Reaching development through visuo-proprioceptive-tactile integration on a humanoid robot - a deep learning approach

The development of reaching in infants has been studied for nearly nine decades. Originally, it was thought that early reaching is visually guided, but more recent evidence is suggestive of “visually elicited” reaching, i.e. infant is gazing at the object rather than its hand during the reaching movement. The importance of haptic feedback has also been emphasized. Inspired by these findings, in this work we use the simulated iCub humanoid robot to construct a model of reaching development. The robot is presented with different objects, gazes at them, and performs motor babbling with one of its arms. Successful contacts with the object are detected through tactile sensors on hand and forearm. Such events serve as the training set, constituted by images from the robot's two eyes, head joints, tactile activation, and arm joints. A deep neural network is trained with images and head joints as inputs and arm configuration and touch as output. After learning, the network can successfully infer arm configurations that would result in a successful reach, together with prediction of tactile activation (i.e. which body part would make contact). Our main contribution is twofold: (i) our pipeline is end-to-end from stereo images and head joints (6 DoF) to armtorso configurations (10 DoF) and tactile activations, without any preprocessing, explicit coordinate transformations etc.; (ii) unique to this approach, reaches with multiple effectors corresponding to different regions of the sensitive skin are possible.

[1]  J. Colombo.  The development of visual attention in infancy. , 2001, Annual review of psychology.

[2]  Gordon Cheng,et al.  Yielding Self-Perception in Robots Through Sensorimotor Contingencies , 2017, IEEE Transactions on Cognitive and Developmental Systems.

[3]  G. Mettaa,et al.  A developmental approach to visually-guided reaching in artificial systems , 1999 .

[4]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[5]  D. Corbetta,et al.  Assessing the Impact of Movement Consequences on the Development of Early Reaching in Infancy , 2016, Front. Psychol..

[6]  C. Hofsten.  Eye–hand coordination in the newborn. , 1982 .

[7]  Andrew J. Davison,et al.  Transferring End-to-End Visuomotor Control from Simulation to Real World for a Multi-Stage Task , 2017, CoRL.

[8]  Giulio Sandini,et al.  The iCub humanoid robot: An open-systems platform for research in cognitive development , 2010, Neural Networks.

[9]  Daniela Corbetta,et al.  Mapping the feel of the arm with the sight of the object: on the embodied origins of infant reaching , 2014, Front. Psychol..

[10]  Jochen J. Steil,et al.  Goal Babbling Permits Direct Learning of Inverse Kinematics , 2010, IEEE Transactions on Autonomous Mental Development.

[11]  Giorgio Metta,et al.  Transferring Visuomotor Learning from Simulation to the Real World for Robotics Manipulation Tasks , 2018, 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[12]  Pierre-Yves Oudeyer,et al.  Active learning of inverse models with intrinsically motivated goal exploration in robots , 2013, Robotics Auton. Syst..

[13]  P. Rochat.  Object manipulation and exploration in 2- to 5-month-old infants , 1989 .

[14]  Emalie G. McMahon,et al.  The Embodied Origins of Infant Reaching: Implications for the Emergence of Eye-Hand Coordination , 2018 .

[15]  D. Parisi,et al.  Learning to Reach by Constraining the Movement Search Space , 2000 .

[16]  L. A. Teixeira,et al.  Early infant's use of visual feedback in voluntary reaching for a spatial target , 2013, Front. Psychol..

[17]  Daniele Caligiore,et al.  The Development of Reaching and Grasping : Towards an Integrated Framework Based on a Critical Review of Computational and Robotic Models , 2018 .

[18]  Jakub W. Pachocki,et al.  Learning dexterous in-hand manipulation , 2018, Int. J. Robotics Res..

[19]  Gordon Cheng,et al.  Adaptive Robot Body Learning and Estimation Through Predictive Coding , 2018, 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[20]  Marco Santello,et al.  Reach-to-grasp behavior: Brain, behavior, and modelling across the life span , 2018 .

[21]  Rolf Pfeifer,et al.  Robots as Powerful Allies for the Study of Embodied Cognition from the Bottom Up , 2018, The Oxford Handbook of 4E Cognition.

[22]  Angelo Cangelosi,et al.  An open-source simulator for cognitive robotics research: the prototype of the iCub humanoid robot simulator , 2008, PerMIS.

[23]  E. Thelen,et al.  The transition to reaching: mapping intention and intrinsic dynamics. , 1993, Child development.

[24]  Jean Piaget,et al.  La naissance de l'intelligence chez l'enfant (Vol. I) , 1938 .

[25]  Chiara Bartolozzi,et al.  Robots with a sense of touch. , 2016, Nature materials.

[26]  Rafael Pérez y Pérez,et al.  Emergence of eye–hand coordination as a creative process in an artificial developmental agent , 2017, Adapt. Behav..

[27]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[28]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[29]  Martina Zambelli,et al.  Online Multimodal Ensemble Learning Using Self-Learned Sensorimotor Representations , 2017, IEEE Transactions on Cognitive and Developmental Systems.

[30]  Stephan K. U. Zibner,et al.  Reaching for objects: a neural process account in a developmental perspective , 2017 .

[31]  Wojciech Zaremba,et al.  Domain randomization for transferring deep neural networks from simulation to the real world , 2017, 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[32]  Giulio Sandini,et al.  Autonomous Online Learning of Reaching Behavior in a humanoid Robot , 2012, Int. J. Humanoid Robotics.

[33]  Masaki Ogino,et al.  Cognitive Developmental Robotics: A Survey , 2009, IEEE Transactions on Autonomous Mental Development.

[34]  P. Rochat,et al.  Multimodal perception in the control of infant reaching. , 1994, Journal of experimental psychology. Human perception and performance.

[35]  Benjamin Kuipers,et al.  Learning and Acting in Peripersonal Space: Moving, Reaching, and Grasping , 2018, Front. Neurorobot..

[36]  Alessandro Roncone,et al.  Peripersonal Space and Margin of Safety around the Body: Learning Visuo-Tactile Associations in a Humanoid Robot with Artificial Skin , 2016, PloS one.

[37]  Stefano Nolfi,et al.  A Robotic Model of Reaching and Grasping Development , 2013, IEEE Transactions on Autonomous Mental Development.

[38]  M. Asada,et al.  Visual attention by saliency leads cross-modal body representation , 2008, 2008 7th IEEE International Conference on Development and Learning.

[39]  Philippe Gaussier,et al.  Reaching and Grasping: what we can learn from psychology and robotics , 2017 .

[40]  J. Piaget,et al.  LA NAISSANCE DE L'INTELLIGENCE CHEZ L'INFANT , 1939 .

[41]  A. Cangelosi,et al.  Developmental Robotics: From Babies to Robots , 2015 .

[42]  Mark H. Lee,et al.  Integration of Active Vision and Reaching From a Developmental Robotics Perspective , 2010, IEEE Transactions on Autonomous Mental Development.