LGSPApr 19

What Physics do Data-Driven MoCap-to-Radar Models Learn?

arXiv:2605.000186.7
Predicted impact top 63% in LG · last 90 daysOriginality Incremental advance
AI Analysis

For researchers using data-driven models for radar signal generation, this work highlights the need for physics-based evaluation beyond reconstruction error.

The paper investigates whether data-driven MoCap-to-radar models learn underlying physics by introducing two physics-based metrics. Experiments show that low reconstruction error does not guarantee physical consistency, and temporal attention is critical for transformers to learn physics.

Data-driven MoCap-to-radar models generate plausible micro-Doppler spectrograms, but do they actually learn the underlying physics? We introduce a physics-based interpretability framework to answer this question via two proposed complementary metrics: one measures alignment between model predictions and the physics-derived Doppler frequency, while the other tests whether predictions preserve the velocity-frequency relationship under velocity intervention. Both metrics require only MoCap input and model predictions, without access to measured radar data. Experiments across several model architectures reveal that low reconstruction error does not guarantee physical consistency: some, but not all, models achieve low error yet perform poorly on the two physics-based metrics. Further analysis shows that temporal attention is critical for transformer-based models to learn the underlying physics.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes