Skip to main content
SpringerLink
Log in

Active Scene Recognition

  • Chapter
  • First Online:
Indoor Scene Recognition by 3-D Object Search

Part of the book series: Springer Tracts in Advanced Robotics ((STAR,volume 135))

  • 429 Accesses

Abstract

Detailed technical presentation of our contributions that are related to Active Scene Recognition. This includes our approaches to Object Pose Prediction and Next-Best-View estimation.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution
eBook
USD 16.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 109.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info
Hardcover Book
USD 109.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    During indirect search, we only look for objects that miss in the available scene models. Objects from scene categories for which no partial scene models have been estimated yet are being looked for during direct search.

  2. 2.

    We used the library SMACH [2] for implementing ASR.

  3. 3.

    We look at Fig. 4.3 in landscape format.

  4. 4.

    Since in relation to learning, the contributions in this thesis are restricted to that of scene category models, such a simplification makes sense.

  5. 5.

    We designate a scene model from a scene category as incomplete when the model does not contain all objects that belong to the category.

  6. 6.

    The pose of the robot is composed of the pose of its mobile base and the orientation of its PTU.

  7. 7.

    Please keep in mind that “NBV estimation” selects both the objects that shall be searched from a viewpoint and the viewpoint itself.

  8. 8.

    Without loss of generality, all variations are supposed to be visible from the same set of robot head poses.

  9. 9.

    We present all operations as performed on the view of the left camera.

  10. 10.

    The confidence in detecting objects from a given viewpoint depends on the absolute number of predictions within the viewing frustum the given viewpoint induces.

  11. 11.

    Please bear in mind that scene model instance and scene category instance are synonyms.

  12. 12.

    This kind of visualization is further detailed out in Sect. 4.6.

  13. 13.

    The subset \([\mathbf {I}_{\{\mathbf {S}\}}^{r}]\) is a multiset [10, p. 29].

  14. 14.

    The color of a cloud changes from red to green with increasing confidence of the represented instance.

  15. 15.

    Instances with low, average and high ratings are displayed as red, yellow and green clouds.

  16. 16.

    This picture displays a dark green frustum of a view at position \(\mathbf {p}\) that contains two predicted poses for a spherical grey object. The same lines of sight, visualized as dark red arrows, are assigned to each of both predictions.

  17. 17.

    For the sake of simplicity, we only consider the left camera on the sensor head of the abstract robot from Sect. 4.4 in this and the following two sections.

  18. 18.

    Throughout the experiments we conducted for this thesis, we set frustum parameters as follows: \(\text {fov}_x = {30}^{\circ }\), \(\text {fov}_y = {20}^{\circ }\), ncp \(= 0.4\) m, fcp \(= 1.5\) m.

  19. 19.

    The dimensionality of the parameter space of our objective function is \(|\{o\}_{P}| + 6\).

  20. 20.

    As earlier in this thesis, we limit the description of our contributions to design decisions that reduce problem complexity, leaving out canonical algorithm optimization strategies such as parallelization.

  21. 21.

    Here, we ignore the specificities of hex grids.

References

  1. Aumann-Cleres, F.: Markerbasiertes Kalibrieren der kinematischen Kette und Aufstellen der Rückwärtstransformation zwischen der Basis und dem Sensorkopf eines mobilen Roboters. Bachelor’s thesis, Advisor: P. Meißner, Reviewer: R. Dillmann, Karlsruhe Institute of Technology (2016)

    Google Scholar 

  2. Bohren, J., Cousins, S.: The smach high-level executive. IEEE Robot. Autom. Mag. (2013)

    Google Scholar 

  3. Bourke, P.: Frustum culling. http://paulbourke.net/miscellaneous/frustum (2000). Accessed 01 Dec 2017

  4. Bronshtein, I., Semendyayev, K., Musiol, G., Muehlig, H.: Handbook of Mathematics, 5th edn. Springer, Berlin (2007)

    Google Scholar 

  5. Devert, A.: Spreading points on a disc and on a sphere—Marmakoide’s Blog. http://blog.marmakoide.org/?p=1 (2012). Accessed 14 Nov 2017

  6. Dillmann, R., Huck, M.: Informationsverarbeitung in der Robotik. Springer, Berlin (1991)

    Google Scholar 

  7. Eidenberger, R., Grundmann, T., Schneider, M., Feiten, W., Fiegert, M., Wichert, G.V., Lawitzky, G.: Scene analysis for service robots. In: Towards Service Robots for Everyday Environments, pp. 181–213. Springer, Berlin (2012)

    Google Scholar 

  8. Gamma, E., Johnson, R., Vlissides, J., Helm, R.: Design Patterns: Elements of Reusable Object-Oriented Software. Addison-Wesley, Boston (1995)

