Research map / Reinforcement learning / Visual and representation RL
Landmark-Based Heuristics for Goal Recognition
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 |
|---|---|---|---|
| landmark-based hgc instead of R&G planner-based recognition | hgc · BLOCKS-WORLD 100% observed | recognition time (s): 2.378 -> 0.163 | Helped |
| landmark-based hgc instead of R&G | hgc · DEPOTS 100% observed | recognition time (s): 15.228 -> 1.397 | Helped |
| landmark-based hgc instead of R&G | hgc · DOCK-WORKER-ROBOTS 100% observed | recognition time (s): 35.306 -> 1.125 | Helped |
| landmark-based hgc instead of R&G | hgc · DOCK-WORKER-ROBOTS 100% observed | goal recognition accuracy: 68.7% -> 100.0% | Helped |
| landmark-based hgc instead of R&G | hgc · SOKOBAN 100% observed | goal recognition accuracy: 81.2% -> 100.0% | Helped |
| hgc (θ=0) instead of R&G at low observability | hgc · BLOCKS-WORLD 10% observed | goal recognition accuracy: 83.8% -> 36.9% | Hurt |
| raise threshold θ from 0 to 30 (ablation) | hgc · BLOCKS-WORLD 10% observed | goal recognition accuracy: 36.9% -> 89.2% | Helped |
| raise threshold θ from 0 to 30 (ablation) | hgc · DOCK-WORKER-ROBOTS 10% observed | goal recognition accuracy: 41.6% -> 100.0% | Helped |
| raise threshold θ from 0 to 30 (more returned goals) (ablation) | hgc · all domains | goal recognition accuracy: 36.9% -> 89.2% | Mixed |
| landmark uniqueness weighting (huniq) instead of plain completion ratio (hgc) (ablation) | huniq vs hgc · BLOCKS-WORLD 10% observed | goal recognition accuracy: 36.9% -> 31.9% | Mixed |
| huniq instead of hgc (ablation) | huniq vs hgc · recognition time (s) across domains | recognition time (s): 1.125 -> 0.987 | Helped |
Closest papers by meaning
- An LP-Based Approach for Goal Recognition as Planning (AAAI 2021 · similarity 0.90)
- Active Goal Recognition (AAAI 2020 · similarity 0.89)
- Integration of Planning with Recognition for Responsive Interaction Using Classical Planners (AAAI 2017 · similarity 0.86)
- Block-Level Goal Recognition Design (AAAI 2024 · similarity 0.85)
- Automatic Selection of Macro-Events for Heuristic-Search Temporal Planning (AAAI 2025 · similarity 0.84)
- Goal-Space Planning with Subgoal Models (ICML 2025 · similarity 0.83)
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.