Abstract
We discuss the problem in which an autonomous vehicle must classify an object based on multiple views. We focus on the active classification setting, where the vehicle controls which views to select to best perform the classification. The problem is formulated as an extension to Bayesian active learning, and we show connections to recent theoretical guarantees in this area. We formally analyze the benefit of acting adaptively as new information becomes available. The analysis leads to a probabilistic algorithm for determining the best views to observe based on information theoretic costs. We validate our approach in two ways, both related to underwater inspection: 3D polyhedra recognition in synthetic depth maps and ship hull inspection with imaging sonar. These tasks encompass both the planning and recognition aspects of the active classification problem. The results demonstrate that actively planning for informative views can reduce the number of necessary views by up to 80 % when compared to passive methods.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
There are a number of additional active vision works relevant to the present paper. We direct the interested reader to Roy et al. [17] for a survey.
- 2.
We formulate the problem for the case of discrete locations. If continuous locations are available, an interpolation function can be used to estimate the informativeness of a location based on the discrete training data (see Sect. 6).
- 3.
Note that the related problem of minimizing expected loss subject to a hard constraint on budget is also relevant. While similar examples show that there is a benefit to acting adaptively in this case, we defer detailed analysis to future work.
References
Y. Aloimonos, I. Weiss, A. Bandopadhyay, Active vision. Int. J. Comput. Vision 1(4), 333–356 (1987)
T. Arbel, F.P. Ferrie, Entropy-based gaze planning. Image Vis. Comput. 19(11), 779–786 (2001)
H. Bay, A. Ess, T. Tuytelaars, L. Gool, SURF: speeded up robust features. Comput. Vis. Image Underst. 110(3), 346–359 (2008)
G. Bradski, A. Kaehler, Learning OpenCV: Computer Vision with the OpenCV Library (O’Reilly Media, 2008)
A. Cameron, H. Durrant-Whyte, A Bayesian approach to optimal sensor placement. Int. J. Robot. Res. 9(5), 70–88 (1990)
C. Connolly, The determination of next best views, in Proceedings of the IEEE Conference on Robotics and Automation (1985)
B. Dean, M. Goemans, J. Vondrak, Approximating the stochastic knapsack: the benefit of adaptivity. Math. Oper. Res. 33(4), 945–964 (2008)
J. Denzler, C. Brown, Information theoretic sensor data selection for active object recognition and state estimation. IEEE Trans. Pattern Anal. Mach. Intell. 24(2), 145–157 (2002)
G. Dobeck, J. Cobb, Fusion of multiple quadratic penalty function support vector machines (QPFSVM) for automated sea mine detection and classification, in Proceedings of the SPIE (2002)
B. Englot, F. Hover, Planning complex inspection tasks using redundant roadmaps, in Proceedings of the International Symposium on Robotics Research (2011)
D. Golovin, A. Krause, Adaptive submodularity: a new approach to active learning with stochastic optimization, in Proceedings of the Conference on Learning Theory (2010)
D. Golovin, D. Ray, A. Krause, Near-optimal Bayesian active learning with noisy observations, in Proceedings of the Neural Information Processing Systems (2010)
G. Hollinger, S. Singh, J. Djugash, A. Kehagias, Efficient multi-robot search for a moving target. Int. J. Robot. Res. 28(2), 201–219 (2009)
A. Krause, C. Guestrin, Near-optimal nonmyopic value of information in graphical models, in Proceedings of the Uncertainty in Artificial Intelligence (2005)
V. Myers, D. Williams, A POMDP for multi-view target classification with an autonomous underwater vehicle, in Proceedings of the IEEE OCEANS Conference, Sept 2010
M. Naghshvar, T. Javidi, Active M-ary sequential hypothesis testing, in Proceedings of the IEEE International Symposium on Information Theory (2010)
S.D. Roy, S. Chaudhury, S. Banarjee, Active recognition through next view planning: a survey. J. Pattern Recognit. 37(3), 429–446 (2004)
A. Singh, A. Krause, C. Guestrin, W. Kaiser, Efficient informative sensing using multiple robots. J. Artif. Intell. Res. 34, 707–755 (2009)
M. Sipe, D. Casasent, Feature space trajectory methods for active computer vision. IEEE Trans. Pattern Anal. Mach. Learn. 24(12),1634–1643 (2002)
D. Steinberg, S. Williams, O. Pizarro, M. Jakuba, Towards autonomous habitat classification using Gaussian mixture models, in Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (2010)
S. Thrun, W. Burgard, D. Fox, Probabilistic Robotics (MIT Press, Cambridge, 2005)
A. Wald, Sequential tests of statistical hypotheses. Ann. Math. Stat. 16(2), 117–186 (1945)
D. Williams, Bayesian data fusion of multiview synthetic aperture sonar imagery for seabed classification. IEEE Trans. Image Process. 18(6), 1239–1254 (2009)
D. Williams, On optimal AUV track-spacing for underwater mine detection, in Proceedings of the IEEE International Conference on Robotics and Automation (2010)
X. Zhou, D. Comaniciu, A. Krishnan, Conditional feature sensitivity: a unifying view on active recognition and feature selection, in Proceedings of the International Conference on Computer Vision (2003)
Acknowledgements
The authors gratefully acknowledge Franz Hover and Brendan Englot at MIT for imaging sonar data and technical support while processing the data. This research has been funded in part by the following grants: ONR N00014-09-1-0700, ONR N00014-07-1-00738, NSF 0831728, NSF CCR-0120778, and NSF CNS-1035866.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer International Publishing Switzerland
About this chapter
Cite this chapter
Hollinger, G.A., Mitra, U., Sukhatme, G.S. (2017). Active Classification: Theory and Application to Underwater Inspection. In: Christensen, H., Khatib, O. (eds) Robotics Research . Springer Tracts in Advanced Robotics, vol 100. Springer, Cham. https://doi.org/10.1007/978-3-319-29363-9_6
Download citation
DOI: https://doi.org/10.1007/978-3-319-29363-9_6
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-29362-2
Online ISBN: 978-3-319-29363-9
eBook Packages: EngineeringEngineering (R0)