- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
A-Flow improves a frozen flow-based driving planner using learned skew-symmetric correction matrices. These matrices modify the final sampling step and are stored in a library for reuse across scenes. A retrieval and selection pipeline chooses suitable corrections at inference time.
Critical strengths
The experiments are extensive, covering two driving benchmarks, component ablations, and an additional flow-based planner. The reported improvements suggest that learned corrections can be useful across different scenes and planner backbones.
Critical weaknesses
- The paper is difficult to follow. The core role of the A matrix should be explained more directly before introducing the theoretical motivation and the full selection pipeline. Some implementation and training details could be moved to the experimental section or appendix.
- The novelty of the correction mechanism needs stronger justification. At a high level, A is a learned transformation that adjusts an initial trajectory proposal. It remains unclear what advantage its skew-symmetric structure offers over a small MLP or another learned residual correction module.
- The practical efficiency benefit is not fully established. Although A is lightweight, learning the correction library still requires reward-based optimization. So how's the time consuming with the GRPO uses on the total trajectory decoder.
Questions for authors
- What distinguishes the proposed correction from an MLP-based trajectory refinement module? Could you compare them under the same frozen planner, reward budget, and candidate-selection pipeline?
- How many trainable parameters does each A matrix require, and how large is the complete correction library relative to the trajectory decoder?
- What is the total training cost, including matrix optimization, library construction, and evaluator training? How does it compare with applying GRPO directly to the trajectory decoder at comparable performance?
6 comments
Skew symmetry and comparison with a residual MLP
Skew symmetry and comparison with a residual MLP
W2. The novelty of the correction mechanism needs stronger justification. At a high level, A is a learned transformation that adjusts an initial trajectory proposal. It remains unclear what advantage its skew-symmetric structure offers over a small MLP or another learned residual correction module.
Q1. What distinguishes the proposed correction from an MLP-based trajectory refinement module? Could you compare them under the same frozen planner, reward budget, and candidate-selection pipeline?
Skew symmetry improves matrix corrections with fewer degrees of freedom, and the correction bank outperforms a similarly sized shared MLP. Table 4 compares unconstrained and skew-symmetric matrices under full A-Flow (including scale adjustment and the second branch); the MLP comparison below evaluates 32-candidate single-branch selection (A-Flow32) alongside full A-Flow.
| Matrix parameterization | Independent degrees of freedom | Oracle@32 | Final PDMS |
|---|---|---|---|
| Learned unconstrained matrix | 576 | 93.18 | 88.81 |
| Learned skew-symmetric matrix | 276 | 95.13 | 90.69 |
The higher scores support the directional constraint, which has a direct geometric interpretation: in normalized trajectory coordinates, before clipping, guarantees . A perturbation along therefore produces zero first-order change in the Euclidean norm of (Appendix B.3).
The correction bank achieves higher oracle and final scores than a shared residual MLP. We reproduce CLEAR on NAVSIM, adapting its trajectory-refinement module to the same frozen planner used by A-Flow. Our implementation uses a 7.94M-parameter shared residual MLP, trained with PPO and vehicle-state feedback until validation performance on the monitoring split plateaued. Its size is comparable to the deployed bank's 7.35M dense elements. The results below are from our NAVSIM reproduction of CLEAR. With 32 candidates and the same single-branch Value+Safety FCT pipeline, A-Flow32 scores 3.50 PDMS points higher across all 12,146 navtest scenes.
| Correction system | Corrector size | Number of corrected candidates | Corrected-candidate Oracle@32 | Final PDMS |
|---|---|---|---|---|
| CLEAR (our NAVSIM reproduction, deterministic output) | MLP: 7.94M parameters | 1 | — | 85.20 |
| CLEAR (our NAVSIM reproduction) + FCT | MLP: 7.94M parameters | 32 | 91.16 | 86.81 |
| A-Flow32 | Deployed correction bank: 7.35M dense elements | 32 | 95.13 | 90.31 |
| Full A-Flow | Deployed correction bank: 7.35M dense elements | 2×32 | — | 90.70 |
The higher oracle score shows that the bank's advantage begins before selection. Similar ego states may require opposite avoidance directions when road geometry or obstacles differ. A lightweight shared MLP may compromise between these modes. The discrete bank instead preserves correction directions from individual source scenes for retrieval and selection in each target scene, helping retain candidate diversity and quality.
Matrix and correction-bank size
Matrix and correction-bank size
Q2. How many trainable parameters does each A matrix require, and how large is the complete correction library relative to the trajectory decoder?
The deployed bank contains roughly one quarter as many dense elements as the full Stage2 decoder has parameters; optimization updates only 576 matrix parameters per scene. Optimized parameters refer to ; independent degrees of freedom refer to . FP32 storage counts only weights or matrices.
| Item | Parameters / dense elements | Independent degrees of freedom | Size relative to decoder | FP32 storage |
|---|---|---|---|---|
| Full Stage2 trajectory decoder (frozen) | 30,876,675 | — | 100% | 117.79 MiB |
| Single matrix | 576 | 276 | 0.0019% | 2.25 KiB |
| Full unfiltered set, 85,109 source matrices | 49.02M | 23.49M | 158.77% | Approx. 187.0 MiB |
| Deployed correction bank, 12,767 matrices | 7.35M | 3.52M | 23.82% | 28.05 MiB |
| 32 matrices retrieved by one branch | 18,432 | 8,832 | 0.06% | 72 KiB |
| 64 matrices retrieved by two branches | 36,864 | 17,664 | 0.12% | 144 KiB |
The bank stores reusable scene corrections. Training maintains gradients and optimizer states only for active matrices, and inference applies only the retrieved subset. Matrix computation at each update or generation step therefore depends on the matrices in use, rather than the full bank size.
How the correction matrix works
Thank you for your suggestions. Matrix A corrects the final sampling step, with skew symmetry constraining the correction direction. The correction bank outperforms a similarly sized shared MLP, and A-Flow outperforms direct Flow-GRPO at comparable training cost, including evaluator training.
How the correction matrix works
The correction matrix A is a 24×24 linear operator applied at the frozen planner's final sampling step, allowing multiple candidates to share one flow rollout. It maps the endpoint proxy q⋆ to the correction term ρAq⋆, applied only at t⋆=0.8. Candidates share the first four steps and the frozen network output at the final step.
Training stores the correction directions learned for each source scene as a matrix. At test time, retrieval supplies relevant matrices, each matrix proposes a correction to the current trajectory, and FCT assesses whether it suits the target scene. We now explain this role of A at the start of Section 4 and present implementation settings in Section 5, with detailed conditioning, optimization, and bank-construction procedures in Appendices A, C.2, and D.1.