Motion tomography via occupation kernels

Benjamin P. Russo; Rushikesh Kamalapurkar; Dongsik Chang; Joel A. Rosenfeld

doi:10.3934/jcd.2021026

Article Contents

2022, Volume 9, Issue 1: 27-45. Doi: 10.3934/jcd.2021026

This issue Previous Article Next Article Model reduction for a power grid model

Motion tomography via occupation kernels

1.
Oak Ridge National Laboratory, Computer Science and Mathematics Division
2.
Oklahoma State University, School of Mechanical and Aerospace Engineering
3.
Oregon State University, Collaborative Robotics and Intelligent Systems Institute
4.
University of South Florida, Department of Mathematics and Statistics

^* Corresponding author: Benjamin P. Russo
^* Corresponding author: Benjamin P. Russo

^§ This work was partially completed while the author was at Farmingdale State College SUNY.
^† Research supported by AFOSR Awards FA9550-20-1-0127 and FA9550-21-1-0134.
^‡ Research supported by NSF grants ECCS-2027976, ECCS-2027999 and NRI 2.0 - 1925147.
^△This work was partially completed while the author was at the University of Michigan.
^** The following YouTube playlist discusses the contents of this manuscript: https://www.youtube.com/playlist?list=PLldiDnQu2phuBWvO40eiqNve6-_GOdSsz.
Notice: This manuscript has been authored, in part, by UT-Battelle, LLC, under contract DEAC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Received: January 31, 2021

Revised: September 30, 2021

Published: January 2022

Abstract Full Text(HTML) Figure(5) / Table(2) Related Papers Cited by

Abstract

The goal of motion tomography is to recover a description of a vector flow field using measurements along the trajectory of a sensing unit. In this paper, we develop a predictor corrector algorithm designed to recover vector flow fields from trajectory data with the use of occupation kernels developed by Rosenfeld et al. [9,10]. Specifically, we use the occupation kernels as an adaptive basis; that is, the trajectories defining our occupation kernels are iteratively updated to improve the estimation in the next stage. Initial estimates are established, then under mild assumptions, such as relatively straight trajectories, convergence is proven using the Contraction Mapping Theorem. We then compare the developed method with the established method by Chang et al. [5] by defining a set of error metrics. We found that for simulated data, where a ground truth is available, our method offers a marked improvement over [5]. For a real-world example, where ground truth is not available, our results are similar results to the established method.

Keywords:

Mathematics Subject Classification: Primary: 93-08; Secondary: 46E22.

Citation:

Full Text(HTML)

Figure 1. Algorithm 1, used for estimation of the vector field in (14). The true trajectories are calculated via RK4 over the time frame $ [0, 1] $. Using Gaussian RBFs with a kernel width of 1, we performed $ 10 $ iterations of Algorithm 1

Download: Full-size image PowerPoint slide

Figure 2. Calculated initial trajectories and output of Algorithm 1 for 5, 10, and 20 iterations on Gliderpalooza data

Download: Full-size image PowerPoint slide

Figure 3. A comparison of Algorithm 1 and the method of [5] for estimation of the vector field in (14)

Download: Full-size image PowerPoint slide

Figure 4. To demonstrate universality of the occupation kernel basis, Algorithm 1 is used to estimate three different flow fields. The flow field in (14) (Flow field 1), a linear flow field (Flow field 2), and a constant flow field (Flow field 3). The plot shows the average estimation error, as defined in 16, plotted against the number of iterates of Algorithm 1 for the three flow fields

Download: Full-size image PowerPoint slide

Figure 5. Simultaneous plots for estimation of the unknown flow field in the Gliderpalooza experiment using Algorithm 1 (green arrows) and the method of [5] (blue arrows)

Download: Full-size image PowerPoint slide

Table 1. Algorithm 1, along with the technique in [5] are used to estimate the vector field in (14). The table shows the maximum and the average estimation error, as defined in (15) and (16), respectively, for the two methods

