Professional Documents
Culture Documents
Bayes Filter For Dynamic Coordinate Measurements - Accuracy PDF
Bayes Filter For Dynamic Coordinate Measurements - Accuracy PDF
Measurement
journal homepage: www.elsevier.com/locate/measurement
a r t i c l e i n f o a b s t r a c t
Article history: This paper presents a novel methodology to improve the measurement accuracy of
Available online 15 April 2013 dynamic measurements. This is achieved by deducing an online Bayes optimal estimate
of the true measurand given uncertain, noisy or incomplete measurements within the
Keywords: framework of sequential Monte Carlo methods. The estimation problem is formulated as
Dynamic coordinate measurements a general Bayesian inference problem for nonlinear dynamic systems. The optimal estimate
Bayesian filtering is represented by probability density functions, which enable an online, probabilistic data
Particle filter
fusion as well as a Bayesian measurement uncertainty evaluation corresponding to the
Sequential Monte Carlo methods
Online measurement uncertainty evaluation
‘‘Guide to the expression of uncertainty in measurement‘‘. The efficiency and performance
and data fusion of the proposed methodology is verified and shown by dynamic coordinate measurements.
Ó 2013 Elsevier Ltd. All rights reserved.
0263-2241/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.measurement.2013.04.001
3738 E. Garcia et al. / Measurement 46 (2013) 3737–3744
The rest of the paper is structured as follows. In Section output quantity. Finally, the Monte-Carlo methods are only
2 the basic principles of measurement uncertainty evalua- exact within the limits of the numerical accuracy and come
tion as described in the GUM and in the ‘‘Supplement 1 to up with the burden of computational complexity. In order
the GUM’’ (GS1) [3,4] are briefly restated. Section 3 intro- to not go beyond the scope of this paper, it is referred to
duces the Bayesian inference and the probabilistic model GUM and GS1 [3,4] for further details of the propagation
of the measurement process using the state-space repre- of distribution.
sentation. The Bayesian estimation problem is formulated In summary, the GUM and the GS1 specify a method to
and a general solution to this problem within the frame- calculate the measurement uncertainty by using the prop-
work of sequential Monte Carlo methods is devised. Sec- agation of probability distributions through a mathemati-
tion 4 presents experimental results of Bayes filter cal model of the measurement to derive the resulting
applied to dynamic coordinate measurements. In the end probability distribution for the output quantity. The mea-
a summary and conclusion are given. surement uncertainty is evaluated based on this PDF and
stated as uncertainty budget, which should include the
measurement model, estimates, and measurement uncer-
2. Basic principles of uncertainty evaluation tainties associated with the quantities in the measurement
model, covariances, type of applied probability density
A measurement result is generally expressed as a single functions, degrees of freedom, type of evaluation of mea-
measured quantity value and a measurement uncertainty surement uncertainty and the coverage interval. Hereby,
[5, definition 2.9]. The de facto international standards to the method of choice would be an analytical solution of
evaluate, calculate and express uncertainty in all kinds of the output density function. But as this is not possible in
measurements are the GUM and its extension the GS1 general, especially for nonlinear dynamic processes,
[3,4]. These documents provide guidance on the uncer- sequential Monte Carlo methods are utilized to achieve a
tainty evaluation as a two-stage process: formulation and Bayes optimal estimation of the output density, as de-
calculation. The formulation stage involves developing a scribed in the next section.
measurement model relating output (measurand) y to in-
put quantities x1, . . ., xN, incorporating corrections and
3. Bayesian inference
other effects as necessary and finally assigning probability
distributions to the input quantities on the basis of avail-
A measurement process generates disperse (uncertain)
able knowledge. The calculation stage consists of propagat-
observations of a true (but unknown) measurand. This is
ing the probability distributions for the input quantities
an erroneous transformation due to ubiquitous errors.
through the measurement model y = f(x1, . . ., xN) to obtain
Thus, a complete, exact or unique reconstruction of the
the probability distribution for the output quantity. From
true measurand is always impossible. But using Bayes’ the-
this PDF the expected value and the standard deviation
orem it is possible to estimate the true measurand from
of the output quantity as well as the coverage interval
observations, since the functional relation between both
are calculated. Several approaches are considered for the
is preserved.
