Research map / Reinforcement learning / Visual and representation RL
Deep Hierarchical Planning from Pixels
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 hierarchical latent-goal planning with the worker receiving goals from the manager | Director · Egocentric Ant Mazes | learning success | Helped |
| Use hierarchical latent-goal planning with the manager selecting subgoals | Director · Visual Pin Pads | task reward | Helped |
| Add task reward to the worker in addition to goal reward | Director · Standard benchmarks | performance | Helped |
| Add task reward to the worker in addition to goal reward | Director · 55 Atari games | human-normalized median score | Helped |
| Use continuous 1024-dimensional world-model goals instead of discrete goal codes (ablation) | Director · Sparse-reward environments | performance | Hurt |
| Replace max-cosine reward with unnormalized inner-product goal reward (ablation) | Director · Goal Reward ablation tasks | performance | No clear effect |
| Replace max-cosine reward with InnerNormed goal reward (ablation) | Director · Goal Reward ablation tasks | performance | No clear effect |
| Replace max-cosine reward with negative Euclidean distance (ablation) | Director · Goal Reward ablation tasks | performance | Hurt |
| Apply exploration bonus to the manager to encourage temporally abstract exploration (ablation) | Director · Ant Maze | learning success | Helped |
| Add low-level exploration bonus to the worker (ablation) | Director · Ant Maze | learning success | Hurt |
| Remove the exploration bonus (ablation) | Director · Standard benchmarks | performance | No clear effect |
Closest papers by meaning
- Long-Horizon Visual Planning with Goal-Conditioned Hierarchical Predictors (NeurIPS 2020 · similarity 0.89)
- Hierarchical Planning and Learning for Robots in Stochastic Settings Using Zero-Shot Option Invention (AAAI 2024 · similarity 0.87)
- Flattening Hierarchies with Policy Bootstrapping (NeurIPS 2025 · similarity 0.87)
- Reconciling Spatial and Temporal Abstractions for Goal Representation (ICLR 2024 · similarity 0.87)
- Diffusion Subgoal Planning for Long-Horizon Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2026 · similarity 0.87)
- Efficient Robotic Policy Learning via Latent Space Backward Planning (ICML 2025 · similarity 0.87)
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.