User: Guest  Login
Document type:
Konferenzbeitrag 
Contribution type:
Textbeitrag / Aufsatz 
Author(s):
Malmir, Mohammadhossein; Josifovski, Josip; Klarmann, Noah; Knoll, Alois 
Title:
Robust Sim2Real Transfer by Learning Inverse Dynamics of Simulated Systems 
Abstract:
This paper presents a data-driven nonlinear disturbance observer to reduce the reality gap caused by the imperfect simulation of the real-world physics. The main focus is on increasing robustness of the closed-loop control without changing the RL algorithm or simulation model to account for the uncertainty of the real world. For this purpose, a DNN representing inverse dynamics of the deterministic source-domain environment is learned by the simulation data. The proposed approach offers a system...    »
 
Keywords:
inverse dynamics, disturbance observer, robotic manipulation, robust reinforcement learning, sim2real transfer 
Dewey Decimal Classification:
000 Informatik, Wissen, Systeme 
Horizon 2020:
ECSEL Joint Undertaking under the H2020 AI4DI project (grant agreement 826060) 
Book / Congress title:
2nd Workshop on Closing the Reality Gap in Sim2Real Transfer for Robotics 
Publisher:
Robotics: Science and Systems (R:SS) 
Date of publication:
30.07.2020 
Year:
2020 
Reviewed:
ja 
Language:
en 
Publication format:
WWW 
Format:
Text