Among parallel robots, spherical robots occupy an important place.
Most applications of spherical manipulators can be found in orienting
devices, such as camera orienting and medical instrument alignment. A
spherical parallel robot is, in general, made up of the base platform and the moving platform. This mobile platform and base are connected by three equally spaced legs, each consisting of revolute joints only. The axes of all joints intersect at a common point, which is called the center of rotation. The motion of the moving platform is confined on the surface of a sphere centered at the rotation center. A spherical parallel robot provides 3 degrees of freedom of pure rotations. These robots have been the subject of many papers dealing with the structure, the problems of position and
velocity, workspace modeling, singularity analysis, and some problems with
the dynamic analysis. However, not all the important problems have been
solved. These concern the problem of accuracy. This paper presents
accuracy of the spherical parallel. In the considered spherical manipulator, each leg consists of five kinematic pairs. The kinematic accuracy is determined on the kinematic problem. The dynamic accuracy is estimated on the equation of motion. Examples of solving the problem of determining the positioning error of the output level are presented.
Introduction
Production automation is ensured by the use of robotic complexes and
systems. One of the important characteristics of assessing the quality of
functioning of robotic systems is the positioning accuracy. The task of ensuring this accuracy should be solved at the design stage of machinery and equipment.
Most of the production of light industry requires laser cutting and surface
treatment, welding, and medical robotics.
The positioning error of the output link, when it is repeatedly withdrawn to a
given point, is expressed for a spherical mechanism in angular units. The
positioning error is determined by systematic (inaccuracy of the control
system and deviations from the nominal dimensions of the links) and random
components (gaps in the joints, loosening in fasteners, temperature
fluctuations, robot vibrations caused by adjacent equipment, residual
vibrations, and fatigue deformations of links). The positioning accuracy is
determined by the positioning error of the output link (capture) when it is
repeatedly brought to a given point and is expressed in angular or linear
dimensions.
Systematic errors can be partially or completely compensated. Reducing
random errors should be considered in design and operation. Most robotic
systems consist of spatial mechanisms, including mechanisms of a parallel
structure (Merlet, 2000; Kong and Gosselin, 2007; Gogu, 2008). Modern control systems allow for the introduction of appropriate amendments. Also, the negative impact of vibrations at the base of the robot on the positioning accuracy can be reduced by organizational measures (i.e., coordination of the manipulation cycle with the cycle of the equipment that caused the vibration). The purpose of cycle matching is to establish the time interval between the impact and the final stage of the manipulation cycle. This allows vibrations to dim before the positioning is complete. Residual oscillations of the output link, resulting from insufficient rigidity of the link drives and instability of the control system, have a significant effect on accuracy. The nature of the residual vibrations depends on the dynamic properties of the manipulator.
When creating robots, they strive to perform links with a uniform, constant
cross section along the length. In this case, there is no need to describe
the links through the distribution of length over mass, moments of inertia,
and stiffness. In this case, the task of dynamic accuracy is simplified,
since, in this case, the Lagrangian equations describe mechanical systems with lumped parameters.
In most cases, the positioning error is determined from solving problems on the position of the mechanism. However, this approach is not universal, since
inaccuracies in the manufacture of links of mechanisms are inevitable, the
temperature of the working environment is not constant, the mechanism can be
located on a moving base, external influences can be present, and
instability of motion occurs at a given law of motion.
An interesting approach is one that takes into account the determination of
the accuracy of finding the executive body while taking into account the above restrictions and the possibility of their compensation. The paper shows a sequential approach to determining the kinematic, then dynamic, accuracy of
the executive body and the positioning errors during control. Thus, an
integrated approach to assessing the performance of the actuator is
presented.
