- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
SEES adapts VLA policies through a failure-driven loop. It decomposes long-horizon tasks, identifies bottleneck skills, and constructs local PPO tasks using saved simulator states and LLM-generated success predicates. Related skills share a family-level LoRA on a frozen backbone. Experiments on RoboCasa365 and LIBERO report improvements over initial policies, with focused comparisons of family sharing and bottleneck selection.
Critical strengths
The system connects recurring skill failures to targeted training and shared updates, providing a coherent approach to improving long-horizon execution without additional expert demonstrations.
Experiments span two backbones and two benchmarks, with direct controls for sharing and bottleneck selection. LIBERO’s disjoint training, approval, and sealed evaluation states strengthen the reported improvements.
Critical weaknesses
Weak starting baselines. LIBERO starts at only 53.1% and reaches 65.4%, while supervised OpenVLA-OFT reports 94.5–97.1% under its own protocols. The low baseline requires explanation, and the results do not establish whether the approach remains useful for stronger starting policies.
Limited effectiveness for a primarily system-level contribution. RoboCasa composite-task gains over the base policies are only 1.1–2.9 percentage points, with final success rates of 6.1–21.8%. The novelty mainly lies in integrating existing components, making stronger effectiveness evidence important. All experiments are simulated; the autonomous improvement loop, which relies on state restoration and simulator-state rewards, remains unvalidated on real robots.
Questions for authors
-
Why do the LIBERO SFT checkpoints start at 53.1%? Please specify their training, input, and evaluation settings, and clarify whether stronger starting policies have been tested.
-
Are matched-interaction controls available against ordinary PPO or uniform/random target selection across multiple RoboCasa tasks? Such evidence could materially change the assessment.
-
Please reconcile the six-to-seven RL targets stated in Section 4.3 with the 7/9/9/9 counts implied by Table 2. Can per-task outcomes and independent evolution repeats or appropriate uncertainty estimates be provided?
2 comments
Reviewer 16nC — Part 2/2
Response to Reviewer 16nC — Part 2/2
RL task counts, selection records, and uncertainty (Q3)
Thank you for pointing out the inconsistency. The correct original counts in manuscript Table 2 are 7/9/9/9: Object selected 1+3+3 targets, whereas the other suites selected 3+3+3. These counts include PPO participation in rejected attempts. We will correct the outdated “six-to-seven” statement.
Our new LIBERO experiment fixes seven adaptation-eligible and three held-out tasks per suite before training, with identical splits across methods and both backbones. Held-out tasks never participate in failure ranking, target selection, PPO, or checkpoint selection, including rejected attempts. They are held out from adaptation, not necessarily unseen during pretraining.
Each round, SEES selects the eligible task with the lowest current diagnostic success rate. Ties use the smallest zero-based task ID within the suite, and reselection is allowed. Table 5 shows round-wise selections from one training run per backbone–suite pair, not a seed aggregate.
We repeat the complete LIBERO evolution process with three independent training seeds, not merely different inference seeds. Table 3 supplements manuscript Table 2, reporting each suite's eligible and held-out results separately. GR00T + SEES achieves 58.07 ± 0.94% overall success versus 54.72 ± 0.72% for Uniform PPO; OpenVLA-OFT achieves 97.69 ± 0.96% versus 96.64 ± 0.71%. All ± values are 95% confidence-interval half-widths in percentage points, not standard deviations. Base receives no additional training and has one evaluation run.
The expanded RoboCasa ablations also use three independent training seeds and report 95% confidence intervals (Tables 1 and 2 in Part 1). Limited rebuttal time and compute prevented repeating the complete main RoboCasa evolution pipeline.
