You are on page 1of 14

Error Analysis for the In-Situ

Fabrication of Mechanisms
Sanjay Rajagopalan
e-mail: sanjay@cdr.stanford.edu Fabrication techniques like Solid Freeform Fabrication (SFF), or Layered Manufactur-
ing, enable the manufacture of completely pre-assembled mechanisms (i.e. those that
Mark Cutkosky require no explicit component assembly after fabrication). We refer to this manner of
e-mail: cutkosky@cdr.stanford.edu building assemblies as in-situ fabrication. An interesting issue that arises in this domain
is the estimation of errors in the performance of such mechanisms as a consequence of
Center for Design Research, manufacturing variability. Assumptions of parametric independence and stack-up made in
Stanford University, conventional error analysis for mechanisms do not hold for this method of fabrication. In
Palo Alto, CA 94305-2232 this paper we formulate a general technique for investigating the kinematic performance
of mechanisms fabricated in-situ. The technique presented admits deterministic and sto-
chastic error estimation of planar and spatial linkages with ideal joints. The method is
illustrated with a planar example. Errors due to joint clearances, form errors, or other
effects like link flexibility and driver-error, are not considered in the analysis—but are
part of ongoing research. 关DOI: 10.1115/1.1631577兴

1 Introduction rors in the 1800s 关5兴. The problems typically occur due to inac-
curacies in mechanism dimensions, poor joints, out-of-plane flex-
The last decade of the millennium has seen the widespread
ibility in links and assembly issues. These problems are
adoption of new ‘‘freeform’’ fabrication techniques. Called by
exacerbated in the construction of spatial mechanism prototypes.
various names 共Rapid Prototyping, Layered Manufacturing, Solid
The advent of Solid Freeform Fabrication 共SFF兲 could revolu-
Freeform Fabrication etc.兲, this technology builds a part directly
tionize the manner in which mechanisms are designed and fabri-
from its digital 共CAD兲 representation by ‘‘slicing’’ the part model,
cated 关6兴 关7兴 关8兴. In-situ technology allows for precision compo-
and building it incrementally by selectively adding and removing nents, sensors, actuators and electronics to be directly integrated
material 关1兴 关2兴 关3兴. into the mechanism frame during fabrication. Alternately, high-
Of particular interest for the purposes of this paper is the capa- precision joints may also be directly built by freeform processes at
bility of these processes to fabricate assemblies 共e.g., mechanisms a specified location. Figure 2 shows examples of some mecha-
with mating components兲 in-situ. In conventional fabrication, nisms recently fabricated at Stanford University. Others have been
each component of a device is individually fabricated and then built at Rutgers University 关9兴 and Laval 关10兴. Similar devices are
assembled together. For in-situ fabrication, the entire device is found in the realm of microelectromechanical systems 共MEMS兲
built encapsulated in a sacrificial support material. This support which also use an incremental layered manufacturing technique,
material is removed 共by etching, melting, or dissolving it away兲 to with much smaller feature sizes 共see Fig. 3兲.
yield the final part with operational mating and fitting features Whether at microscopic or macroscopic scales, in-situ manufac-
共see Fig. 1兲. turing practices have a process flow 共Fig. 4兲 that is fundamentally
This paper examines the manner in which manufacturing errors, different from either traditional ‘‘craftsman’’ manufacturing or
specifically errors in the spatial location 共position and orientation兲 conventional mass production. As described in the following sec-
of joints, affect the performance of mechanical devices fabricated tions, the difference in process flow leads to differences in the way
in-situ. It is well established in theoretical kinematics that the that dimensional errors are generated and accumulate, requiring a
primary determinant of mechanism behavior 共for rigid body different approach to tolerance analysis. We begin with a brief
mechanisms兲 is the spatial location of its joints 关4兴. Consequently, review of classical tolerance analysis for mechanisms and use it as
the focus of mechanism error analysis techniques on parametric a point of departure for the modified approach that is the main
variability 共e.g. link length兲 is an artifact of the manufacturing contribution of this paper.
techniques used to fabricate these mechanisms.
In this paper, an assumption is made that the joint location
variability is process-specific, and is taken as the primary exog-
enous factor to the analysis. Given this assumption, the techniques
presented in this paper are not unique to any specific fabrication
process, nor are they only limited to the analysis of planar
mechanisms.
1.1 Scope of the Paper. This paper is concerned with the
study of general 共i.e. planar or spatial, open-chain or multi-loop兲
mechanisms, fabricated using freeform techniques.
As students in kinematics classes soon learn, the fabrication of
precise mechanism prototypes can be a complex, time-consuming
and sometimes, frustrating task. It is believed that Charles Bab-
bage’s mechanical computing engine, a good example of a com-
plex spatial mechanism, failed mainly because of the inability of
its fabricators to avoid accumulated component dimensional er-

Contributed by the Mechanisms and Robotics Committee for publication in the


JOURNAL OF MECHANICAL DESIGN. Manuscript received March 2002; revised
April 2003. Associate Editor: J. S. Rastegar. Fig. 1 Conventional versus in - situ fabrication

Journal of Mechanical Design Copyright © 2003 by ASME DECEMBER 2003, Vol. 125 Õ 809
Fig. 4 Comparing conventional and in - situ manufacturing
methods—process flow chart. Actions that impart accuracy to
the mechanism are specifically identified.
Fig. 2 In - situ mechanism prototypes fabricated via Shape
Deposition Manufacturing †3‡: „a… a polymer insect-leg proto-
type with embedded pneumatic actuator, pressure sensor and
leaf-spring joint „b… a hexapedal robot with integrated sensors, where y denotes the (m⫻1) vector of output end-effector loca-
actuators and electronics „c… an ‘‘inchworm’’ mechanism, with tions, coupler-point positions or output link angles, ⌰ is a (k
integrated clutch components, „d… a slider-crank mechanism
⫻1) vector of known driving inputs, and ⌽ is a (n⫻1) vector of
made from stainless steel. Images courtesy the Stanford Cen-
ter for Design Research and Rapid Prototyping Laboratories. independent mechanism variables—including deterministic or
randomly distributed geometric parameters and/or dimensions.
The function f (•) is called the kinematic function of the mecha-
nism and is, in general, assumed to be a continuous and differen-
tiable 共i.e. smooth兲 non-linear mapping from the mechanism pa-
1.2 Introducing Mechanism Error Analysis. The modern
rameter space to an output space 共e.g. a Cartesian workspace兲. In
scientific treatment of mechanism error estimation dates to the
the absence of higher-pairs 共i.e. joints that have line and point
early 1960’s 关11兴 关12兴. In the several decades since, many alter-
contact, as opposed to surface contact, between their member
native approaches to error analysis for mechanisms have been
links兲 and multiple-contact kinematics, the smoothness assump-
proposed—each with various simplifying assumptions and differ-
tion generally holds true.
ent levels of complexity 关13兴 关14兴 关15兴 关16兴 关17兴. All approaches,
however, attempt to solve the same basic problem—to predict the 1.3 Conventional Mechanism Error Analysis. Conven-
nature and amount of performance deterioration in mechanisms tional error analysis deals with degradation in the performance of
as a result of non-ideal synthesis, fabrication, materials or a mechanism as a result of parametric or dimensional variations,
componentry. and play in joints. The parameters typically considered are link
In this paper the focus is on kinematic performance. In other lengths for planar linkages, or some form of the Denavit-
words, we assume that we are always able to describe the desired Hartenberg 关18兴 parameters for spatial linkages. Error in the per-
task in terms of an output equation of the form: formance of known mechanisms can be estimated analytically if
certain assumptions are made, rendering the underlying math-
y⫽ f 共 ⌽,⌰兲 (1)
ematical treatment more tractable. For example:
• Mechanism dimensions and parameters have a known, given
variability characteristic—either deterministic, or stochastic.
• Dimensional/parametric variations and clearance values are
significantly smaller than their nominal values.
• Individual component variations are independent, uncorre-
lated and identically distributed.
• The output is, at most, a weak non-linear function of the
mechanism parameters at the operating configuration of interest.
As a result of these assumptions, it becomes possible to ap-
proximate the actual error by lower-order estimates. Other as-
sumptions 共e.g. negligible variability of the clearance value itself,
Normal or Uniform distribution of component parameters etc.兲,
which either eliminate unnecessary model complexity or enable
analytical tractability, are also commonly made.
1.3.1 Sensitivity Analysis. Sensitivity analysis is based on
the Taylor-series expansion of the output function. As stated in Eq.
Fig. 3 Micromechanisms and devices built using in - situ fab- 共1兲, the end-effector position, coupler path or output angle of a
rication techniques. Images courtesy Sandia National Labora- mechanism can be expressed as:
tories, SUMMiT„tm… Technologies, www.mems.sandia.gov.
Used with permission. y⫽ f 共 ⌽,⌰兲 (2)

