紀要論文

Range Aware Recognition of Hand Gestures Using FMCW Radar

eng

24321192

https://doi.org/10.15002/00030606

法政大学大学院情報科学研究科

法政大学大学院紀要. 情報科学研究科編

19

1

8

2024-03-24

Version of Record

This paper introduces a hand gesture classification implementation that combines FMCW radar and deep learning. Unlike previous work, this research implements a range-aware methodology to automatically locate the hand range bin enabling more accurate classification. Our proposal is based on the observations made about the impact of the subject chest in the Time-Range map and the signature of the hand movement in the Range-Doppler map. Then, three distinct methods are proposed for hand range bin localization. Method I and II take advantage of the first observation by locating the chest and using it as a reference to select the hand bin. On the other hand, Method III exploits the second observation about the Doppler signature of the hand movement to directly select the hand bin. Both subject-dependent and independent evaluations are performed to classify six hand gestures and investigate the impact of several schemes including the type of input data, the use of a CNN-only and CNN-LSTM layer, an increased range of up to 150 cm, three types of angles, and presence of nearby person scenarios. The evaluation showed an average accuracy of 98.72% for subject-independent and 98.69% for subject-independent, with automatic hand position detection even at distances of up to 150 cm.

Article

AA12746425