robotics
The Simulator Learns Which Failures to Rebuild
CoDA shares a memory between the data engine and robot policy so new environments target diagnosed weaknesses.
Summary
CoDA shares a memory between the data engine and robot policy so new environments target diagnosed weaknesses.
Physically valid simulated scenes are not necessarily useful training examples. CoDA stores environment programs, policy failures and learning progress in one heterogeneous memory. A vision-language critic turns validated diagnoses into new generation programs, while staged reward potentials supervise policy updates. Across four manipulation families, the authors report out-of-distribution success rising from 5.3 percent after supervised fine-tuning to 27.6 percent, 9.5 points above the strongest closed-loop baseline at the same generation budget. On LIBERO-Pro, fifteen rounds produced larger gains under position and instruction perturbations. The measurements remain simulator- and benchmark-specific.
Why it matters
CoDA shares a memory between the data engine and robot policy so new environments target diagnosed weaknesses.
Limits and context
- Physically valid simulated scenes are not necessarily useful training examples.
- The measurements remain simulator- and benchmark-specific.
Key claims
CoDA shares a memory between the data engine and robot policy so new environments target diagnosed weaknesses.
Qualification: Physically valid simulated scenes are not necessarily useful training examples.
Evidence: source-2026-10-17-005
Sources
- arXiv preprint 2610.22362, original v1arXiv · primary research
Corrections
No corrections have been recorded for this story.