Kinematic accuracy of the spherical mechanism
The article presents a solution to the problem of determining the
positioning error of the working body, using the example of a spherical
manipulator with 3 degrees of freedom of a parallel structure (Fig. 1). Such manipulators are used in orienting devices, in test benches,
medical robotic complexes, for processing spherical surfaces, in mixers, and for performing rotational movements (Huda and Takeda, 2007; Mianovski, 2007;
Leguay-Durand and Reboulet, 1997; Bai et al., 2016; Chaker et al., 2011; Nosova
et al., 2018). The paper considers a spherical manipulator with 3 degrees of freedom and investigates the issues of the kinematic and dynamic accuracy of the positioning of the executive body.
In the considered mechanism, each input chain link is connected to a rotary
motor. The output link is a platform rotating at point O around three
coordinate axes. The output coordinates are the angles of rotation of the
platform, with the angle α of rotation around the x axis, the angle
β of rotation around the y axis, and the angle γ of rotation
around the z axis. The generalized coordinates are the angles ϕ11, ϕ21, and ϕ31 of rotation of the input links, respectively, of the first kinematic chain and second and third kinematic chain.
Mechanism of a parallel manipulator with 3 degrees of freedom.
The solution to the positioning problem determines the relationship between the input and output coordinates and is presented in the following form (Kheylo and
Glazunov, 2016; Nhan et al., 2019):
F1=tgϕ11-cosγ⋅sinγ⋅sinβ+cosγ⋅sinαcosα⋅cosβ=0;F2=sinβcosγ⋅cosβ-tgϕ21=0;F3=cosγ⋅sinβ⋅sinα-cosα⋅sinγcosα⋅cosγ+sinα⋅sinβ⋅sinγ+tgϕ31=0.
In general, the positioning problem is specified in the form of implicit
functions, as follows:
Fi=(α,β,γ,θi1,θi2,ϕi1),
where θi1, θi2 is the angle between the axes of
the kinematic pairs (Fig. 2).
Angles between the axes of kinematic pairs θ11,
θ22 in the first kinematic chain.
The total differential of function (2) is written as follows:
∂Fi∂αδα+∂Fi∂βδβ+∂Fi∂γδγ+∂Fi∂θi1δθi1+∂Fi∂θi2δθi2+∂Fi∂ϕi1δϕi1=0.
According to the linear accuracy theory, the increment in the drives is zero, where δϕ11=δϕ21=δϕ31=0.
The implicit function equations can be written as follows (Kheylo and
Glazunov, 2014):
∂F1∂αδα+∂F1∂βδβ+∂F1∂γδγ=-∂F1∂θ12δθ12+∂F1∂θ11δθ11∂F2∂αδα+∂F2∂βδβ+∂F2∂γδγ=-∂F2∂θ22δθ22+∂F2∂θ21δθ21∂F3∂αδα+∂F3∂βδβ+∂F3∂γδγ=-∂F3∂θ32δθ32+∂F3∂θ31δθ31.
In the considered mechanism, the angles between the axes of the kinematic
pairs θi1, θi2 are not equal to 90∘. The
sequence of rotations of the output link from a movable coordinate system to
a fixed one is shown in Fig. 3.
Sequence of turns of the output link.
The transition matrix, from a moving coordinate system to a fixed one, will
have the following form:
B1=Fz⋅Fx⋅Bx′⋅Cy′⋅By′⋅Cx′⋅Bz′,
where, in the following:
Bx′=cosϕ11-sinϕ110sinϕ11cosϕ110001,By′=cosϕ12-sinϕ120sinϕ12cosϕ120001,andBz′=cosϕ13-sinϕ130sinϕ13cosϕ130001,
which are the rotation matrices of the first kinematic pairs and second and third
kinematic pairs around the movable axis z by the angles φ11,
φ12, and φ13, respectively.
Cx′=1000cosθ11-sinθ110sinθ11cosθ11,andCy′=cosθ120-sinθ12010sinθ120cosϕ12,
which are the matrices of rotations around the x and y axes by the angles
θ11 and θ12, respectively. θ11 and θ12 are the angles between the adjacent pairs.
