliably grip the object to be gripped is set as the second torque target value, and contact with this value can be reliably detected, and when contact is made, It is possible to determine the first torque target value by multiplying it by a certain percentage so as not to damage the gripped object.

Therefore, it is possible to modify this ratio while the robot hand is in operation, and the time required for the teaching stage can be significantly shortened.

Normally, setting the torque target value in two stages when grasping the object to be grasped is sufficiently practical, and if all the joints of each part detect the first torque target value, each operator's joint Since positioning is considered to have been completed, it is also possible to control each joint of each part to generate the second torque target value at the same time, and the time required for the gripping operation can be shortened.

It is also possible to set the torque target value in three or more stages, and to switch to the next stage torque target value each time each joint of each part reaches the previous stage torque target value.

Further, the torque target value can be set to a different value for each joint of each part in relation to the center of gravity of the object to be grasped.

Furthermore, in this embodiment, the gripping force applied to each joint of the operator is detected by a torque detector, but the gripping force can also be detected by a pressure sensor attached to the fingertip or a strain gauge that detects stress in each part. The present invention can be similarly applied to this case as well.

〔Effect of the invention〕

According to the present invention, when grasping and grasping an object with a robot hand, the position of the object, the grasping motion of each part and joint, and multiple stages of grip force can be taught. Even if the robot hand deviates from the taught position, all fingers can be brought into contact with the grasped object at the taught position and orientation in advance and then a grasping force can be generated, so there is no chance of damage to the robot hand or the object. This makes it possible to simplify teaching.

[Brief Description of the Drawings] Fig. 1 is a diagram showing an example of a robot hand, Fig. 2 is a diagram showing the configuration of a joint control device of the embodiment, and Fig. 3 is a diagram of the overall control device of the robot hand of the embodiment. FIG. 4 is a diagram showing the principle of grip control of the present invention, and FIG. 5 is a flow chart showing the grip control operation.

Claims (1)

[Claims] A multi-jointed, multi-fingered robot hand having a plurality of fingers each having one or more joints grasps an object by sequentially bending each finger from the joint near the palm. A grasping control method for each finger and each joint when grasping, wherein the grasping force of the joint of each finger near the palm reaches a first grasping force among a plurality of pre-taught grasping forces. At the same time, the control is repeated to move the next joint, and after all the joints of each finger reach the first gripping force of the plurality of gripping forces taught in advance for all the joints of each finger, A gripping control method for a multi-joint, multi-fingered robot hand, characterized in that the next stage of gripping force is generated at the fully open joint of each finger, and by repeating this operation, a predetermined gripping force is finally generated. .