Research map / Reinforcement learning / Visual and representation RL
Liquid Pouring Monitoring via Rich Sensory Inputs
Tz-Ying Wu, Juan-Ting Lin, Tsun-Hsuang Wang, Chan-Wei Hu, Juan Carlos Niebles, Min Sun
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 initial object state classification auxiliary task (ablation) | Vanilla RNN · Liquid pouring monitoring (cross user) | success/failure accuracy: 81.95 -> 89.25 | Improved |
| add trajectory forecasting auxiliary task (ablation) | Vanilla RNN · Liquid pouring monitoring (cross user) | success/failure accuracy: 81.95 -> 90.82 | Improved |
| add both auxiliary tasks (regression only) (ablation) | Vanilla RNN · Liquid pouring monitoring (cross user) | success/failure accuracy: 81.95 -> 92.97 | Improved |
| add adversarial training loss to trajectory forecasting (ablation) | Ours w/o adv. · Liquid pouring monitoring (cross user) | success/failure accuracy: 92.97 -> 93.25 | Improved |
| add adversarial training loss to trajectory forecasting (ablation) | Ours w/o adv. · Liquid pouring monitoring (cross user) | initial state classification accuracy: 64.15 -> 75.69 | Improved |
| add adversarial training loss to trajectory forecasting (ablation) | Ours w/o adv. · Liquid pouring monitoring (cross user) | rotation error: 14.20 -> 14.06 | Improved |
| use hierarchical late-fusion LSTM instead of 2-layer early-fusion LSTM (ablation) | Ours · Liquid pouring monitoring (cross user) | success/failure accuracy: 87.06 -> 93.25 | Improved |
| use hierarchical late-fusion LSTM instead of 2-layer early-fusion LSTM (ablation) | Ours · Liquid pouring monitoring (cross user) | initial state classification accuracy: 58.92 -> 75.69 | Improved |
| add both auxiliary tasks with adversarial loss (ablation) | Vanilla RNN · Liquid pouring monitoring (cross container) | success/failure accuracy: 89.16 -> 97.11 | Improved |
| add adversarial training loss to trajectory forecasting (ablation) | Ours w/o adv. · Liquid pouring monitoring (cross container) | initial state classification accuracy: 63.92 -> 67.69 | Improved |
| add adversarial training loss to trajectory forecasting (ablation) | Ours w/o adv. · Liquid pouring monitoring (cross container) | rotation error: 11.11 -> 11.30 | Hurt |
| add both auxiliary tasks with adversarial loss (ablation) | Vanilla RNN · Liquid pouring monitoring (cross trial) | success/failure accuracy: 99.65 -> 100 | No clear effect |
Closest papers by meaning
- PourIt!: Weakly-Supervised Liquid Perception from a Single Image for Visual Closed-Loop Robotic Pouring (ICCV 2023 · similarity 0.88)
- The Sound of Water: Inferring Physical Properties From Pouring Liquids (TPAMI 2026 · similarity 0.87)
- Phys-Liquid: A Physics-Informed Dataset for Estimating 3D Geometry and Volume of Transparent Deformable Liquids (AAAI 2026 · similarity 0.84)
- SparseDFF: Sparse-View Feature Distillation for One-Shot Dexterous Manipulation (ICLR 2024 · similarity 0.82)
- Touch in the Wild: Learning Fine-Grained Manipulation with a Portable Visuo-Tactile Gripper (NeurIPS 2025 · similarity 0.82)
- ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning (CVPR 2025 · similarity 0.81)
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.