propagation of distributions:
This estimation problem can be solved as a sequential
probabilistic inference problem for nonlinear dynamic sys-
(a) the GUM uncertainty framework, constituting the
tems. The unknown measurand (e.g. coordinates) is mod-
application of the law of propagation of uncertainty,
elled as a state of a dynamic system (e.g. moving object)
(b) analytic methods, in which mathematical analysis is
that evolves over time and is observed with a particular
used to derive an algebraic form for the probability
measurement process. This allows the estimation of mea-
distribution for the output quantity and
surable states (e.g. position) as well as derived or not di-
(c) a Monte Carlo method (MCM), in which an approxi-
rectly measurable states (e.g. velocities, acceleration and
mation to the distribution function for the output
orientations) of arbitrary dynamic systems in a statistically
quantity is established numerically by making ran-
sound way given uncertain, noisy or incomplete
dom draws from the probability distributions for
observations.
the input quantities, and evaluating the model at
The nonlinear dynamic system is described by the gen-
the resulting values.
eral discrete-time stochastic state space model
4. Experimental results for known dynamics Mean (error) in lm Max (error) in Std (error) in
lm lm
Bayes filter is a general term for all algorithmic imple- CMM 10.1, 16.8, 24.1 19.8, 25.8, 40.1 0.9, 2.1, 3.1
EKF 8.96, 15.4, 20.7 15.9, 19.8, 30.3 0.7, 1.8, 2.1
mentations of the Bayesian recursion Eqs. (2)–(6). In this
SIR PF 8.90, 13.8, 17.1 15.6, 19.8, 23.7 0.7, 1.5, 1.9
section the extended Kalman filter (EKF) and the sequen-
tial importance resampling particle filter (SIR PF) – as the
two mainly used Bayes filters – were analyzed and verified process was modelled as nonlinear estimation problem of
with respect to efficiency and performance applied to dy- 3-dimensional Cartesian coordinates. The measured quan-
namic coordinate measurements of a marker-based close tity yk (i.e. coordinates of the circular trajectory) is gener-
range photogrammetric (i.e. stereo vision based) CMM. ated by
The experiments were conducted in a high precision 2 3 2 3
measuring room (accuracy class 1 according to VDI/VDE xk r cos uk sin hk
6 7 6 7
2627 part 1 [11]) by constant temperature # = 20 ± 0.1 °C. xk ¼ 4 yk 5 ¼ 4 r sin uk sin hk 5 þ v k ;
The dynamic CMM consists of two Gigabit Ethernet inter- zk r cos hk
line progressive scan charge-coupled device (CCD) cameras
using infrared illumination. To block out visible light, opti- yk ¼ xk þ wk :
cal infrared band-pass filters, cutting off at 810 nm and a
spectral bandwidth of 35 nm, were used. Both cameras The PF with stratified resampling (based on the effec-
are mounted on a mechanical frame fulfilling the normal tive sample size), 200 particles6 Gaussian process noise
stereo case. The system calibration, i.e. estimating the in- with standard deviation 5 lm, and uniform measurement
ner and outer orientation of each camera, was performed noise with standard deviation 50 lm was used. The EKF al-
by bundle adjustment using a 3D field of calibrated refer- ways uses Gaussian noise with the same noise parameters as
ence points. Both cameras were synchronized by hardware the PF. The applied process noise based on an uncertainty
trigger and acquire images with 776 582 pixels at maxi- budget that incorporates the positioning error of the parallel
mal 64 frames per seconds. All tracking targets were cali- kinematic, the calibration uncertainty and environmental
brated with a traditional CMM that is specified with a influences. The measurement noise corresponds to the Max-
bidirectional Maximum Permissible Error E3 = (0.75 + L/ imum Permissible Error of the dynamic CMM, which was
500) lm, where the measuring length L is given in mm determined by calibration measurements as specified in
comparable to ISO 10360 [12]. The targets were moved VDI 2634 (derived from ISO 10360) [11,12]. The realized
by a high accuracy parallel kinematic (hexapod) that is measurement error is given in Table 1. It was computed as
specified with a smallest linear and angular increment of Euclidean distance between the least square circles fitted
1 nm and 1 lrad as well as a linear and angular repeatabil- in the measured coordinates as well as the unfiltered and fil-
ity of 200 nm and 10 lrad. The target was moved along a tered coordinates of the moved tracking target. The mea-
circular trajectory r = 5 mm in an inclined plane of 45° sured coordinates, the fitted circular trajectory and the
according to ISO 9283 [13]. The experimental setup is de- histogram of deviations for a single measurement run are
picted in Fig. 3. shown in Fig. 4. An example of an estimated posterior Bayes
The measurement rate of the camera system was set to posterior PDF is given in Fig. 5.
f = 30 Hz. Three independent measurement runs were per-
formed with v = 1, 3, 5 mm/s. The SIR PF and EKF were used 6
More particles were computed but did not change the results signif-
for recursive forward–backward smoothing with a sliding icantly, due to the well-known process dynamic and process noise PDF
window of 500 ms length of time. The measurement (Gaussian).