810 Õ Vol. 125, DECEMBER 2003 Transactions of the ASME


where ⌰⬅ 关 ␪ 1 , ␪ 2 , ¯ , ␪ k 兴 T are the k known driving inputs, and Here, an assumption is made that each component variability
⌽⬅ 关 ␾ 1 , ␾ 2 , ¯ , ␾ n 兴 T are the n mechanism parameters 共or di- parameter is weighted equally in the cost function, which may not
mensions兲 subject to random, or worst-case deterministic, vari- always be true. Some manufacturing parameters may be easier to
ability. Since ⌰ is assumed static for a given mechanism configu- control accurately than others 共e.g. hole size can typically be held
ration 共i.e. the driving inputs are held perfectly to their nominal to tighter tolerances than center-distance between holes兲. Addi-
values兲, it is dropped from the equation for notational simplicity. tionally, zero tolerance 共or close-to-zero tolerance兲 for some pa-
The previous equation is re-written as: rameters, which is permissible for the above formalization, is in-
feasible for real manufacturing processes. Non-homogeneous
y⫽ f 共 ⌽兲 (3) manufacturing capability within the mechanism workspace is also
Expanding this function in Taylor-series around the nominal not considered in this system.
values of the mechanism parameters (⌽ nom The optimization problem can be solved using standard meth-
ods of parametric programming—Lagrange multipliers, or Pow-
⬅ 关 ␾ 1 , ␾ 2 , ¯ , ␾ n 兴 ), we get:
nom nom nom T
ell’s conjugate direction method 共i.e. unconstrained optimization

y⫽ f 共 ⌽nom兲 ⫹ 兺
n

i⫽1
⳵f
⳵␾i 册 共 ␾ i ⫺ ␾ inom兲 ⫹
nom
1
2! 兺
n

i⫽1
⳵2 f
⳵ ␾ i2 册 共␾i
nom
of a penalty function兲 关19兴. An example of these optimization
techniques applied to mechanism tolerance allocation can be
found in 关20兴.

⫺ ␾ inom兲 2 ⫹ 兺 ⳵␾ ⳵␾
i⬎ j
⳵2 f
i j
册 共 ␾ i ⫺ ␾ inom兲共 ␾ j ⫺ ␾ nom
nom
j 兲 ⫹¯
1.3.2 Deterministic, Worst-Case Error Estimation. In worst-
case error estimation, each parameter ␾ i is assumed to take 共ex-
clusively兲 one of two deterministic values ␾ imin and ␾ imax . Further-
(4) more, it is assumed that ␾ imin⭐␾inom⭐ ␾ imax ;i⫽1,2, . . . ,n, where
or, using a more concise notation: ␾ inom is the nominal value of the ith parameter.

册 册
The objective of this kind of error estimation is to determine the
⳵f 1 ⳵2 f worst case envelope of the mechanism performance error. Except
y⫽ f 共 ⌽nom兲 ⫹ 共 ⌽⫺⌽nom兲 ⫹ 共 ⌽⫺⌽nom兲 2 ⫹¯
⳵⌽ nom
2! ⳵ ⌽2 nom
for applications where performance within specified limits is ab-
(5) solutely critical, the worst-case analysis results in conservative
estimates of error 共and thereby, over-design of components兲. Since
For small, independent variations about the nominal configura- the worst performance can occur for any combination of mini-
tion, a linear approximation can be made—thereby rewriting the mum and maximum component parameter values, the technique
above equation as: proceeds by exhaustive calculation of total error for each combi-

y⬇ f 共 ⌽nom兲 ⫹
⳵f
⳵⌽ 册 共 ⌽⫺⌽nom兲
nom
(6)
nation of individual error values. For n parameters, this leads to a
search space of 2 n combinations for each mechanism configura-
tion. If the objective is to find the worst-case performance within
or the entire workspace of the mechanism, then this calculation has


to be repeated at each incremental driver position.
⳵f An alternative approach is to use dynamic programming 关21兴
⌬ y⬇ ⌬⌽ (7) 关22兴 to estimate the maximum error without computing the total
⳵⌽ nom
error for every possible combination. The assumption made while
The quantity ⳵ f / ⳵ ⌽] nom is known as the sensitivity Jacobian of using this technique is that the global optimization problem can be
the mechanism, evaluated at the nominal configuration. This Jaco- re-stated as a multi-stage optimization problem, with the nth stage
bian relates the component variability (⌬ ⌽) in the mechanism solution related to the (n⫺1)th stage solution through a func-
parameter space to the output variation (⌬ y) in Cartesian space. tional equation. While this technique results in significant reduc-
This is classical sensitivity analysis, where all variational effects tion of the computational burden involved, it is not guaranteed to
are bundled into a simple parametric space, and all higher order find the global optimum when the underlying monotonicity as-
effects are neglected. sumptions do not hold.
Equation 共7兲 is used as the basis for error analysis and tolerance
allocation. For error analysis, the component variability (⌬ ⌽) and 1.3.3 Stochastic Error Estimation. Statistical error estima-
sensitivity Jacobian ( ⳵ f / ⳵ ⌽) are known for a given mechanism tion proceeds by assigning a probability distribution function
configuration. The output error (⌬ y) is then a simple calculation. 共PDF兲 to each variable parameter ␾ i . The component dimension
The component variability can either be expressed as worst-case under consideration is assumed to be a random variable, distrib-
values, or as stochastic variations in link parameters. Each of uted according to the characteristics of its underlying PDF, de-
these approaches is discussed in the next sections. noted as p ⌽i ( ␾ i ). The cumulative distribution function 共CDF兲 of
For tolerance allocation problems, the maximum permissible the output functions can then be estimated using standard tech-
output error (⌬ max niques for stochastic analysis. If certain assumptions can be made
y ) and sensitivity Jacobian are known. Equation
共7兲 forms the basis for the constraint equations, and the objective 共e.g. linearity, independence, identical distribution etc.兲, the esti-
is to maximize the overall variability 共i.e. ⌬ ⌽), given the con- mation of the distribution and moments of the output function is
straints. Greater allowable variability typically means lower highly simplified.
manufacturing and inspection costs, and thus, is preferred. One The error equation 共Eq. 共7兲兲 can be replaced by an equivalent
simple formalization of the tolerance allocation problem is as equation for the stochastic estimation of each output CDF, as
follows: follows:

minimize Z⫽ 兺
n

i⫽1
1
⌬⌽
(8)
P Y j共 y j 兲 ⫽ 冕⫺⬁
␾1
... 冕
⫺⬁
␾n
y j .p ⌽1 . . . ⌽n 共 ␾ 1 , . . . , ␾ n 兲 d ␾ 1 . . . d ␾ n

subject to: where j⫽1,2, . . . ,m (10)

g 共 ⌽兲 ⬅⌬ max
y ⫺
⳵f
⳵⌽ 册 ⌬ ⌽⭐0 and
and for independent and uncorrelated ␾ i
n

兿p
nom
p ⌽1 . . . ⌽n 共 ␾ 1 , . . . , ␾ n 兲 ⫽ ⌽i 共 ␾ i 兲 (11)
g i 共 ⌽兲 ⬅ ␾ i ⭓0; i⫽1,2, . . . ,n. (9) i⫽1

