RF-HOI: Radio Frequency Framework for Human-Object Interaction Recognition
A new framework named RF-HOI has been developed by researchers to identify human-object interactions (HOI) using radio frequency (RF) signals, effectively addressing the challenges of privacy and lighting that affect vision-based techniques. This innovative system integrates millimeter-wave radar with RFID technology to recognize actions and objects simultaneously, improving upon the limitations of single-modality RF sensing. To overcome the issue of limited training data, the team created a simulator that generates diverse multimodal RF data for various HOIs, facilitating the fine-tuning of their recognition model. The findings are presented in the paper 'RF-HOI: Recognize Human-Object Interaction with Radio Frequency Signals', accessible on arXiv (2608.00289v1). This method has promising applications in fields such as virtual and augmented reality, embodied AI, and assistive robotics.
Key facts
- RF-HOI is the first framework that uses only radio frequency (RF) signals for HOI recognition.
- It combines mmWave radar and RFID for simultaneous action recognition and target identification.
- A simulator synthesizes multimodal RF data to address limited training data.
- The framework addresses privacy concerns and poor light conditions of vision-based methods.
- Potential applications include virtual and augmented reality, embodied AI, and assistive robotics.
- The paper is available on arXiv with identifier 2608.00289v1.
Entities
Institutions
- arXiv