Research map / Reinforcement learning / Visual and representation RL
Predicate Invention for Bilevel Planning
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 only goal predicates, without predicate invention (ablation) | Bilevel planning system · PickPlace1D, Blocks, Painting, and Tools | success rate (%): 98.6 -> 39.6; 98.4 -> 3.2; 100.0 -> 0.0; 96.8 -> 0.0 | Hurt |
| Limit abstract-plan generation to nabstract = 1 at evaluation (ablation) | Bilevel planning system · PickPlace1D | success rate (%): 98.6 -> 98.6 | No clear effect |
| Limit abstract-plan generation to nabstract = 1 at evaluation (ablation) | Bilevel planning system · Blocks | success rate (%): 98.4 -> 98.2 | No clear effect |
| Limit abstract-plan generation to nabstract = 1 at evaluation (ablation) | Bilevel planning system · Painting | success rate (%): 100.0 -> 98.8 | No clear effect |
| Limit abstract-plan generation to nabstract = 1 at evaluation (ablation) | Bilevel planning system · Tools | success rate (%): 96.8 -> 42.8 | Hurt |
| Optimize predicates for the surrogate planning-time objective rather than the bisimulation criterion | Predicate invention · Four robotic planning environments | success rate (%) | Helped |
| Use learned predicates rather than manually designed predicates | Bilevel planning system · PickPlace1D | planning time (seconds): 0.045 -> 0.006 | Helped |
Closest papers by meaning
- Learning Planning Abstractions from Language (ICLR 2024 · similarity 0.88)
- VisualPredicator: Learning Abstract World Models with Neuro-Symbolic Predicates for Robot Planning (ICLR 2025 · similarity 0.87)
- Learning Neuro-Symbolic Abstractions for Robot Planning and Learning (AAAI 2024 · similarity 0.86)
- UniDomain: Pretraining a Unified PDDL Domain from Real-World Demonstrations for Generalizable Robot Task Planning (NeurIPS 2025 · similarity 0.85)
- Generalized Planning for the Abstraction and Reasoning Corpus (AAAI 2024 · similarity 0.85)
- Zero-Shot Trajectory Planning for Signal Temporal Logic Tasks (NeurIPS 2025 · similarity 0.85)
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.