Research map / Vision-language and multimodal / Robot action and navigation
Spatial Forcing: Implicit Spatial Representation Alignment for Vision-language-action Model
Fuhao Li, Wenxuan Song, Han Zhao, Jingbo Wang, Pengxiang Ding, Donglin Wang, Long ZENG, Haoang Li
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 |
|---|---|---|---|
| add spatial representation alignment (Spatial Forcing) | OpenVLA-OFT · LIBERO average | success rate (%): 97.1 -> 98.5 | Improved |
| add spatial representation alignment (Spatial Forcing) | OpenVLA-OFT · LIBERO-Long | success rate (%): 94.5 -> 96.0 | Improved |
| align to SigLIP features instead of no alignment (ablation) | OpenVLA-OFT · LIBERO average (component analysis, 1 GPU) | success rate (%): 92.7 -> 94.0 | Improved |
| use VGGT as alignment target instead of SigLIP (ablation) | OpenVLA-OFT · LIBERO average (component analysis, 1 GPU) | success rate (%): 94.0 -> 96.9 | Improved |
| add positional embedding to VGGT target (ablation) | OpenVLA-OFT · LIBERO average (component analysis, 1 GPU) | success rate (%): 94.7 -> 96.9 | Improved |
| align at layer 24 instead of layer 1 (ablation) | OpenVLA-OFT · LIBERO average (component analysis, 1 GPU) | success rate (%): 94.6 -> 96.9 | Improved |
| align at layer 24 instead of layer 32 (ablation) | OpenVLA-OFT · LIBERO average (component analysis, 1 GPU) | success rate (%): 94.8 -> 96.9 | Improved |
| train 150K instead of 2K iterations with alignment (ablation) | OpenVLA-OFT · LIBERO average (component analysis, 1 GPU) | success rate (%): 72.7 -> 96.9 | Improved |
| add alignment to speed up training (ablation) | OpenVLA-OFT · LIBERO training efficiency | speedup: 3.8x more quickly | Improved |
| use 5% of data with alignment (ablation) | OpenVLA-OFT · LIBERO data efficiency | success rate (%): 75.8 | Improved |
| use full 100% data instead of 1% with alignment (ablation) | OpenVLA-OFT · LIBERO average (component analysis, 1 GPU) | success rate (%): 42.3 -> 96.9 | Improved |
| set alignment loss weight 0.5 instead of 0 (ablation) | OpenVLA-OFT · LIBERO alpha sweep | success rate (%): 73.2 -> 93.6 | Improved |
Later papers that compared with this method
1 later accepted paper reports Spatial Forcing as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- Spatial Representation Distillation and Knowledge Routing for Vision-Language-Action Models (NeurIPS 2026 · better on 0 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- CoT-VLA — CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (CVPR 2025) · better on 1 of 1 shared result
- TraceVLA — TraceVLA: Visual Trace Prompting Enhances Spatial-Temporal Awareness for Generalist Robotic Policies (ICLR 2025) · better on 1 of 1 shared result
- Dita — Dita: Scaling Diffusion Transformer for Generalist Vision-Language-Action Policy (ICCV 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ConsisVLA-4D: Advancing Spatiotemporal Consistency in Efficient 3D-Perception and 4D-Reasoning for Robotic Manipulation (CVPR 2026 · similarity 0.91)
- Beyond Data Scaling: Representation-Centric Pre-training for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- 4D-VLA: Spatiotemporal Vision-Language-Action Pretraining with Cross-Scene Calibration (NeurIPS 2025 · similarity 0.90)
- BridgeVLA: Input-Output Alignment for Efficient 3D Manipulation Learning with Vision-Language Models (NeurIPS 2025 · similarity 0.90)
- Seeing Across Views: Benchmarking Spatial Reasoning of Vision-Language Models in Robotic Scenes (ICLR 2026 · similarity 0.90)
- SpatialRGPT: Grounded Spatial Reasoning in Vision-Language Models (NeurIPS 2024 · 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.