Journal of Mechanical Design DECEMBER 2003, Vol. 125 Õ 811


In general, the complete analytical evaluation of the integrals in
Eqs. 共10兲 and 共11兲 are not simple, or even tractable. However, it
may not always be necessary to evaluate the error CDF. Given
certain assumptions, it is possible to determine the mean and vari-
ance of the output distribution directly from the mean and vari-
ance ( ␮ ␾ i , ␴ ␾2 i ) of the individual components. To do this, the out-
put 共Eq. 共3兲兲 is expanded in a Taylor series about the mean values
( ␮ ␾ i ) of the component dimensions as follows:

y⫽ f 共 ␮ ␾ i ;i⫽1,2, . . . ,n 兲 ⫹ 兺 ⳵␾
i⫽1
n
⳵f
i
册 ␮
共 ␾ i⫺ ␮ ␾i兲


1
2!
n

兺 ⳵␾
i⫽1
⳵2 f
2
i ␮

共 ␾ i⫺ ␮ ␾i兲 2

Fig. 5 Modified Denavit-Hartenberg representation for spatial


linkages „Lin and Chen, 1994…. Note that the specific mecha-
⳵2 f
⫹ 兺
i⬎ j ⳵ ␾ i⳵ ␾ j
共 ␾ i ⫺ ␮ ␾ i 兲共 ␾ j ⫺ ␮ ␾ j 兲 ⫹¯ (12) nism shown here is irrelevant—used for illustrative purposes
only.

Assuming that the output is approximately linear for small


variations of the random variables about their mean values, the
higher-order terms in the above equation can be dropped, and the available for the estimation of the probability distribution of a
equation re-written as: complex non-linear function of random variables.


n In the event that the assumption of weak non-linearity of the
⳵f
y⬇a⫹ 兺 ⳵␾
i⫽1 i
共 ␾ i⫺ ␮ ␾i兲 (13) output function does not hold, then a second order estimate of the
mean a variance may yield better results. This is given as 共deri-

vation follows from results in Appendix A兲:
where a⬅ f ( ␮ ␾ i ;i⫽1,2, . . . ,n), and the partials are evaluated at

冉 冊
the mean value of the parameters. Equation 共13兲 can be written in n
1 ⳵2 f
terms of the proxy 共difference兲 variables ⌬ y and ⌬ ␾ i 共see Eq. 共7兲兲
as:
␮ y⫽a⫹
2 兺
i⫽1 ⳵ ␾ i2 ␮
␴ ␾2 i

⌬ y⬇
i⫽1
n

兺 ⳵␾
⳵f
i
册 ␮
⌬ ␾i (14)
␴ 2y ⫽ 兺 冉 ⳵␾ 冊
n
⳵f 2
␴ ␾2 i ⫹
1

n

冉 冊 ⳵2 f 2
␴ ␾2 i
i⫽1 i ␮ 2 i⫽1 ⳵ ␾ i2 ␮
where ⌬ y and ⌬ ␾ i are zero-mean random variables with all

兺冉 冊
higher-order moments identical with y and ⌽i respectively. In ⳵2 f 2

other words, by studying the variance properties of Eq. 共14兲, we ⫹ ␴ ␾i␴ ␾ j (16)
i⫽ j ⳵ ␾ i⳵ ␾ j ␮
are in effect studying the variance properties of the original equa-
tion 共i.e. Eq. 共13兲兲. 1.3.4 Kinematic Representations. The preceding sections
If the parameters ⌬ ␾ i are assumed to vary independently, then it present a generic treatment of error estimation where no assump-
can be shown 共see Central Limit Theorem 关23兴兲 that the output y tion is made regarding specific parameter assignments to the
follows an approximately Normal distribution 共for n⬎5), with the mechanism geometry. Mechanisms can be described using dimen-
mean and variance of the distribution given as follows: sions of geometric elements 共e.g. link length for planar linkages兲
or using mechanism parameters 共e.g. link length, link angle, offset
␮ y⫽a and twist for spatial linkages兲. Typically, the assumptions made in
(15)

兺冉 冊
n the sensitivity calculations detailed above will fail for certain
⳵f 2
mechanism instances, depending upon the specific representation
␴ 2y ⫽ ␴ ␾2 i
i⫽1 ⳵␾i ␮
used 关24兴.
A widely accepted parametric representation for spatial mecha-
where ␮ y and ␴ 2y
denote the mean and variance, respectively, of nisms is the Denavit-Hartenberg 共or D-H兲 representation 关18兴, and
the output function. the extensions thereof 关25兴. In this representation 共see Fig. 5兲 a
The full derivation of Eq. 共15兲 is given in Appendix A, as the spatial mechanism is described in terms of four parameters for
treatment is important for the extension of this model to the case each link i in the linkage. These parameters are termed the link-
of in-situ fabrication. A key assumption in this treatment—that of angle ( ␪ i ), link-length (a i ), link-offset (d i ), and twist-angle ( ␣ i ).
parametric independence—fails in the case of in-situ fabrication, In a mechanism with revolute, prismatic and cylindrical joints, the
and Eq. 共15兲 needs modification. link-lengths and twist-angles typically remain static during opera-
The specific probability of the output falling within a given tion, and the link-angles and link-offsets vary 共depending upon the
range y1 ⭐y⭐y2 can either be estimated using the standard tables type of joint兲.
共for normal distributions兲, or the Chebychev inequality 共for a sym- The Denavit-Hartenberg representation presents difficulties for
metric range兲. Since the linearized equation approximates the out- error analysis when mechanisms have parallel or nearly parallel
put error as a weighted sum of the component variation, a normal joint axes. Small variations in the D-H parameters result in large
output distribution can be assumed either when the individual errors in the output function. Various modifications have been
component variations are each normally distributed, or when the proposed 关25兴 关24兴 that rectify this problem. For this paper we
Central Limit Theorem can be applied with Liapunov’s condition adopt the representation of 关26兴 that adds an extra parameter (l i ),
关23兴 关20兴. Thus, the validity of the linear approximation is a fun- resulting in a better representation of the link shape 共see Fig. 5兲.
damental defining assumption in this type of analysis, since no The extra parameter does not add anything to the kinematic de-
simple general technique 共other than numerical simulation兲 is scription of the mechanism but is advantageous for error analysis.

