PULSE: Doppler prompting cuts mmWave pose jitter by 40%
New technique filters out fake motion cues from radar for stable tracking.
Researchers propose PULSE, which converts mmWave Doppler signatures into confidence-aware motion prompts and injects them into spatial magnitude reasoning through constrained interactions. By screening Doppler prompts before they influence prediction, PULSE first suppresses spurious spectral motion cues and then uses the screened prompts to stabilize prediction. Across three datasets spanning single- and multi-person settings, PULSE consistently improves pose accuracy and temporal stability, indicating that controlled Doppler prompting is a practical direction for stable mmWave HPE.
- PULSE converts Doppler signatures into confidence-aware motion prompts before injecting them into spatial magnitude reasoning.
- It suppresses non-human Doppler signatures that cause jittery pose trajectories, improving temporal stability by up to 40%.
- Validated across three datasets (single and multi-person), PULSE outperforms naive fusion methods on both accuracy and stability.
Why It Matters
Stable radar-based pose tracking enables reliable AR/VR, health monitoring, and autonomous systems without privacy concerns.