Researchers have developed mmHRI, a novel framework for privacy-preserving human-robot interaction (HRI) that utilizes millimeter-wave (mmWave) radar instead of traditional RGB cameras. This approach addresses privacy concerns in sensitive environments by sensing human motion without capturing identifiable imagery. The mmHRI system employs a dual-stream architecture and a memory-based state-space model to effectively process sparse and inconsistent radar data, estimating human actions and 3D poses. These estimations are then translated into instructions for a vision-language-action policy, enabling closed-loop robot manipulation and human-aware reactions, achieving 85.09% action-recognition accuracy in privacy-preserving scenarios. AI
IMPACT Enables robots to interact with humans in privacy-sensitive environments without compromising user data.
RANK_REASON The cluster describes a research paper detailing a new framework for human-robot interaction.
Read on Hugging Face Daily Papers →
- Hugging Face
- millimeter-wave radar
- mmHRI
- RGB cameras
- vision-language-action policy
- arXiv
- Junqiao Fan
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →