Research map / Probabilistic and scientific modeling / Dynamical systems
PIPHEN: Physical Interaction Prediction with Hamiltonian Energy Networks
Kewei Chen, Yayu Long, Mingsheng Shang
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 |
|---|---|---|---|
| PIPHEN vs LLaMAR | PIPHEN · MAP-THOR 2-agent | Success rate (%): 66 -> 75 | Improved |
| add Hybrid Physical Representation (ablation) | PIPHEN · MAP-THOR ablation | Task completion rate (%): 78.4 -> 92.6 | Improved |
| add Hamiltonian Energy Network (ablation) | PIPHEN · MAP-THOR ablation | Control precision (cm): 5.7 -> 3.2 | Improved |
| add Hamiltonian Energy Network (ablation) | PIPHEN · MAP-THOR ablation | Task completion rate (%): 82.1 -> 92.6 | Improved |
| add uncertainty decomposition (ablation) | PIPHEN · MAP-THOR ablation | Task completion rate (%): 89.1 -> 92.6 | Improved |
| add Micro-brain ecosystem (ablation) | PIPHEN · MAP-THOR ablation | Comm. load (MB/s): 5.4 -> 1.8 | Improved |
| add LLM-enhanced knowledge pipeline (ablation) | PIPHEN · MAP-THOR ablation | Task completion rate (%): 86.3 -> 92.6 | Improved |
| swap deploy model to Qwen2.5-0.5B (ablation) | PIPHEN · MAP-THOR | Success rate (%): 75 -> 68 | Hurt |
| replace Claude/GPT-4o Generate/Purify models (ablation) | PIPHEN · MAP-THOR | Success rate (%): 75 -> 73 (GPT-4o/GPT-4o); 75 -> 70 (Claude/Claude) | Hurt |
Closest papers by meaning
- PIP: Physical Interaction Prediction via Mental Simulation with Span Selection (ECCV 2022 · similarity 0.85)
- PhysBench: Benchmarking and Enhancing Vision-Language Models for Physical World Understanding (ICLR 2025 · similarity 0.83)
- SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulation (ICML 2025 · similarity 0.83)
- R2C: Mapping Room to Chessboard to Unlock LLM As Low-Level Action Planner (CVPR 2025 · similarity 0.83)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.83)
- Scaling Proprioceptive-Visual Learning with Heterogeneous Pre-trained Transformers (NeurIPS 2024 · similarity 0.82)
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.
