Fine-tuning transferred RL policies for peg-in-hole tasks improves sample efficiency and success rates over zero-shot transfer or training from scratch on new robot hardware.
A generalist agent
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
-
Sustainable Transfer Learning for Adaptive Robot Skills
Fine-tuning transferred RL policies for peg-in-hole tasks improves sample efficiency and success rates over zero-shot transfer or training from scratch on new robot hardware.