VisualPredicator: Learning Abstract World Models with Neuro-Symbolic Predicates for Robot Planning

arXiv 24.10 2024 42.5 method

TLDR

Introduces Neuro-Symbolic Predicates for learning abstract world models, improving sample efficiency, generalization, and interpretability in robot planning.

Reasoning

The paper presents a novel neuro-symbolic approach for abstract world models, with strong empirical results across simulated domains. However, it lacks real-world experiments and does not address several related keywords like generative or video world models.

Read-first score

Read-first score 42.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 23.

Recency 8%

75.1

Uses a gentle age decay so recent papers surface without erasing older foundations. 2024

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=baseline,result

Topical relevance 42%

32.9

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

Candidate

Rank sensitivity

Stability: volatile; rank range: 227.

Keyword Scores

world model

model-based reinforcement learning world model

world dynamics prediction

interactive world model

world simulator

generative world model

video world model

Deep Analysis

Innovations

Neuro-Symbolic Predicates as a first-order abstraction language combining symbolic and neural knowledge representations
Online algorithm for inventing such predicates and learning abstract world models for robot planning

Methodology

The paper proposes Neuro-Symbolic Predicates, a first-order abstraction language that integrates symbolic and neural knowledge. An online algorithm is introduced to invent these predicates and learn abstract world models. The approach is evaluated against hierarchical reinforcement learning, vision-language model planning, and symbolic predicate invention across five simulated robotic domains on both in- and out-of-distribution tasks.

Key Results

The proposed approach demonstrates better sample complexity, stronger out-of-distribution generalization, and improved interpretability compared to the baselines.

Resources

Paper