DynaWM: Dynamics-Aware Distillation with World Model and Momentum Targets for Smooth Locomotion over Continuous Stairs

arXiv 2026 47.5 method, application

TLDR

DynaWM uses a world model regularizer and momentum targets to improve terrain encoding and motion smoothness for bipedal-wheeled robots on stairs.

Reasoning

The paper presents a novel dynamics-aware representation learning framework with clear methodology (world model regularizer, momentum targets) and strong empirical validation (simulation and real hardware). Weakness: limited detail on the world model architecture and comparison to baselines in the abstract.

Read-first score

Read-first score 47.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 24.

Recency 5%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Citation impact 15%

92.6

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.92574086

Methodology quality 15%

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

Reproducibility 15%

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

Topical relevance 25%

34.3

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

Graph prestige 15%

5.7

PageRank-style prestige over the local citation graph built from available OpenAlex references. local_pagerank=0.0016

Citation velocity 10%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

FoundationFrontierBridgeMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 374.

Keyword Scores

world model

world dynamics prediction

model-based reinforcement learning world model

world simulator

generative world model

interactive world model

video world model

Deep Analysis

Innovations

World model as a regularizer to enforce forward-dynamics awareness, preserving comprehensive terrain geometry and enabling hierarchical encoding visualization
Momentum target encoder to provide consistent distillation targets, preventing dimensional collapse from non-stationary teacher updates

Methodology

DynaWM is a dynamics-aware representation learning framework built on a teacher-student paradigm. It incorporates a world model regularizer to enforce forward-dynamics awareness and a momentum target encoder to stabilize knowledge transfer. Evaluation is conducted via PCA visualization and quantitative metrics on terrain encoding, with experiments in simulation and on real hardware.

Key Results

The method achieves superior terrain adaptability and motion smoothness, enabling bipedal-wheeled robots to traverse diverse continuous stairs in both simulation and real hardware experiments.

Resources

Paper