Fz′=cosξ12sinξ120-sinξ12cosξ120001,andFx′=1000cosξ11sinξ110-sinξ11cosξ11,
which are matrices of additional rotations for aligning the moving and
stationary coordinate system around the z and x axes by tje angles ξ11, and ξ12, respectively, with ξ11=90∘, and
ξ12=90∘.
The unit vector of the axis of the pair of the output links of the first
chain has the following coordinates: 001, which are substituted into the following constraint equation
A⋅001=B1⋅001. Using the found values of the matrices A and B′, we obtain the following equation:
sinγ⋅sinα+cosα⋅cosγ⋅sinβcosα⋅sinγ⋅sinβ-cosγ⋅sinαcosβ⋅cosα=cosθ12cosθ11+sinϕ12sinθ11sinθ12cosϕ11cosθ12sinθ11-sinθ12(cosϕ12sinϕ11+cosϕ11cosθ11sinθ12)sinθ12cosϕ11⋅cosϕ12-cosθ11sinϕ11sinϕ12+cosθ12sinϕ11sinθ11.
For the second kinematic chain, the transition matrix from a moving
coordinate system to a fixed one will have the following form:
B2=Fx⋅Fy⋅By′′⋅Cz′′⋅Bz′′⋅Cy′′⋅Bx′′,
where, in the following:
By′′=1000cosϕ21-sinϕ210sinϕ21cosϕ21,Bz′′=1000cosϕ22-sinϕ220sinϕ22cosϕ22,andBx′′=1000cosϕ21-sinϕ210sinϕ21cosϕ21,
which are rotation matrices of the first kinematic pairs and second and third
kinematic pairs around the movable axis x by the angles φ21,
φ22, and φ23, respectively.
Cz′=cosθ22-sinθ220cosθ22sinθ220001,andCy′=cosθ210sinθ12010-sinθ210cosϕ12,
which are matrices of rotations around the z, y axes by the angles θ22 and θ21, respectively, and θ21 and θ22 are the angles between adjacent pairs.
Fy′′=cosξ220sinξ22010sinξ220cosξ22,andFx′′=1000cosξ21sinξ210-sinξ21cosξ21,
which are matrices of additional rotations for aligning the moving and
stationary coordinate systems around the z and x axes by the angles ξ22 and ξ21, respectively, with ξ21=90∘ and ξ22=90∘.
For the third kinematic chain, the transition matrix from a moving
coordinate system to a fixed one will have the following form:
B3=Fy′′′⋅Fz′′′⋅Bz′′′⋅Cx′′′⋅Bx′′′⋅Cz′′′⋅By′′′Bz′′′=cosϕ310sinϕ31010-sinϕ310cosϕ31,By′′′=cosϕ320sinϕ32010-sinϕ320cosϕ32,andBx′′′=cosϕ330sinϕ33010-sinϕ330cosϕ33,
which are matrices of rotations of the first kinematic pairs and second and third kinematic pairs around the movable axis y by the angles φ31, φ32, and φ33, respectively.
Cz′′′=cosθ31-sinθ310cosθ31sinθ310001,andCx′′′=1000cosθ32sinθ320-sinθ32cosθ32,
which are matrices of rotations around the z and x axes by the angles θ31 and θ32, respectively. θ31 and θ32 are the angles between adjacent pairs.
Fy′′′=cosξ320-sinξ32010sinξ320cosξ32,andFz′′′=cosξ31sinξ310-sinξ31cosξ310001,
which are matrices of additional rotations for aligning the moving and
stationary coordinate systems around the z and x axes by the angles ξ31 and
ξ32, respectively, with ξ31=90∘ and ξ32=90∘.
