Skip to main content
Springer Nature Link
Log in

Keyframe-based Learning from Demonstration

Method and Evaluation

  • Published:
International Journal of Social Robotics Aims and scope Submit manuscript

Abstract

We present a framework for learning skills from novel types of demonstrations that have been shown to be desirable from a Human–Robot Interaction perspective. Our approach—Keyframe-based Learning from Demonstration (KLfD)—takes demonstrations that consist of keyframes; a sparse set of points in the state space that produces the intended skill when visited in sequence. The conventional type of trajectory demonstrations or a hybrid of the two are also handled by KLfD through a conversion to keyframes. Our method produces a skill model that consists of an ordered set of keyframe clusters, which we call Sequential Pose Distributions (SPD). The skill is reproduced by splining between clusters. We present results from two domains: mouse gestures in 2D and scooping, pouring and placing skills on a humanoid robot. KLfD has performance similar to existing LfD techniques when applied to conventional trajectory demonstrations. Additionally, we demonstrate that KLfD may be preferable when demonstration type is suited for the skill.

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

Access this article

Log in via an institution

Subscribe and save

Springer+ Basic
$34.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or eBook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15

Similar content being viewed by others

Notes

  1. Using velocity and acceleration data along with position data helps to keep some of the dynamics of the demonstration.

  2. Not all LfD techniques have this problem, e.g. [17].

  3. C k continuity for a function means that the function’s 1…k derivatives exist and are all continuous.

  4. See e.g. http://www.youtube.com/watch?v=5lVxOKSeYsk.

References

  1. Abbeel P, Ng AY (2004) Apprenticeship learning via inverse reinforcement learning. In: Proceedings of the twenty-first international conference on machine learning. ACM Press, New York

    Google Scholar 

  2. Akgun B, Cakmak M, Wook Yoo J, Thomaz LA (2011) Augmenting kinesthetic teaching with keyframes. In: ICML workshop on new developments in imitation learning

    Google Scholar 

  3. Akgun B, Cakmak M, Wook Yoo J, Thomaz LA (2012) Trajectories and keyframes for kinesthetic teaching: a human–robot interaction perspective. In: CM/IEEE intl conference on human–robot interaction (HRI)

    Google Scholar 

  4. Amor HB, Berger E, Vogt D, Jun B (2009) Kinesthetic bootstrapping: teaching motor skills to humanoid robots through physical interaction. In: Lecture notes in computer science. Advances in artificial intelligence, vol 58. Springer, Berlin, pp 492–499

    Google Scholar 

  5. Argall B, Chernova S, Browning B, Veloso M (2009) A survey of robot learning from demonstration. Robot Auton Syst 57(5):469–483

    Article  Google Scholar 

  6. Billard A, Calinon S, Guenter F (2006) Discriminative and adaptive imitation in uni-manual and bi-manual tasks. Robot Auton Syst 54(5):370–384

    Article  Google Scholar 

  7. Bitzer S, Howard M, Vijayakumar S (2010) Using dimensionality reduction to exploit constraints in reinforcement learning. In: IEEE/RSJ international conference on intelligent robots and systems (IROS), pp 3219–3225

    Google Scholar 

  8. Calinon S, Billard A (2007) What is the teacher’s role in robot programming by demonstration? Toward benchmarks for improved learning. Interaction Studies 8(3):441–464

    Google Scholar 

  9. Calinon S, Billard A (2009) Statistical learning by imitation of competing constraints in joint space and task space. Adv Robot 23(15):2059–2076

    Article  Google Scholar 

  10. Calinon S, Guenter F, Billard A (2007) On learning, representing and generalizing a task in a humanoid robot. IEEE Trans Syst Man Cybern, Part B, Cybern 37(2):286–298. Special issue on robot learning by observation, demonstration and imitation

    Article  Google Scholar 

  11. Eddy S (1998) Profile hidden Markov models. Bioinformatics 14(9):755–763

    Article  Google Scholar 

  12. Flash T, Hogan N (1985) The coordination of arm movements: an experimentally confirmed mathematical model. J Neurosci 5(7):1688–1703

    Google Scholar 

  13. Gribovskaya E, Billard A (2009) Learning nonlinear multi-variate motion dynamics for real-time position and orientation control of robotic manipulators. In: Proceedings of IEEE-RAS international conference on humanoid robots

