Research map / Probabilistic and scientific modeling / Dynamical systems
Learning to Poke by Poking: Experiential Learning of Intuitive Physics
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a forward-dynamics objective to the inverse model (ablation) | Inverse-only model · Real-robot long-distance displacement | generalization performance | Helped |
| Add a forward-dynamics objective to the inverse model (ablation) | Inverse-only model · Real-robot object displacement | pose error | Helped |
| Use learned object-interaction features and dynamics instead of the blob model's location-only rule | Blob model · Real-robot object displacement | pose error | Helped |
| Add a forward-dynamics objective to the inverse model (ablation) | Inverse-only model · 2-D simulation, 10K training examples | actions to reach goal | Helped |
| Add a forward-dynamics objective to the inverse model (ablation) | Inverse-only model · 2-D simulation, 20K training examples | actions to reach goal | Helped |
| Add a forward-dynamics objective to the inverse model (ablation) | Inverse-only model · 2-D simulation, 100K training examples | actions to reach goal | No clear effect |
| Use greedy planning to move an object around an obstacle | Joint model · Real-robot obstacle avoidance | successful obstacle avoidance | Hurt |
| Apply the learned model to objects with novel geometry and texture | Joint model · Real-robot object displacement | goal displacement | Helped |
Closest papers by meaning
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.86)
- SOLD: Slot Object-Centric Latent Dynamics Models for Relational Manipulation Learning from Pixels (ICML 2025 · similarity 0.86)
- RoboScape: Physics-informed Embodied World Model (NeurIPS 2025 · similarity 0.86)
- Predictive Inverse Dynamics Models are Scalable Learners for Robotic Manipulation (ICLR 2025 · similarity 0.86)
- VidMan: Exploiting Implicit Dynamics from Video Diffusion Model for Effective Robot Manipulation (NeurIPS 2024 · similarity 0.85)
- InfoCon: Concept Discovery with Generative and Discriminative Informativeness (ICLR 2024 · similarity 0.85)
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.