JPH06131018A - Control method for robot - Google Patents

Abstract

PURPOSE:To realize a high speed follow-up property to a command value by adopting non-linear control on a locus using a ring buffer. CONSTITUTION:A command value ring buffer has the capacity which can accumulate request command positions theta<1>...thetaimax outputted by a command position arithmetic part extending from '0' to the maximum PMAX frequency portion at every prescribed calculation period required for movement of a distance between smaller distances Ak1 and Ak2 than a distance between middle points Ak and Ak+1 in middle points A1...Am at an equal interval of a locus on a route coordinate system (p) extending from the present point P0 to a target point P1. A non-linear control processing routine on the locus stores one after another a request command value on the ring buffer 6 at every calculation period, calculates an actual command position which can be realized with regard to each joint, and searches in what part on the ring buffer the derived actual command position exists. When an index increment is derived with regard to all the joints, it is determined as the index increment by selecting their minimum value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention expresses the tool tip position of an articulated robot in a Cartesian coordinate system, and the tool tip position at a specified speed along a straight line connecting two points defined in the Cartesian coordinate system. The present invention relates to a trajectory control method for a multi-joint robot in which a joint drive system including an actuator receives a joint angle or a joint angular velocity as a command value so as to move and independently configures a servo system for each joint. The present invention relates to a robot control method characterized in that a position command is controlled so as to always follow each joint so as to minimize the amount of deviation from the trajectory.

[0002]

2. Description of the Related Art Generally, a robot controller calculates a motion angle of each joint in advance in time series so that the tip of a robot arm moves on a target locus and performs servo control so as to realize this motion angle. However, it is already known in Japanese Patent Laid-Open No. 64-42713 to increase the position loop gain and to employ a non-linear servo control method for each joint in order to shorten the moving time. However, in this non-linear servo control method for each joint, there is a problem that the target trajectory and the robot arm tip trajectory deviate greatly as the speed becomes higher due to the non-uniformity and reduction of the position loop gain of each joint. It was inherent.

[0003]

In order to solve such a problem, in place of the non-linear servo control for each joint, the applicant of the present invention has disclosed an unpublished Japanese Patent Application No. 4-75634 in which the arm tip trajectory A method of adopting nonlinear control is proposed. This is because, by performing non-linear control on the locus of the arm tip, the position on the target locus that can be realized for any joint to the position servo loop system is always output as the command position, and in what speed range Even in this case, high trajectory accuracy can be realized without reducing the position loop gain of each joint.

The object of the present invention is to realize a position loop gain of each joint in any speed range by adopting a new robot control method called a non-linear control on a trajectory, which embodies this control method. A robot control that can output a commanded position on the road that can always be realized without reducing unevenly and that can achieve high trajectory accuracy if it follows the commanded value, that is, if it is a linear interpolation operation. To provide a method.

[0005]

Therefore, the present invention has solved the above-mentioned problems by providing a robot control method described in the claims.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the attached FIGS. FIG. 6 is an explanatory view of an exemplary articulated robot having two arms used in the control method of the robot according to the embodiment of the present invention. The joint angles of the robot shown in FIG. 6 are defined as θ ¹ and θ ² as shown in the figure, and a first coordinate system (x ₁ , x ₂ ) whose origin is the rotation center of the first arm is defined. A second coordinate system independent of the joint coordinate system of the robot is defined, a trajectory determined by the current position and the target position is given on the second coordinate system, and the joint drive system including the actuator is configured to detect the current point P ₀ and the target point. The trajectory determined by the point P ₁ is the second
On the path coordinate system p, the intermediate points A _{1 ...} of the loci on the path coordinate system p from the commanded moving speed v to the target point P ₁ commanded at each predetermined command cycle _. A _m
The servo amplifier for driving the angular joint determines the joint angles θ ¹ and θ ²
Alternatively, it is assumed that the joint angular velocity is received as a command value and the servo system is configured independently for each joint.

FIG. 1 shows a block diagram of the whole servo system independent for each joint including non-linear control on the locus. Reference numeral 1 is a command position calculation unit that calculates and determines a command position, 3 is a servo amplifier unit including a position loop, a velocity loop, and a current loop, 4 is an AC servomotor, and 5 is a position detector. Non-linear control processing unit 2 for performing non-linear control on a locus
Is between the command position calculation unit 1 and the servo amplifier unit 3, and the command position output from the command position calculation unit 1 (hereinafter referred to as “request command position”) is input to the servo amplifier by adding non-linear processing on the locus. It is output to the unit 3 (hereinafter referred to as actual command position).

