Research map / Point clouds / Point cloud classification
EquAct: An SE(3)-Equivariant Multi-Task Transformer for 3D Robotic Manipulation
Xupeng Zhu, Yu Qi, Yizhe Zhu, Robin Walters, Robert Platt
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 |
|---|---|---|---|
| keep SE(3) equivariance in the network layers (ablation) | EquAct · RLBench 4 tasks, 10 demos | avg. success rate (%): 12.3 -> 52.8 | Improved |
| use spherical harmonic degree 3 instead of 2 (ablation) | EquAct · RLBench 4 tasks, 10 demos | avg. success rate (%): 45.5 -> 52.8 | Improved |
| use iFiLM instead of standard FiLM (ablation) | EquAct · RLBench 4 tasks, 10 demos | avg. success rate (%): 50.3 -> 52.8 | Improved |
| use iFiLM instead of FiLM on place_cups (ablation) | EquAct · RLBench place_cups, 10 demos | success rate (%): 24 -> 62 | Improved |
| add rotation data augmentation (ablation) | EquAct · RLBench 4 tasks, 10 demos | avg. success rate (%): 50.5 -> 52.8 | Improved |
| use EquAct instead of baselines in SE(3) pose setting | EquAct · RLBench 18 tasks, 10 demos SE(3) | avg. success rate (%): 37.9 -> 53.3 | Improved |
| use EquAct instead of baselines in 100-demo SE(2) setting | EquAct · RLBench 18 tasks, 100 demos | avg. success rate (%): 86.8 -> 89.4 | Improved |
| use EquAct on tasks with fixed object pose | EquAct · RLBench sweep_to_dustpan | success rate (%): 95 -> 59 (vs 3DDA, SE(3)) | Hurt |
| use EquAct on physical SE(3) tasks | EquAct · 4 physical tasks | avg. success rate (%): 12.5 -> 65.0 | Improved |
Closest papers by meaning
- Deep SE(3)-Equivariant Geometric Reasoning for Precise Placement Tasks (ICLR 2024 · similarity 0.88)
- Localizing, Structuring, and Rendering: Bridging 3D and 2D Vision-Language-Action Models for Robotic Manipulation (CVPR 2026 · similarity 0.87)
- SAM-E: Leveraging Visual Foundation Model with Sequence Imitation for Embodied Manipulation (ICML 2024 · similarity 0.87)
- SE(3)-Equivariant Diffusion Policy in Spherical Fourier Space (ICML 2025 · similarity 0.86)
- OmniManip: Towards General Robotic Manipulation via Object-Centric Interaction Primitives as Spatial Constraints (CVPR 2025 · similarity 0.86)
- ET-SEED: EFFICIENT TRAJECTORY-LEVEL SE(3) EQUIVARIANT DIFFUSION POLICY (ICLR 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.