Table 3. LIBERO success rates (%) under the fixed 7/3 split: supplementary results for manuscript Table 2. Three-seed means; Mean averages four suites.
| Suite | Method | GR00T: Eligible (7) | GR00T: Held-out (3) | GR00T: Overall | OpenVLA-OFT: Eligible (7) | OpenVLA-OFT: Held-out (3) | OpenVLA-OFT: Overall |
|---|---|---|---|---|---|---|---|
| Object | Base Model | 68.38 | 60.89 | 66.13 | 97.14 | 100.00 | 98.00 |
| Random | 68.89±1.79 | 65.04±2.55 | 67.73±2.01 | 96.70±0.98 | 99.41±1.27 | 97.51±1.06 | |
| Uniform | 72.57±1.25 | 70.07±3.19 | 71.82±1.49 | 96.38±1.89 | 99.85±0.64 | 97.42±1.34 | |
| SEES | 78.86±1.42 | 76.59±2.30 | 78.18±1.57 | 97.97±1.97 | 99.41±2.55 | 98.40±1.75 | |
| Spatial | Base Model | 41.71 | 41.33 | 41.60 | 97.33 | 100.00 | 98.13 |
| Random | 39.17±1.09 | 42.67±2.21 | 40.22±1.38 | 97.08±1.52 | 98.22±1.91 | 97.42±1.49 | |
| Uniform | 39.68±0.27 | 44.15±1.69 | 41.02±0.69 | 97.59±1.37 | 99.11±1.10 | 98.04±1.25 | |
| SEES | 42.67±2.50 | 43.26±2.30 | 42.84±1.06 | 98.22±1.91 | 99.26±1.27 | 98.53±1.72 | |
| Goal | Base Model | 41.14 | 48.44 | 43.33 | 96.76 | 99.11 | 97.47 |
| Random | 43.17±1.79 | 50.96±0.64 | 45.51±1.34 | 95.62±1.42 | 96.00±2.92 | 95.73±1.84 | |
| Uniform | 44.19±2.37 | 47.85±1.69 | 45.29±1.57 | 95.81±0.82 | 96.59±2.30 | 96.04±1.16 | |
| SEES | 45.59±0.72 | 53.63±3.37 | 48.00±1.19 | 97.65±1.45 | 98.07±1.69 | 97.78±0.51 | |
| Long | Base Model | 62.48 | 59.56 | 61.60 | 95.24 | 92.89 | 94.53 |
| Random | 61.90±1.71 | 62.67±2.92 | 62.13±2.01 | 93.27±1.97 | 96.00±2.21 | 94.09±1.67 | |
| Uniform | 60.13±1.19 | 62.22±1.10 | 60.76±1.16 | 95.75±1.45 | 93.48±0.64 | 95.07±1.19 | |
| SEES | 64.06±3.15 | 61.33±1.91 | 63.24±2.78 | 96.13±2.69 | 95.85±3.19 | 96.04±1.88 | |
| Mean | Base Model | 53.43 | 52.56 | 53.17 | 96.62 | 98.00 | 97.03 |
| Random | 53.29±1.05 | 55.33±0.96 | 53.90±1.02 | 95.67±1.28 | 97.41±1.15 | 96.19±1.24 | |
| Uniform | 54.14±1.13 | 56.07±0.42 | 54.72±0.72 | 96.38±0.59 | 97.26±1.04 | 96.64±0.71 | |
| SEES | 57.79±1.58 | 58.70±0.57 | 58.07±0.94 | 97.49±1.32 | 98.15±0.57 | 97.69±0.96 |
Table 5. Per-round bottleneck selections. One run per backbone–suite pair. IDs are zero-based within each suite; ties use the smallest ID.
| Backbone | Suite | Round 1 | Round 2 | Round 3 |
|---|---|---|---|---|
| GR00T | Object | 5 | 2 | 4 |
| GR00T | Spatial | 4 | 3 | 1 |
| GR00T | Goal | 3 | 6 | 0 |
| GR00T | Long | 0 | 9 | 7 |
| OpenVLA-OFT | Object | 8 | 6 | 1 |
| OpenVLA-OFT | Spatial | 4 | 5 | 1 |
| OpenVLA-OFT | Goal | 3 | 0 | 2 |
| OpenVLA-OFT | Long | 0 | 6 | 9 |
Task descriptions. Object: 1 = cream cheese into basket; 2 = salad dressing into basket; 4 = ketchup into basket; 5 = tomato sauce into basket; 6 = butter into basket; 8 = chocolate pudding into basket. Spatial: 1 = black bowl next to ramekin onto plate; 3 = black bowl on cookie box onto plate; 4 = black bowl from drawer onto plate; 5 = black bowl on ramekin onto plate. Goal: 0 = open middle drawer; 2 = wine bottle onto cabinet top; 3 = open top drawer and place bowl inside; 6 = cream cheese into bowl. Long: 0 = alphabet soup and tomato sauce into basket; 6 = white mug onto plate and chocolate pudding to the right of the plate; 7 = alphabet soup and cream cheese into basket; 9 = yellow-and-white mug into microwave and close its door.
