Advanced search
1 file | 2.45 MB

Modular reservoir computing networks for imitation learning of multiple robot behaviors

Tim Waegeman (UGent) , Eric Antonelo (UGent) , Francis wyffels (UGent) and Benjamin Schrauwen (UGent)
Author
Organization
Abstract
Autonomous mobile robots must accomplish tasks in unknown and noisy environments. In this context, learning robot behaviors in an imitation based approach would be desirable in the perspective of service robotics as well as of learning robots. In this work, we use Reservoir Computing (RC) for learning robot behaviors by demonstration. In RC, a randomly generated recurrent neural network, the reservoir, projects the input to a dynamic temporal space. The reservoir states are mapped into a readout output layer which is the solely part being trained using standard linear regression. In this paper, we use a two layered modular structure, where the first layer comprises two RC networks, each one for learning primitive behaviors, namely, obstacle avoidance and target seeking. The second layer is composed of one RC network for behavior combination and coordination. The hierarchical RC network learns by examples given by simple controllers which implement the primitive behaviors. We use a simulation model of the e-puck robot which has distance sensors and a camera that serves as input for our system. The experiments show that, after training, the robot learns to coordinate the Goal Seeking (GS) and the Object Avoidance (OA) behaviors in unknown environments, being able to capture targets and navigate efficiently.
Keywords
robots, Autonomous mobile robots, Reservoir Computing, Robotics

Downloads

  • (...).pdf
    • full text
    • |
    • UGent only
    • |
    • PDF
    • |
    • 2.45 MB

Citation

Please use this url to cite or link to this publication:

Chicago
Waegeman, Tim, Eric Antonelo, Francis wyffels, and Benjamin Schrauwen. 2009. “Modular Reservoir Computing Networks for Imitation Learning of Multiple Robot Behaviors.” In IEEE International Symposium on Computational Intelligence in Robotics and Automation, 8th, Proceedings, ed. Keun-Ho Rew, 27–32. New York, NY, USA: IEEE.
APA
Waegeman, T., Antonelo, E., wyffels, F., & Schrauwen, B. (2009). Modular reservoir computing networks for imitation learning of multiple robot behaviors. In K.-H. Rew (Ed.), IEEE International symposium on Computational Intelligence in Robotics and Automation, 8th, Proceedings (pp. 27–32). Presented at the 8th IEEE International symposium on Computational Intelligence in Robotics and Automation (CIRA 2009), New York, NY, USA: IEEE.
Vancouver
1.
Waegeman T, Antonelo E, wyffels F, Schrauwen B. Modular reservoir computing networks for imitation learning of multiple robot behaviors. In: Rew K-H, editor. IEEE International symposium on Computational Intelligence in Robotics and Automation, 8th, Proceedings. New York, NY, USA: IEEE; 2009. p. 27–32.
MLA
Waegeman, Tim, Eric Antonelo, Francis wyffels, et al. “Modular Reservoir Computing Networks for Imitation Learning of Multiple Robot Behaviors.” IEEE International Symposium on Computational Intelligence in Robotics and Automation, 8th, Proceedings. Ed. Keun-Ho Rew. New York, NY, USA: IEEE, 2009. 27–32. Print.
@inproceedings{814608,
  abstract     = {Autonomous mobile robots must accomplish tasks in unknown and noisy environments. In this context, learning robot behaviors in an imitation based approach would be desirable in the perspective of service robotics as well as of learning robots. In this work, we use Reservoir Computing (RC) for learning robot behaviors by demonstration. In RC, a randomly generated recurrent neural network, the reservoir, projects the input to a dynamic temporal space. The reservoir states are mapped into a readout output layer which is the solely part being trained using standard linear regression. In this paper, we use a two layered modular structure, where the first layer comprises two RC networks, each one for learning primitive behaviors, namely, obstacle avoidance and target seeking. The second layer is composed of one RC network for behavior combination and coordination. The hierarchical RC network learns by examples given by simple controllers which implement the primitive behaviors. We use a simulation model of the e-puck robot which has distance sensors and a camera that serves as input for our system. The experiments show that, after training, the robot learns to coordinate the Goal Seeking (GS) and the Object Avoidance (OA) behaviors in unknown environments, being able to capture targets and navigate efficiently.},
  author       = {Waegeman, Tim and Antonelo, Eric and wyffels, Francis and Schrauwen, Benjamin},
  booktitle    = {IEEE International symposium on Computational Intelligence in Robotics and Automation, 8th, Proceedings},
  editor       = {Rew, Keun-Ho},
  isbn         = {9781424448081},
  keyword      = {robots,Autonomous mobile robots,Reservoir Computing,Robotics},
  language     = {eng},
  location     = {Daejeon, South Korea},
  pages        = {27--32},
  publisher    = {IEEE},
  title        = {Modular reservoir computing networks for imitation learning of multiple robot behaviors},
  url          = {http://dx.doi.org/10.1109/CIRA.2009.5423194},
  year         = {2009},
}

Altmetric
View in Altmetric
Web of Science
Times cited: