Advanced search
1 file | 2.20 MB

Body randomization reduces the sim-to-real gap for compliant quadruped locomotion

Author
Organization
Abstract
Designing controllers for compliant, underactuated robots is challenging and usually requires a learning procedure. Learning robotic control in simulated environments can speed up the process whilst lowering risk of physical damage. Since perfect simulations are unfeasible, several techniques are used to improve transfer to the real world. Here, we investigate the impact of randomizing body parameters during learning of CPG controllers in simulation. The controllers are evaluated on our physical quadruped robot. We find that body randomization in simulation increases chances of finding gaits that function well on the real robot.
Keywords
DESIGN, compliant robotics, quadruped control, knowledge transfer, simulation-reality gap, dynamics randomization

Downloads

  • DS211.pdf
    • full text
    • |
    • open access
    • |
    • PDF
    • |
    • 2.20 MB

Citation

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

Chicago
Vandesompele, Alexander, Gabriel Urbain, Hossain Mahmud, Francis wyffels, and Joni Dambre. 2019. “Body Randomization Reduces the Sim-to-real Gap for Compliant Quadruped Locomotion.” Frontiers in Neurorobotics 13.
APA
Vandesompele, A., Urbain, G., Mahmud, H., wyffels, F., & Dambre, J. (2019). Body randomization reduces the sim-to-real gap for compliant quadruped locomotion. FRONTIERS IN NEUROROBOTICS, 13.
Vancouver
1.
Vandesompele A, Urbain G, Mahmud H, wyffels F, Dambre J. Body randomization reduces the sim-to-real gap for compliant quadruped locomotion. FRONTIERS IN NEUROROBOTICS. Lausanne: Frontiers Media Sa; 2019;13.
MLA
Vandesompele, Alexander et al. “Body Randomization Reduces the Sim-to-real Gap for Compliant Quadruped Locomotion.” FRONTIERS IN NEUROROBOTICS 13 (2019): n. pag. Print.
@article{8614197,
  abstract     = {Designing controllers for compliant, underactuated robots is challenging and usually requires a learning procedure. Learning robotic control in simulated environments can speed up the process whilst lowering risk of physical damage. Since perfect simulations are unfeasible, several techniques are used to improve transfer to the real world. Here, we investigate the impact of randomizing body parameters during learning of CPG controllers in simulation. The controllers are evaluated on our physical quadruped robot. We find that body randomization in simulation increases chances of finding gaits that function well on the real robot.},
  articleno    = {9},
  author       = {Vandesompele, Alexander and Urbain, Gabriel and Mahmud, Hossain and wyffels, Francis and Dambre, Joni},
  issn         = {1662-5218},
  journal      = {FRONTIERS IN NEUROROBOTICS},
  keywords     = {DESIGN,compliant robotics,quadruped control,knowledge transfer,simulation-reality gap,dynamics randomization},
  language     = {eng},
  pages        = {9},
  publisher    = {Frontiers Media Sa},
  title        = {Body randomization reduces the sim-to-real gap for compliant quadruped locomotion},
  url          = {http://dx.doi.org/10.3389/fnbot.2019.00009},
  volume       = {13},
  year         = {2019},
}

Altmetric
View in Altmetric
Web of Science
Times cited: