Proposes a multi-phase imitation learning framework using supplementary demonstrations with a discriminator offline and self-supervised adaptation online to mitigate distribution shift in robotic control, claiming superior robustness in MuJoCo evaluations.
URL https://proceedings.mlr
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
How to Mitigate the Distribution Shift Problem in Robotics Control: A Robust and Adaptive Approach Based on Offline to Online Imitation Learning
Proposes a multi-phase imitation learning framework using supplementary demonstrations with a discriminator offline and self-supervised adaptation online to mitigate distribution shift in robotic control, claiming superior robustness in MuJoCo evaluations.