Position function F1 will look as follows:
F1=cosαsinγsinβ-cosγsinαcosβcosα-cosϕ11cosθ12sinθ11-sinθ12(cosϕ12sinϕ11+cosϕ11cosθ11sinθ12)sinθ12(cosϕ11cosϕ12-cosθ11sinϕ11sinϕ12)+cosθ12sinϕ11sinθ11.F2=-sinβcosγcosβ-cosϕ21cosθ21sinθ22-sinθ21(cosϕ22sinϕ21+cosϕ21cosθ22sinϕ22)sinθ21(cosϕ21cosϕ22-cosθ22sinϕ21sinϕ22)+cosθ21sinϕ21sinθ22.F3=cosγsinβsinα-cosαsinγcosγcosα+sinγsinβsinα-cosϕ31cosθ31sinθ32-sinθ31(cosϕ32sinϕ31+cosϕ31cosθ32sinθ32)sinθ31(cosϕ31cosϕ32-cosθ32sinϕ31sinϕ32)+cosθ31sinϕ31sinθ32.
Partial derivatives will be equal to the following:
∂F1∂θ11=-cosϕ11cosθ11cosθ12+cosϕ11sinϕ12sinθ11sinθ12sinθ12cosϕ11cosϕ12-cosθ11sinϕ11sinϕ12+cosθ12sinϕ12sinθ11-(cosθ11cosθ12sinϕ11+sinϕ11sinϕ12sinθ11sinθ12)(sinθ12(cosϕ11cosϕ12-cosθ11sinϕ11sinϕ12)+cosθ12sinϕ11sinθ11)2×(sinθ12cosϕ12sinϕ12+cosϕ11cosθ11sinϕ12-cosϕ11cosθ12sinθ11).∂F1∂θ12=-cosθ12cosϕ21sinϕ12+cosϕ11cosθ11sinϕ12+cosϕ11sinθ11sinθ12sinθ12cosϕ11cosϕ12-cosθ11sinϕ11sinϕ12+cosθ12sinϕ12sinθ11-sinθ12(cosθ12sinϕ11+cosϕ11sinϕ12cosθ11)-cosϕ11cosθ12sinθ11(sinθ12(cosϕ11cosϕ12-cosθ11sinϕ11sinϕ12)+cosθ12sinϕ11sinθ11)2×(cosθ12cosϕ11cosϕ12-sinϕ11cosθ11sinϕ12-sinϕ11sinθ12sinθ11).∂F2∂θ21=-cosθ21(cosϕ22sinϕ21+cosϕ21cosθ22sinϕ22)+cosϕ21sinθ21sinθ22sinθ21cosϕ21cosϕ22-cosθ22sinϕ21sinϕ22+cosθ21sinϕ21sinθ22-sinθ21cosϕ22sinϕ21+cosϕ21sinθ22sinϕ22-cosϕ21cosθ21sinθ22(sinθ21cosϕ21cosϕ22-cosθ22sinϕ21sinϕ22+cosθ21sinϕ21sinθ22)2×(cosθ21(cosϕ21cosϕ22-sinϕ22cosθ22sinϕ12)-sinϕ21sinθ21sinθ22).∂F2∂θ22=(cosϕ21cosθ21cosθ22+cosϕ21sinθ22sinϕ22sinθ21)sinθ21cosϕ21cosϕ22-cosθ22sinϕ21sinϕ22+cosθ21sinϕ21sinθ22-(cosθ21sinϕ21cosθ22+sinϕ21sinθ21sinθ22sinϕ22)(sinθ21cosϕ21cosϕ22-cosθ22sinϕ21sinϕ22+cosθ21sinϕ21sinθ22)2×(sinθ21cosϕ22sinϕ21+cosϕ21cosθ22sinϕ22-cosϕ21cosθ21sinθ22).∂F3∂θ31=-cosθ31cosϕ32sinϕ31+cosϕ31cosθ32sinϕ32+cosϕ31sinθ31sinθ32sinθ31cosϕ31cosϕ32-cosθ32sinϕ31sinϕ32+cosθ31sinϕ31sinθ32-sinθ31cosϕ32sinϕ31+cosϕ31cosθ32sinϕ32-cosϕ31cosθ31sinθ32(sinθ31cosϕ31cosϕ32-cosθ32sinϕ31sinϕ32+cosθ31sinϕ31sinθ32)2×(cosθ31cosϕ31cosϕ32-sinϕ32cosθ32sinϕ31-sinϕ31sinθ31sinθ32).∂F3∂θ32=(cosϕ31cosθ31cosθ32+cosϕ31sinθ32sinθ31sinϕ32)sinθ31cosϕ31cosϕ32-cosθ32sinϕ31sinϕ32+cosθ31sinϕ31sinθ32-(cosθ32cosθ31sinϕ31+sinϕ31sinθ31cosθ32sinϕ32)(sinθ31cosϕ31cosϕ32-cosθ32sinϕ31sinϕ32+cosθ31sinϕ31sinθ32)2×(sinθ31cosϕ32sinϕ31+sinϕ32cosθ32cosϕ31-cosϕ31cosθ31sinθ32).
