Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Planning With Pixels in (Almost) Real Time
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 Rollout IW(1) instead of breadth-first IW(1) (ablation) | IW(1) planning over B-PROST screen features · Atari games (58 games), 0.5s budget | games scoring at least human: 7 (14.2%) -> 19 (38.7%) | Helped |
| Add risk-averse reward shaping (ablation) | Rollout IW(1) · Atari games (58 games), 0.5s budget | games scoring at least human: 19 (38.7%) -> 22 (44.8%) | Helped |
| Add subscoring (ablation) | RA Rollout IW(1) · Atari games (58 games), 0.5s budget | games scoring at least human: 22 (44.8%) -> 25 (51.0%) | Helped |
| Add subscoring (ablation) | RA Rollout IW(1) · Atari games (58 games), 32s budget | games scoring at least human: 35 (71.4%) -> 37 (75.5%) | Helped |
| Add risk-averse reward shaping and subscoring (ablation) | Rollout IW(1) · Montezuma's Revenge, 0.5s | score: 0.0 -> 100.0 | Helped |
| Increase the decision-time budget | RAS Rollout IW(1) · Atari games (58 games) | games scoring at least human: 25 (51.0%) -> 37 (75.5%) | Helped |
| Use screen B-PROST features rather than RAM states | IW(1) · Atari games, fixed time budget | score | Mixed |
Closest papers by meaning
- Width-based Lookaheads with Learnt Base Policies and Heuristics Over the Atari-2600 Benchmark (NeurIPS 2021 · similarity 0.88)
- Planning from Pixels in Atari with Learned Symbolic Representations (AAAI 2021 · similarity 0.88)
- Planning from Pixels in Environments with Combinatorially Hard Search Spaces (NeurIPS 2021 · similarity 0.85)
- Best-First Width Search: Exploration and Exploitation in Classical Planning (AAAI 2017 · similarity 0.85)
- Investigating Pre-Training Objectives for Generalization in Vision-Based Reinforcement Learning (ICML 2024 · similarity 0.82)
- Visual Planning: Let's Think Only with Images (ICLR 2026 · similarity 0.82)
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.