Learning efficient haptic shape exploration with a rigid tactile sensor array


Autoři: Sascha Fleer aff001;  Alexandra Moringen aff001;  Roberta L. Klatzky aff002;  Helge Ritter aff001
Působiště autorů: Neuroinformatics Group, Bielefeld University, Bielefeld, Germany aff001;  Department of Psychology, Carnegie Mellon University, Pittsburgh, Pennsylvania, United States of America aff002
Vyšlo v časopise: PLoS ONE 15(1)
Kategorie: Research Article
doi: 10.1371/journal.pone.0226880

Souhrn

Haptic exploration is a key skill for both robots and humans to discriminate and handle unknown objects or to recognize familiar objects. Its active nature is evident in humans who from early on reliably acquire sophisticated sensory-motor capabilities for active exploratory touch and directed manual exploration that associates surfaces and object properties with their spatial locations. This is in stark contrast to robotics. In this field, the relative lack of good real-world interaction models—along with very restricted sensors and a scarcity of suitable training data to leverage machine learning methods—has so far rendered haptic exploration a largely underdeveloped skill. In robot vision however, deep learning approaches and an abundance of available training data have triggered huge advances. In the present work, we connect recent advances in recurrent models of visual attention with previous insights about the organisation of human haptic search behavior, exploratory procedures and haptic glances for a novel architecture that learns a generative model of haptic exploration in a simulated three-dimensional environment. This environment contains a set of rigid static objects representing a selection of one-dimensional local shape features embedded in a 3D space: an edge, a flat and a convex surface. The proposed algorithm simultaneously optimizes main perception-action loop components: feature extraction, integration of features over time, and the control strategy, while continuously acquiring data online. Inspired by the Recurrent Attention Model, we formalize the target task of haptic object identification in a reinforcement learning framework and reward the learner in the case of success only. We perform a multi-module neural network training, including a feature extractor and a recurrent neural network module aiding pose control for storing and combining sequential sensory data. The resulting haptic meta-controller for the rigid 16 × 16 tactile sensor array moving in a physics-driven simulation environment, called the Haptic Attention Model, performs a sequence of haptic glances, and outputs corresponding force measurements. The resulting method has been successfully tested with four different objects. It achieved results close to 100% while performing object contour exploration that has been optimized for its own sensor morphology.

Klíčová slova:

Employment – Learning – Machine learning – Machine learning algorithms – Recurrent neural networks – Robots – Tactile sensation – Touch


Zdroje

1. Szegedy C, Liu W, Jia Y, Sermanet P, Reed SE, Anguelov D, et al. Going Deeper with Convolutions. CoRR. 2014;abs/1409.4842.

2. He K, Zhang X, Ren S, Sun J. Deep Residual Learning for Image Recognition. CoRR. 2015;abs/1512.0.

3. Levine S, Pastor P, Pastor P, Krizhevsky A, Ibarz J, Ibarz J, et al. Learning hand-eye coordination for robotic grasping with deep learning and large-scale data collection. The International Journal of Robotics Research. 2017;37(4-5):421–436. doi: 10.1177/0278364917710318

4. Okamura A, Cutkosky M. Feature Detection for Haptic Exploration with Robotic Fingers. vol. 20; 2001.

5. Martins R, Ferreira JF, Dias J. Touch attention Bayesian models for robotic active haptic exploration of heterogeneous surfaces. CoRR. 2014;abs/1409.6.

6. Tian S, Ebert F, Jayaraman D, Mudigonda M, Finn C, Calandra R, et al. Manipulation by Feel: Touch-Based Control with Deep Predictive Models. arxiv. 2019;.

7. Lee MA, Zhu Y, Srinivasan K, Shah P, Savarese S, Fei-Fei L, et al. Making Sense of Vision and Touch: Self-Supervised Learning of Multimodal Representations for Contact-Rich Tasks. arxiv. 2019;.

8. Kalagher H, Jones SS. Young children’s haptic exploratory procedures. Journal of Experimental Child Psychology. 2011;110(4):592–602. https://doi.org/10.1016/j.jecp.2011.06.007 21783203

9. Klatzky RL, Lederman SJ, Mankinen JM. Visual and haptic exploratory procedures in children’s judgments about tool function. The Development of Haptic Perception. 2005;28(3):240–249.

10. Klatzky RL, Lederman SJ, Reed CL. There’s more to touch than meets the eye: the salience of object attributes for hpatics with and without vision. Journal of Experimental Psychology. 1987;.

11. Klatzky RL, Lederman SJ. Identifying objects from a haptic glance. Perception & Psychophysics. 1995;57(8):1111–1123. doi: 10.3758/BF03208368

12. Fishel JA, Loeb GE. Bayesian Exploration for Intelligent Identification of Textures. Frontiers in Neurorobotics. 2012;6. doi: 10.3389/fnbot.2012.00004

13. Chu V, McMahon I, Riano L, McDonald CG, He Q, Perez-Tejada JM, et al. Using robotic exploratory procedures to learn the meaning of haptic adjectives. In: 2013 IEEE International Conference on Robotics and Automation (ICRA). IEEE; 2013. p. 3048–3055.

