- Rating
- 1
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
StructRL uses an LLM to decompose long-horizon tasks into subtasks with dependency stages, grounds each subtask to a binary predicate on simulator state, and rewards each completion once, only after prerequisites are met, scaled by completion speed relative to demonstrations. Trained with PPO on RoboCasa365 and LIBERO-Long with GR00T-N1.5 and π0.5, it beats the strongest online RL baseline by 2.2 to 7.6 points.
Critical strengths
- The paper is clearly written and easy to follow.
- Reasonable breadth in simulation: two benchmarks, two backbones, GRPO variant, and a decomposition-density is a useful practical finding.
Critical weaknesses
- No real-world experiments, and the method depends on privileged simulator state. Every intermediate reward is a predicate over the benchmark's native state and success-check utilities.
- The core argument is that scalar progress estimates cannot tell which prerequisite makes a completion valid. Stage-aware reward models already address this: VICtoR (ICLR 2025) and SARM (ICLR 2026) both predict the current stage plus within-stage progress. Neither is cited or compared.
- Table 1 reports single runs. Figure 3's "three evaluation runs" measure evaluation noise, not training variance. Many deltas (+0.4, +0.5, +0.8, +0.9) and both the gating and pacing contributions are of this magnitude.
- The method rely on privileged state and success checks which is unavailable on real environment
Questions for authors
- How would subtask verification work on a real robot?
- Why were VICtoR, SARM not compared, at least on RoboCasa365?
- How many training seeds, and what are the standard deviations?