Research map / Vision-language and multimodal / Robot action and navigation
LaST$_{0}$: Latent Spatio-Temporal Chain-of-Thought for Robotic Vision-Language-Action Model
Zhuoyang Liu, Jiaming Liu, Hao Chen, Jiale Yu, Ziyu Guo, Chengkai Hou, Chenyang Gu, Xiangju Mi, Renrui Zhang, Kun Wu, Zhengping Che, Jian Tang and 2 more
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 latent tokens (one per modality) (ablation) | LaST0 · RLBench ablation | success rate: 68% -> 82% | Improved |
| increase latent tokens beyond one per modality (ablation) | LaST0 · RLBench ablation | success rate | No clear effect |
| extend latent temporal coverage from 0 to 4 steps (ablation) | LaST0 · RLBench ablation | success rate: 68% -> 82% | Improved |
| extend latent temporal coverage beyond 4 steps (ablation) | LaST0 · RLBench ablation | success rate | No clear effect |
| combine image, point cloud and state latents instead of image only (ablation) | LaST0 · RLBench ablation | success rate: 74% -> 82% | Improved |
| combine latent modalities instead of point cloud only (ablation) | LaST0 · RLBench ablation | success rate: 76% -> 82% | Improved |
| mixed fast-slow ratio training instead of fixed 1:8 collaboration (ablation) | LaST0 · RLBench ablation | success rate: 74% -> 82% | Improved |
| use MoT dual-system instead of a single backbone (ablation) | LaST0 · RLBench ablation | success rate: 74% -> 82% | Improved |
| reorder latent modalities from [Visual, Geometric, State] (ablation) | LaST0 · RLBench ablation | success rate: 82 -> 78 | Hurt |
| increase latent stride from 1 to 8 in dense action settings (ablation) | LaST0 · LIBERO | success rate: 97.4 -> 99.2 | Improved |
| increase latent stride from 8 to 16 (ablation) | LaST0 · LIBERO | success rate: 99.2 -> 99.0 | No clear effect |
| add LaST CoT to the model (ablation) | LaST0 · Action feature PCA | action inter-class distance: 1.03 -> 1.34 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CoT-VLA — CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (CVPR 2025) · better on 4 of 4 shared results
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
- LaST-R1: Reinforcing Robotic Manipulation via Adaptive Physical Latent Reasoning (NeurIPS 2026 · similarity 0.93)
- A Latent World-Action Model with Jointly Aligned Reasoning (NeurIPS 2026 · similarity 0.92)
- CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (CVPR 2025 · similarity 0.91)
- LaST-VLA: Thinking in Latent Spatio-Temporal Space for Vision-Language-Action in Autonomous Driving (NeurIPS 2026 · similarity 0.90)
- ActionUNet: Improving Robustness of VLA Models with Efficient Multi-scale Fine-tuning (NeurIPS 2026 · similarity 0.90)
- RotVLA: Rotational Latent Action for Vision-Language-Action Model (NeurIPS 2026 · 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.