E. Garcia et al. / Measurement 46 (2013) 3737–3744 3741
Fig. 4. Analysis of a measurement run. A least square circle was fitted in the measured coordinates of the CMM.
Fig. 7. Superior accuracy of proposed APF for the estimation of a highly nonlinear helix trajectory.
E. Garcia et al. / Measurement 46 (2013) 3737–3744 3743
onds (see Fig. 6), whereas the APF shows superior estima-
tion accuracy (see Fig. 7) for all dynamics and kinematics.
In order to study the WPAM the experiments were re-
peated with a less nonlinear trajectory generated with
the initial true state x0 = [0, 1, 0, 2p/(100T), 0, 1/(100T),
0, 0, 0.4]T and the same noise sources. The resulting root-
Fig. 9. Workflow of Bayesian filtering and data fusion for dynamic
mean-square error (RMSE) with increasing number of
coordinate measurements. For the sake of visualization, the case of
particles NP = 300, . . ., 10,000 is depicted in Fig. 8. The APF measuring a two-dimensional motion of two objects with a photogram-
realized always significant better estimation accuracy metric CMM is illustrated, i.e. the state vector xk = [sx, sy]T.
compared to SIR-PF with WPAM. Especially for low num-
ber of particles the WPAM model failed to compute an esti-
mate. That results in NaN estimations. These results show density functions and subsequent evaluation of the result-
the robustness, outstanding estimation accuracy and easi- ing density function, see e.g. [17]. This could be further im-
ness of the proposed methodology for arbitrary nonlinear proved by using maximum a posteriori, maximum
dynamics and kinematics (see Fig. 8). likelihood, minimax, mean-shift, kernel density estimation
or entropy based criterions. In general the data fusion
problem can be considered as an optimization problem
6. Fusing dynamic coordinate measurements for a given metric. Several approaches exist and can be
found in the literature, e.g. [18,19].
The usage of the Bayesian estimation for dynamical Unfortunately, there is no general solution to this kind
coordinate measurements requires the software integra- of data fusion since every fusion approach highly depends
tion in existing coordinate measurement systems as well on the specific measurement task.
as straightforward and easy to use models for various mea-
surement tasks without laborious determination of model
parameters. Since Bayes filter need as inputs only sequen- 7. Summary and conclusions
tial coordinate measurements, their addition and integra-
tion in the data processing chain of existing coordinate In this paper the problem of improving and fusing coor-
measurement systems should be possible without dinate measurements of an arbitrary dynamic process was
problems. considered. In order to deal with uncertainty and estimate
The above given APF allows the measurement of arbi- the true measurand, it was solved as a Bayesian sequential
trary trajectories and kinematics of multiple objects and/ probabilistic inference problem for nonlinear dynamic sys-
or sensors by simply extending the state vector. This re- tems. Benefit of the developed methodology is a statisti-
sults automatically in Bayesian estimation for the com- cally sound, real-time and data-driven estimation of
bined state vector, i.e. a Bayesian data fusion is achieved. dynamic system variables of interest (measurand, parame-
Hereby, the number of the state dimensions is only limited ters, etc.) in form of the Bayes posterior PDF. Based on the
by the available computing power. The resulting workflow posterior density the measurement accuracy can be im-
and data processing chain is given in Fig. 9. proved by filtering of ubiquitous noise and deducing an
Another kind of data fusion can be achieved by combin- optimal quantity value with respect to any optimality cri-
ing the estimated posterior densities of separate filter teria. Furthermore it is possible to realize a measurement
based on probabilistic, information theoretical or other uncertainty evaluation and probabilistic data fusion in
metrics. The most simple data fusion approach would be every time step. This approach allows a Bayesian uncer-
the multiplication or convolution of multiple posterior tainty analysis. Depending on the process under investiga-
3744 E. Garcia et al. / Measurement 46 (2013) 3737–3744
tion this measurement uncertainty evaluation yields to terms (VIM), in: Joint Committee for Guides in Metrology, JCGM 200,
2008.
identical posterior PDFs like the GUM, which is also based
[6] O. Cappé, E. Moulines, T. Ryden, Inference in hidden markov models,
on the model-based Bayesian evaluation of probabilities. Springer, 2010.
