Abstract
Semidefinite relaxation has recently emerged as a powerful technique for machine perception, in many cases enabling the recovery of certifiably globally optimal solutions of generally-intractable nonconvex estimation problems. However, the high computational cost of standard interior-point methods for semidefinite optimization prevents these algorithms from scaling effectively to the high-dimensional problems frequently encountered in machine perception tasks. To address this challenge, in this paper we present an efficient algorithm for solving the large-scale but low-rank semidefinite relaxations that underpin current certifiably correct machine perception methods. Our algorithm preserves the scalability of current state-of-the-art low-rank semidefinite optimizers that are custom-built for the geometry of specific machine perception problems, but generalizes to a broad class of convex programs over spectrahedral sets without the need for detailed manual analysis or design. The result is an easy-to-use, general-purpose computational tool capable of supporting the development and deployment of a broad class of novel certifiably correct machine perception methods.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
More precisely, it requires knowledge of an efficiently-computable retraction on that manifold [2].
- 2.
In coordinates, \(([x]_{+})_i = \max (x_i, 0)\) for all \(i \in [n]\).
- 3.
Here the LICQ is necessary to ensure that the Lagrange multipliers \((\lambda , \gamma )\) associated with Y are unique [44]. Without this condition, there could conceivably exist some other set of multipliers \((\lambda ', \gamma ') \in \mathbb {R}^{m_1} \times \mathbb {R}_{+}^{m_2}\) (that we do not have in hand) satisfying (8) and also (5), in which case X would be optimal for Problem 1. However, any multipliers satisfying (5) for \(X = YY^\mathsf {T}\) satisfy (8) a fortiori. Therefore, the uniqueness of \((\lambda , \gamma )\) implies that there cannot exist alternative multipliers \((\lambda ', \gamma ')\) satisfying (5), and therefore that X is not optimal (by Theorem 1).
- 4.
- 5.
Note that the argument used to prove Theorem 4 does not directly apply to (19) and (20) because the inequality term \(\Vert \left[ \mathcal {B}(X) - u\right] _{+} \Vert ^2\) renders \(\varphi \) only \(C^1\). In the supplementary material, we show how to derive a \(C^\infty \) reformulation of (19) and (20).
- 6.
- 7.
Available at https://github.com/david-m-rosen/LowRankSDP.
- 8.
Version 3.13.1, available at https://github.com/coin-or/Ipopt.
- 9.
Available at https://spectralib.org/.
- 10.
Version 6.2, available at https://www.pardiso-project.org/.
- 11.
Available at https://github.com/david-m-rosen/SE-Sync.
- 12.
This termination criterion only affects Alg1-T, since SE-Sync-S and SE-Sync-T employ Riemannian optimization methods that automatically enforce feasibility.
- 13.
References
Åström, K.J., Murray, R.M.: Feedback Systems: An Introduction for Scientists and Engineers. Princeton University Press, Princeton, NJ, USA (2008)
Absil, P.-A., Mahony, R., Sepulchre, R.: Optimization Algorithms on Matrix Manifolds. Princeton University Press, Princeton, NJ (2008)
Bandeira, A.S.: A note on probably certifiably correct algorithms. Comptes Rendus Math. 354(3), 329–333 (2016)
Boumal, N., Voroninski, V., Bandeira, A.S.: The non-convex Burer-Monteiro approach works on smooth semidefinite programs. In: Advances in Neural Information Processing Systems (NIPS), pp. 2757–2765, December 2016
Boyd, S., Vandenberghe, L.: Convex Optimization. Cambridge University Press, Cambridge (2004)
Briales, J., Gonzalez-Jimenez, J.: Cartan-Sync: fast and global SE(d)-synchronization. Robot. Autom. Lett. 2(4), 2127–2134 (2017)
Briales, J., Kneip, L., Gonzalez-Jimenez, J.: A certifiably globally optimal solution to the non-minimal relative pose problem. In: IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), pp. 145–154 (2018)
