JPH0429517B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application Many of the force sensors used in systems in which a force sensor is attached to a robot are constructed using a strain gauge. In many cases,
There is a drift in the force sensor output, and the drift fluctuates as the operating environment changes, such as temperature, and as the operating time elapses. In order to obtain the absolute value of the force acting upon force detection, it is necessary to remove varying drift from the force sensor output. The present invention relates to a method for compensating for drift in a force sensor attached to a robot.

Prior Art In order to compensate for the drift of a force sensor, the drift is first determined by some means and then
A method used was to subtract a constant value of the drift from the force sensor output.

Problems to be Solved by the Invention When the drift of a force sensor changes over time, the method of subtracting the drift from the force sensor output by setting the drift as a constant value causes an error equal to the change in drift. The drift of a force sensor does not change significantly over a short period of time, but in systems that operate continuously for long periods of time, such as industrial robots, stopping operation to determine the drift is disadvantageous in terms of operating efficiency. Furthermore, since the drift does not necessarily change in the same pattern every day, it is necessary to compensate for the varying drift while operating the system.

Means for Solving the Problems In order to solve the above problems, in the drift compensation method of the present invention, an accurate drift is determined in advance and stored as an initial drift, and at the same time, the force at the operating point that appears during work during robot operation is calculated. After measuring and storing the sensor output by causing the robot to operate under the same load conditions as during the robot's work, the force sensor output is measured when the robot operates at the aforementioned operating point during the work, and this force sensor output and the memorized value are The difference between the force sensor outputs is stored as a drift change, and when force is detected, the initial drift and the drift change are subtracted from the force sensor output at that time, thereby removing the drift from the force sensor output. It's getting better.

Effect The force sensor output that is measured at the calibration measurement point at the same time as the initial drift is measured under the same load conditions as when the force sensor output is measured at the calibration measurement point during robot operation. The difference between the force sensor output and the force sensor output is considered to be a drift change in the force sensor. In other words, d(t _c ) = d(t _p ) + {c(t _c ) − c(t _p )} where d(t): Drift at time t c(t): At the calibration measurement point at time t Force sensor output t _p : Initial drift measurement time t _c : Force sensor output measurement time of calibration measurement point When detecting force, remove drift using the following calculation.

f (t _c + Δt) = s (t _c + Δt) - d (t _c ) where f (t): Force sensor output after removing drift at time t s (t): Force sensor output at time t Δt: For calibration The elapsed time from the time when the force sensor output at the measurement point was measured and the drift change was last updated to the time when the force sensor output was measured for force detection.

Even if a long time has passed since the initial drift measurement, if the elapsed time Δt from the update of the drift change performed at the calibration measurement point is short, drift compensation can be performed with less error when detecting force.

Conversely, errors can be reduced by measuring the force sensor output at the calibration measurement point and updating the drift change at sufficiently short intervals.

Embodiment FIG. 1 is a block diagram of a robot system with a force sensor according to an embodiment of the present invention, and FIG. 2 is a perspective view of the robot. The robot 1 is a vertically articulated robot capable of movement in six degrees of freedom, and can be operated by arbitrarily specifying the position and posture of the hand 10 within the allowable range of motion. A force sensor 2 is attached between the robot tip 9 and the hand 10. The force sensor 2 is a 6-axis force sensor, and calculates the force and moment applied to the hand 10 in three orthogonal axes directions of the sensor coordinate system as shown in FIG.
f _x , f _y , f _z and moments about the three axes
It can be detected and measured as m _x , m _y , and m _z . The operation of the robot 1 is controlled by a control device 8, and the force sensor output is measured by the control device 8. The control device 8 operates according to a built-in computer program, and the robot repeats operations according to the work according to the programs and data taught and set in the control device.

Select one or more appropriate points during the operation and use them as measurement points for calibration. Using the teaching/setting program and data, the following procedure is executed at the calibration measurement point during execution of the work operation. That is, the force sensor output is measured, the difference from the initial force sensor output at the calibration measurement point is taken as a drift change, and the result obtained by adding this to the initial drift is stored in the memory element 7 in the control device 8 as the drift at that point. Update. A flowchart of this procedure is shown in Figure 4.

Next, the drift compensation method according to the present invention will be explained based on FIG. Before starting the work operation of the robot, the drift is first determined by the means (initial drift determination means) 3 to be described later, and stored as the initial drift in the memory element 7 of the control device 8, and at the same time, the force sensor output at the calibration measurement point is calculated. is measured (calibration measuring point force sensor output measuring means 4) under the same load conditions (speed, load, etc.) as when the teaching program is executed and the robot 1 is operated, and the calibration measuring point initial force sensor output is measured. The data is stored in the memory element 7 in the control device 8 as a. The drift change at the start of the work operation is stored in the memory element 7 in the control device 8 as zero. Then let the robot do the work. This work consists of repeated operations, and one cycle of operation includes the calibration measurement points, and each procedure shown in the flowchart of Fig. 4 is performed according to the set program and data. Performed at calibration measurement points. When detecting force during this work, the drift is removed by subtracting it from the force sensor output (drift removal means 5). In this way, the drift is updated every time the procedure in the flowchart in Figure 4 is executed, so even if the drift changes while the robot is continuously operated, the error in force detection due to the drift change will be reduced. Can be made smaller. In addition, in FIG. 1, 6 is a robot operation control means.

Next, the means for determining the initial drift will be explained. First, three appropriate hand postures are selected and the force sensor output in each hand posture is measured. Then, perform the following calculations.

Consider the relationship between the acting force and the force sensor output using the following equation.