    Google Scholar 

  9. Garvey, T.D.: Perceptual strategies for purposive vision. Tech-Technical Note 117, SRI International (1976)

    Google Scholar 

  10. Hein, J.L.: Discrete Mathematics, 2nd edn. Jones and Bartlett Publishers, Inc, Burlington (2002)

    Google Scholar 

  11. Isard, M., Blake, A.: Condensation–conditional density propagation for visual tracking. Int. J. Comput. Vis. 29(1), 5–28 (1998)

    Article  Google Scholar 

  12. Karrenbauer, O.: Realisierung und komparative Analyse von alternativen Methoden zum uninformierten Generieren optimaler Folgen von Ansichten für die 3D-Objektsuche. Bachelor’s thesis, Advisor: P. Meißner, Reviewer: R. Dillmann, Karlsruhe Institute of Technology (2017)

    Google Scholar 

  13. Kunze, L., Doreswamy, K.K., Hawes, N.: Using qualitative spatial relations for indirect object search. In: Proceedings of the 2014 IEEE International Conference on Robotics and Automation (ICRA), pp. 163–168. IEEE (2014)

    Google Scholar 

  14. Lehmann, A., Leibe, B., Van Gool, L.: Fast prism: branch and bound hough transform for object class detection. Int. J. Comput. Vis. 94(2), 175–197 (2011)

    Article  Google Scholar 

  15. Lorbach, M., Hofer, S., Brock, O.: Prior-assisted propagation of spatial information for object search. In: Proceedings of the 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2014), pp. 2904–2909. IEEE (2014)

    Google Scholar 

  16. Meißner, P., Reckling, R., Wittenbeck, V., Schmidt-Rohr, S., Dillmann, R.: Active scene recognition for programming by demonstration using next-best-view estimates from hierarchical implicit shape models. In: Proceedings of the 2014 IEEE International Conference on Robotics and Automation (ICRA), pp. 5585–5591. IEEE (2014)

    Google Scholar 

  17. Meißner, P., Schleicher, R., Hutmacher, R., Schmidt-Rohr, S., Dillmann, R.: Scene recognition for mobile robots by relational object search using next-best-view estimates from hierarchical implicit shape models. In: Proceedings of the 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 137–144. IEEE (2016)

    Google Scholar 

  18. Patel, A.: Hexagonal grids. https://www.redblobgames.com/grids/hexagons (2013 & 2015). Accessed 11 Nov 2017

  19. Potthast, C., Sukhatme, G.S.: A probabilistic framework for next best view estimation in a cluttered environment. J. Vis. Commun. Image Represent. 25(1), 148–164 (2014)

    Article  Google Scholar 

  20. Quigley, M., Conley, K., Gerkey, B., Faust, J., Foote, T., Leibs, J., Wheeler, R., Ng, A.Y.: Ros: an open-source robot operating system. In: ICRA Workshop on Open Source Software, Kobe, p. 5 (2009)

    Google Scholar 

  21. Russell, S.J., Norvig, P.: Artificial Intelligence: A Modern Approach, 3rd international edn. Prentice Hall Press, Upper Saddle River (2010)

    Google Scholar 

  22. Siciliano, B., Khatib, O.: Springer Handbook of Robotics. Springer Science + Business Media, Berlin (2008)

    Google Scholar 

  23. Thrun, S., Burgard, W., Fox, D.: Probabilistic Robotics (Intelligent Robotics and Autonomous Agents). The MIT Press, Cambridge (2005)

    Google Scholar 

  24. Vasquez-Gomez, J.I., Sucar, L.E., Murrieta-Cid, R.: View planning for 3d object reconstruction with a mobile manipulator robot. In: Proceedings of the 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2014), pp. 4227–4233. IEEE (2014)

    Google Scholar 

  25. Wixson, L.E., Ballard, D.H.: Using intermediate objects to improve the efficiency of visual search. Int. J. Comput. Vis. 12(2–3), 209–230 (1994)

    Article  Google Scholar 

  26. Ye, Y., Tsotsos, J.K.: Sensor planning for 3d object search. Comput. Vis. Image Underst. 73(2), 145–168 (1999)

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. IAR-IPR, Karlsruhe Institute of Technology, Karlsruhe, Germany

    Pascal Meißner

Authors
  1. Pascal Meißner
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Pascal Meißner .

Rights and permissions

Reprints and permissions

Copyright information

© 2020 Springer Nature Switzerland AG

About this chapter

Check for updates. Verify currency and authenticity via CrossMark

Cite this chapter

Meißner, P. (2020). Active Scene Recognition. In: Indoor Scene Recognition by 3-D Object Search. Springer Tracts in Advanced Robotics, vol 135. Springer, Cham. https://doi.org/10.1007/978-3-030-31852-9_4

Download citation

Publish with us

Policies and ethics

Search

Navigation