Research map / Vision-language and multimodal / Robot action and navigation
Mantis: A Versatile Vision-Language-Action Model with Disentangled Visual Foresight
Yi Yang, Xueqi Li, Yiyang Chen, Jin Song, Yihan Wang, Zipeng Xiao, Jiadi Su, You Qiaoben, Pengfei Liu, Zhijie Deng
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 residual connection of current frame to the DiT head (ablation) | Mantis DVF · LIBERO (avg) | success rate (%): 94.4 -> 95.7 | Improved |
| add the disentangled visual foresight head (ablation) | Mantis · LIBERO (avg) | success rate (%): 91.3 -> 95.7 | Improved |
| pretrain the DVF on human and robot videos (ablation) | Mantis · LIBERO (avg) | success rate (%): 95.7 -> 96.2 | Improved |
| add residual connection (Long suite) (ablation) | Mantis DVF · LIBERO Long | success rate (%): 89.8 -> 92.2 | Improved |
| add DVF (Spatial suite) (ablation) | Mantis · LIBERO Spatial | success rate (%): 93.4 -> 98.2 | Improved |
| replace temporal ensemble with adaptive temporal ensemble (ablation) | Mantis · LIBERO (4 suites) | inference count: reduces the IC by nearly 50% | Improved |
| replace temporal ensemble with adaptive temporal ensemble (ablation) | Mantis · LIBERO (4 suites) | success rate | No clear effect |
| add language supervision (vs Mantis-LU) (ablation) | Mantis · Real-world OOD instructions | success | Improved |
| decouple foresight prediction from action learning | Mantis vs UnifiedVLA · LIBERO Spatial fine-tuning | convergence speed: success rate of zero during the first ten epochs | Improved |
| Mantis with language supervision vs pi0.5 without, OOD scenario 1 | Mantis · Real-world scenario 1 OOD | avg successes: 3.5 -> 8.25 | Improved |
| Mantis vs pi0.5, OOD scenario 2 | Mantis · Real-world scenario 2 OOD | avg successes: 2.5 -> 7.75 | Improved |
| Mantis vs pi0.5, ID scenario 1 | Mantis · Real-world scenario 1 ID | avg successes: 7.75 -> 8.25 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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
- DreamVLA — DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (NeurIPS 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
- DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (NeurIPS 2025 · similarity 0.89)
- ATI-VLA: Action-Centric Predictive Vision–Language–Action Models via Actionable Alignment Then Adaptive Injection (NeurIPS 2026 · similarity 0.89)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.89)
- CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models (CVPR 2025 · similarity 0.89)
- CofactVLA: Deconfounding Vision-Language-Action Models via Counterfactual Intervention (NeurIPS 2026 · similarity 0.89)
- Exploring the Limits of Vision-Language-Action Manipulation in Cross-task Generalization (NeurIPS 2025 · similarity 0.89)
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.
