Research map / Vision-language and multimodal / Robot action and navigation
ManiSoft: Towards Vision-Language Manipulation for Soft Continuum Robotics
Ziyu Wei, Luting Wang, Chen Gao, Li Wen, Si Liu
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 stability reward R_s (beta=1, D=0.3) to executor (ablation) | SAC executor · control stability | pose-difference variance: 0.176 -> 0.053 | Improved |
| increase R_s strength beta from 0 to 1.5 (average over D) (ablation) | SAC executor · control stability | pose-difference variance: 0.176 -> 0.087 | Improved |
| set pose-distance rotation weight alpha to 0.2 instead of 0.04 (ablation) | SAC executor · executor reaching, lambda=1 | success rate (%): 16.0 -> 33.0 | Improved |
| scale d1,d2 by lambda=1 versus lambda=0.4 (ablation) | SAC executor · executor reaching (avg over alpha) | success rate (%): 4.7 -> 12.5 | Improved |
| randomized scenes instead of clean scenes | Diffusion Policy · ManiSoft average | success rate (%): 31.6 -> 2.2 | Hurt |
| randomized scenes instead of clean scenes | OpenVLA-OFT · ManiSoft average | success rate (%): 30.4 -> 27.0 | Hurt |
| randomized scenes instead of clean scenes | RDT · ManiSoft average | success rate (%): 9.2 -> 1.6 | Hurt |
| use pretrained OpenVLA-OFT instead of DP trained from scratch | OpenVLA-OFT vs DP · ManiSoft Randomized average | success rate (%): 2.2 -> 27.0 | Improved |
| use DP instead of OpenVLA-OFT on COLL (clean) | Diffusion Policy · ManiSoft COLL (Clean) | success rate (%): 45.4 -> 63.0 | Improved |
| use 1B-parameter RDT instead of ~400M models | RDT · ManiSoft Clean average | success rate (%): 30.4 -> 9.2 | Hurt |
Closest papers by meaning
- ForceVLA: Enhancing VLA Models with a Force-aware MoE for Contact-rich Manipulation (NeurIPS 2025 · similarity 0.88)
- PAC Bench: Do Foundation Models Understand Prerequisites for Executing Manipulation Policies? (NeurIPS 2025 · similarity 0.88)
- Beyond World-Frame Action Heads: Motion-Centric Action Frames for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.88)
- Object-Centric Prompt-Driven Vision-Language-Action Model for Robotic Manipulation (CVPR 2025 · similarity 0.87)
- ActionUNet: Improving Robustness of VLA Models with Efficient Multi-scale Fine-tuning (NeurIPS 2026 · similarity 0.87)
- RealVLG-R1: A Large-Scale Real-World Visual-Language Grounding Benchmark for Robotic Perception and Manipulation (CVPR 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.
