Warp: Adapting the Structured Output Space of Frozen Policies
What happened
Ethan Hirschowitz and Fabio Ramos propose Warp, a context-conditioned monotone bijective spline adapter that nonlinearly reshapes the sampled outputs of a frozen base policy without modifying it, while preserving per-dimension ordering. On four controlled deployment-shift tasks in ManiSkill, the authors report that Warp achieves the strongest recovery among matched evolution-strategy adapter studies; however, when switching to PPO, the ordering relationship varies by task. In a physical SO101 peg-insertion task and in interaction-limited RLPD/SAC pipelines, Warp's success rate is comparable to a strong stochastic residual, with median time per successful episode reduced by about 30%.
AI 根据报道生成 · 6 days ago更新
Reporting timeline
Follow the reporting to explore different perspectives.
- arXiv cs.ROWarp: Structured Output-Space Adaptation for Frozen Robot Policies
The authors propose Warp, a context-conditioned monotone bijective spline adapter that nonlinearly reshapes the sampled outputs of a frozen base policy while preserving per-dimension ordering. Across four ManiSkill tasks, Warp achieves the strongest recovery in a matched evolution-strategy adapter study; after switching to PPO, the ordering relationships vary by task. In a physical SO101 peg insertion task, Warp's success rate is comparable to a strong stochastic residual, with median time per successful cycle reduced by about 30%.
Activity for this event
- Current
- 3
- Peak
- 10Nov 5 14:00
- Past 24 hours
- -49%
The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.