Research map / Reinforcement learning / Visual and representation RL
Track2Act: Predicting Point Tracks from Internet Videos enables Generalizable Robot Manipulation
Homanga Bharadhwaj, Roozbeh Mottaghi, Abhinav Gupta, Shubham Tulsiani
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 |
|---|---|---|---|
| predict point tracks instead of flow or generated video for goal-conditioned motion prediction | Track2Act track prediction · EpicKitchens | track AUC Delta: 0.21 (Flow) / 0.30 (Video) -> 0.67 | Improved |
| add residual policy on top of open-loop plan (ablation) | Track2Act · Robot manipulation, MG | success rate: 35% -> 70% | Improved |
| add residual policy on top of open-loop plan (ablation) | Track2Act · Robot manipulation, TG | success rate: 25% -> 40% | Improved |
| predict residuals over the open-loop plan rather than actions directly (ablation) | Track2Act residual policy · Robot manipulation, CG | success rate: 30% -> 55% | Improved |
| predict residuals over the open-loop plan rather than actions directly (ablation) | Track2Act residual policy · Robot manipulation, MG | success rate: 70% -> 70% | No clear effect |
| use web-video-derived track plans instead of goal-conditioned BC | Track2Act · Robot manipulation, TG | success rate: 0% -> 40% | Improved |
| use track-based plan instead of hand-object mask conditioning | Track2Act · Robot manipulation, MG | success rate: 70% -> 70% | No clear effect |
| use track-based plan instead of hand-object mask conditioning | Track2Act · Robot manipulation, CG | success rate: 25% -> 55% | Improved |
Closest papers by meaning
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.90)
- Zero-Shot Robotic Manipulation with Pre-Trained Image-Editing Diffusion Models (ICLR 2024 · similarity 0.89)
- Learning to Act from Actionless Videos through Dense Correspondences (ICLR 2024 · similarity 0.88)
- PointZero: 3D Point Track Completion for Learning Transferable 3D Dynamics (NeurIPS 2026 · similarity 0.88)
- Predictive Inverse Dynamics Models are Scalable Learners for Robotic Manipulation (ICLR 2025 · similarity 0.88)
- AffordDP: Generalizable Diffusion Policy with Transferable Affordance (CVPR 2025 · similarity 0.88)
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.