F _O = F + F _W (θ) + D where F _O : Output vector of the force sensor F: Vector of force acting on the hand other than gravity F _W (θ): Force sensor coordinates of the factor due to gravity applied to the hand at hand posture θ System vector D: Drift vector of force sensor When only gravity is applied to the hand, F _O = F _W (θ) + D.

The gravity acting on the hand always has a constant magnitude and direction in the absolute coordinate system, and F _W (θ) is the gravity acting on the hand transformed according to the hand posture θ.

It is assumed that the hand coordinate system and the sensor coordinate system match. If they do not match, the hand coordinate system and the force sensor coordinate system have a certain relative positional relationship, so coordinates are exchanged. In the hand posture θ _i (i=1, 2, 3), the unit vectors in the positive directions of the X, Y, and Z axes of the hand coordinate system are N _i =(n _xi , n _yi , n _zi ) in the absolute coordinate system, respectively. ) ^T O _i = (o _xi , o _yi , o _zi ) ^T A _i = (a _xi , a _yi , a _zi ) ^T. The weight on the hand is W
Assume that the center of gravity of the hand is at Gr=(g _hx , g _hy , g _hz ) ^T in the hand coordinate system. At this time, F _W (θ) is F _W (θ) = F _x 〓 F _y 〓 F _z 〓 M _x 〓 M _y 〓 M _z 〓 = −Wn _zi o _zi a _zi a _zi g _hy −o _zi g _hz n _zi g _hz −a _zi g _hx o _zi g _hx −n _zi g _hy , and the force sensor output at hand posture θ _i (i=1, 2, 3) is f _pi (f _xi , f _yi , f _zi , m _xi , m _yi , m _zi
)
Then, the drift D=(d _fx , d _fy , d _fz , d _nx , d _ny , d _nz ) ^T is obtained by, for example, the following calculation.

W=(f _x1 +f _y1 +f _z1 −f _x2 −f _y2 −f _z2 )/(n _z1 +o _z1
+a _z1 −n _z2 −o _z2 −a _z2 ) d _fx =1/2{f _x1 +f _x2 +w・(n _z1 +n _z2 )} d _fy =1/2{f _y1 +f _y2 +w・(o _z1 +o _z2 )} d _fz = 1/2 {f _zi +f _z2 +w・(a _z1 +a _z2 )} H=θ −a _z1 +a _z2 o _z1 −o _z2 θ −a _z1 +a _z3 o _z1 −o _z3 a _z1 −a _z2 θ −n _z1 +n _z2 a _z1 −a _z3 θ −n _z1 +n _z3 −o _z1 +o _z2 n _z1 −n _z2 θ −o _z1 +o _z3 n _z1 −n _z3 θ q _hx q _hy q _hz =H ⁺ m _x1 −m _x2 m _y1 −m _y2 m _z1 −m _z2 m _x1 −m _x3 m _y1 −m _y3 m _z1 −m _z3 d _nx =1/3{ ₃ Σ ⁱ⁼¹ m _xi +w ₃ 〓 ⁱ⁼¹ ( a _zi g _hy −o _zi g _hz )} d _ny =1/3{ ₃ Ζ ⁱ⁼¹ m _yi +w ₃ 〓 ⁱ⁼¹ (n _zi g _hz −a _zi g _hx )} d _nz =1/3{ ₃ Σ ⁱ⁼¹ m _xi + w ₃ 〓 ⁱ⁼¹ (o _zi g _hx −n _zi g _hy )} However, H ⁺ is the pseudo inverse matrix of H (Moore−
Penrose's inverse matrix).

Effects of the Invention As described above, the present invention employs a method of updating the drift value to be subtracted during force detection while the robot is working with the force sensor still installed in the robot. Even if sensor drift changes over time, it can be removed within tolerance.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of the configuration of the present invention, FIG. 2 is a perspective view of a robot system with a force sensor according to an embodiment of the present invention, and FIG. 3 is an explanatory diagram of a six-axis force sensor.
4 and 5 are flowcharts of one embodiment of the present invention. DESCRIPTION OF SYMBOLS 1... Robot, 2... Force sensor, 3... Initial drift determining means, 4... Calibration measurement point force sensor output measuring means, 5... Drift removal means, 6...
Robot operation control means, 7... Memory element, 8...
Control device, 9... 6-axis force sensor, 10... hand.

Claims (1)

[Scope of Claims] 1. In a robot system with a force sensor consisting of a robot, a force sensor attached to the robot, and a control device, an accurate drift of the force sensor is determined in advance and the control device calculates the accurate drift of the force sensor as an initial drift. At the same time, one or more points at which the robot operates or stops during work are selected as calibration measurement points, and the robot is operated or stopped under the same load conditions as during work to perform the calibration measurements. After measuring the force sensor output at a point and storing it in a memory element in the control device, when the robot moves or stops at the calibration measurement point during actual robot operation, the force sensor output at the calibration measurement point is measured. The difference between this force sensor output and the force sensor output stored in the memory element is stored and updated in the element as a drift change, and when detecting force, the initial drift and the force sensor output at that time are updated. A method for compensating for force sensor drift in a robot, characterized in that a time-varying drift is removed from a force sensor output by subtracting the drift change. 2 The robot can be operated by arbitrarily specifying the hand posture within the operating range, and a force sensor is installed between the tip of the robot and the hand, and the initial drift is caused by multiple 2. The force sensor drift compensation method for a robot according to claim 1, wherein the robot is caused to assume a hand posture, the force sensor output in each hand posture is measured, and calculation is performed based on the output value.