FIG. 2 is a block diagram showing a detailed configuration of the non-linear control processing unit 2 on the locus. First, the maximum acceleration in the moving direction of the locus is obtained in advance from the allowable angular acceleration of each joint.
A feasible speed in the moving direction of the trajectory is obtained from the maximum acceleration, and it is determined whether or not the calculated midpoint on the trajectory can be achieved by the limitation of the maximum acceleration. Ε ⁱ in the figure
Is the command value deviation (required command position minus the actual command position), v ⁱ is the distance over which the command position can advance per unit time, that is, command value speed (subscript i means joint number, and v ⁱ is indicating a command value speed for number of joints), alpha ⁱ is a good amount or command value acceleration also command value speed v ⁱ per unit time of the joints and changes in joint number i. Here, for each constant calculation cycle required to move the distance between the intermediate points,
The non-linear processing routine is assumed to execute c scans.

First, in step 21, the command value deviation ε ⁱ is calculated for each axis, and in step 22 which is the control switching section, this and the value α ⁱ / of the linear non-linear switching point which is the allowable angular acceleration are calculated.
Compare with c ² . If ε ⁱ ≤ (α ⁱ / c ² ), the process proceeds to step 23, where linear processing is performed, and ε ⁱ > (α ⁱ /
If c ² ), the process proceeds to step 27 and nonlinear processing is performed.
As shown in FIG. 3, these processes are performed by the command value deviation ε during damping.
^This is to enable smoother positioning by switching from the non-linear control to the linear control when ⁱ reaches α ⁱ / c ² . The actual command position is determined by applying an acceleration limit to the command value speed v ⁱ obtained by the linear process or the non-linear process in step 25 and then adding the integration process in step 26, and the actual command position is input to step 28.

The series of processes on the flow chart will be described as above, but the Applicant has invented a "command value ring buffer" as one measure for actually realizing this. A conceptual diagram of the structure is shown in FIG. Between which a middle point between the A _k and A _{k + 1} at equidistant midpoint A _{1 ...} A _m of the trajectory on the path coordinate system p of the current point P ₀ in Fig. 6 to the target point P ₁ Distances A _k1 and A smaller than
_The required command position θ ¹ , θ output from the command position calculation unit 1 at every constant calculation cycle required to move the distance to _k2.
² , .. θ ^imax (imax = maximum number of joints) from 0 to maximum P _MAX
It has the capacity to accumulate up to the number of times.

The non-linear control processing routine on the locus stores request command values in the ring buffer 6 one after another at every constant calculation cycle (the storage location is hereinafter referred to as index: 0 to P _max ). The actual command position that can be realized is calculated by performing the process of FIG. 2 for each joint, and the position of the obtained actual command position on the ring buffer is searched. At the time of the search, the position data of the indexes that are adjacent to each other on the ring buffer are simply interpolated to derive the increment of the index that the joint may advance.
When the index increments that can proceed for all joints are obtained in this way, these minimum values are selected and this is determined as the current index increment. Non-linear control on the locus is established by selecting the minimum value here. When the index increment is determined, the actual command position is derived by simply interpolating the position data of the adjacent indexes, and this is output to the servo amplifier unit 3. In this way, during the reproduction operation, the required command position index advances on the ring buffer at a constant speed, and the actual command position index advances so as to chase after that.

The present invention will be specifically described below with reference to examples. FIG. 5 is a flowchart in which the block diagram of FIG. 2 is written down into an actual algorithm using a command value ring buffer. The control switching unit 7 discriminates between linear and non-linear, and if it is linear, it moves to the linear processing unit 8, where the actual command position is allowed to advance by the same amount as the distance that the request command position has traveled on the ring buffer. . That is, step 1
In step 4, the index increment ΔP ⁱ is output to the index determination unit 12 as ΔP ⁱ = 1 index. On the other hand, if it is non-linear, the process proceeds to the non-linear processing unit 9, the speed v ⁱ that may proceed first is calculated, and the acceleration limiting unit 10 imposes acceleration limitation. The acceleration limit value α ⁱ used here should be as close as possible to the actual ability value of the joint.

When v ⁱ is determined, the previous actual command position ACTUAL
^The position on the ring buffer where the data to be the new actual command position advanced from ⁱ to v ⁱ is searched for. In the flowchart, PD is an index in which the requested command position is stored, PA is an index immediately before that used in determining the actual command position, and R _ing ⁱ (N) is the position of the joint stored in address N of the index. The data. If there is desired position data between the addresses (PA + J-1) and (PA + J) as a result of the search, the index is obtained by simply interpolating the preceding and following position data. The index increment determination unit 11 determines the index increment ΔP ⁱ in this way.

More specifically, the position data that should be the new actual command position is the position that has changed from ACTUAL ⁱ by the speed v ⁱ . If the argument is J, the sign of v ⁱ is determined by the magnitude relationship between the stored R _ing ⁱ (PA + J) and ACTUAL ⁱ , so δ ⁱ = S _ign (R _ing ⁱ (PA + J)- ACTUAL ⁱ ) ・ v ⁱ
Then, the new actual command position can be expressed as ACTUAL ⁱ + δ ⁱ . What you want is not this value, but the index value of where this value is. Now ACTUAL ⁱ + δ ⁱ is R
It is assumed that the search has revealed that it is between _ing ⁱ (PA + J-1) and R _ing ⁱ (PA + J). Considering the loop with the argument J in FIG. 5, the index increment ΔP ⁱ to be derived is
^{ΔP i = (J-1)} + [ACTUAL i + δ i - R ing i (PA + J-
1)] / [ _Ring ⁱ (PA + J) _-Ring ⁱ (PA + J-1)] is obtained by adding the increments obtained by the simple interpolation calculation.