In future work it is planned to apply this methodology [7] AM Fraser, Hidden Markov Models and Dynamical Systems, 1st ed.,
to simultaneous utilization of multiple sensors with subse- Society for Industrial Mathematics, 2009.
[8] D. Simon, Optimal State Estimation: Kalman H1 and Nonlinear
quent data fusion and measurement uncertainty evalua- Approaches, John Wiley and Sons, 2006.
tion based on the estimated probability density functions [9] A. Doucet, N. de Freitas, N. Gordon, Sequential Monte Carlo Methods
for each sensor. The presented APF will be enhanced to a in Practice, Springer, 2001.
[10] E. Garcia, A. Weckenmann, Sequential bayesian estimation for data
completely data-driven and self-adapting tracking filter fusion and data-driven measurement uncertainty evaluation, in: J.
without the need of laborious modelling and parameteri- Stadek, W. Jakubiec (Eds.), Advances in Coordinate Metrology:
zation of the measurement process under investigation. Fi- University of Bielsko Biala, 2010. – 9th International Scientific
Conference ‘‘Coordinate Measuring Technique CMT’’ (14–16 April
nally, all developed software will be released as open- 2010, Bielsko-Biala, Poland) – ISBN 978-83-62292-56-1, 2010, pp.
source software for reducing the burden of implementing 83–93.
Bayesian filters. [11] The Association of German Engineers: VDI guideline: VDI/VDE 2627
Blatt 1: Measuring rooms – Classification and Characteristics. VDI/
VDE, 1998, <http://www.vdi.eu/index.php?id=44061&tx_vdirili_
Acknowledgement pi2[showUID]=90286>.
[12] International Organization for Standardization (ISO): ISO 10360-
2:2009 – Geometrical Product Specifications (GPS) – Acceptance and
The research was supported by the German Research
Reverification Tests for Coordinate Measuring Machines (CMM) –
Foundation (Deutsche Forschungsgemeinschaft e.V.) under Part 2: CMMs Used for Measuring Linear Dimensions. ISO, 2009,
the DFG-Project Number WE 918/34-1. <http://www.iso.org/iso/iso_catalogue/catalogue_tc/
catalogue_detail.htm?csnumber=40954>.
[13] International Organization for Standardization (ISO): ISO 9283:1998
References – Manipulating Industrial Robots – Performance Criteria and Related
Test Methods. ISO, 1998, <http://www.iso.org/iso/iso_catalogue/
[1] H. Schwenke, U. Neuschaefer-Rube, T. Pfeifer, H. Kunzmann, Optical catalogue_tc/catalogue_detail.htm?csnumber=40954>.
methods for dimensional metrology in production engineering, CIRP [14] G. Box, G.M. Jenkins, G. Reinsel, Time Series Analysis: Forecasting &
Annals – Manufacturing Technology 51 (2) (2002) 685–699. Control, third. ed., Prentice Hall, 1994. ISBN 0130607746.
[2] W.T. Estler, K.L. Edmundson, G.N. Peggs, D.H. Parker, Large-scale [15] J.D. Hamilton, Time Series Analysis, first ed., Princeton University
metrology – an update, CIRP Annals – Manufacturing Technology 51 Press, 1994. ISBN 0691042896.
(2) (2002) 587–609. [16] X. Rong Li, V.P. Jilkov, Survey of maneuvering target tracking. Part I.
[3] BIPM, IEC, IFCC, ILAC, ISO, IUPAC, IUPAP and OIML, Evaluation of Dynamic models, IEEE Transactions on Aerospace and Electronic
measurement data – guide to the expression of uncertainty in Systems 39 (4) (2003) 1333–1364.
measurement GUM 1995 with minor corrections, in: Joint [17] D. Elsaesser, Sensor data fusion using a probability density grid, in:
Committee for Guides in Metrology, JCGM 100, 2008. IEEE 10th International Conference on, Information Fusion, 2007.
[4] BIPM, IEC, IFCC, ILAC, ISO, IUPAC, IUPAP and OIML: Evaluation of [18] B.D.O. Anderson, J.B. Moore, Optimal Filtering, Dover Publications,
measurement data – supplement 1 to the ‘Guide to the Expression of 2005.
Uncertainty in Measurement’ – propagation of distributions using a [19] H.B. Mitchell, Multi-Sensor Data Fusion: An Introduction, Springer,
Monte Carlo method. in: Joint Committee for Guides in Metrology, 2007.
JCGM 101, 2008.
[5] BIPM, IEC, IFCC, ILAC, ISO, IUPAC, IUPAP and OIML, International
vocabulary of metrology—basic and general concepts and associated