Awesome World Model Hub Papers · Datasets · Projects
← Back to papers

Learning World Models for Unconstrained Goal Navigation

NeurIPS 24 2024 63.3 method

TLDR

Introduces MUN algorithm for goal-conditioned RL that models state transitions between arbitrary subgoal states, improving world model generalization.

Reasoning

Strengths: Addresses key generalization challenges in world models for goal navigation, with experimental results showing improved reliability and policy generalization. Weaknesses: Abstract lacks details on experimental setup and does not mention real-world validation, limiting assessment of practical applicability.

Read-first score

Read-first score 63.3, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 40.

Recency 8%
75.1

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

Reproducibility 25%
73

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=True; dataset=False; markers=github

Methodology quality 25%
60

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

Topical relevance 42%
57.1

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

Field roles

Reproducibility anchor

Rank sensitivity

Stability: volatile; rank range: 221.

Keyword Scores

world model
10
model-based reinforcement learning world model
9
world dynamics prediction
8
world simulator
5
generative world model
4
interactive world model
3
video world model
1

Deep Analysis

Innovations

  • Introduces MUN, a novel goal-directed exploration algorithm that models state transitions between arbitrary subgoal states in the replay buffer.
  • Addresses the challenge of generalizing world models to backward transitions along recorded trajectories and transitions between states across different trajectories.
  • Facilitates learning policies to navigate between any key states, improving generalization to new goal settings.

Methodology

The paper proposes MUN, a goal-directed exploration algorithm that learns world models to model state transitions between arbitrary subgoal states in the replay buffer. It uses a replay buffer to store trajectories and aims to generalize to backward and cross-trajectory transitions. The algorithm is evaluated in goal-conditioned reinforcement learning with sparse rewards.

Key Results

Experimental results demonstrate that MUN strengthens the reliability of world models and significantly improves the policy's capacity to generalize across new goal settings.

Tags