Burer, S., Monteiro, R.D.C.: A nonlinear programming algorithm for solving semidefinite programs via low-rank factorization. Math. Program. 95, 329–357 (2003)
Burer, S., Monteiro, R.D.C.: Local minima and convergence in low-rank semidefinite programming. Math. Program. 103, 427–444 (2005)
Cadena, C., Carlone, L., Carrillo, H., Latif, Y., Scaramuzza, D., Neira, J., Reid, I., Leonard, J.J.: Past, present, and future of simultaneous localization and mapping: toward the robust-perception age. IEEE Trans. Robot. 32(6), 1309–1332 (2016)
Carlone, L., Rosen, D.M., Calafiore, G., Leonard, J.J., Dellaert, F.: Lagrangian duality in 3D SLAM: verification techniques and optimal solutions. In: IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Hamburg, Germany, September 2015
Carlone, L., Tron, R., Daniilidis, K., Dellaert, F.: Initialization techniques for 3D SLAM: a survey on rotation estimation and its use in pose graph optimization. In: IEEE International Conference on Robotics and Automation (ICRA), Seattle, WA, pp. 4597–4605, May 2015
Cifuentes, D.: On the Burer-Monteiro method for general semidefinite programs. arXiv preprint arXiv:1904.07147 (2019)
Cifuentes, D., Moitra, A.: Polynomial time guarantees for the Burer-Monteiro method. arXiv preprint arXiv:1912.01745 (2019)
Conn, A.R., Gould, N.I.M., Toint, Ph.L.: Trust-Region methods. The Society for Industrial and Applied Mathematics (SIAM) (2000)
Eriksson, A., Olsson, C., Kahl, F., Chin, T.-J.: Rotation averaging and strong duality. In: IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, UT, USA, pp. 127–135, June 2018
Fallon, M., Kuindersma, S., Karumanchi, S., Antone, M., Schneider, T., Dai, H., D’Arpino, C.P., Deits, R., DiCicco, M., Fourie, D., Koolen, T., Marion, P., Posa, M., Valenzuela, A., Yu, K.-T., Shah, J., Iagnemma, K., Tedrake, R., Teller, S.: An architecture for online affordance-based perception and whole-body planning. Intl. J. Robot. Res. 32(2), 229–254 (2015)
Fan, T., Wang, H., Rubenstein, M., Murphey, T.: Efficient and guaranteed planar pose graph optimization using the complex number representation. In: IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Macau, China, pp. 1904–1911, November 2019
Gallier, J.: The Schur complement and symmetric positive semidefinite (and definite) matrices. unpublished note, December 2010. http://www.cis.upenn.edu/~jean/schur-comp.pdf
Grisetti, G., Kümmerle, R., Stachniss, C., Burgard, W.: A tutorial on graph-based SLAM. IEEE Intell. Transp. Syst. Mag. 2(4), 31–43 (2010)
Hu, S., Carlone, L.: Accelerated inference in Markov random fields via smooth Riemannian optimization. Robot. Autom. Lett. 4(2), 1295–1302 (2019)
Johnson, M., Shrewsbury, B., Bertrand, S., Wu, T., Duran, D., Floyd, M., Abeles, P., Stephen, D., Mertins, N., Lesman, A., Carff, J., Rifenburgh, W., Kaveti, P., Straatman, W., Smith, J., Griffioen, M., Layton, B., de Boer, T., Koolen, T., Neuhaus, P., Pratt, J.: Team IHMC’s lessons learned from the DARPA robotics challenge trials. Intl. J. Robot. Res. 32(2), 192–208 (2015)
Journée, M., Bach, F., Absil, P.-A., Sepulchre, R.: Low-rank optimization on the cone of positive semidefinite matrices. SIAM J. Optim. 20(5), 2327–2351 (2010)
Kaess, M., Johannsson, H., Roberts, R., Ila, V., Leonard, J.J., Dellaert, F.: iSAM2: Incremental smoothing and mapping using the Bayes tree. Intl. J. Robot. Res. 31(2), 216–235 (2012)
Kümmerle, R., Grisetti, G., Strasdat, H., Konolige, K., Burgard, W.: g2o: A general framework for graph optimization. In: IEEE International Conference on Robotics and Automation (ICRA), Shanghai, China, pp. 3607–3613, May 2011
