Research map / Vision-language and multimodal / Robot action and navigation
HAMLET: Switch Your Vision-Language-Action Model into a History-Aware Policy
Myungkyu Koo, Daewon Choi, Taeyoung Kim, Kyungmin Lee, Changyeon Kim, Younggyo Seo, Jinwoo Shin
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 HAMLET (moment tokens + memory module) | GR00T N1.5 · Real-world average | success rate (%): 29.2 -> 76.4 | Improved |
| Add naive multi-frame input | GR00T N1.5 · LIBERO avg | success rate (%): 95.6 -> 86.8 | Hurt |
| Add naive multi-frame input | GR00T N1.5 · RoboCasa Kitchen (100 demos) | success rate (%): 62.6 -> 59.3 | Hurt |
| Add HAMLET | GR00T N1.5 · RoboCasa Kitchen (100 demos) | success rate (%): 62.6 -> 65.4 | Improved |
| Add HAMLET | GR00T N1.5 · LIBERO avg | success rate (%): 95.6 -> 97.6 | Improved |
| Add HAMLET | CogACT · SimplerEnv-Bridge | avg full success rate (%): 52.1 -> 63.5 | Improved |
| Add memory module on top of moment token and TCL (ablation) | HAMLET · RoboCasa Kitchen (100 demos) | success rate (%): 63.4 -> 65.4 | Improved |
| Add TCL initialization of moment tokens (ablation) | HAMLET · RoboCasa Kitchen (100 demos) | success rate (%): 64.8 -> 65.4 | Improved |
| Use Transformer memory instead of Moment Concat. (ablation) | HAMLET · RoboCasa Kitchen (100 demos) | success rate (%): 62.7 -> 65.4 | Improved |
| Use Transformer memory instead of LSTM (ablation) | HAMLET · RoboCasa Kitchen (100 demos) | success rate (%): 65.0 -> 65.4 | Improved |
| Increase moment token length from 4 to 8 (ablation) | HAMLET · RoboCasa Kitchen (100 demos) | success rate (%): 65.4 -> 66.4 | Improved |
| Increase moment token length from 8 to 64 (ablation) | HAMLET · RoboCasa Kitchen (100 demos) | success rate (%): 66.4 -> 62.5 | Hurt |
Closest papers by meaning
- AVA-VLA: Improving Vision-Language-Action models with Active Visual Attention (CVPR 2026 · similarity 0.89)
- RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies (ICML 2026 · similarity 0.89)
- Rethinking Progression of Memory State in Robotic Manipulation: An Object-Centric Perspective (AAAI 2026 · similarity 0.89)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.88)
- DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge (NeurIPS 2025 · similarity 0.87)
- ATI-VLA: Action-Centric Predictive Vision–Language–Action Models via Actionable Alignment Then Adaptive Injection (NeurIPS 2026 · similarity 0.87)
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.