The values of the remaining partial derivatives are known from solving the
positioning problem. Let us determine the positioning error of the output link in different positions with deviations of the angles between the axes equal to θ11=θ12=θ21=0.50 and
θ22=θ31=θ32=0.40.
The deviations of the angles of the output link are expressed, from Eq. (4), as follows:
δα=∂F1∂β∂F2∂γW3-∂F1∂β∂F3∂γW2-∂F1∂γ∂F2∂βW3+∂F1∂γ∂F3∂βW2+∂F2∂β∂F2∂γW1-∂F2∂γ∂F3∂βW1∂F1∂α∂F2∂β∂F3∂γ-∂F1∂β∂F2∂γ∂F3∂β-∂F1∂β∂F2∂α∂F3∂γ+∂F1∂β∂F3∂α∂F2∂γ+∂F2∂α∂F1∂γ∂F3∂β-∂F1∂γ∂F2∂β∂F3∂α,δβ=∂F1∂α∂F2∂γW3-∂F1∂α∂F3∂γW2-∂F2∂α∂F1∂γW3+∂F2∂α∂F3∂γW1+∂F1∂γ∂F3∂αW1-∂F3∂α∂F2∂γW1∂F1∂α∂F2∂β∂F3∂γ-∂F1∂β∂F2∂γ∂F3∂β-∂F1∂β∂F2∂α∂F3∂γ+∂F1∂β∂F3∂α∂F2∂γ+∂F2∂α∂F1∂γ∂F3∂β-∂F1∂γ∂F2∂β∂F3∂α,andδα=∂F1∂α∂F2∂βW3-∂F1∂α∂F3∂βW2-∂F1∂β∂F2∂αW3+∂F1∂β∂F3∂αW2+∂F2∂α∂F3∂χW1-∂F2∂β∂F3∂αW1∂F1∂α∂F2∂β∂F3∂γ-∂F1∂β∂F2∂γ∂F3∂β-∂F1∂β∂F2∂α∂F3∂γ+∂F1∂β∂F3∂α∂F2∂γ+∂F2∂α∂F1∂γ∂F3∂β-∂F1∂γ∂F2∂β∂F3∂α,
where, in the following:
W1=-∂F1∂θ12δθ12+∂F1∂θ11δθ11,W2=-∂F2∂θ22δθ22+∂F2∂θ21δθ21,andW3=-∂F3∂θ32δθ32+∂F3∂θ31δθ31.
Thus, the solution to the problem of determining the deviation of the
actuator of the mechanism in the case of the inaccuracy of its manufacture is
presented. The values of the deviation of the coordinates of the output
link at a given deviation between the axes are presented in Table 1.
When assessing the dynamic positioning accuracy of the grip, small
fluctuations around the stable equilibrium position are investigated.