Leonard, J., How, J., Teller, S., Berger, M., Campbell, S., Fiore, G., Fletcher, L., Frazzoli, E., Huang, A., Karaman, S., Koch, O., Kuwata, Y., Moore, D., Olson, E., Peters, S., Teo, J., Truax, R., Walter, M., Barrett, D., Epstein, A., Maheloni, K., Moyer, K., Jones, T., Buckley, R., Antone, M., Galejs, R., Krishnamurthy, S., Williams, J.: A perception-driven autonomous urban vehicle. Intl. J. Robot. Res. 25(10), 727–774 (2008)
Majumdar, A., Hall, G., Ahmadi, A.A.: Recent scalability improvements for semidefinite programming with applications in machine learning, control and robotics. Annu. Rev. Control Robot. Auton. Syst. 3, 331–360 (2020)
Mandel, J.: On block diagonal and Schur complement preconditioning. Numerische Mathematik 58(1), 79–93 (1990)
Mangelson, J.G., Liu, J., Eustice, R.M., Vasudevan, R.: Guaranteed globally optimal planar pose graph and landmark SLAM via sparse-bounded sums-of-squares programming. In: IEEE International Conference on Robotics and Automation (ICRA), Montreal, Canada, pp. 9306–9312, May 2019
Nocedal, J., Wright, S.J.: Numerical Optimization, 2nd edn. Springer Science+Business Media, New York (2006)
Rockafellar, J.T., Wets, R.J.-B.: Variational Analysis. Springer Science+Business Media, (2009)
Rosen, D.M.: Towards provably robust machine perception. Presented at Robotics: Science and Systems (RSS) in the workshop “RSS Pioneers", June 2019
Rosen, D.M., Carlone, L.: Computational enhancements for certifiably correct SLAM. In: Presented at the International Conference on Intelligent Robots and Systems (IROS) in the workshop “Introspective Methods for Reliable Autonomy", September 2017
Rosen, D.M., Kaess, M., Leonard, J.J.: RISE: an incremental trust-region method for robust online sparse least-squares estimation. IEEE Trans. Robot. 30(5), 1091–1108 (2014)
Rosen, D.M., Carlone, L., Bandeira, A.S., Leonard, J.J.: A certifiably correct algorithm for synchronization over the special Euclidean group. In: International Workshop on the Algorithmic Foundations of Robotics (WAFR), San Francisco, CA (2016)
Rosen, D.M., Carlone, L., Bandeira, A.S., Leonard, J.J.: SE-Sync: a certifiably correct algorithm for synchronization over the special Euclidean group. Intl. J. Robot. Res. 38(2–3), 95–125 (2019)
Russel, S., Norvig, P.: Artifical Intelligence: A Modern Approach, 3rd edn. Prentice Hall, Upper Saddle River, NJ, USA (2010)
Stengel, R.F.: Optimal Control and Estimation. Dover Publications (1994)
Taylor, R.H.: A perspective on medical robotics. Proc. IEEE 94(9), 1652–1664 (2006)
Todd, M.J.: Semidefinite optimization. Acta Numerica 10, 515–560 (2001)
Triggs, B., McLauchlan, P., Hartley, R., Fitzgibbon, A.: Bundle adjustment – a modern synthesis. In: Triggs, W., Zisserman, A., Szeliski, R., (eds.), Vision Algorithms: Theory and Practice volume 1883 of Lecture Notes in Computer Science, pp. 298–372. Springer Verlag (2000)
Vandenberghe, L., Boyd, S.: Semidefinite programming. SIAM Rev. 38(1), 49–95 (1996)
Vlasic, B., Boudette, N.E.: As U.S. investigates fatal Tesla crash, company defends Autopilot system. The New York Times, 12 July 2016
Wachsmuth, W.: On LICQ and the uniqueness of Lagrange multipliers. Oper. Res. Lett. 41, 78–80 (2013)
Wächter, A., Biegler, L.T.: On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Math. Program. 106, 25–57 (2006)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
1 Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
Copyright information
© 2021 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Rosen, D.M. (2021). Scalable Low-Rank Semidefinite Programming for Certifiably Correct Machine Perception. In: LaValle, S.M., Lin, M., Ojala, T., Shell, D., Yu, J. (eds) Algorithmic Foundations of Robotics XIV. WAFR 2020. Springer Proceedings in Advanced Robotics, vol 17. Springer, Cham. https://doi.org/10.1007/978-3-030-66723-8_33
Download citation
DOI: https://doi.org/10.1007/978-3-030-66723-8_33
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-66722-1
Online ISBN: 978-3-030-66723-8
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)