Skip to content
arXiv cs.RO·· 6 days agoAI score58

Code as Policy Agents' Self-Repair: What Drives Retry Success

Summary

This study compared retry mechanisms of code-as-policy agents on simulated robot manipulation tasks: on the same 252 initial failure cases, Qwen2.5-Coder-7B, using feedback templates selected by failure labels, recovered 85.3% within five retries, versus 68.3% for resampling with the same retry cap, and the advantage persisted under a fixed total token budget.

This page provides a summary and source link only. Read the full article at the source.

Source: arXiv cs.RO · arxiv.org

Grasping and manipulationRobot models and learningResearch papers#Research#Robot models and learning#Grasping and manipulation