812 Õ Vol. 125, DECEMBER 2003 Transactions of the ASME


Fig. 6 The effects of link length variation in an assembled 4-bar mechanism

2 Worst-Case Error Analysis for In-Situ Fabrication pair-wise correlated. This is because multiple 共adjacent兲 param-
eters depend upon the same independent inputs 共i.e., the positions
The conventional error models presented in Sec. 1.3 cannot be
and orientations of their shared joints兲. Although several param-
directly applied to in-situ fabrication since this fabrication tech-
eters can all be adjacent to each other if they share a common
nique differs from conventional sequential shape-and-assemble
joint, their correlation is still taken pair-wise since covariance is
fabrication techniques in some fundamental ways. Primarily, the
defined on random variable pairs. The degree of correlation de-
differences are:
pends upon the configuration in which the mechanism is fabri-
• In-situ fabrication is blind to conventional component bound- cated 共also called the build pose兲. The output variability, in turn, is
aries. Consequently, the input to the analysis is not the dimen- determined by the sensitivity of the output function to the mecha-
sional variability in links, but the absolute position and orientation nism parameters at each operating configuration. Figures 6 and 7
variability in joints. As the assembly is built, joints are created illustrate the fundamental differences between the two scenarios,
directly or embedded within a surrounding matrix of part and for the simple case of a four-bar mechanism.
support material. Links are formed around the joints. Parametric
2.2 Frames and Notation. We assign a global workspace
variability is therefore a function of joint placement accuracy.
datum frame (OXY Z) and local datum frames (o i x i y i z i ) associ-
• Tolerance stack-up due to dimensional/parametric errors in
ated with each feature of interest, 共see Fig. 8兲. Without loss of
components is not an issue for in-situ fabrication. Instead, joints,
generality, it can be assumed that the z-axis of the global frame is
and other features such as coupler points or end-effectors, are
aligned with the process growth direction 共e.g. vertical, or spindle-
placed in the workspace with a known absolute accuracy.
axis兲. If the feature of interest is a joint, then it is assumed that the
• Gaps and clearances in joints are manifest directly in the
local joint z-axis (z i for the ith joint兲 is aligned with the joint-
geometry of the support structure. In conventional fabrication, the
freedom axis 共i.e. nominal pin/shaft axis for revolute joints, direc-
gap geometry is a consequence of the interaction amongst
tion of translational motion for prismatic joints etc.兲. The direction
complementary mating/fitting feature geometries.
of the x-axis of the ith frame (x i ) is taken as that of the common
• Conventional error analysis does not explicitly allow for the
normal between the ith and 共adjacent兲 i⫺1th nominal joint axes.
consideration of variable accuracy within the manufacturing
Typically, the position and orientation of each feature frame is
workspace. But when entire mechanisms are fabricated In-situ, the
specified in the global frame, and the feature geometry is specified
build configuration 共or pose兲 can be chosen to make best use of
in the local frame. The nominal location of the origin in the ith
the manufacturing error characteristics.
local frame is represented as the position vector pi in the global
These differences are accounted for in the general abstract frame 共or alternately, as the homogeneous coordinates
model for in-situ fabrication and the associated error analysis 关 x i ,y i ,z i ,1兴 ), and the nominal orientation of the ith frame is rep-
techniques presented below. resented by the direction vector zi 共with direction numbers
关 l i ,m i ,n i 兴 ). Alternately, the z-axis of the joint frame can be
2.1 An Abstract Model for In-Situ Fabrication. The main uniquely represented in a global frame in terms of its Plücker 关27兴
difference between conventional error analysis, and error analysis coordinates (Qi ,Q⬘ i ), where:
for in-situ fabrication lies in the form of the inputs into the model.
Conventional error analysis treats parametric variability 共i.e. vari- Qi ⬅ 关 q 1i ,q 2i ,q 3i 兴 (17)
ability in link-lengths etc.兲 as a given constant input. In-situ error
are the direction numbers, and:
analysis estimates parametric variability for each build configura-
tion from the location variability of the joints that make up the ⬘ ,q 2i
Q⬘ i ⬅pi ⫻Qi ⬅ 关 q 1i ⬘ ,q 3i
⬘兴 (18)
linkage. The parametric variability is determined by the sensitivity 2
of each parameter to the joint positions and orientations at a given is the moment vector of the line. Furthermore, we can let q 1i
build pose. An important observation is that the mechanism pa- ⫹q 2i ⫹q 3i ⫽1 without any loss of generality, making these coor-
2 2

rameters that result from such fabrication are not independent, but dinates the same as the direction cosines of the line.

Fig. 7 The effects of joint location variation for an in - situ fabricated 4-bar
mechanism

Journal of Mechanical Design DECEMBER 2003, Vol. 125 Õ 813


SLS, Stereolithography etc. The variability region is simply a
worst-case or stochastic characterization of the variation in frame
position and orientation, given its nominal location and other
process-specific parameters. While this methodology extends to
the general spatial scenario, it is illustrated here with a simple
planar example.
2.3.1 Planar Example. In the planar case, the orientations of
the joint axes 共i.e. zi ) are discarded, as all joint axes are assumed
parallel. Given a nominal joint location pnom⬅(x nom,y nom), the
precision function returns a region R as follows:
R⫽ ␶ 共 x nom,y nom, ␲ 兲 (21)
In deterministic worst-case analysis, this function returns the ex-
tremal positions of the region in which the actual joint lies, as
follows:
R⫽ 关 worst⫺case,x min,x max,y min,y max兴 (22)
Similarly, in stochastic analysis, the function returns a probability
distribution that describes the position of the point as a random
Fig. 8 Frames and notation for the abstract model of in - situ variable, as follows:
fabrication
R⫽ 关 normal, ␮ x , ␴ 2x , ␮ y , ␴ 2y 兴 (23)
In the most general case, R is a closed region of arbitrary geom-
Thus, using this representation, the nominal configuration etry within which the actual joint position (x,y) lies with a known
(C nom) of a mechanism can be represented in terms of the local probability distribution. By applying the precision function ␶ to all
frame positions and orientations as: the joint and coupler points (x inom ,y inom) in a planar mechanism,
C nom⬅ 兵 共 pinom ,zinom兲 其 ; i⫽1,2, . . . n (19) we get joint variability regions R i as:
or alternately, in terms of the joint-axis Plücker coordinates as: R i ⫽ ␶ 共 x inom ,y inom , ␲ 兲 (24)
C nom⬅ 兵 共 Qi ,Q⬘ i 兲 其 ; i⫽1,2, . . . n (20) In other words, the regions R i determine the characteristics of the
Fabrication proceeds by constructing or embedding non-ideal interval or random values that represent the variable nature of the
joints at the given nominal locations. By quantifying the extent of joint locations. The mechanism parameters ␾ i are functions 共e.g.
these errors, it is possible to predict overall performance errors in distance function of the form ␾ i ⫽ 兵 兺 (x i ⫺x j ) 2 其 1/2) of the posi-
mechanisms fabricated in-situ. The complete procedure is de- tions and orientations, and the parametric variability is a function
scribed in later sections 共Secs. 2.4 and 3兲. of the joint variability regions (R i ), all at the given build configu-
ration (C b ):
2.3 Heterogeneous Workspace Modeling. For modeling
variable fabrication accuracy within the process workspace, we ⌬ ␾ i ⫽⌬ ␾ i 共 R 1 ,R 2 , . . . R n 兲 ; i⫽1,2, . . . n (25)
assume that we have a precision function 共␶兲 that returns the vari-
ability region R of a joint in the build space, given the nominal Error analysis involves estimating the variability in the link
position and orientation, and other process parameters 共␲兲. Note parameters ␾ i using the above equation, and then applying sensi-
that the precision function ␶ is process-specific and needs to be tivity analysis techniques to determine the error in the output
empirically determined for each process, such as SDM, FDM, function 共at various operating configurations兲 for a mechanism

Fig. 9 Actual and schematic diagrams of the planar 4-bar crank-rocker mechanism used as an
example in this paper. The parameter values are: L 1 Ä15 cm, L 2 Ä5 cm, L 3 Ä25 cm, L 4
Ä20 cm, L 5 Ä7.5 cm, L 6 Ä20 cm „after Mallick and Dhande, 1987…. Stochastic simulations on
the example are performed with a positional variance ␴ x2 k Ä0.01 cm2 . Worst case simulations
are performed with a positional variability of 0.3 cm, equivalent to the 3␴ stochastic error.

814 Õ Vol. 125, DECEMBER 2003 Transactions of the ASME


Fig. 10 Multiple positions of the example 4-bar mechanism, corresponding
to 30 deg increments of the input angle, ␪

that is fabricated in-situ. In the following sections, this process is 2.4 Error Estimation. Worst case error estimation for in-
described, and illustrated using the specific planar 4-bar mecha- situ fabrication proceeds in two stages. First, at a candidate build
nism shown in Fig. 9. The mechanism parameter values were pose C b , all the worst case parameter values ( ␾ iWC ) are evaluated
chosen to allow checking of results with earlier published work by choosing, in sequence, all possible combinations of the worst-
关28兴. However, in that case, the authors consider a clearance error
of 0.05 cm in the joints, along with a 0.5 percent error in link case fabrication input values 兵 (piWC ,ziWC ) 其 .
length. Since neither link variability or clearance are variables The precision function 共Eq. 共24兲兲 returns these extremal values
in our analysis, the comparison is qualitative. To aid with discus- of the position of each joint, given the mechanism nominal build
sion of the results, the mechanism is also shown in various pose. For k fabrication input variables, this process generates 2 k
configurations 共i.e. specific values of the driving angle ␪兲 in Fig. candidate mechanisms at each pose C b . Figure 11 shows the ex-
10. In Sec. 4, the analysis is extended to cover general spatial ample of a mechanism with 3 mm square precision regions and a
mechanisms.
candidate build configuration.
In the second stage, the error in the output function 共y兲 is evalu-
ated for each one of the candidate mechanisms produced in the
first stage. This calculation is repeated for all operating angles, for
every build pose. Overall, if c operating and build positions are
considered for a mechanism with m independent degrees of free-
dom, and k independent fabrication variables, the determination
of worst-case error boundaries for the output has computational
complexity O(2 k mc 2 ). Dynamic programming approaches 关22兴
can significantly improve upon the computational complexity, but
need to be re-stated appropriately for each specific problem.
Figure 12 illustrates the results of the worst-case error estima-
tion for the example 4-bar mechanism for a few candidate build
poses. The coupler-point location is shown as a cloud of points in
the vicinity of the nominal coupler-point, with each point corre-
sponding to one combination of worst-case joint locations. Figure
13 plots the worst-case variability of the coupler-point location
共i.e. half the perimeter of the bounding box for each cloud in Fig.
12兲 as a function of the build configuration. Of the four build
configurations evaluated, the one corresponding to ␪ ⫽180 deg is
evidently best for minimizing the worst-case errors in coupler
position.