The equations of motion for a manipulator with 3 degrees of freedom will
be described by a system of differential equations (Kheylo and Glazunov, 2014; Nhan et al., 2019) as follows:
ddt∂T∂ϕ˙i-∂T∂ϕi=-∂∏∂ϕi,i=1,…3,
where T is the kinetic energy of the system, and Π is potential energy of the system.
The potential energy of the system is defined as follows:
∏=12∑i=1nci⋅ϕi2,
where ci is the chain stiffness.
Kinetic energy is a quadratic function of generalized velocities.
T=12⋅Jx⋅ωx2+12⋅Jy⋅ωy2+12⋅Jz⋅ωz2,
where ωx, ωy, and ωz are the angular
velocities of the output link around the x, y, and z axes, respectively. Jz=12⋅m⋅r2 and
Jy=Jz=112⋅m⋅r2 are the moments of inertia of
the output link around the axes x, y, and z, respectively. m is the mass of the output link (we take m=0.5 kg, and r=0.1 m). r is the radius of the platform of the output link. Then, Eq. (6) takes the following form:
T=12⋅112⋅m⋅r2⋅ωx2+12⋅112⋅m⋅r2⋅ωy2+12⋅12⋅m⋅r2⋅ωz2.
The problem of determining the deviation of the coordinates of the input
link when removing it from the equilibrium position is solved numerically
(Fig. 4). Reducing the dynamic error can be achieved by increasing the
rigidity of the drive or by introducing damping devices.
Automatic compensation of movements, using digital or pulse controllers, is
also an effective method of damping the vibrations.
Deviation of the output link along the coordinates α,
β, and γ.
Evaluation of the positioning of the output link when controlling the mechanism
The reduction of residual vibrations is possible by increasing the stiffness
of the drives. This must also be taken into account when constructing a
control algorithm (Nhan et al., 2019).
Let us set the required movement of the output link in the form of the law
of coordinate change, i.e., αT(t), βT(t), and γT(t). The required speeds are
α˙T(t), β˙T(t), and γ˙T(t), and
acceleration is α¨T(t), β¨T(t), and γ¨T(t), which we obtain after the differentiation.
The control problem is to minimize the coordinate error is as follows: Δ1(t)=αT(t)-α(t), Δ2(t)=βT(t)-β(t), Δ3(t)=γT(t)-γ(t), with the speed of Δ˙1(t)=α˙T(t)-α˙(t),
Δ˙2(t)=β˙T(t)-β˙(t), and Δ˙3(t)=γ˙T(t)-γ˙(t), and acceleration of
Δ¨1(t)=α¨T(t)-α¨(t), Δ¨2(t)=β¨T(t)-β¨(t), and Δ¨3(t)=γ¨T(t)-γ¨(t), where αt, βt, and γt are the actual values of the coordinates of the output link.
We then simulate the movement of the output link of the parallel structure
mechanism along a given trajectory using the developed algorithm for optimal
motion control (Fig. 5).
Control algorithm.
To measure the magnitude of the deviations, we use a quadratic integral assessment as follows:
JS=∫t0TΔi2+k1⋅Δ˙i2+k2⋅Δ¨i2dt.
There must be the following:
Δ¨+γ1⋅Δ˙+γ0⋅Δ=0.
We then rewrite Eq. (8) in a form appropriate to the oscillatory link, as follows:
τ2Δ¨+2ζτ⋅Δ˙+Δ=0τ2=1γ0;2ζτ=γ1γ0,
where τ is time, and ζ is the damping ratio.
The law of acceleration change corresponds to the oscillatory link as follows:
α¨=α¨T+γ1⋅(α˙T-α˙)+γ0⋅(αT-α)β¨=β¨T+γ1⋅(β˙T-β˙)+γ0⋅(βT-β)γ¨=γ¨T+γ1⋅(γ˙T-γ˙)+γ0⋅(γT-γ).
