Research map / Vision-language and multimodal / Robot action and navigation
TDSR-VLA: Transition-aware Denoising Sequence Representations for Vision-Language-Action
Dong-Woo Kim, KEUNHO SONG, Seungmin Lee, Hwanhee Ju, Eun Cha, Daekyum Kim
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use USR instead of SR as keys in subgoal layers (ablation) | TDSR-VLA · OpenArm real-world | average score: 0.70 -> 0.83 | Improved |
| use three subgoal injection layers instead of zero (ablation) | TDSR-VLA · OpenArm real-world | average score | Improved |
| increase subgoal injection layers from six to nine (ablation) | TDSR-VLA · OpenArm real-world | average score | Hurt |
| three versus six subgoal layers (ablation) | TDSR-VLA · OpenArm real-world | average score | No clear effect |
| planner-side transition cues plus subgoal image versus GR00T N1.5 | GR00T N1.5 · OpenArm real-world | average score: 0.74 -> 0.83 | Improved |
| TDSR-VLA versus GR00T N1.5 on object-cooperative tasks | GR00T N1.5 · OpenArm object-cooperative | average score: 0.70 -> 0.88 | Improved |
| TDSR-VLA versus GR00T N1.5 on Dual Cup Sorting | GR00T N1.5 · OpenArm Dual Cup Sorting | score: 0.58 -> 0.55 | Hurt |
| TDSR-VLA versus OpenVLA | OpenVLA · LIBERO Goal | success rate: 79.2 -> 82.0 | Improved |
| TDSR-VLA versus OpenVLA | OpenVLA · LIBERO Long | success rate: 53.7 -> 56.0 | Improved |
| TDSR-VLA versus pi0 without pretraining | pi0 · LIBERO Goal | success rate: 89.0 -> 82.0 | Hurt |
| TDSR-VLA versus pi0 without pretraining | pi0 · LIBERO Long | success rate: 48.0 -> 56.0 | Improved |
| add planner subgoal generation before acting | TDSR-VLA · Per-chunk latency | latency | Hurt |
Closest papers by meaning
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.91)
- Beyond Data Scaling: Representation-Centric Pre-training for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.91)
- VLA-Corrector: Lightweight Detect-and-Correct Inference for Adaptive Action Horizon (NeurIPS 2026 · similarity 0.90)
- NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching (NeurIPS 2025 · similarity 0.90)
- EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models (NeurIPS 2025 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
