Research map / Vision-language and multimodal / Robot action and navigation
RoboAnnotatorX: A Comprehensive and Universal Annotation Framework for Accurate Understanding of Long-horizon Robot Demonstration
Longxin Kou, Fei Ni, Yan Zheng, Peilong Han, Jinyi Liu, Haiqin Cui, Rui Liu, Jianye Hao
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 video-level token stream (ablation) | RoboAnnotatorX · RoboX-VQA | temporal localization | Improved |
| add clip-level token stream on top of scene tokens (ablation) | RoboAnnotatorX · RoboX-VQA | temporal reasoning | Improved |
| add scene-level tokens for semantic understanding (ablation) | RoboAnnotatorX · RoboX-VQA | scene understanding | Improved |
| add simulated data to training (ablation) | RoboAnnotatorX · RoboX-VQA | accuracy | Improved |
| add QA-based instruction tuning (ablation) | RoboAnnotatorX · RoboX-VQA | accuracy | Improved |
| use curriculum three-stage training instead of mixed training (ablation) | RoboAnnotatorX · RoboX-VQA | accuracy | Improved |
| fine-tune LLaMA-VID on RoboX-VQA | LLaMA-VID · RoboX-VQA | Avg IoU: 13.43 -> 25.38 | Improved |
| multi-scale encoder versus fine-tuned LLaMA-VID | RoboAnnotatorX · RoboX-VQA | Avg IoU: 25.38 -> 69.29 | Improved |
| fine-tune LLaMA-VID on RoboX-VQA | LLaMA-VID · RoboX-VQA OOD embodiment-level | accuracy (%): 12.51 -> 35.73 | Improved |
| multi-scale encoder versus fine-tuned LLaMA-VID | RoboAnnotatorX · RoboX-VQA OOD scene-level | accuracy (%): 30.03 -> 58.32 | Improved |
| add training-free foundation models to MLLM (NILS, MVU) | general MLLM · RoboX-VQA | Avg IoU: 2.01 and 17.47 vs 42.94 for GPT-4o | No clear effect |
| use RoboAnnotatorX annotations for planning | OpenVLA · LohoRavens | plan accuracy (%): 65.4% -> 83.5% | Improved |
Closest papers by meaning
- KISA: A Unified Keyframe Identifier and Skill Annotator for Long-Horizon Robotics Demonstrations (ICML 2024 · similarity 0.91)
- Robotic Visual Instruction (CVPR 2025 · similarity 0.88)
- RoboCerebra: A Large-scale Benchmark for Long-horizon Robotic Manipulation Evaluation (NeurIPS 2025 · similarity 0.88)
- DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation (NeurIPS 2026 · similarity 0.88)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.88)
- RoboDreamer: Learning Compositional World Models for Robot Imagination (ICML 2024 · similarity 0.88)
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.