The equation of motion for a spherical mechanism with 3 degrees of
freedom has the following form:
Jξ⋅ϕ¨ξ=M1⋅∂ϕ11∂ϕξ+M2⋅∂ϕ21∂ϕξ+M3⋅∂ϕ31∂ϕξ+ϕ˙η⋅ϕ˙ζ⋅Jζ-JηJη⋅ϕ¨η=M1⋅∂ϕ11∂ϕη+M2⋅∂ϕ21∂ϕη+M3⋅∂ϕ31∂ϕη+ϕ˙ξ⋅ϕ˙ζ⋅Jξ-JζJζ⋅ϕ¨ζ=M1⋅∂ϕ11∂ϕζ+M2⋅∂ϕ21∂ϕζ+M3⋅∂ϕ31∂ϕζ+ϕ˙ξ⋅ϕ˙η⋅Jη-Jξ,
where Jξ=Jη,Jζ are the moments of inertia about the axes ξ, η, and ζ. M1,M2, and M3 are the moments in drives. ∂ϕij∂ϕξ are the variable coefficients.
ϕ¨ξ, ϕ˙ξ, ϕ¨η,
ϕ˙η, ϕ¨ζ, and ϕ˙ζ are the projection of accelerations and velocities on the
moving axes of ξ, η, and ζ.
The dynamic properties of the system are determined by transient processes.
Transient processes (the overshoot time and deviations of the output link from a given trajectory) are regulated by feedback coefficients γ0 and γ1.
Let us set the law of motion of the executive link, as follows:
αT(t)=0,1⋅sin(ωt);βT(t)=0,1⋅sin(ωt);γT(t)=0,1⋅sin(ωt).
The differential acceleration in Eq. (9) takes the following form:
α¨=α¨T+120⋅(α˙T-α˙)+7200⋅(αT-α)β¨=β¨T+120⋅(β˙T-β˙)+7200⋅(βT-β)γ¨=γ¨T+120⋅(γ˙T-γ˙)+7200⋅(γT-γ).
When moving with feedback, the graph of the changes in the position error and the torque in the drive is shown in Fig. 6.
Graph of the changes in the position error Δα,
Δβ, and Δγ of the output link.
Thus, by changing the feedback coefficients, it is possible to regulate the
deviations of the output link from the desired trajectory, while the
overshoot time τ changes.
Conclusion
An integrated approach allows one to evaluate and determine the positioning
error of the output link from the standpoint of kinematics, dynamics, and
control. This approach will allow, when synthesizing mechanisms, us to propose approaches to compensate for or completely eliminate errors.
So, with the kinematic estimate, the proposed approach to assessing the
kinematic accuracy allows one to determine the deviations in the output link
using the theory of accuracy. This allows one to determine the deviations in
the output link at any point in the working area and propose constructive
solutions for its compensation. This approach to determining the positioning
error makes it possible to calculate the deviations in the output link for
similar mechanisms of a parallel structure.
From the perspective of dynamics with free oscillations of the output link,
or when it stops, a numerical estimate of the positioning error is given,
which makes it possible to assess the technological requirements for the
accuracy of operations and propose solutions to reduce it. In this case, an
increase in the accuracy of the manipulation from the position of dynamics can be carried out by optimally selecting the rigidity of the drive or by
introducing additional damping devices. In control, the minimization of the
error during the transient process is ensured by the choice of feedback
coefficients.
Code availability
Code used in the article is available at https://drive.google.com/drive/folders/1FTdrY5xAoc_A-4FIvr-CIQLTS2xSCJ-y?usp=sharing (Vo, 2022a).
Data availability
Data used in the article are available at https://drive.google.com/drive/folders/1-bKDoZlJDiLQ4ufK0BcWhQLCw6O-mpzl?usp=sharing (Vo, 2022b).
Author contributions
DTV conducted the research and wrote the paper. SK and VQN analyzed the data and checked paper. All authors had approved the final version.
Competing interests
The contact author has declared that neither they nor their co-authors have any competing interests.