Please ask further questions if anything remains unclear or inadequately addressed; we welcome further discussion.
Response to Reviewer 16nC — Part 1/2
Response to Reviewer 16nC — Part 1/2
Starting checkpoints and stronger baselines (W1 / Q1)
Thank you for your thoughtful review and constructive suggestions. Manuscript Table 3 tests family-level adapter sharing, while Table 4 tests failure-guided bottleneck selection. We added simple multi-task PPO controls to both: one globally shared LoRA over nine atomic tasks (Table 1, expanding manuscript Table 3), and uniform training over reached subtasks (Table 2, expanding manuscript Table 4). The proposed designs exceed these controls by 5.34 and 6.27 percentage points overall, respectively, under matched budgets within each experiment.
Our original LIBERO policies are GR00T SFT checkpoints released by RLinf; we did not train them ourselves. To address weak starting policies, the new fixed-7/3 comparison includes both GR00T and official OpenVLA-OFT checkpoints, with matched interactions within each backbone and suite (Table 3 in Part 2, supplementing manuscript Table 2). OpenVLA-OFT + SEES reaches 97.69% overall success versus 97.03% for Base, 96.64% for Uniform PPO, and 96.19% for Random: gains of 0.66, 1.05, and 1.50 points. It exceeds Uniform in all four suites, including Long (96.04% vs. 95.07%). These results support effectiveness with strong starting policies, although gains are modest near the performance ceiling.
Evaluation uses native LIBERO success predicates on 15 sealed initial states × five inference seeds per task. GR00T checkpoints are suite-specific. In Object/Spatial/Goal/Long order, GR00T uses horizons of 240/240/320/480, action chunks of 5/5/5/10, and four denoising steps. OpenVLA-OFT uses horizons of 280/220/300/520 and action chunks of eight.
Matched-interaction controls (Q2)
Within each RoboCasa ablation, methods share the starting checkpoint, LoRA configuration, PPO recipe, parallel-environment count, rollout epochs, horizon, and termination/loss-masking rules. Total PPO outer steps summed across adapters are matched, ensuring equal fixed-horizon interaction budgets. Target selection across five composite tasks uses 80 steps per task and method, in two 40-step rounds. Simple multi-task distributes this total across reached subtasks, not 40 steps per subtask or family. Random uniformly selects one reached subtask per round; Bottleneck selects the one with the most attributed failures. Policies are evaluated on paired held-out scenes.
Simple multi-task raises mean success from 20.40% to 50.80%, showing that extra RL helps. Bottleneck reaches 57.07% under the same budget, exceeding Simple multi-task, Random (41.07%), and Highest-success (21.33%) by 6.27, 16.00, and 35.74 points. This supports failure-guided allocation over uniform, random, or highest-success selection. Table 1 separately matches 90 total PPO outer steps across sharing schemes: Semantic family reaches 73.87% overall versus 68.53% for Simple multi-task and 71.42% for Random groups, supporting semantic sharing beyond joint RL alone.
Table 1. RoboCasa365 LoRA sharing: expanded manuscript Table 3. Success (%) over nine trained and six RL-held-out sibling tasks, 50 rollouts/task. Adapted values are means over three independent training seeds; ± denotes the 95% CI half-width in percentage points. Base has no further training and one evaluation run. Best means are bold.
Table 2. RoboCasa365 target selection: expanded manuscript Table 4. Mean success (%) across five composites; the same three-seed and CI convention applies.
System-level effectiveness and physical validation (W2)
Together, Tables 1–3 support semantic sharing and failure-guided allocation beyond additional RL alone. In a preliminary RoboDojo real-to-sim check, we reconstructed a scene from real-robot SFT images and successfully executed the corresponding SFT actions in simulation. This establishes simulation-side feasibility, not real-robot adaptation or sim-to-real transfer. The complete failure-guided adaptation loop remains unvalidated on physical robots, which we acknowledge as a key limitation.
Please ask further questions if anything remains unclear or inadequately addressed; we welcome further discussion.