Research map / Reinforcement learning / Visual and representation RL
Efficient Hybrid SE(3)-Equivariant Visuomotor Flow Policy via Spherical Harmonics for Robot Manipulation
Qinglun Zhang, Shen Cheng, Tian Dan, Haoqiang Fan, Guanghui Liu, Shuaicheng 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 FEM image feature fusion over point cloud only (rectified flow) (ablation) | E3Flow · MimicGen 8 tasks | success rate: 75.88 -> 79.00 | Improved |
| naive concatenation of image and point cloud features instead of point cloud only (ablation) | E3Flow · MimicGen 8 tasks | success rate: 75.88 -> 72.36 | Hurt |
| use FEM instead of concatenation (ablation) | E3Flow · MimicGen 8 tasks | success rate: 72.36 -> 79.00 | Improved |
| rectified flow instead of diffusion (point cloud only) (ablation) | E3Flow · MimicGen 8 tasks | success rate: 75.23 -> 75.88 | No clear effect |
| rectified flow instead of diffusion (with FEM) (ablation) | E3Flow · MimicGen 8 tasks | success rate: 77.58 -> 79.00 | Improved |
| 10 rectified flow steps instead of 1 (ablation) | E3Flow · MimicGen 8 tasks | success rate: 69.00 -> 79.00 | Improved |
| rectified flow 1 step instead of MeanFlow 1 step (ablation) | E3Flow · MimicGen 8 tasks | success rate: 54.50 -> 69.00 | Improved |
| use E3Flow vs SDP (DDPM) | E3Flow · MimicGen 8 tasks | success rate: 75.88 -> 79.00 | Improved |
| switch SDP solver to DDiM | SDP · MimicGen 8 tasks | success rate: 75.88 -> 69.75 | Hurt |
| E3Flow vs SDP under 10 deg tilt | E3Flow · MimicGen Nut Assembly D0 tilted | success rate: 39 -> 52 | Improved |
Closest papers by meaning
- 3D Equivariant Visuomotor Policy Learning via Spherical Projection (NeurIPS 2025 · similarity 0.92)
- SE(3)-Equivariant Diffusion Policy in Spherical Fourier Space (ICML 2025 · similarity 0.91)
- ET-SEED: EFFICIENT TRAJECTORY-LEVEL SE(3) EQUIVARIANT DIFFUSION POLICY (ICLR 2025 · similarity 0.88)
- DynaMo: In-Domain Dynamics Pretraining for Visuo-Motor Control (NeurIPS 2024 · similarity 0.87)
- Diffusion-EDFs: Bi-equivariant Denoising Generative Modeling on SE(3) for Visual Robotic Manipulation (CVPR 2024 · similarity 0.86)
- FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistency (NeurIPS 2025 · similarity 0.86)
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.