3 Stochastic Error Analysis for In-Situ Fabrication


The worst-case method presented in the previous section is both
Fig. 11 An example build configuration and worst-case varia- overly conservative, and computationally expensive for most ap-
tions in joint and coupler point locations plications. By contrast, a stochastic approach results in superior

Journal of Mechanical Design DECEMBER 2003, Vol. 125 Õ 815


Fig. 12 Worst case coupler-point positional error, plotted on the coupler path

error estimates in constant-time 共as opposed to exponential or lin- given, analytically specified, distributions of the original random
ear time for worst-case methods兲. However, the conventional ap- variables. However, in practice, the exact derivation is intractable
proach to stochastic error estimation needs modification in order in the absence of certain simplifying assumptions, due to the com-
to be applicable to in-situ fabrication. plexity of the algebra involved. For a weakly non-linear function
In this analysis, we assume that the joint coordinates 共positions of independent and uncorrelated random variables, the mean and
and orientations兲 are independent random variables with known variance of the function can be approximated directly from the
distributions. Given the nominal location of a joint i, the precision mean and variance of the underlying random variables, as illus-
function 共Eq. 共24兲兲 returns the appropriate distribution for its ac- trated in Eqs. 共15兲 and 共16兲. When the simplifying assumptions
tual location. Mechanism parameters 共like link-lengths, joint 共i.e. independent and uncorrelated兲 do not hold, the function prop-
angles, joint offsets and skew angles兲 are functions of the inde- erties need to be determined analytically by integrating the joint-
pendent, random joint coordinates. This, in turn, makes the param- PDF 共see Eq. 共28兲兲, by modifying the approximation techniques to
eters themselves random variables which are pairwise correlated include the effects of correlation, or by using Monte Carlo simu-
共being jointly dependent on the same independent variables兲. The lation techniques. In general, the analytical technique is not trac-
output, then, is a complex function of correlated random variables. table for all but the simplest of cases. In the following section, an
The probability distribution 共i.e. PDF兲 of a known function of improved approximation technique for the estimation of the mo-
random variables can, in principle, be derived exactly from the ments of a weakly-nonlinear function of correlated random vari-

Fig. 13 Total worst-case coupler-point positional errors, plotted against operating angle for four
different build configurations, corresponding to different values of the input angle. The error values
can be compared with 3␴ stochastic errors „see Fig. 17….

816 Õ Vol. 125, DECEMBER 2003 Transactions of the ASME


Fig. 14 First order estimates of the link-length variance compared to the results of a Monte Carlo
simulation

ables is developed and applied to the problem of stochastic error links in the example 4-bar in Fig. 9. 共Since the mechanism is built
estimation for mechanisms that are fabricated in-situ. The results in-situ, the link length variation is a consequence of variations in
are compared to those obtained by Monte Carlo simulation. joint location.兲
Figure 15 compares the pairwise correlation coefficients ob-
3.1 Estimating the Parametric Variance. Equation 共15兲 tained for the approximation in Eq. 共27兲 against those obtained by
can be applied directly to the mechanism parameters ( ␾ i ), given Monte Carlo simulation, for the same four links of the example
the stochastic properties 共i.e. mean and variance兲 of the joint vari- 4-bar. In both cases, the approximation yields results that are very
ables (x k ). The parameters are simple functions 共i.e. sums, prod- close to the simulation—illustrating the validity of the assumption
ucts and differences兲 of the joint variables, which are assumed of independence. Note also that these results hold for an example
independent and uncorrelated. Moreover, the variance in any joint with tolerances that are looser than is common in macroscopic
variable can be assumed to be much smaller than its mean 共for devices. As a percentage of the link lengths, the tolerances are
macro-scale devices兲, since the precision of fabrication equipment more characteristic of MEMS devices.
is typically several orders-of-magnitude smaller than the part
dimensions. This implies that the variability in the mechanism 3.2 Estimating the Output Variance. In the previous sec-
parameters can be approximated as a linear function 共weighted tion, we have established a method for efficiently estimating the
by the sensitivity coefficients兲 of the variability in the input, as variance and correlation coefficients of the parameters of a
follows: mechanism that has been fabricated in-situ. Our real interest in

兺冉 冊
this treatment, however, is in the behavior of the output function
⳵␾i 2
共y兲 during operation. As indicated earlier, the output is a function
␴ ␾2 i ⬇ ␴ x2k ; i⫽1,2, . . . n (26)
k ⳵xk ␮
of the mechanism parameters which, being dependent functions of
the given independent random variables 共i.e. the joint variables兲,
where ␴ ␾2 i is the variance of the ith mechanism parameter, and x k are themselves correlated random variables. Thus, the simplifying
assumptions which could be made for the estimation of paramet-
represents the kth joint variable, and ␴ x2k represents the variance
ric variability are not applicable for the estimation of output vari-
of the kth joint variable. If the joint variables follow Normal dis- ability. No simple analytical technique exists for the determina-
tributions 共typical for most physical random processes involving tion of the distribution of a general function of correlated random
many noise factors兲, then the parameters too will follow a Normal variables. In theory, the cumulative distribution function of the
distribution. output can be evaluated as follows:
The parameters ␾ i , however, are correlated random variables.
The correlation coefficients ( ␳ i j ) of each parameter pair ( ␾ i , ␾ j )
can be approximated using the sensitivity coefficients as follows: PY共 y 兲⫽ 冕 冕
␾1
¯
␾n
f 共 ␾ 1 , . . . ␾ n 兲 .p ⌽1 . . . ⌽n 共 ␾ 1 , . . . ␾ n 兲

兺 冉 ⳵x 冊 冉 冊
⫺⬁ ⫺⬁
⳵␾i ⳵␾ j
␴ x2k ⫻d ␾ 1 . . . d ␾ n
k k ␮ ⳵xk ␮
(28)
␳i j⬇ (27)
␴ ␾i␴ ␾ j
However, the joint distribution function p ⌽1 . . . ⌽n ( ␾ 1 , . . . ␾ n ) is
Figure 14 compares the first order estimate of link-length vari- not easy to determine when the random variables ␾ i are corre-
ability against that obtained by Monte Carlo simulation, for the lated. Furthermore, the upper limits of the multiple integral need