Method in [5] Algorithm 1

Max Error 1.1849 0.25321

Mean Error 0.51549 0.025642

| Show Table

DownLoad: CSV

Table 2. Norm difference statistics between estimates of the Gliderpalooza flow field, approximated using Algorithm 1 and the method in [5]

max norm difference 0.14877

mean norm difference 0.0088628

variance 0.0001869

| Show Table

DownLoad: CSV

Related Papers

Cited by

References

[1]	N. Aronszajn, Theory of reproducing kernels, Trans. Amer. Math. Soc., 68 (1950), 337-404. doi: 10.1090/S0002-9947-1950-0051437-7.
[2]	A. S. Aweiss, B. D. Owens, J. Rios, J. R. Homola and C. P. Mohlenbrink, Unmanned aircraft systems (UAS) traffic management (UTM) national campaign Ⅱ, AIAA Inf. Syst. - AIAA Infotech @ Aerosp., Kissimmee, FL, 2018. doi: 10.2514/6.2018-1727.
[3]	Y. Benyamini and J. Lindenstrauss, Geometric Nonlinear Functional Analysis. Vol. 1, American Mathematical Society Colloquium Publications, 48, American Mathematical Society, Providence, RI, 2000. doi: 10.1090/coll/048.
[4]	R. L. Burden and J. D. Faires, Numerical Analysis, $7^{th}$ edition, Brooks/Cole, 2001.
[5]	D. Chang, W. Wu, C. R. Edwards and F. Zhang, Motion tomography: Mapping flow fields using autonomous underwater vehicles, Internat. J. Robotics Res., 36 (2017), 320-336. doi: 10.1177/0278364917698747.
[6]	R. E. Moore, Computational Functional Analysis, Ellis Horwood Series: Mathematics and its Applications, Ellis Horwood Ltd., Chichester; Halsted Press [John Wiley & Sons, Inc.], New York, 1985.
[7]	V. I. Paulsen and M. Raghupathi, An Introduction to the Theory of Reproducing Kernel Hilbert Spaces, Cambridge Studies in Advanced Mathematics, 152, Cambridge University Press, Cambridge, 2016. doi: 10.1017/CBO9781316219232.
[8]	J. Petrich, C. A. Woolsey and D. J. Stilwell, Planar flow model identification for improved navigation of small AUVs, Ocean Engrg., 36 (2009), 119-131. doi: 10.1016/j.oceaneng.2008.10.002.
[9]	J. A. Rosenfeld, R. Kamalapurkar, L. F. Gruss and T. T. Johnson, Dynamic mode decomposition for continuous time systems with the Liouville operator, preprint, J. Nonlinear Sci., 32 (2022). doi: 10.1007/s00332-021-09746-w.
[10]	J. A. Rosenfeld, B. Russo, R. Kamalapurkar and T. T. Johnson, The occupation kernel method for nonlinear system identification, preprint, arXiv: 1909.11792.
[11]	V. Stepanyan and K. S. Krishnakumar, Estimation, navigation and control of multi-rotor drones in an urban wind field, AIAA Inf. Syst. - AIAA Infotech @ Aerosp., Grapevine, TX, 2017. doi: 10.2514/6.2017-0670.
[12]	W. Walter, Ordinary Differential Equations, Graduate Texts in Mathematics, 182, Readings in Mathematics, Springer-Verlag, New York, 1998. doi: 10.1007/978-1-4612-0601-9.
[13]	H. Wendland, Scattered Data Approximation, Cambridge Monographs on Applied and Computational Mathematics, 17, Cambridge University Press, Cambridge, 2005. doi: 10.1017/CBO9780511617539.
[14]	W. Wu, D. Chang and F. Zhang, Glider CT: Reconstructing flow fields from predicted motion of underwater gliders, Proceedings of the Eighth ACM International Conference on Underwater Networks and Systems, (2013), 1-8. doi: 10.1145/2532378.2532403.