14. Chu V, McMahon I, Riano L, McDonald CG, He Q, Perez-Tejada JM, et al. Robotic learning of haptic adjectives through physical interaction. Robotics and Autonomous Systems. 2015;63:279–292. https://doi.org/10.1016/j.robot.2014.09.021

15. Pape L, Oddo CM, Controzzi M, Cipriani C, Förster A, Carrozza MC, et al. Learning tactile skills through curious exploration. Frontiers in Neurorobotics. 2012;6. doi: 10.3389/fnbot.2012.00006 22837748

16. Sutton RS, Barto AG. Reinforcement learning: An introduction. second edtion ed. MIT Press; 2018.

17. van Hoof H, Hermans T, Neumann G, Peters J. Learning robot in-hand manipulation with tactile features. In: 2015 IEEE-RAS 15th International Conference on Humanoid Robots (Humanoids); 2015. p. 121–127.

18. Rajeswaran A, Kumar V, Gupta A, Schulman J, Todorov E, Levine S. Learning Complex Dexterous Manipulation with Deep Reinforcement Learning and Demonstrations. CoRR. 2017;abs/1709.10087.

19. Mnih V, Heess N, Graves A, Kavukcuoglu K. Recurrent Models of Visual Attention. CoRR. 2014;abs/1406.6247.

20. Ba J, Mnih V, Kavukcuoglu K. Multiple Object Recognition with Visual Attention. CoRR. 2014;abs/1412.7755.

21. Hayhoe M, Ballard D. Eye movements in natural behavior. Trends in Cognitive Sciences. 2005;9(4):188–194. doi: 10.1016/j.tics.2005.02.009 15808501

22. Mathe S, Sminchisescu C. Action from still image dataset and inverse optimal control to learn task specific visual scanpaths. In: Advances in neural information processing systems; 2013. p. 1923–1931.

23. Itti L, Koch C, Niebur E. A model of saliency-based visual attention for rapid scene analysis. IEEE Transactions on pattern analysis and machine intelligence. 1998;20(11):1254–1259. doi: 10.1109/34.730558

24. Itti L, Koch C. Computational modelling of visual attention. Nature Reviews Neuroscience. 2001;2(3):194–203. doi: 10.1038/35058500 11256080

25. Schurmann C, Koiva R, Haschke R, Ritter H. A modular high-speed tactile sensor for human manipulation research. In: 2011 IEEE World Haptics Conference (WHC 2011). IEEE; 2011. p. 339–344.

26. Moringen A, Krieger K, Haschke R, Ritter H. Haptic Search for Complex 3D Shapes Subject to Geometric Transformations or Partial Occlusion. In: IEEE World Haptics; 2017.

27. Krieger K, Moringen A, Haschke R, Ritter H. Shape Features of the Search Target Modulate Hand Velocity, Posture and Pressure during Haptic Search in a 3D Display. In: Lecture Notes in Computer Science. Springer; 2016.

28. Moringen A, Haschke R, Ritter H. Search Procedures during Haptic Search in an Unstructured 3D Display. In: IEEE Haptics Symposium; 2016.

29. Williams RJ. Toward a theory of reinforcement-learning connectionist systems. Technical Report NU-CCS-88-3, Northeastern University. 1988;.

30. Williams RJ. Simple statistical gradient-following algorithms for connectionist reinforcement learning. Machine Learning. 1992;8(3-4):229–256. doi: 10.1007/BF00992696

31. Hochreiter S, Schmidhuber J. Long short-term memory. MIT Press. 1997;9(8):1735–1780.

32. Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks. In: Advances in neural information processing systems; 2012. p. 1097–1105.

33. Goodfellow I, Bengio Y, Courville A. Deep learning. MIT press; 2016.

34. Larochelle H, Hinton GE. Learning to combine foveal glimpses with a third-order Boltzmann machine. In: Lafferty JD, Williams CKI, Shawe-Taylor J, Zemel RS, Culotta A, editors. Advances in Neural Information Processing Systems 23. Curran Associates, Inc.; 2010. p. 1243–1251.

35. Dugas C, Bengio Y, Bélisle F, Nadeau C, Garcia R. Incorporating second-order functional knowledge for better option pricing. In: Advances in neural information processing systems; 2001. p. 472–478.

36. Nesterov Y. A method for solving the convex programming problem with convergence rate O(1/k2). In: Dokl. Akad. Nauk SSSR; 1983. p. 543–547.

37. Nesterov Y. Introductory Lectures on Convex Optimization: A Basic Course. Applied Optimization. Springer US; 2013.

38. Bergstra J, Bengio Y. Random search for hyper-parameter optimization. Journal of Machine Learning Research. 2012;13(Feb):281–305.

39. He K, Zhang X, Ren S, Sun J. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. In: Proceedings of the IEEE international conference on computer vision; 2015. p. 1026–1034.

40. Graves A, Wayne G, Danihelka I. Neural Turing Machines. CoRR. 2014;abs/1410.5401.


Článek vyšel v časopise

PLOS One


2020 Číslo 1