Research map / Vision-language and multimodal / Robot action and navigation
AutoSpeed: Annotation-Free Stage-Adaptive Motion Speed Learning for Robot Manipulation
Qingda Hu, Ziheng Qiu, Jieru Zhao, Zhongxue Gan, Wenchao Ding
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 |
|---|---|---|---|
| train with AutoSpeed stage-adaptive speed | ACT · ALOHA Transfer Cube | episode length: 272 -> 160 | Improved |
| train with AutoSpeed (standard control) | ACT · ALOHA Transfer Cube | success rate (%): 72% -> 64% | Hurt |
| add high-gain controller to AutoSpeed (ablation) | ACT AutoSpeed · ALOHA Transfer Cube | success rate (%): 64% -> 78% | Improved |
| train with AutoSpeed | ACT · ALOHA Insertion | success rate (%): 22% -> 24% | Improved |
| train with AutoSpeed | BAKU-DiT · Meta-World | success rate (%): 43% -> 62% | Improved |
| train with AutoSpeed | BAKU-Flow · Meta-World | success rate (%): 57% -> 65% | Improved |
| train with AutoSpeed | BAKU-Flow · LIBERO-10 | episode length: 287 -> 171 | Improved |
| train with AutoSpeed | BAKU-DiT · LIBERO-10 | success rate (%): 50% -> 49% | No clear effect |
| train with AutoSpeed + NTA (real robot) | BAKU-Flow · Real: Transfer the Banana | completion time (s): 20.80s -> 11.24s | Improved |
| train with AutoSpeed + NTA (real robot) | BAKU-Flow · Real: Place the Toy | success rate (%): 33% -> 43% | Improved |
| narrow speed range to [1x,2x] vs aggressive [2x,4x] (ablation) | ACT AutoSpeed · ALOHA Transfer Cube | success rate / length: SR 76% vs 70%; length down to 145 at [2x,4x] | Mixed |
| use smaller length penalty w (ablation) | AutoSpeed · ALOHA Transfer Cube | speed | Improved |
Closest papers by meaning
- STEP: Warm-Started Visuomotor Policies with Spatiotemporal Consistency Prediction (ICML 2026 · similarity 0.90)
- Chain-of-Action: Trajectory Autoregressive Modeling for Robotic Manipulation (NeurIPS 2025 · similarity 0.90)
- FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistency (NeurIPS 2025 · similarity 0.88)
- FreqPolicy: Frequency Autoregressive Visuomotor Policy with Continuous Tokens (NeurIPS 2025 · similarity 0.88)
- PACE: Phase-Aware Chunk Execution for Robot Policies with Action Chunking (NeurIPS 2026 · similarity 0.88)
- SkiP: When to Skip and When to Refine for Efficient Robot Manipulation (NeurIPS 2026 · 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.