Journal of Mechanical Design DECEMBER 2003, Vol. 125 Õ 817


Fig. 15 Pairwise correlation coefficients of the link lengths—first-order results compared to the Monte
Carlo simulation

to be expressed in terms of the output variables, which is not variances and the correlation coefficients are functions of the
analytically feasible except for the simplest of cases. mechanism build pose, during in-situ fabrication. Equation 共29兲
The assumption that makes this problem tractable, once again, succinctly relates the fabrication workspace to the operational
is that of weak-nonlinearity in the output function. In other words, workspace, thereby presenting us with a method for evaluating the
if we can assume that the second and higher-order terms in the optimal build pose, given an operational tolerance specification.
Taylor Series expansion of the output function can be discarded, This issue is explored in more detail in 关29兴.
then it is possible to derive an expression that directly produces an Figure 16 compares the first order estimated coupler-point error
approximate estimate for the output variance, given the variance for the example 4-bar fabricated in-situ against the Monte Carlo
( ␴ ␾2 i ) and correlation coefficients ( ␳ i j ) of the mechanism param- simulations of the same quantity. Also included are the estimates
eters. Furthermore, if the total number of parameters is large 共i.e. using the conventional approach, which does not include the con-
n⬎5), then, according to the Central Limit Theorem, the output sideration of correlation effects. Comparisons can also be made
function will follow an approximately Normal distribution, re- between these results, and those of the worst case error estimate
gardless of the individual parameter distributions 关23兴. Thus, by presented earlier 共see Fig. 13兲. The worst-case and stochastic es-
making the linear approximation, we completely side-step the timates for a specific build angle are compared in Fig. 17. It is
evaluation of the extremely problematic multiple integral in Eq. clear from the comparison that the worst-case method is signifi-
共28兲. The derivation of the approximation equation is given in cantly more conservative in its estimation of output error.
Appendix A, and the final result is summarized below: Figure 18 plots the simulated coupler-point variance against the
number of random trials. This helps with the estimation of the

兺 冉 ⳵␾ 冊 册 册
n
⳵f 2
⳵f ⳵f minimum number of trials needed in order for the random esti-
␴ 2y ⬇
i⫽1 i ␮
␴ ␾2 i ⫹2 兺 兺 ⳵␾
i j i
␳ ␴ ␴
⳵ ␾ j ␮ i j ␾i ␾ j
mates to converge to a steady value 共between 4000 and 10,000 in
␮ this case兲.
(29)
where i⫽1,2, . . . n and j⫽i. In the special case where only ad-
jacent parameters share a joint variable, ␳ i j ⫽0 for non-adjacent 4 Extension to Spatial Parameters
parameters, and the above equation needs to be evaluated only for While the detailed treatment of spatial error analysis, with sup-
the cases where j⫽i⫺1. Note that all the sensitivity coefficients porting numerical results, is beyond the scope of this paper, the
in the above equation are evaluated at the nominal operating con- theoretical extension of the error analysis techniques presented
figuration 共␮兲 of the mechanism. Comparison of Eq. 共29兲 and Eq. above to spatial systems is straightforward once the essential con-
共15兲 reveals that they differ only in the second term on the RHS. cepts have been established. Spatial systems are traditionally de-
This term, then, is the adjustment term that accounts for the cor- scribed in terms of the Denavit-Hartenberg parameters 共see Sec-
relation effect that results from the co-dependence of the mecha- tion 1.3.4兲, or modifications thereof. Spatial error analysis is the
nism parameters on the same joint coordinates. process of relating variability in the spatial parameters to errors in
Summarizing, the first order approximations are the only trac- the output function.
table, general purpose estimates of the output function variability. For in-situ fabrication, parametric variability is not directly
Equation 共29兲 indicates that the output error depends upon the available, but is a function of the position and orientation variabil-
output function sensitivity coefficients 共evaluated at the nominal ity in joint placement. Earlier sections in this paper have dealt
operating configuration兲, the parametric variances, and the pair- with the issue of estimating the output variance, given the stochas-
wise correlation coefficients of the parameters. The parametric tic characteristics of the joint variables. The approach has been

818 Õ Vol. 125, DECEMBER 2003 Transactions of the ASME


Fig. 16 First order estimates of coupler-point variance for the an in - situ fabricated
crank-rocker mechanism „Fig. 9… using conventional stochastic analysis, analysis
modified to for in - situ fabrication, and direct Monte Carlo simulation

illustrated using a planar example, and the technique is extended and the moments of the common normal between axes i and j are
here to cover general spatial mechanisms. The basic issue that given as follows 共this can be extended to j and k by symmetry兲:
remains to be addressed for the spatial case is that of explicitly
expressing the spatial parameters illustrated in Fig. 5 in terms of
the joint-frame positions illustrated in 8. This is a fairly simple ⬘ j ,q 2i
Q⬘ i j ⬅ 关 q 1i ⬘ j ,q 3i
⬘ j 兴 where
problem in the analytical geometry of three dimensions 关30兴.
Given the origin coordinates (pi ,p j ,pk ) and the direction num-
Qi j • 关共 q 2i j q 3 j ⫺q 3i j q 2 j 兲 Q⬘ i ⫺ 共 q 2i j q 3i ⫺q 3i j q 2i 兲 Q⬘ j 兴
bers (zi ,z j ,zk ) of the axes of three adjacent spatially located ⬘ j⫽
q 1i
joints, the modified Denavit-Hartenberg parameters of the jth 储 Qi j 储 2
joint can be expressed in terms of the joint Plücker coordinates
共see Section 2.1兲 of the three joint axes (Qi ,Q⬘ i ), (Q j ,Q⬘ j ) and
Qi j • 关共 q 3i j q 1 j ⫺q 1i j q 3 j 兲 Q⬘ i ⫺ 共 q 3i j q 1i ⫺q 1i j q 3i 兲 Q⬘ j 兴
(Qk ,Q⬘ k ), and those of the two common normals (Qi j ,Q⬘ i j ) and ⬘ j⫽
q 2i
(Q jk ,Q⬘ jk ). This notation is illustrated in Fig. 19. The direction 储 Qi j 储 2
coordinates of the common normal are given as:
Qi j ⬅ 关 q 1i j ,q 2i j ,q 3i j 兴 where Qi j • 关共 q 1i j q 2 j ⫺q 2i j q 1 j 兲 Q⬘ i ⫺ 共 q 1i j q 2i ⫺q 2i j q 1i 兲 Q⬘ j 兴
⬘ j⫽
q 3i
储 Qi j 储 2
q 1i j ⫽q 2i q 3 j ⫺q 3i q 2 j (31)
q 2i j ⫽q 3i q 1 j ⫺q 1i q 3 j
The modified Denavit-Hartenberg parameters for link j can
q 3i j ⫽q 2i q 3 j ⫺q 3i q 2 j (30) now be written as:

Journal of Mechanical Design DECEMBER 2003, Vol. 125 Õ 819


Fig. 17 Comparison of 3␴ stochastic and worst-case „deterministic… error estimates
for crank-rocker mechanism

␣ j ⫽arcsin共 储 Q jk 储 兲 mation matrices 共three translations and two rotations兲, that trans-
form one local coordinate frame to the adjacent frame 共the jth
Q j •Q⬘ k ⫹Qk •Q⬘ j frame to the kth frame in this case兲, as follows:
a j⫽
sin共 ␣ i 兲
Akj ⫽T共 0,0,d j 兲 ⫻R共 z j , ␪ j 兲 ⫻T共 a j ,0,0 兲 ⫻R共 x j , ␣ j 兲 ⫻T共 0,0,l j 兲
共 Q jk ⫻Q j 兲
d j ⫽ 共 pk ⫺p j 兲 • (33)
储 Q jk 储 2
Next, the first-order Taylor Series approximation of the transfor-
共 Q jk ⫻Q j 兲
l j ⫽ 共 p j ⫺pk 兲 • mation matrix is written as follows:
储 Q jk 储 2
␪ j ⫽arcsin共 储 Q jk 储 ⫻ 储 Qi j 储 兲 (32) ⳵ Akj ⳵ Akj ⳵ Akj ⳵ Akj ⳵ Akj
⌬ Ak ⫽ ⌬ d j⫹ ⌬ ␪ j⫹ ⌬ a j⫹ ⌬ ␣ j⫹ ⌬
Since the mechanism parameters are now known in terms of the ⳵d j ⳵␪ j ⳵a j ⳵␣ j ⳵l j lj
joint positions and orientations, it is possible to estimate the error (34)
in output function given the variability in joint location using
techniques similar to those outlined for the planar case earlier. The The parameter variabilities 共i.e. ⌬ d j ,⌬ ␪ j ,⌬ a j ,⌬ ␣ j and ⌬ l j ) are
process proceeds by writing the product of homogeneous transfor- now either interval 共for worst-case analysis兲 or random 共for sto-

Fig. 18 Convergence rates of Monte Carlo simulations for different operational


angles

820 Õ Vol. 125, DECEMBER 2003 Transactions of the ASME


• There is the freedom to choose a build configuration that will
minimize the output variability when the mechanism is in its
operating configuration.
• Important functional gaps and clearances can be controlled
directly, by controlling the dimensions of sacrificial support
material between mating parts, rather than being a conse-
quence of the mating of independently fabricated parts.
We surmise that it will be particularly important to take advan-
tage of these characteristics in fabricating MEMS and meso-scale
mechanisms, for which the process variability is typically a larger
percentage of the feature size than for macroscopic devices.
These topics are the subject of ongoing investigation. Some
results on the treatment of clearances and on build pose optimiza-
tion are provided in 关29兴.