After repeating the above process for all joints, the index determination unit 12 finds the minimum value of ΔP ⁱ and sets it as ΔP min. When ΔPmin is obtained, the processing unit 12 also obtains a new actual command position index PA. Non-linear control on the locus can be realized by adopting the minimum value ΔPmin and calculating the actual commanded positions of all axes with PA determined by this. Once the new index PA and ΔPmin have been obtained, the actual command position calculation unit 13 then uses R _ing ⁱ (P
A) and R _ing ⁱ (PA + J) are [1- ΔPmin + int (ΔPmin)] vs [Δ
Pmin-int (ΔPmin)] and simply interpolate the actual command position ACTUAL ⁱ
Can be calculated. The actual command position ACTUAL ⁱ obtained in this way is a position that can be realized for all joints, and, of course, is a position on the target locus. And it goes without saying that the position loop gain never changes. The obtained actual command position ACTUAL ⁱ is output to the servo amplifier unit 3. The above processing is repeated at regular calculation intervals. In the (P _max +1) and subsequent processes, the index 0 on the ring buffer is sequentially overwritten.

[0016]

As described above, according to the present invention, by adopting a new robot control method called a non-linear control on a trajectory using a ring buffer, it is possible to accelerate or decelerate the next intermediate point on the ring buffer. By selecting and outputting data, it is possible to output the command position on the road that can always be realized without unevenly reducing the position loop gain of each joint in any speed range. It is possible to realize high locus accuracy with high value followability, that is, linear interpolation operation.

[Brief description of drawings]

FIG. 1 is a block diagram of an entire servo system that is independent for each joint and that includes non-linear control on a locus according to an embodiment method of the present invention.

FIG. 2 is a block diagram showing a detailed configuration of a nonlinear control processing unit 2 on the locus of FIG.

FIG. 3 is a graph for explaining control in which the nonlinear control processing unit 2 in FIG. 2 switches from nonlinear control to linear control when the command value deviation ε ⁱ reaches α ⁱ / c ² during damping.

FIG. 4 is a conceptual diagram of the structure of a command value ring buffer of the present invention.

5 is a flowchart in which the block diagram of FIG. 2 is written into an actual algorithm using the command value ring buffer of FIG.

FIG. 6 is an illustration of an exemplary two-arm articulated robot used in embodiments of the present invention.

Claims (2)

[Claims] 1. A joint including an actuator for expressing a tool tip position of a robot in a rectangular coordinate system and moving the tool tip position at a designated speed along a straight line connecting two points defined in the rectangular coordinate system. The drive system receives the joint angle and the joint angular velocity as command values, and defines a second coordinate system independent of the joint coordinate system of the multi-joint robot that independently configures the servo system for each joint, and determines it according to the current position and the target position. A locus is given on the second coordinate system, and intermediate points at equal intervals of the locus on the second coordinate system from a commanded moving speed to a target position commanded in each predetermined cycle are calculated, The maximum acceleration in the moving direction of the locus is obtained in advance from the allowable angular acceleration of each joint, the feasible speed in the moving direction of the locus is obtained from the maximum acceleration, and It is determined whether the intermediate point can be achieved by the limitation of the maximum acceleration, and if it is not possible, it is changed to the next intermediate point that can accelerate or decelerate the intermediate point commanded immediately before, and the new intermediate point is set. In the robot trajectory control method, the required command positions θ ¹ , θ ² , .. θ ^imax (commanded at each constant calculation cycle required to move a distance smaller than the calculated distance between the intermediate points on the trajectory. imax = maximum number of joints) is provided with a command value ring buffer having a capacity capable of accumulating 0 to the maximum number of P _MAX times, and the next intermediate point data that can be accelerated / decelerated on the ring buffer is selected and output. A robot control method characterized by the above. 2. A non-linear control processing routine on a locus for selecting and outputting the next intermediate point data capable of acceleration / deceleration on the ring buffer is stored on the ring buffer one after another at every constant calculation cycle. The required command value is stored as an index: 0 to _Pmax in a location to calculate the actual command position that can be realized for each joint, and search for where the obtained actual command position is on the ring buffer. However, at the time of search, by simply interpolating the position data of the indexes that are adjacent to each other on the ring buffer, it is derived as an increment of the index that the joint may advance, and the index increment that may advance for all the joints. When is obtained, select these minimum values, determine this as the index increment of this time, and simply interpolate the position data of the adjacent indexes. The method of claim 1, wherein the robot it derives the actual command position and outputs to the servo amplifier unit by the.