Disclaimer
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Review statement
This paper was edited by Guowu Wei and reviewed by Daniel Condurache and one anonymous referee.
References
Bai,S., Hansen, M. R., and Angeles, J.: A robust forward-displacement analysis of spherical parallel robots, Mech. Mach. Theory., 44, 2204–2016, 2009.
Chaker, M., Laribi, A., Romdhane, L., and Zeghloul, S.: Synthesis of
spherical parallel manipulator for dexterous medical task 2nd IFToMM Int.
Sym. on Rob. and Mech., ISRM, 3–5 November 2011, Shanghai, China, 2011.Gogu, G.: Structural Synthesis of Parallel Robots, Part 1: Methodology
(Solid Mechanics and Its Applications), edited by: Gladwell, G. M. L., 706 pp., Springer, Vol. 149, 10.1007/978-1-4020-5710-6_1, 2008.
Huda, S. and Takeda, Y.: Dimension Syntesis of 3-URU Pure Rotation Parallel
Mechanism with Respect to Singularity and Workspace, 12th IFToMM World
Congress, 17–21 June 2007, Becasson, 235–242, 2007.
Kheylo, S. V. and Glazunov, V. A.: Kinematics, Dynamics, Control and Accuracy
of spherical parallel robot. Advanced on theory and practice of robots and
manipulators, ROMANSY 2014 – XX CISM-IFToMM Symposium on theory and
practice of robots and manipulators, Russia, Moscow, 133–144,
2014.Kheylo, S. V. and Glazunov, V. A.: Dynamics and control of planar,
translational, and spherical parallel manipulators, in: Dynamic
Balancing of Mechanisms and Synthesizing of Parallel Robots, edited by: Zhang, D. and Wei, B., 365–403, 10.1007/978-3-319-17683-3,
Springer, 2016.
Kheylo, S. V. and Glazunov, V. A.: Kinematics, Dynamics, Control and Accuracy of spherical parallel robot, Advanced on theory and practice of robots and manipulators. ROMANSY 2014 – XX CISM-IFToMM Symposium on theory and practice of robots and manipulators, Russia, Moscow, 133–144, 2014.
Kong, X. and Gosselin, C.: Type Synthesis of Parallel Mechanisms, Tracts in Advanced Robotics, Vol. 33, Springer, Heidelberg, Berlin, Germany, 275 pp., 2007.
Leguay-Durand, S. and Reboulet, C.: Optimal design of a redundant
spherical parallel manipulator, Robotica, 15, 399–405, 1997.Merlet, J. P.: Parallel robots, 44, 2204–2016,
Kluwer Academic Publishers, 10.1007/978-94-010-9587-7, 2000.Mianovski, K.: Singularity analysis of parallel manipulator POLMAN 3×2 with six degrees of freedom, 12th IFToMM World Congress, Becasson, 126–132, 2007.
Nosova, N. Y., Kheilo, S. V., Glazunov, V. A., and Tsar’kov, A. V.: Dynamic Analysis of the Spherical Part of the Parallel Manipulator Taking into Account the Control Law, Journal of Machinery Manufacture and Reliability, 47, 205–212, 2018.
Nhan, N. H. K., Tung, V. D., Kheylo, S., and Victor, G.: Oscillations and control of spherical parallel manipulator, Int. J. Adv. Robot. Syst., 16, 2019.Vo, D. T.: Code for the article “Kinematic and dynamic accuracy of spherical mechanisms”, Google Drive [code], available at: https://drive.google.com/drive/folders/1FTdrY5xAoc_A-4FIvr-CIQLTS2xSCJ-y?usp=sharing, last access: 12 January 2022a.Vo, D. T.: Data for the article “Kinematic and dynamic accuracy of spherical mechanisms”, Google Drive [data set], available at: https://drive.google.com/drive/folders/1-bKDoZlJDiLQ4ufK0BcWhQLCw6O-mpzl?usp=sharing, last access: 21 January 2022b.