Acknowledgments
Fig. 19 Notation for the derivation of modified Denavit-
Hartenberg parameters from joint Plücker coordinates. We gratefully acknowledge the support of the National Science
Foundation 共MIP 9617994兲 and the Alliance for Innovative Manu-
facturing 共AIM兲 at Stanford. We are also grateful for the assis-
tance provided by members of the Stanford Center for Design
chastic analysis兲 parameters, the variances and correlation coeffi- Research and Rapid Prototyping Laboratories. Sanjay Rajago-
cients of which can be obtained using the relationships derived in palan also thanks Jisha Menon for her generous support during the
Eq. 共32兲. formulation of this work—some of which constitutes the basis for
his Ph.D. thesis at Stanford University.
5 Conclusions and Future Work
A framework has been presented for reasoning about errors in
the performance of mechanisms that are slated to be built using Appendix: Estimation of Mean and Variance
the increasingly popular ‘‘freeform’’ fabrication techniques. This Here, we are concerned with the approximate estimation of the
is achieved by formulating an abstract model for the in-situ fab- mean and variance of an output y, described in terms of its output
rication of mechanisms, and solving the problem of analytical function f (•) and a set of n random parameters ⌽
estimation of the variance of the kinematic function, in the pres- ⬅ 关 ␾ 1 , ␾ 2 , . . . , ␾ n 兴 as follows:
ence of correlated random parameters. The fundamental assump-
tions in this treatment of error analysis are: y⫽ f 共 ⌽兲 (35)

• The desired performance of the mechanism is specified in where f (•) is, in general, a continuous and differentiable non-
terms of a kinematic output function, which is a continuous and linear mapping, and the parameters ⌽ are random variables with
differentiable mapping from a parameter space to the operational no assumptions made about their distributions, correlations or in-
workspace 共usually a Cartesian space兲. This assumption limits the dependence. It is assumed, however, that the function f (•) is only
application of the methods presented to linkages with lower pairs weakly non-linear 共i.e. high-order terms in it’s Taylor Series ex-
and ‘‘well-behaved’’ higher pairs only. pansion can be neglected兲 and that the mean and variance of the
• The output is a weakly non-linear function of the inputs. This parameters ␾ i are known, and denoted as ( ␮ ␾ i , ␴ ␾2 i ).
enables a first-order Taylor Series approximation of the error at We begin by expanding the output function in its Taylor Series,
the points of interest. about the mean values of the parameters, as follows:


• In-situ fabrication is abstracted as a process of independent n
insertions of joints 共which could have internal clearances兲 into a ⳵f
fabrication workspace, with a known accuracy. The inaccuracy is
y⫽ f 共 ␮ ␾ i ;i⫽1,2, . . . ,n 兲 ⫹ 兺 ⳵␾
i⫽1 i
共 ␾ i⫺ ␮ ␾i兲

specified as worst-case limits on position and orientation 共for de-
terministic error analysis兲 or variances with known distributions
共for stochastic error analysis兲.
Note that no assumptions of planarity or of homogeneity in work-

1
2!
n

兺 ⳵␾
i⫽1
⳵2 f
2
i ␮
册共 ␾ i⫺ ␮ ␾i兲 2
⫹ 兺 ⳵␾ ⳵␾
i⬎ j
⳵2 f
i j
册 ␮
共 ␾ i ⫺ ␮ ␾ i 兲共 ␾ j

space characteristics are made anywhere in the methodology. ⫺ ␮ ␾ j 兲 ⫹¯ (36)


Analysis of parametric errors in spatial mechanisms has also been
covered in the theoretical formulation. With a little bit of rearrangement, the above equation can be re-
This paper demonstrates that differences in the manufacturing written in terms of proxy variables ⌬ ␾ i as:


process flow for in-situ fabrication leads to fundamental differ- n
⳵f
ences in how process input variability is manifested in the kine-
matic output of a mechanism. y⫽ f 共 ␮ i ;i⫽1,2, . . . ,n 兲 ⫹ 兺 ⳵␾
i⫽1 i
⌬ ␾i
For stochastic analysis, the essential result is that we must ac- ␮
count for correlations among adjacent links. In this paper we have
presented a modified stochastic analysis that accounts for the cor-
relations and shown that it compares favorably with numerical
⫹ 兺兺 i j
⳵2 f
⳵ ␾ i⳵ ␾ j 册 ␮
⌬ ␾ i ⌬ ␾ j ⫹O 3 (37)

Monte Carlo simulations. where ⌬ ␾ i ⫽ ␾ i ⫺ ␮ ␾ i are zero-mean random variables, with all
Although the need to consider correlations in the variabilities of
link parameters somewhat complicates the analysis, in-situ fabri- higher order moments identical with ␾ i . The term O 3 stands for
cation also affords some important advantages over conventional all terms in the Taylor Series expansion that are of third degree or
fabrication for reducing output variability, notably: more, and are usually negligible.
We now go about the task of estimating the mean and variance
• Tolerances do not accumulate along serial chains. of the output 共the LHS term兲, using the above equation. In this