    Google Scholar 

  14. Halit C, Capin T (2011) Multiscale motion saliency for keyframe extraction from motion capture sequences. Comput Animat Virtual Worlds 22(1):3–14

    Article  Google Scholar 

  15. Hersch M, Guenter F, Calinon S, Billard A (2008) Dynamical system modulation for robot learning via kinesthetic demonstrations. IEEE Trans Robot 24(6):1463–1467

    Article  Google Scholar 

  16. Hsiao K (2006) Imitation learning of whole-body grasps. In: IEEE/RJS international conference on intelligent robots and systems (IROS), pp 5657–5662

    Google Scholar 

  17. Khansari-Zadeh SM, Billard A (2011) Learning stable non-linear dynamical systems with Gaussian mixture models. IEEE Trans Robot 27(5):943–957

    Article  Google Scholar 

  18. Lipman D, Carrillo H (1988) The multiple sequence alignment problem in biology. SIAM J Appl Math 48:1073–1082

    Article  MathSciNet  MATH  Google Scholar 

  19. Liu Y, Zhou F, Liu W, De la Torre F, Liu Y (2010) Unsupervised summarization of rushes videos. In: Proceedings of the international conference on multimedia, MM ’10. ACM Press, New York, pp 751–754

    Chapter  Google Scholar 

  20. Lowe D (1987) Three-dimensional object recognition from single two-dimensional images. Artif Intell 31(3):355–395

    Article  Google Scholar 

  21. Miyamoto H, Schaal S, Gandolfo F, Gomi H, Koike Y, Osu R, Nakano E, Wada Y, Kawato M (1996) A Kendama learning robot based on bi-directional theory. Neural Netw 9:1281–1302

    Article  Google Scholar 

  22. Nair N, Sreenivas T (2007) Joint decoding of multiple speech patterns for robust speech recognition. In: IEEE workshop on automatic speech recognition understanding, ASRU, pp 93–98. doi:10.1109/ASRU.2007.4430090

    Chapter  Google Scholar 

  23. Parent R (2002) Computer animation: algorithms and techniques. Morgan Kaufmann series in computer graphics and geometric modeling. Morgan Kaufmann, San Mateo

    Google Scholar 

  24. Pastor P, Hoffmann H, Asfour T, Schaal S (2009) Learning and generalization of motor skills by learning from demonstration. In: IEEE intl conference on robotics and automation

    Google Scholar 

  25. Ratliff N, Ziebart B, Peterson K, Bagnell JA, Hebert M, Dey AK, Srinivasa S (2009) Inverse optimal heuristic control for imitation learning. In: Proc AISTATS, pp 424–431

    Google Scholar 

  26. Todorov E, Jordan M (1998) Smoothness maximization along a predefined path accurately predicts the speed profiles of complex arm movements. J Neurophysiol 80(2):696–714

    Google Scholar 

  27. Wada Y, Kawato M (1993) A neural network model for arm trajectory formation using forward inverse dynamics models. Neural Netw 6:919–932

    Article  Google Scholar 

  28. Weiss A, Igelsboeck J, Calinon S, Billard A, Tscheligi M (2009) Teaching a humanoid: a user study on learning by demonstration with hoap-3. In: Proceedings of the IEEE symposium on robot and human interactive communication (RO-MAN), pp 147–152

    Chapter  Google Scholar 

Download references

Acknowledgements

This research is supported by NSF CAREER grant IIS-1032254.

Author information

Authors and Affiliations

  1. School of Interactive Computing, Georgia Institute of Technology, Atlanta, GA, 30332, USA

    Baris Akgun, Maya Cakmak, Karl Jiang & Andrea L. Thomaz

Authors
  1. Baris Akgun
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Maya Cakmak
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Karl Jiang
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Andrea L. Thomaz
    View author publications

    You can also search for this author inPubMed Google Scholar

Corresponding author

Correspondence to Baris Akgun.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Akgun, B., Cakmak, M., Jiang, K. et al. Keyframe-based Learning from Demonstration. Int J of Soc Robotics 4, 343–355 (2012). https://doi.org/10.1007/s12369-012-0160-0

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s12369-012-0160-0

Keywords