Research map / Reinforcement learning / Visual and representation RL
Sub-Goal Trees a Framework for Goal-Based Reinforcement Learning
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 |
|---|---|---|---|
| Use SGT-PG instead of SeqSG with 3 sub-goals | 7-DoF robot arm planning · wall | success rate: 0.593±0.047 -> 0.973±0.007 | Helped |
| Use SGT-PG instead of SeqSG with 7 sub-goals | 7-DoF robot arm planning · wall | success rate: 0.487±0.037 -> 0.973±0.007 | Helped |
| Increase SGT-PG from 1 to 3 sub-goals (ablation) | 7-DoF robot arm planning · wall | success rate: 0.896±0.016 -> 0.973±0.007 | Helped |
| Increase SGT-PG from 3 to 7 sub-goals (ablation) | 7-DoF robot arm planning · wall | success rate: 0.973±0.007 -> 0.973±0.007 | No clear effect |
| Increase SeqSG from 1 to 7 sub-goals (ablation) | 7-DoF robot arm planning · wall | success rate: 0.676±0.034 -> 0.487±0.037 | Hurt |
| Use SGT-PG instead of SeqSG with 7 sub-goals | 7-DoF robot arm planning · self-collision | success rate: 0.88±0.03 -> 0.996±0.007 | Helped |
| Increase SGT-PG from 1 to 7 sub-goals (ablation) | 7-DoF robot arm planning · self-collision | success rate: 1.±0. -> 0.996±0.007 | No clear effect |
| Use sub-goal predictions with an inverse model instead of FQI alone | 2D particle motion planning · 2D particle motion planning | average distance to goal: 0.58 -> 0.13 | Helped |
| Use the FQI controller with SGT sub-goals instead of FQI alone | 2D particle motion planning · 2D particle motion planning | average distance to goal: 0.58 -> 0.29 | Helped |
| Use the FQI controller with SGT sub-goals instead of the inverse model | approx. SGTDP controller · 2D particle motion planning | average collision rate: 0.25 -> 0.06 | Helped |
| Use the inverse model with SGT sub-goals instead of the FQI controller | approx. SGTDP controller · 2D particle motion planning | average distance to goal: 0.29 -> 0.13 | Helped |
Closest papers by meaning
- Stitching Sub-trajectories with Conditional Diffusion Model for Goal-Conditioned Offline RL (AAAI 2024 · similarity 0.87)
- Flattening Hierarchies with Policy Bootstrapping (NeurIPS 2025 · similarity 0.87)
- Diffusion Subgoal Planning for Long-Horizon Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2026 · similarity 0.87)
- Are Expressive Models Truly Necessary for Offline RL? (AAAI 2025 · similarity 0.87)
- Efficient Reinforcement Learning by Discovering Neural Pathways (NeurIPS 2024 · similarity 0.86)
- Exploring the Edges of Latent State Clusters for Goal-Conditioned Reinforcement Learning (NeurIPS 2024 · 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.