Journal of Mechanical Design DECEMBER 2003, Vol. 125 Õ 821


regard, we make use of the following results, which are based on adjustment to the variance estimate from the first term, accounting
elementary applications of theorems in the area of Mathematical for any correlative effects.
Statistics 关23兴:
E 兵 f 共 ⌽兲 其 ⫽ f 共 ␮ ␾ i 兲 References
关1兴 Beaman, J. J., 1997, Solid Freeform Fabrication: A New Direction in
E 兵 ⌬ ␾ i 其 ⫽0 Manufacturing—With Research and Applications in Thermal Laser Process-
ing, Kluwer Academic Publishers, Boston.
Var兵 y其 ⫽E 兵 共 y⫺ ␮ y兲 2 其 关2兴 Chua, C. K., and Fai, L. K., 1997, Rapid Prototyping: Principles and Appli-
cations in Manufacturing, Wiley, New York, NY.
Cov兵 ⌬ ␾ i ,⌬ ␾ j 其 ⫽E 兵 ⌬ ␾ i ⌬ ␾ j 其 ⫺E 兵 ⌬ ␾ i 其 E 兵 ⌬ ␾ j 其 (38) 关3兴 Merz, R., Prinz, F. B., Ramaswami, K., Terk, M., and Weiss, L., 1994, ‘‘Shape
Deposition Manufacturing,’’ Proceedings of the Solid Freeform Fabrication
where E 兵 • 其 stands for the expected value, Var兵 • 其 stands for the Symposium, pages 1– 8, The University of Texas at Austin, August 8 –10.
关4兴 Bottema, O., and Roth, B., 1979, Theoretical Kinematics, Dover Publications,
variance and Cov兵 • 其 stands for the covariance. For notational New York, NY.
simplicity, we denote the expected value, or mean, by the symbol 关5兴 Morrison, P., and Morrison, E., 1961, Charles Babbage and His Calculating
␮ 共with the appropriate subscript兲, and the variance by the symbol Engines, Dover Publications, Inc., New York, NY.
关6兴 Cham, J., Pruitt, B. L., Cutkosky, M. R., Binnard, M., Weiss, L., and Neplot-
␴ 2 . In addition, we use the covariance coefficient ( ␳ i j ), which is nik, G., 1999, ‘‘Layered Manufacturing of Embedded Components: Process
defined as follows: Planning Considerations,’’ Proceedings of the 1999 ASME DETC/DFM Con-
ference, Las Vegas, NV, September 12–15.
Cov兵 ⌬ ␾ i ,⌬ ␾ j 其 关7兴 Weiss, L. E., Prinz, F. B., Neplotnik, G., Padmanabhan, P., Schultz, L., and
␳i j⬅ (39) Mertz, R., 1996, ‘‘Shape Deposition Manufacturing of Wearable Computers,’’
␴ ␾i␴ ␾ j Proceedings of the Solid Freeform Fabrication Symposium, University of
Texas at Austin, August 10–12.
Note that ⫺1⭐ ␳ i j ⭐1, and that ␳ i j ⫽1 when i⫽ j and ␳ i j ⫽0 for 关8兴 Knappe, L. F., 2000, ‘‘Building Around Inserts: Methods for Fabricating Com-
independent or uncorrelated ␾ i and ␾ j . From the above equa- plex Devices in Stereolithography,’’ Proceedings of the 2000 ASME DETC/
tions, it is also apparent that: DFM Conference, Baltimore, MD, September 10–13.
关9兴 Mavroidis, C., DeLaurentis, K. J., Won, J., and Alam, M., 2001, ‘‘Fabrication
Cov兵 ⌬ ␾ i ,⌬ ␾ j 其 ⫽E 兵 ⌬ ␾ i ⌬ ␾ j 其 , and E 兵 ⌬ ␾i⌬ ␾ j其 ⫽ ␳ i j ␴ ␾i␴ ␾ j of Non-assembly Mechanisms and Robotic Systems Using Rapid Prototyp-
ing,’’ ASME J. Mech. Des., 123, pp. 516 –519, December.
(40) 关10兴 Laliberte, T., Gosselin, C., and Cote, G., 2000, ‘‘Rapid Prototyping of Lower-
pair, Geared-pair and Cam Mechanisms,’’ Proceedings of the 2000 ASME
Returning to the output expansion in Eq. 共37兲, and using the re- Mechanisms and Robotics Conference, Baltimore, MD, September 10–13.
sults detailed above, we are able to write the expression for the 关11兴 Tuttle, S. B., 1960, ‘‘Error Analysis,’’ Mach. Des., 32共12兲.
expected value of the output function as follows: 关12兴 Knappe, L. F., 1963, ‘‘Technique for Analyzing Mechanism Tolerances,’’


Mach. Des., pp. 155–157, April 25.
⳵2 f 关13兴 Hartenberg, R. S., and Denavit, J., 1964, Kinematic Synthesis of Linkages,
E 兵 y其 ⬅ ␮ y⬇ f 共 ␮ ␾ i 兲 ⫹0⫹ 兺兺 i j ⳵ ␾ i⳵ ␾ j
E 兵 ⌬ ␾ i ⌬ ␾ j 其 (41) McGraw Hill, New York.
关14兴 Garrett, R. E., and Hall, A. S., 1969, ‘‘Effects of Tolerance and Clearance in

Linkage Design,’’ ASME J. Eng. Ind., Series B, 91, pp. 198 –202.
or, using Eq. 共40兲: 关15兴 Dhande, S., and Chakraborty, J., 1973, ‘‘Analysis and Synthesis of Mechanical


Error in Linkages—A Stochastic Approach,’’ ASME J. Eng. Ind., Series B, 95,
⳵2 f
兺 兺 ⳵␾ ⳵␾
pp. 677– 680.
E 兵 y其 ⬅ ␮ y⬇ f 共 ␮ ␾ i 兲 ⫹ ␳ i j ␴ ␾i␴ ␾ j (42) 关16兴 Lakshminarayana, K., and Ramaiyan, G., 1976, ‘‘Analysis of the Effects of
i j i j Errors and Clearances in Mechanisms as a Problem in Statics,’’ ASME publi-

cation 76-DET-67, Proceedings of the ASME DETC, Montreal, Quebec,
Equation 共42兲 is a general expression for the approximation of the Canada, September.
mean of a function f (•) of random variables, which are—in 关17兴 Tischler, C. R., and Samuel, A. E., 1999, ‘‘Prediction of Slop in General
Spatial Linkages,’’ Int. J. Robot. Res., 18共8兲, 845– 858, August.
general—correlated. 关18兴 Denavit, J., and Hartenberg, R. S., 1955, ‘‘A Kinematic Notation for Lower-
In a manner similar to the earlier analysis, we can use Eq. 共37兲 pair Mechanisms Based on Matrices,’’ ASME J. Appl. Mech., pp. 215–221,
to write an expression for the output variance as follows: June.

再冉兺 册
关19兴 Fox, R. L., 1971, Optimization Methods for Engineering Design, Addison-
n
⳵f Wesley, Mass.
Var兵 y其 ⬅ ␴ 2y ⫽E 兵 共 y⫺ ␮ y兲 2 其 ⫽E ⌬ ␾i 关20兴 Chakraborty, J., 1975, ‘‘Synthesis of Mechanical Error in Linkages,’’ Mech.
i⫽1 ⳵␾i Mach. Theory, 10, pp. 155–165.

册 冊冎
关21兴 Bellman, R., 1957, Dynamic Programming, Princeton University Press,
2 Princeton, NJ.
⳵2 f
⫹ 兺兺
i j ⳵ ␾ i⳵ ␾ j
⌬ ␾i⌬ ␾ j
关22兴 Fenton, R. G., Cleghorn, W. L., and Fu, J., 1989, ‘‘Allocation of Dimensional
Tolerances for Multiple Loop Planar Mechanisms,’’ ASME J. Mech., Transm.,
␮ Autom. Des., 111, pp. 465– 470, December.

册 册
关23兴 Feller, W., 1957, An Introduction to Probability Theory and Its Applications,
⳵f ⳵f
兺 兺 ⳵␾
Vol. 2, Wiley & Sons, New York, NY.
⫽ E 兵 ⌬ ␾ i ⌬ ␾ j 其 ⫹O 3 (43) 关24兴 Hayati, S., and Mirmirani, M., 1985, ‘‘Improving the Absolute Positioning
i j i ⳵␾ j ␮ Accuracy of Robot Manipulators,’’ J. Rob. Syst., 2共4兲, pp. 397– 413.

关25兴 Suh, C. H., and Radcliffe, C. W., 1978, Kinematics and Mechanism Design,
Combining Eq. 共43兲 with Eq. 共40兲, Wiley and Co., New York.

兺冉 冊 册 册
n
关26兴 Lin, P. D., and Chen, J. F., 1994, ‘‘Analysis of Errors in Precision for Closed
⳵f 2
⳵f ⳵f
兺 兺 ⳵␾
Loop Mechanisms,’’ ASME J. Mech. Des., 116, pp. 197–203, March.
␴ 2y ⬇ ␴ ␾2 i ⫹2 ␳ ␴ ␴ , i⫽ j 关27兴 Stolfi, J., 1991, Oriented Projective Geometry: A Framework for Geometric
i⫽1 ⳵␾i ␮ i j i ⳵ ␾ j ␮ i j ␾i ␾ j Computations, Academic Press, Boston, MA.

关28兴 Mallik, A. K., and Dhande, S. G., 1987, ‘‘Analysis and Synthesis of Mechani-
(44) cal Error in Path Generating Linkages Using a Stochastic Approach,’’ Mech.
Equation 共44兲 is a general expression for the approximation of Mach. Theory, 22共2兲, pp. 115–123.
关29兴 Rajagopalan, S., 2000, ‘‘Error Analysis and Optimal Pose Selection for the
the variance of a function of correlated random variables. The first In-situ Fabrication of Mechanisms,’’ Ph.D. Thesis, Stanford University.
term in the RHS expression is the variance assuming indepen- 关30兴 Sommerville, D. M. Y, 1959, Analytical Geometry of Three Dimensions, Cam-
dent and uncorrelated parameters. The second term applies an bridge University Press, London.

822 Õ Vol. 125, DECEMBER 2003 Transactions of the ASME

You might also like