| Title |
A Single Channel UWB through Wall People Counting Method for Disaster Response Environments |
| Authors |
김태민(Taemin Kim) ; 고진환(Jinhwan Koh) |
| Keywords |
Collapsed structure; UWB radar; Through wall; People counting; Artificial intelligence |
| Abstract |
To rapidly and accurately detect survivors located behind collapsed structures in disaster environments, technologies capable of operating in non-contact and non-line-of-sight conditions are essential. However, existing through wall people counting studies mostly rely on multi-channel radar systems, which inevitably lead to high power consumption as well as increased size and weight, making practical deployment difficult. Moreover, most of these studies are limited to scenarios involving three or fewer individuals. To address these limitations, this study proposes a neural network model designed to compensate for the lack of spatial resolution inherent to single channel measurements, using a single channel UWB radar. Through experiments, we demonstrate that the proposed approach enables reliable multi person detection even under a single channel configuration. As a result of comparing the performance of the proposed model along with SVM (Support Vector Machine), CNN (Convolutional Neural Network), LSTM (Long Short-Term Memory), and Transformer for through-wall people counting, the proposed model recorded the highest accuracy of 91.8%. These results highlight that, unlike prior work relying on multi-antenna systems, high accuracy through wall people counting is feasible even with a single-channel radar, demonstrating strong potential for practical deployment in disaster rescue scenarios. |