Research map / Reinforcement learning / Visual and representation RL
Universal Planning Networks: Learning Generalizable Representations for Visuomotor Control
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 |
|---|---|---|---|
| increase test-time planning steps from 40 to 160 (ablation) | UPN trained on 20000 demos · Reacher VOVG | test success rate %: 38.1% -> 64.44% | Helped |
| use 160 planning steps instead of 40 across demo counts (ablation) | UPN · Reacher VOVG | success rate | Helped |
| double training demos from 20000 to 40000 at 40 steps (ablation) | UPN · Reacher VOVG | test success rate %: 38.1% -> 64.78% | Helped |
| use UPN feature space for RL reward instead of VAE features | PPO RL on reacher · Fig 4d varying goals | success rate %: 24.67% -> 71.12% | Helped |
| use RIL or AIL encoder features as RL reward space instead of UPN | PPO RL on reacher · Fig 4d fixed goals | success rate %: 69.84% -> 0% | Hurt |
| RL with UPN rewards versus the imitating expert | UPN-RL · Fig 4d varying goals | success rate %: 51.1 % -> 71.12% | Helped |
| more demonstrations for AIL | AIL imitation · four generalization tasks | success rate | Helped |
| replace UPN reward with naive spatial distance reward | RL on transfer tasks · Point to Ant and Reacher transfer | success | Hurt |
Closest papers by meaning
- H$^3$DP: Triply‑Hierarchical Diffusion Policy for Visuomotor Learning (ICLR 2026 · similarity 0.89)
- Latent Policy Barrier: Learning Robust Visuomotor Policies by Staying In-Distribution (NeurIPS 2025 · similarity 0.88)
- Latent Diffusion Planning for Imitation Learning (ICML 2025 · similarity 0.87)
- DynaMo: In-Domain Dynamics Pretraining for Visuo-Motor Control (NeurIPS 2024 · similarity 0.87)
- Zero-Shot Robotic Manipulation with Pre-Trained Image-Editing Diffusion Models (ICLR 2024 · similarity 0.87)
- FreqPolicy: Frequency Autoregressive Visuomotor Policy with Continuous Tokens (NeurIPS 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.